UBC Faculty Research and Publications

Molecular adaptation and expression evolution following duplication of genes for organellar ribosomal… Liu, Shao-Lun; Adams, Keith Jan 26, 2008

Your browser doesn't seem to have a PDF viewer, please download the PDF to view this item.

Item Metadata


52383-12862_2007_Article_594.pdf [ 2.38MB ]
JSON: 52383-1.0223577.json
JSON-LD: 52383-1.0223577-ld.json
RDF/XML (Pretty): 52383-1.0223577-rdf.xml
RDF/JSON: 52383-1.0223577-rdf.json
Turtle: 52383-1.0223577-turtle.txt
N-Triples: 52383-1.0223577-rdf-ntriples.txt
Original Record: 52383-1.0223577-source.json
Full Text

Full Text

ralssBioMed CentBMC Evolutionary BiologyOpen AcceResearch articleMolecular adaptation and expression evolution following duplication of genes for organellar ribosomal protein S13 in rosidsShao-Lun Liu and Keith Adams*Address: UBC Botanical Garden & Centre for Plant Research and Botany Department, University of British Columbia, Vancouver, BC, V6T1Z4, CanadaEmail: Shao-Lun Liu - shaolun@interchange.ubc.ca; Keith Adams* - keitha@interchange.ubc.ca* Corresponding author    AbstractBackground: Gene duplication has been a fundamental process in the evolution of eukaryoticgenomes. After duplication one copy (or both) can undergo divergence in sequence, expressionpattern, and function. Two divergent copies of the ribosomal protein S13 gene (rps13) ofchloroplast origin are found in the nucleus of the rosids Arabidopsis, Gossypium, and Glycine. Oneencodes chloroplast-imported RPS13 (nucp rps13), while the other encodes mitochondria-imported RPS13 (numit rps13). The rps13 gene has been lost from mitochondrial DNA (mt rps13)of many rosids.Results: We studied sequence evolution of numit rps13 in comparison with nucp rps13 in sevenrosid genera. Ka/Ks analysis and likelihood ratio tests showed considerably higher Ka values and Ka/Ks ratios in numit rps13 than in nucp rps13, indicating increased amino acid sequence divergence innumit rps13. Two positively selected codons were detected in numit RPS13 in regions that areinferred to interact with the 16S rRNA. Several amino acids in numit RPS13 have changed from theone present in nucp RPS13 to the one present in mt RPS13, showing that numit rps13 is becomingmore like mt rps13. Comparison of expression patterns and levels of numit rps13 and nucp rps13in Arabidopsis using microarray data indicated divergence in gene expression. We discovered thatin addition to numit rps13, Malus (apple) contains a transcribed mt rps13 gene. To determine ifpartitioning of expression takes place between numit rps13 and mt rps13, expression of both copiesand RNA editing of mt rps13 were examined by RT-PCR, qRT-PCR, and sequencing from 14different organ types plus seedlings subjected to five different abiotic stresses. Co-expression ofnumit rps13 and mt rps13 was observed in all the organs and various stress treatments. Wedetermined that purifying selection is acting on both numit rps13 and mt rps13 in Malus.Conclusion: Our data provide evidence that numit rps13 genes in rosids have experiencedadaptive sequence evolution and convergent evolution with mt rps13. Co-expression of numitrps13 and mt rps13 and purifying selection on both genes in Malus suggest that both are functional.The three organellar rps13 genes in rosids provide a distinctive case of gene duplication involvingthe co-evolution of the nuclear and cytoplasmic genomes.Published: 26 January 2008BMC Evolutionary Biology 2008, 8:25 doi:10.1186/1471-2148-8-25Received: 7 August 2007Accepted: 26 January 2008This article is available from: http://www.biomedcentral.com/1471-2148/8/25© 2008 Liu and Adams; licensee BioMed Central Ltd. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.Page 1 of 16(page number not for citation purposes)BMC Evolutionary Biology 2008, 8:25 http://www.biomedcentral.com/1471-2148/8/25BackgroundGene duplication has been an ongoing process duringeukaryotic evolution that has provided genetic raw mate-rial for the evolution of new gene functions that can leadto morphological and physiological novelty. Duplicatedgenes can undergo sequence divergence caused by positiveselection or neutral drift [1-3] and divergence in expres-sion patterns and function. Two common fates of retainedduplicated genes are neofunctionalization – gain of a newfunction or expression pattern by one copy [4] and sub-functionalization – partitioning of ancestral function orexpression pattern between both copies [5,6]. Plantgenomes contain large numbers of duplicated genes,derived by polyploidy, segmental duplications, tandemduplications, and retroposition of cDNAs. Many dupli-cated genes in plant genomes have been preserved andundergone purifying selection, a few have undergone pos-itive selection and functional diversification, and somehave experienced subfunctionalization [7-10].There are three genes that code for organellar ribosomalS13 genes among rosid species. Analysis of rps13 genes inthe rosid species Arabidopsis thaliana, cotton (Gossypiumarboreum), and soybean (Glycine max) revealed the pres-ence of two expressed copies of rps13 in the nucleus thatwere derived by gene duplication [11,12]. Both in vitroand in vivo RPS13 protein import experiments indicatedthat one copy encodes the chloroplast-imported protein(nucp rps13) while the other encodes mitochondria-imported RPS13 (numit rps13) [11,12]. It was inferredthat the missing mt rps13 gene product has been function-ally replaced by the product of numit rps13 in a commonancestor of Arabidopsis, cotton, and legumes. Thus thefunction of numit rps13 has been modified after geneduplication, and one could argue that numit rps13 hasgained a new function because it is operating in a new cel-lular context (the mitochondrial ribosome instead of thechloroplast ribosome). Subsequently mt rps13 was lostfrom mitochondrial DNA many times during the evolu-tionary history of rosids, as inferred from a Southern blothybridization survey [13](see Additional File 1). Surpris-ingly, however, there were many species of rosids that doappear to retain rps13 in the mitochondrion, based onSouthern blot hybridizations, but the gene in some spe-cies may not be intact or functional.The organellar rps13 genes in rosids provide an intriguingsystem to study gene duplication because the subcellularlocation and site of action of numit RPS13 has changedafter gene duplication from the chloroplast to the mito-chondrion. We have studied sequence evolution of numitrps13 among rosids to determine what kinds of aminoacid changes have taken place and where those aminoAlso we have examined the divergence of expression pat-terns between numit rps13 and nucp rps13. After findingintact and expressed numit rps13 and mt rps13 genes inMalus we tested the hypothesis that there has been expres-sion partitioning of the two genes in different organ typesand/or stress conditions to preserve both genes.ResultsIdentification of numit rps13 in Malus, Populus, and Citrus and phylogenetic analysisTo study the evolution of numit rps13 sequences in therosids, we identified sequences homologous to numitrps13 in Malus, Populus, and Citrus by BLAST searches ofthe NCBI expressed sequence tag (EST) database using thenumit rps13 from Arabidopsis as a query. ESTs (see Materi-als and Methods) were aligned and the open readingframes were identified. The genes in each species are pre-dicted with high probability to encode mitochondrialproteins by four prediction programs: MitoProt [14], Tar-getP v1.1 [15], Predotar v1.03 [16], and iPSORT [17].Three of the four programs discriminate mitochondrialfrom chloroplast proteins and no support was obtainedfor chloroplast targeting. Alignment of the N-termini withmitochondrial targeting presequences from numit rps13in other rosid species shows sequence conservation andindicates that they are homologous genes (Figure 1).Because numit rps13 in Arabidopsis, Gossypium, and Glycinehave been experimentally shown to be imported intomitochondria but not chloroplasts (Adams et al., 2002),we infer that the products from the homologous genes inMalus, Populus, and Citrus also are targeted to mitochon-dria. We found numit rps13 sequences only in speciesbelonging to the rosid lineage, suggesting that the geneduplication event that created numit rps13 likely occurredafter the emergence of the rosid lineage. We did not find anumit rps13 sequence in Vitis, a group at the base of rosids[18,19], despite the mostly sequenced genome [20] and alarge EST collection available at NCBI. We infer that thegene duplication event occurred at the base of the eurosidsafter separation from the Vitaceae lineage (see AdditionalFile 1).Mitochondrial targeting sequence alignment of numit RPS13Figure 1Mitochondrial targeting sequence alignment of numit RPS13. Predicted targeting presequences of numit RPS13 from seven rosid species are aligned. Identical amino acid residues are marked in white on the black background. Page 2 of 16(page number not for citation purposes)acids are located in the tertiary structure, as well as to testthe hypothesis that there has been adaptive evolution.Dots indicate gaps inserted to improve the alignment.BMC Evolutionary Biology 2008, 8:25 http://www.biomedcentral.com/1471-2148/8/25Nucp rps13 sequences from seven rosid species wereobtained from GenBank by BLAST searches using the pre-viously characterized nucp rps13 from Arabidopsis [21] andaligned with the numit rps13 sequences. Phylogeneticanalysis of the sequences verified the orthologous rela-tionships of numit rps13 and nucp rps13 sequences (Fig-ure 2). The phylogenetic relationships inferred from nucprps13 and numit rps13, however, did not follow estab-lished relationships of rosid species [18]. For numit rps13,the positions of Populus and Gossypium in the tree areswitched; for nucp rps13 the gene positions in the tree arecompletely scrambled. The lack of congruence betweenthe gene trees and organismal phylogeny is probably dueto the short sequences being analyzed (about 242 bp afterexcluding the third codon position). Alternatively, itmight be caused by the complex history of multiple poly-ploidy events and subsequent loss of duplicated genesduring rosid evolution. For example, after polyploidiza-tion the lineage containing legumes and Malus might haveretained copy 1 while Arabidopsis, Citrus, Gossypium andPopulus retained copy 2 (Figure 2). Comparisons ofbranch lengths showed that numit rps13 sequences havediverged rapidly with much longer branch lengths thanthe nucp rps13 sequences, suggesting there has been accel-erated evolution of numit rps13 in each lineage, and con-sistent with results from Adams et al. [11] that includedfewer species of rosids.Accelerated nucleotide substitution rates and positive selection on numit rps13To better quantify the rate increase and to determine ifboth Ka and Ks (or just Ka) are accelerated in numit rps13sequences, the Ka, Ks values, and Ka/Ks ratios were com-pared between nucp rps13 and numit rps13 in the rosids.Significantly higher Ka values and Ka/Ks ratios wereobserved in the numit rps13 genes than in the nucp rps13genes (P < 0.05; Table 1; see Additional file 2). The Ka val-ues in the numit rps13 genes were 3–4 times greater thanin the nucp rps13 genes. No significant difference in Ks val-ues, however, was found between the nucp rps13 andnumit rps13 copies (P > 0.05; see Additional file 2). Theseresults show that numit rps13 has been experiencing aconsiderably accelerated rate of non-synonymous substi-tutions compared with the nucp rps13 in all seven lineagesof rosids.Although simple pair-wise comparison of Ka and Ks analy-sis provides some insights into the accelerated amino acidsubstitution in numit rps13, branchwise estimation of Ka,Ks values, and Ka/Ks ratios can provide additional informa-tion such as positive selection and adaptive molecularevolution along certain branches and clades [22-26]. Todetect if there has been positive selection acting on numitrps13 and nucp rps13 in different rosid species, site specificmodel analysis was conducted using PAML [23].(Although synonymous and non-synonymous substitu-tions are abbreviated as dS and dN in PAML analysis, weuse the common designations Ks and Ka to refer to synon-ymous and non-synonymous substitutions, respectively.)Both the rps13 tree (Figure 2) and the species tree thatreflects our current understanding of rosid phylogeny [18]were used for detection of positive selection to determineif the tree topology influenced detection of positivelyselected sites, and no differences were found in thisregard.One ratio model (M0) shows that the Ka/Ks ratio of numitrps13 is about 4 times higher than nucp rps13 (Table 2),congruent with observations from simple Ka and Ks analy-sis. For each dataset, likelihood ratio tests (LRT) for detec-tion of positive selection were examined using M7-M8and M8a-M8 comparisons. For numit rps13, M8 is signifi-cantly better than M7 and M8a in numit rps13 (P < 0.05)and 2.3% of sites are inferred to undergo positive selec-tion (Table 2), suggesting there has been positive selectionacting on numit rps13. In addition, based on Bayes Empir-ical Bayes (BEB) analysis, two codon sites in numit rps13were identified as showing strong positive selection (pos-terior probability > 0.8; Table 2). However, there is no evi-dence for sites under positive selection in nucp rps13. Innucp rps13, M8 is not significantly better than M7 andPhylogenetic analysis of numit rps13 and nucp rps13 from rosid speciesFigure 2Phylogenetic analysis of numit rps13 and nucp rps13 from rosid species. Shown is an unrooted phylogram derived from maximum likelihood analyses of the first and second nucleotide positions. Bootstrap values from 100 rep-Page 3 of 16(page number not for citation purposes)M8a (Table 2). These results indicate that strong positivelicates of ML analyses are labeled on the internodes.BMC Evolutionary Biology 2008, 8:25 http://www.biomedcentral.com/1471-2148/8/25selection acts on the evolution of numit rps13among therosid species, particularly at codons 28 and 114.Structural location of positively selected amino acidsRelative locations of the two positively selected sites wereplotted on the tertiary amino acid structures of RPS13from Escherichia coli and Thermus thermophilus (Figure 3)to infer approximate locations in numit RPS13. Except forthe C-terminal end which is longer and contains onemore α-helix structure in T. thermophilus, the overall struc-ture of RPS13 from E. coli and T. thermophilus are relativelysimilar (Figure 3). RPS13 residues at positions 28 and 114in E. coli (28 and 116 in T. thermophilus) correspond to thepositively selected sites in numit RPS13 based on theamino acid alignment (Figures 3, 4). Previous structuraland functional studies of Thermus RPS13 showed thatthere are two structurally important regions where RPS13interacts with the 16S rRNA [27-29]. One of them is theloop region between helix 1 and turn 1 (residues 22–25)[29]. One of the positively selected residues found in ouranalysis is close to that region. The second positivelyselected residue in numit Rps13 is located in the highlybasic COOH-terminal extended region (Figure 3). Thisregion is virtually devoid of secondary structure andfound to interact with the 16S rRNA at the P-site and A-site (residues 116–120 for P-site and residues 120–122 forA-site) [27-29]. Positively selected sites in or near regionsthat interact with the 16S rRNA suggest that there has beenfunctional refinement of numit RPS13 to interact betterwith the 16S rRNA.Amino acid changes in numit rps13 that increase identity to mt rps13Because numit rps13 was derived from nucp rps13 andencodes a RPS13 protein that functions in the mitochon-dria, we were interested in determining if the numitRPS13 has become more like the mt RPS13 amino acidsequence. We determined if there have been mutations inthe numit rps13 genes in any of the seven rosid speciesthat change an amino acid to the residue that is present inTable 2: LRT statistics of site specific model for numit rps13 and nucp rps13.Gene Number of sequencesTree Lengtha Ksa Kaa Ka/Ksa 11th class from M8 M7-M8 comparison M8a-M8 comparison Selected positive sitesp, % Ka/Ks 2δL P Value 2δL P Valuenumit rps13 7 6.93 5.35 1.30 0.24 2.7 9.7400 9.0670 0.0107b 7.5048 0.0031 28R, 114R (P > 0.8)nucp rps13 7 4.08 5.16 0.31 0.06 1.6 1.6268 4.7952 0.0909 0.8448 0.1790 NoneTable 1: Comparison of Ka/Ks ratios between nucp rps13 and numit rps13.Taxon Arabidopsis Citrus Glycine Gossypium Malus Medicago Populusnucp rps13 (0.06 ± 0.02)aArabidopsis -Citrus 0.019761 -Glycine 0.031462 0.111523 -Gossypium 0.045554 0.051540 0.076516 -Malus 0.037126 0.049357 0.081350 0.059803 -Medicago 0.040972 0.084293 0.059997 0.057981 0.085442 -Populus 0.048739 0.112004 0.062117 0.049504 0.061190 0.067945 -numit rps13 (0.28 ± 0.08)Arabidopsis -Citrus 0.276419 -Glycine 0.331700 0.253453 -Gossypium 0.257117 0.308765 0.299647 -Malus 0.299325 0.317365 0.238339 0.468253 -Medicago 0.267372 0.179926 0.032697 0.336763 0.273923 -Populus 0.382829 0.261000 0.241346 0.294584 0.301918 0.170266 -aThe mean and standard deviation are shown in parentheses.Page 4 of 16(page number not for citation purposes)aEstimated using M0 model in PAML.bBoldface indicates the statistically significant difference (P < 0.05).BMC Evolutionary Biology 2008, 8:25 http://www.biomedcentral.com/1471-2148/8/25mt RPS13. In our alignment of nucp RPS13, numit RPS13,and mt RPS13 across different plant species, sixteen siteswere identified where numit RPS13 in one or more spe-cies has the same amino acid at the corresponding site inmt RPS13 from seven angiosperms, and the amino acid isdifferent from the amino acid(s) present in nucp RPS13(Figure 4A). The sixteen sites are relatively evenly distrib-uted in the RPS13 tertiary structure (Figure 3). Althoughlittle is known about the exact functions of those regions,some of the mutations might help improve the functionof numit RPS13 in the mitochondrial ribosome. We plot-occurred (Figure 4B). Mutations are inferred to haveoccurred along most of the branches, suggesting continu-ous refinement of the numit rps13 sequence. We infer thatthree mutations occurred in the common ancestor of allthe species, with a subsequent mutation at site 80 in Ara-bidopsis from D to E (a conservative substitution) and atsite 110 in Gossypium from A to S, although scenarios withmultiple recent mutations cannot be ruled out.Expression evolution of numit rps13Having studied sequence evolution of numit rps13 wenext tested the hypothesis that there have been changes inexpression patterns and levels of expression of numitrps13 relative to nucp rps13. Extensive microarray data areavailable for Arabidopsis thaliana including a single studythat examined expression in 51 organs and developmen-tal stages using the ATH1 array [30]. We analyzed expres-sion data for numit rps13 and nucp rps13 using an ANOVAapproach (see Methods). Nucp rps13 is expressed a higherlevel than numit rps13 in most organs (P < 0.05; Figure5A; Additional file 3). Notable exceptions were roots,senescing leaves and pollen where numit rps13 is morehighly expressed (P < 0.05; Figure 5A). When comparingexpression levels of the two genes among organs, some-times the levels of both genes go up or down together, butsometimes the levels go in opposite directions. Overallthe organ-specific expression patterns between the twogenes show both similarities and differences (R = 0.31, P< 0.05), depending on the organs compared.To compare expression patterns of numit rps13 to othernuclear encoded ribosomal protein genes, we analyzedexpression data for three other nuclear encoded mito-chondrial ribosomal protein genes: rps9, rps10, and rps11[31]. The expression levels and patterns of all four geneswere highly similar (R = 0.73–0.94, P < 0.0001), althoughrps11 expression levels were lower in some organs (P <0.05; Figure 5B). Thus numit rps13 has evolved an expres-sion pattern similar to that of other nuclear-encodedgenes for mitochondrial ribosomal proteins, and itsexpression has diverged from that of nucp rps13.Malus contains an expressed and RNA-edited copy of rps13 in the mitochondrionHaving studied the sequence and expression evolution ofnumit rps13 we next consider the fate of mt rps13 in rosids.A large number of rosids, including Arabidopsis, Gossypium,Glycine, and Citrus have lost mt rps13, as judged by DNAgel blot hybridization [13] (see Additional file 1), butsome rosids retain rps13 in mitochondrial DNA. We iden-tified a transcribed copy of rps13 in Malus domestica(apple) from BLAST searches of the NCBI EST databasethat is 90–92% identical with the rps13 gene in the mito-Tertiary structure of numit RPS13Figure 3Tertiary structure of numit RPS13. The structures of RPS13 from E. coli (A) and T. thermophilus (B) are shown with amino acids from numit RPS13 plotted, as in the alignment (Figure 4). Red regions and plus signs indicate relative posi-tions of positively selected sites. Green regions and asterisks indicate relative positions of amino acids in numit Rps13 in at least one rosid species that have mutated to the amino acid present in mt rps13 genes (Figure 4). Among them, sites 4 and 10 (Figure 4) are located in the coil region of the N-ter-minal end, site 15 is in the beginning of the first β-sheet, site 21 is in the first α-helix, site 32 is in the second α-helix, sites 42 and 43 are in the coil region between the second α-helix and the second β-sheet, site 53 are in the third α-helix, site 67 is in the coil region between the third α-helix and the fourth α-helix, sites 74, 79, and 80 are in the fourth α-helix, and sites 97, 107, 108, and 110 are in the C-terminal coil region.Page 5 of 16(page number not for citation purposes)ted the sixteen amino acid changes on the phylogeny ofthe seven rosid species to infer when they might havechondrion of several eudicots. The sequence was derivedfrom a study of ESTs in Malus [32]. A similar gene (97%BMC Evolutionary Biology 2008, 8:25 http://www.biomedcentral.com/1471-2148/8/25Page 6 of 16(page number not for citation purposes)Amino acid evolution in numit RPS13Figure 4Amino acid evolution in numit RPS13. (A) Alignment of mt RPS13, numit RPS13, nucp RPS13 sequences, and RPS13 from E. coli and Thermus thermophilis. Amino acids in numit RPS13 that are identical to those in mt RPS13 or nucp RPS13 are shown on a black background. Dashes indicate gaps inserted to improve alignment. Red plus signs indicate positions of positively selected amino acids. Black arrows indicate amino acids in numit rps13 in at least one rosid species that have mutated to the amino acid present in all eight mt rps13 genes, summarized in panel B. Panel B also includes a phylogeny of the rosid species and our hypothesized evolutionary timing of each amino acid change. Numbers in panel B indicate positions in the alignment shown in panel A.BMC Evolutionary Biology 2008, 8:25 http://www.biomedcentral.com/1471-2148/8/25identical) was found in Prunus persica, another member ofthe Rosaceae family. We evaluated the rps13 sequencefrom Malus for sites of C-to-U RNA editing by PCR ampli-fying and sequencing rps13 from genomic DNA and com-paring the gDNA sequence to the EST sequences. C-to-Uto those that are evolutionarily conserved [33]. Amongangiosperm rps13 genes, nine possible edited sites havebeen identified (see Additional file 4). Four of those sitesare already T's instead of C's in the genomic DNAsequence of mt rps13 from Malus, and thus RNA editingmight be expected at five sites in the Malus cDNAs. Noediting was observed in the two ESTs from Malus, and theEST from Prunus had editing at only one site (the 100thbase). To verify the lack of RNA editing, we amplified anddirectly sequenced mt rps13 cDNAs from leaves and petalsof Malus. Unexpectedly five RNA editing sites were discov-ered in both leaves and petals at five sites (see Additionalfile 4). After transcription, RNA editing converts codonsfrom serine to leucine (the 26th, 56th, and 287th bases)and from arginine to cysteine (the 100th and 256thbases). The changes made by RNA editing would restorean evolutionarily conserved amino acid sequence andmake the resulting protein likely to be functional, shouldthe transcripts be translated.Co-expression of numit rps13 and mt rps13 in 14 different organ types and under five different stress conditionsWe conducted expression assays of numit rps13 and mtrps13, and RNA editing examination of mt rps13, in Malusto test the hypothesis that expression patterns have beenpartitioned between the two genes in different organtypes. RT-PCR was performed with (RT+) or without (RT-)reverse transcriptase to check for DNA contamination(Figure 6). Transcripts of numit rps13 and mt rps13 wereobserved in roots (from seedlings), stems, stigmas +styles, and ovaries. Transcripts of these two genes werealso observed in 10 additional organ types: hypocotyls,cotyledons, young leaves (from seedlings), mature leaves,peduncles, petals, seeds, sepals, stamens, and young fruit.Mt rps13 cDNAs were sequenced from each organ type todetermine if any sites were edited. Different organ types inapple all showed the five RNA editing sites mentionedabove. The results of the RT-PCR experiments show thatboth copies are co-expressed in all examined organ types.Although both copies of rps13 are expressed in manyorgan types it is possible that there might be partitioningof expression between the two genes under stress condi-tions. We tested the hypothesis by examining the expres-sion patterns of numit rps13 and mt rps13 under fivedifferent stresses including cold, dark, heat, salt, and watersubmersion treatments to determine if expression parti-tioning occurs under different environmental stresses.Apple seedlings were independently subjected to each ofthese five stresses (see Materials and Methods). Fourorgan types, including roots, hypocotyls, cotyledons, andleaves were examined for expression of numit rps13 andmt rps13. Based on RT-PCR results, transcripts of numitExpression patterns of numit rps13 in Arabidopsis thalianaFigure 5Expression patterns of numit rps13 in Arabidopsis thaliana. Shown are graphs from ANOVA analysis of micro-array data from 51 organs and developmental stages (see Methods). Organ types and developmental stages are listed in Additional File 3. Error bars show 3 biological replicates. The Y-axis indicates the expression level normalized by log2. (A) Numit rps13 compared with nucp rps13. (B) Numit rps13 compared with three other nuclear-encoded mitochondrial ribosomal protein genes, rps9, rps10, and rps11 [31].Page 7 of 16(page number not for citation purposes)RNA editing plays an important role in the expression ofplant mitochondrial genes to restore certain amino acidsrps13 and mt rps13 were observed under each of the fivestress conditions: cold, dark, heat, salt, and water submer-BMC Evolutionary Biology 2008, 8:25 http://www.biomedcentral.com/1471-2148/8/25sion treatments. Sequencing of RT-PCR products showedthat the transcripts were RNA edited at 5 sites, although insome cases there was partial editing at one or more sites.We have shown transcription of both numit rps13 and mtrps13 in many organs and under several stress conditions.However it is possible that transcripts from one genemight be present at low levels. PCR of cDNA template(RT-PCR) is not a quantitative technique and low levels oftranscripts might not be distinguishable from high levels.To determine if abundant levels of steady-state transcriptsare derived from both genes, we used real-time PCR (qRT-PCR) to assay transcript levels of numit rps13 and mtrps13. We compared levels of numit rps13 to those ofanother nuclear gene for a mitochondrial ribosomal pro-tein, rps10 (for ribosomal protein S10), and we comparedlevels of mt rps13 to those of another mitochondrial gene,cob (for cytochrome b). We assayed expression levels in aabove the levels of rps10 in ovaries and roots, and mostorgans showed at least half as many transcripts fromnumit rps13 as rps10 (Figure 7A). Cold stress was an excep-tion, where the numit rps13/rps10 ratios were less than 0.4(Figure 7A). Mt rps13 transcripts were present at slightlyover twice the levels of cob transcripts in leaves (Figure7B). The mt rps13/cob transcript ratio was above 0.5 inmost other organs and under cold stress. Overall there isevidence for relatively high levels of steady-state tran-scripts from numit rps13 and mt rps13 in most organs.qRT-PCR of numit rps13 and mt rps13 in MalusFigure 7qRT-PCR of numit rps13 and mt rps13 in Malus. (A) Ratios of numit rps13 to rps10 transcripts. (B) Ratios of mt rps13 to cob transcripts. Error bars show variation from two Expression of numit rps13 and mt rps13 from in Malus in dif-ferent organ typesFigure 6Expression of numit rps13 and mt rps13 from in Malus in different organ types. Plus signs indicate reactions con-taining reverse transcriptase (RT) and minus signs indicate reactions without RT. Abbreviations: coty, cotyledon; g, genomic; hypo, hypocotyl; m, marker. (A) A gel showing a subset of the RT-PCR products of numit rps13 (457 bp). Sequencing of a larger band in RT-PCR assays of numit rps13 revealed that it was due to non-specific primer binding, whereas the smaller band was the numit rps13. (B) A gel showing a subset of the RT-PCR products of mt rps13 (305 bp).Page 8 of 16(page number not for citation purposes)subset of 9 organs and two stress conditions that wereassayed above. Numit rps13 transcripts were present at orreplicates.BMC Evolutionary Biology 2008, 8:25 http://www.biomedcentral.com/1471-2148/8/25Purifying selection acting on numit rps13, nucp rps13, and mt rps13 in MalusDetecting purifying selection, neutral evolution, and pos-itive selection on each of the three rps13 genes would pro-vide insights into the selective forces at work and couldindicate if one copy is a pseudogene. A Ka/Ks ratio lessthan one would indicate purifying selection and be evi-dence for a functional gene, and a Ka/Ks ratio of about onewould be evidence of neutral evolution and thus probablypseudogenization. The branch specific model in PAMLwas used to detect adaptive molecular evolution in numitrps13, nucp rps13, and mt rps13 for the Malus branches bycomparing with other plant species. For each LRT, M1 isthe model assuming neutral evolution among sites,MAtest1 assumes that there are positively selected aminoacid sites for the foreground lineage, and MAtest2, is usedto test if the detection of positive selection is an artifact.LRTs showed that MAtest1 is not significantly better thanM1 and MAtest2 in Malus numit rps13 and Malus mt rps13,indicating that there is no positive selection for these twobranches (Table 3). For nucp rps13, MAtest1 is significantlybetter than M1, but not better than MAtest2 (Table 3), sug-gesting that there is no positive selection in nucp rps13. Inaddition, branch specific models can provide detailedinformation about how many sites are undergoing purify-ing selection, neutral evolution, and positive selection.For numit rps13, 65% of the sites are under purifyingselection with Ka/Ks less than one, 28% of sites are underneutral evolution with Ka/Ks close to one, and 7% of sitesare under positive selection with Ka/Ks greater than one(Table 3) without statistical support by LRTs. For nucprps13, 90% of sites are under purifying selection, 7% ofsites are under neutral selection, and 3% of sites are underpositive selection (Table 3) although the positivelyselected sites are not supported by LRTs. For mt rps13,66% of sites are under purifying selection and 34% of sitesare under neutral selection (Table 3). The number of sitesexperiencing each type of selection in numit rps13 inMalus is similar to numit rps13 in other rosids that do nothave mt rps13 (Table 4). Overall our results reveal that thethree ribosomal protein S13 genes in Malus are undergo-ing higher purifying selection than neutral evolution, sug-gesting that their functionality is maintained by purifyingselection.DiscussionContinuous accelerated evolution and molecular adaptation of numit rps13 among rosidsAfter gene duplication some retained duplicated genesundergo asymmetric rate divergence and the faster evolv-ing copy experiences relaxed constraint or positive selec-tion [34,35]. The increased Ka rate, Ka/Ks ratio, andpositively selected sites detected in numit rps13, comparedwith nucp rps13, across different rosid species show therehas been accelerated sequence evolution of numit rps13.The increased Ka rate is likely to be correlated with themodified function of numit rps13 – from encoding a chlo-roplast ribosomal protein to a mitochondrial ribosomalprotein. Some amino acid sequence changes were proba-bly necessary for numit RPS13 to interact well with otherproteins in the mitochondrial ribosome. The increasedrate of non-synonymous substitutions is a continuingprocess instead of there being a burst of sequence changeupon formation of numit rps13 followed by a prolongedperiod of lower rates of non-synonymous substitutions.The continued higher Ka rate, and the two positivelyselected codons in numit rps13 near the positions whereRPS13 interacts with 16S RNA domain, show that thesequence, and suggest that the function, of numit RPS13in mitochondria is continuing to be refined. Site-specificpositive selection has been detected in other duplicatedgenes [36-38]. For example, in duplicated AP3 and PIgenes there has been functional diversification driven bypositive selection acting on different sites within a func-tional domain involved in heterodimerization [39].The sixteen amino acids in one or more rosid species thathave changed to the amino acid present in mt RPS13 (Fig-ure 4) indicate that numit RPS13 is becoming more likemt RPS13. We propose that this is a type of convergentsequence evolution of numit rps13 that possibly improvesthe function of numit RPS13 in the ribosome. The aminoacid changes appear to have been taking place continu-ously during rosid evolution (Figure 4B), with the largestTable 3: LRT statistics of branch specific model for Malus branch.Gene Parameters from MAtest1 M1-MAtest1 comparison MAtest2-MAtest1 comparison Selected positive sites2δL P Value 2δL P Valuenumit rps13 p0 = 0.6485, p1 = 0.2817, (p2 + p3 = 0.0698)ω0 = 0.12, ω1 = 1, ω2 = 112.981.7396 0.4190 1.2140 0.2705 31H, 32Q, 117R (0.5 <P < 0.8)nucp rps13 p0 = 0.9005, p1 = 0.0733, (p2 + p3 = 0.0262)ω0 = 0.04, ω1 = 1, ω2 = 96.79.7274 0.0077a 3.5026 0.0613 32Q, 105Cmt rps13 p0 = 0.6646, p1 = 0.3354, (p2 + p3 = 0) ω0 = 0, ω1 = 1, ω2 = 10 1 0 1 NonePage 9 of 16(page number not for citation purposes)aBoldface indicates the statistically significant difference (P < 0.05).BMC Evolutionary Biology 2008, 8:25 http://www.biomedcentral.com/1471-2148/8/25number (five) having occurred on the branch leading tothe legumes Glycine and Medicago. We speculate that someof the amino acid changes that have taken place along theterminal or subterminal branches of the tree might haveallowed numit rps13 to be selected for, over mt rps13, andallowed multiple independent losses of mt rps13 in differ-ent lineages (see Additional file 1). In species where mtrps13 has been lost the product of numit rps13 presumablyfunctions as well as, or perhaps better than, mt rps13 orelse there would have been selection for retention of mtrps13. Overall the adaptive and convergent evolutionaryforces that seem to be acting on numit rps13 have contin-ued during rosid evolution instead of being factors onlysoon after gene duplication in a common ancestor ofmost rosids.Expression evolution of numit rps13 and nucp rps13 in ArabidopsisExpression patterns of some duplicated genes have beenshown to evolve in a divergent and sometimes asymmet-ric manner [40-42]. For example, many of the genesderived by an ancient polyploidy event in the Arabidopsislineage have undergone considerable expression diver-gence [43-45], some of which may have been asymmetricbetween the two duplicates. Considering that numit rps13and nucp rps13 show asymmetric divergence in sequence,and different subcellular locations of their protein prod-ucts, we predicted that they also would show asymmetricdivergence in expression, with the expression pattern ofnumit rps13 being similar to that of other nuclear-encoded genes for mitochondrial ribosomal proteins.Numit rps13 and nucp rps13 did show some differences inexpression patterns although the differences were onlydramatic in roots, senescing leaves, and pollen. Numithaps by gaining regulatory elements from another genefor a mitochondrial protein, as have several mitochon-drial genes that have been transferred to the nucleus [46].Alternatively there may have been mutations in the regu-latory elements of numit rps13 soon after gene duplica-tion that produced an expression pattern similar to that ofother genes for mitochondrial ribosomal proteins. Con-sidering that numit rps13 was formed in a common ances-tor of most rosids, distinguishing between the abovepossibilities about the origin of its regulatory elements isimpossible. Overall numit rps13 and nucp rps13 add tothe growing number of duplicated genes reported in Ara-bidopsis thaliana that have experienced divergence inexpression patterns.Co-expression of numit rps13 and mt rps13 in MalusWe have shown that numit rps13 and mt rps13 in Malusare both transcribed in 14 different organ types and underfive abiotic stress conditions, mt rps13 is RNA edited atsites to make the transcripts contain an evolutionarilyconserved sequence, and abundant steady-state tran-scripts from both genes are present in a variety of organs.Both genes are experiencing purifying selection. Takentogether, the data obtained in this study indicate that it islikely that both numit rps13 and mt rps13 in Malus arefunctional genes and not pseudogenes. However there isthe possibility, albeit unlikely, that transcripts from onegene (particularly mt rps13) might not be translated. Evenif both proteins are present, there is the possibility thatonly one is assembled into mitochondrial ribosomes, andthat process could vary among organs and environmentalconditions. Showing that both numit RPS13 and mtRPS13 proteins are assembled into mitochondrial ribos-omes, especially in a variety of organ types and under var-Table 4: Selection on numit RPS13 in seven rosid species.Taxon Purifying Selection (%, Ka/Ks < 1)aNeutral Evolution (%, Ka/Ks = 1)aPositive Selection (%, Ka/Ks > 1)aM1-MAtest1 comparison MAtest2-MAtest1 comparison Selected positive sites2δL P Value 2δL P ValueMalus 65 28 7 (120.33)b 1.7396 0.4190 1.2140 0.2705 31H, 32Q, 117RCitrus 65 30 5 (154.19) 4.3230 0.1152 3.9182 0.0478c 50K, 92HGlycine 67 33 0 0 1 0 1 35CMedicago 67 33 0 0 1 0 1 NoneGossypium 64 36 0 0.6314 0.7293 0.0024 0.9609 108KPopulus 66 30 4 (18.05) 3.6624 0.1603 3.4746 0.0623 39I, 88S, 102SArabidopsis 52 48 0 6.1770 0.0456 0 1 18G, 36H, 60G, 83DaEstimated using MAtest1.bBracket indicates Ka/Ks ratio. If Ka/Ks ratio is close to one for the third and fourth class of MAtest1 without statistical support to be greater than one, it will be counted into the category of neutral evolution.cBoldface indicates a statistically significant difference (P < 0.05).Page 10 of 16(page number not for citation purposes)rps13 has evolved a similar expression pattern to threeother genes for mitochondrial ribosomal proteins, per-ious environmental conditions, is beyond the scope of thecurrent evolutionary study.BMC Evolutionary Biology 2008, 8:25 http://www.biomedcentral.com/1471-2148/8/25Although we have provided evidence for the functionalityof the gene products from numit rps13 and mt rps13 inMalus, might numit rps13 have experienced functionalreversion to encode a chloroplast protein? Several lines ofevidence do not support that possibility. The mitochon-drial targeting presequence of numit RPS13 is similar tothose of numit RPS13 in Arabidopsis, Gossypium, and Gly-cine (Figure 1), each of which was experimentally deter-mined to be imported into mitochondria but notchloroplasts [11,12]. Numit rps13 in Malus does not havea greatly accelerated rate of sequence evolution or showmore sites under positive selection, compared with numitrps13 in other rosids, as might be expected if its productnow functions in the chloroplast. Instead purifying selec-tion is operating on numit rps13 in Malus. Finally, there isno evidence to suggest that nucp rps13 in Malus is a pseu-dogene and indeed the gene is experiencing strong purify-ing selection.Co-expression of numit rps13 and mt rps13 genes in Maluscontrasts with the atp9 genes in Neurospora crassa wherethe nuclear and mitochondrial copies are expressed dur-ing different stages of the life cycle [47] and expression hasbeen partitioned between the two genes. Presumably thenuclear copy of atp9 was derived from transfer of the mito-chondrial gene to the nucleus in an ancestor of Neu-rospora, and the current availability of several ascomycetegenome sequences could shed light on the evolutionarytiming of the gene transfer. Only three other cases of co-expression of nuclear and mitochondrial genes have beenreported, to our knowledge, including cox2 genes in mul-tiple legume species within the Phaseoleae tribe [48], rpl5genes in wheat (Triticum aestivum) [49], and sdh4 genes inPopulus [50]. Those genes contrast to rps13 in rosidsbecause the nuclear copies were derived by transfer of themitochondrial gene to the nucleus.Considering that numit rps13 was created by gene dupli-cation in the common ancestor of most rosids, co-expres-sion of mt rps13 and numit rps13 in Malus has presumablyoccurred for a long period of evolutionary time. Othercases of co-expression of nuclear and mitochondrial genesin plants represent evolutionarily recent gene transfers tothe nucleus. Indeed following transfer of a mitochondrialgene to the nucleus the mitochondrial copy is often lost ina relatively short amount of time [13]. Thus it was unex-pected that numit rps13 and mt rps13 have been preservedwithout partitioning of expression patterns. Why wouldboth copies continue to be retained and expressed? Onepossibility is that there is partitioning of expression inorgans, tissues, or cell types, or under environmental con-ditions that were not examined in this study. Another pos-sibility is that, despite the considerable level of sequencemitochondrial ribosomes of Malus. That possibility is sup-ported by the fact that some rosids (such as Arabidopsis,Gossypium, Glycine, and Citrus) have only the numit rps13and non-rosid angiosperms have only mt rps13. Anotherpossibility is that mutations have not occurred in numitrps13 from Malus that cause it to be selected for over mtrps13, as presumably occurred in other lineages of rosidsthat have lost mt rps13. In this regard it is notable thatthere are no amino acid changes in numit RPS13 to theresidue present in mt RPS13 that occurred along the ter-minal branch leading to Malus (Figure 4B), unlike all ofthe other terminal branches on the tree leading to speciesthat have lost mt rps13.ConclusionThe accelerated rate of amino acid evolution, positiveselection on specific sites, and amino acid changes to theresidue present in mt RPS13 provide evidence that numitrps13 genes in rosids have experienced adaptive evolutionand convergent sequence evolution with mt rps13. Numitrps13 provides an example of a gene that has experiencedsite-specific positive selection while the gene as a wholehas been under purifying selection. Expression patterns ofnumit rps13 have diverged to become similar to those ofother genes for mitochondrial ribosomal proteins. Maluscontains intact genes for both numit rps13 and mt rps13that are transcribed in a range of organs and under severalabiotic stresses. Abundant levels of steady state transcriptsfrom both numit rps13 and mt rps13 in Malus in a varietyof organ types, as well as purifying selection acting onboth genes, suggest that both genes are functional and notpseudogenes. The three organellar rps13 genes in rosidsprovide a distinctive case of gene duplication involvingthe co-evolution of the nuclear and cytoplasmic genomes.MethodsDatabase searchesThe following expressed sequence tags (EST) for numitrps13 were obtained from GenBank by BLAST searchesusing numit rps13 from Arabidopsis thaliana (DR380621)as a query: Citrus sinensis (CX672493), Malus domestica(DR995890, CN494589, CN925904, and CX023021),Populus trichocarpa (DT496554), Medicago truncatula(BI272420), Glycine max (BM188020), and Gossypium hir-sutum (DW488419). The following ESTs for nucp rps13were obtained from GenBank by BLAST searches usingnucp rps13 from Arabidopsis thaliana (DR367899) as aquery: Citrus sinensis (CX053776), Glycine max(EH261685), Gossypium hirsutum (DW226103), Malusdomestica (DT000985), Medicago truncatula (BI265445)and Populus trichocarpa (DT486949). Mt rps13 ESTs fromMalus (CN871057 and CN875489) and Prunus persica(AJ873687) were identified by BLAST searches of Gen-Page 11 of 16(page number not for citation purposes)divergence between numit rps13 and mt rps13 (about 42%identity), the product of either gene functions well in theBank using mt rps13 from Beta vulgaris as a query. Mt rps13genes used in this study include: Triticum aestivumBMC Evolutionary Biology 2008, 8:25 http://www.biomedcentral.com/1471-2148/8/25(Y00520), Zea mays (AF079549), Magnolia spp. (Z49799),Helianthus annuus (AJ243789), Daucus carota (X54417),Oenothera berteriana (X54416), Beta vulgaris (DQ381464)and Marchantia polymorpha (M68929).Sequence alignment/analysis and phylogenetic analysisSequence alignment was done using transAlign, whichaligns protein-coding DNA sequence based on the align-ment of amino acids [51]. Aligned sequences were refinedwith BioEdit [52] for further phylogenetic and codon sub-stitution analysis (see Additional file 5). For phylogeneticanalysis, maximum likelihood (ML) analysis was con-ducted with MultiPhyl v1.0.6 [53] with SPR (subtreepruning and regrafting) branch swapping. The optimalevolutionary model selected for ML was K81uf+I+G(Kimura three parameter with unequal base frequencies +proportion of sites + gamma distribution) using the fol-lowing parameters: assumed nucleotide frequencies A =0.36736, C = 0.20950, G = 0.27698, T = 0.14616; expectedtransition/transversion ratio = 0.94; expected pyrimidinetransition/purine transition ratio = 0.30; proportion sitesassumed to be invariable = 0.29; rates for variable sitesassumed to follow the gamma distribution with shapeparameter = 4.28. Bootstrapping was performed usingMultiPhyl v1.0.6 with neighbor-joining algorithm and100 replicates [53].Ka/Ks analyses and likelihood ratio tests (LRT) for positive selectionNonsynonymous (Ka) and synonymous (Ks) nucleotidesubstitution rates were calculated by using program yn00in PAML v3.15 [23] (see Additional file 2). The t-test wasused for testing for significant differences of Ka, Ks, and Ka/Ks ratios between the nucp rps13 and numit rps13 cDNAsequences. The statistically significant level was set at95%. The statistical analysis in this study was imple-mented by using the statistical package R [54].For detection of positive selection, codon-based analysiswas implemented using codeml in PAML v3.15 [23]. Themature coding region, without the mitochondrial target-ing presequences, was included in our analysis. Site spe-cific models were used for testing positive selection onnumit rps13 and nucp rps13 [25]. Two LRTs were used forthe detection of positive selection: M7-M8 and M8a-M8[22,25,26]. Because comparison of M7 and M8 provides amore powerful test of positive selection and has less sen-sitivity to large evolutionary distances and G+C contentthan comparison of M1 and M2 [55,56], only comparisonof M7 and M8 was used for the detection of positive selec-tion in this study. M7 and M8a models are the null mod-els without positive selection (no codon with Ka/Ks > 1)and the M8 model is the alternative model with positiverps13, nucp rps13, and mt rps13 [25]. For this analysis, thebranch which we are interested in testing positive selec-tion was assigned as foreground lineage. Therefore, Malusbranch in numit rps13, nucp rps13, and mt rps13 were setas foreground lineage and the rest of other rosid specieswere designated as background lineage. For branch spe-cific model, two LRTs were used for detection of positiveselection as follows: M1-Model A test1 (MAtest1) andModel A test2 (MAtest2)-MAtest1 [25,26]. M1 and MAtest2models are the null hypothesis without positive selection(no codon with Ka/Ks > 1) and the MAtest1 model is thealternative selection with positive selection. For site-spe-cific and branch-specific models used for detection of pos-itive selection, M8a is more stringent than M7 and MAtest2is more stringent than M1 because their ω2 (Ka/Ks) is fixedat 1. For all LRTs, the first model is simpler than the sec-ond one, with fewer parameters and poor fit to the data.Therefore, the first model has a lower maximum likeli-hood index. To test if there is statistically better maximumlikelihood for the second model, twice difference of logmaximum likelihood values between the two comparedmodels [2δL = 2(Ln2-Ln1), where Ln1 and Ln2 representfor log of maximum likelihood value in the first modeland the second model] was compared against χ2 distribu-tion. The degrees of freedom (d.f.) equals to the addi-tional parameters used in more complex model. The d.f.is 2 for M7-M8 and M1-MAtest1, while the d.f. is 1 for M8a-M8 and MAtest2-MAtest1. However, it was argued that theappropriate model comparison for M8a-M8 would be touse 50:50 mixture of d.f. = 0 and d.f. = 1 [22]. In this study,the calculation of P value for M8a-M8 was followed asdescribed in Kapralov and Filatov [57]. P value was firstobtained using d.f. = 1 and then divided by 2. Bayesempirical bayes (BEB) approach was used to determinepositive selected sites in M8 model and MAtest1 model. InBEB analysis, the posterior probabilities were calculated toselect the codon with Ka/Ks greater than 1. Because BEBanalysis is more powerful to detect positive selection thannaive empirical Bayes (NEB) analysis [58], only BEB anal-ysis was considered in this study.Structural analysis of RPS13Using first approach mode in SWISS-MODEL [59], nucp,numit, and mt RPS13 amino acid sequences in Malus wereselected to search for a suitable template for further struc-tural analysis. Based on the results, RPS13 of Escherichiacoli and Thermus thermophilus were chosen for the follow-ing structural analysis. RPS13 structural data file for E. coli(PDB ID: 2gy9M) [60] and T. thermophilus (PDB ID:1fjgM) [27] were obtained from RCSB Protein Data Bank[61]. Location and property of each amino acid in E. coliand T. thermophilus RPS13 was analyzed using DeepView-Swiss-PdbViewer v.3.7 [59]. Selected positive sitesPage 12 of 16(page number not for citation purposes)selection. Branch specific model was used to test if therehas been positive selection for Malus branch in numitobtained by BEB analysis and sites shared with similaritybetween numit RPS13 and mt RPS13 were plotted in theBMC Evolutionary Biology 2008, 8:25 http://www.biomedcentral.com/1471-2148/8/25relative position of RPS13 in E. coli and T. thermophilusaccording to amino acid alignment (Figure 4).Microarray data analysisRaw Affymetrix ATH1 microarray data [30] were down-loaded from the TAIR website (TAIR accession number:ME00319)[62]. Raw data were processed and normalizedbased on the GC-RMA method [63]. The expression val-ues were converted into log2 numbers by only consider-ing values of perfect-match probes [63]. Normalizationand analysis of the microarray data were implementedusing Bioconductor. Pearson correlation analysis was con-ducted to statistically compare the similarity of expressionprofile among rps9, rps10, rps11, and numit rps13, andbetween numit rps13 and nucp rps13. The correlationcoefficient (R) values correspond to the similarity of theexpression profile between two genes. A one-way ANOVAwith Bonferroni post hoc tests was used to test if there is asignificant expression difference between two differentgenes or any two different organ types. All statistical anal-yses were implemented using statistical package R [54].The statistically significant level was set at 95%.Plant materials and abiotic stress treatmentsSeveral organ types, including stems, leaves, peduncles,sepals, petals, stamens, stigmas+styles, and young fruit ofapple (Malus domestica Borkh.) were collected from theUniversity of British Columbia's Botanical Garden. Seed-lings were used for abiotic stress experiments. Prior tosowing, apple seeds were soaked in distilled water at 4°Cfor 6 weeks for vernalization. Seedlings were cultivated ina peat-vermiculite soil mixture under fixed day/night (16h day/8 h night) and temperature of 20–23°C. After ger-mination and emergence from the soil for 7 days, theplants were subjected to five different abiotic stresses: 4°Cfor 7 days (cold treatment), 37–40°C for 12 hours (heatstress), 100 mM NaCl solution for 7 days (salt treatment),submerged in distilled water for 4 days (water submersiontreatment), and dark for 7 days (dark treatment). Afterstress treatments, roots, hypocotyls, cotyledons, andleaves were collected and immediately frozen in liquidnitrogen and stored at -80°C until nucleic acid extraction.Nucleic acid extraction, gene amplification, and sequencingGenomic DNA extraction was done using the DNeasyPlant Mini Kit (Qiagen) following the manufacturer's pro-tocol. Total RNA extraction was performed as describedpreviously [64]. The extracted nucleic acid concentrationsand purities were determined by using a NanoDrop spec-trophotometer. The quality was checked by running on1.5% agarose gels. Before reverse transcription, 3 μg ofRNA (500 μg/μl) was treated with 1 unit of DNaseI (Newby using M-MLV reverse transcriptase (Invitrogen). Thereverse transcription conditions were 25°C for 10 min,37°C for 60 min, and 70°C for 15 min. Finally, thereverse-transcribed samples were treated with RNase (Inv-itrogen) at 37°C for 20 min.The rps13 genes were amplified from genomic DNA andcDNA by polymerase chain reaction (PCR) using gene-specific primers (see Additional file 6). The PCR was per-formed in a reaction mixture (10 μl) consisting of 4.88 μlof ddH2O, 1 μl of genomic DNA/cDNA solution, 1 μl ofPCR buffer, 1 μl of 2.5 mM MgCl2 solution, 1 μl of 0.2 mMdNTPs, 0.5 μl of 0.4 μM each primer, and 0.12 units ofTaq DNA Polymerase (Sigma). The PCR conditions were96°C for 4 min, and 30 cycles of 96°C for 40 s, 60°C for40 s, 72°C for 1 min, and 72°C for 10 min. The PCR prod-ucts were run on a 1.5% agarose gel and extracted from thegel using QIAquick Gel Extraction Kit (Qiagen). PCRproducts amplified from genomic DNA and cDNA weresequenced directly. The sequencing was performed in areaction mixture containing 0.4 μl of ABI BigDye Version3.1 (Applied Biosystems), 3.6 μl of BigDye buffer, 5.5 μlof 50 ng template, and 0.5 μl of 0.4 μM forward or reverseprimer. The sequencing reaction was carried out with thefollowing program: 1 min at 96°C, and 25 cycle of 10 s at96°C, 5 s at 50°C, and 4 min at 60°C. The sequencingproducts were run on an ABI 377 DNA Sequencer(Applied Biosystems) at the UBC Centre for PlantResearch, or on an ABI 3730 Sequencer at the NucleicAcids Protein Service unit at UBC. The GenBank accessionnumber for mt rps13 in Malus, including informationabout RNA editing sites, is [GenBank: EU084692].Real-time qRT-PCRQuantitative real-time RT-PCR was performed with a Bio-Rad iQ5 system using SYBR green master mix (BioRad)following the manufacturer's instructions, except that25μl total reaction volumes were used. The PCR condi-tions were 96°C for 3 min, and then 35 cycles of 96°C for10 s, 58°C for 30 s, and 72°C for 30 s. Gene-specific prim-ers are listed in Additional file 6. For each sample, twotechnical replicates were performed. Reactions for a stand-ard curve were run with each set of experimental reactions.After the completion of PCR, the melting curves were ana-lyzed to distinguish the true product from artifacts such asprimer dimers. The iQ5 software and Microsoft Excel wereused for data analysis. Normalization was done using theactin gene ACT2. GenBank accession numbers forsequences used to design primers are: Malus ACT2:CN903171, CN902302, and N917499; Malus cob:CN872477, CN856986, CN856316; Malus rps10:CV882925.Page 13 of 16(page number not for citation purposes)England Biolabs) and incubated at 37°C for 30 min twice.Then 4 μg of DNase-treated RNA was reverse-transcribedBMC Evolutionary Biology 2008, 8:25 http://www.biomedcentral.com/1471-2148/8/25AbbreviationsBEB, Bayes Empirical Bayes analysis; EST, expressedsequence tag; LRT, likelihood ratio test; ML, maximumlikelihood; NCBI, National Center for BiotechnologyInformation; qRT-PCR, quantitative real-time polymerasechain reaction; RT-PCR, reverse transcription-polymerasechain reactionAuthors' contributionsSLL planned and carried out the experiments and analy-ses, and wrote the manuscript. KA conceived the study,helped with the experimental design, and wrote the man-uscript. Both authors read and approved of the final man-uscript.Additional materialAcknowledgementsWe thank Nolan Kane and Michael Barker for advice on the Ka/Ks analysis and LRTs, two anonymous reviewers for helpful comments on an earlier version of the manuscript, and the UBC Botanical Garden for providing tis-sue from their apple trees. This study was funded by a grant from the National Science and Engineering Research Council of Canada, by start-up funds from the Faculty of Land and Food Systems at the University of British Columbia, and by infrastructure funds from the Canadian Foundation for Innovation.References1. Wagner A: Selection after gene duplication: a view from thegenome.  Genome Biol 2002, 3:1012.1-1012.3.2. Zhang J: Evolution by gene duplication: an update.  Trends EcolEvol 2003, 18:292-298.3. Li WH, Yang J, Gu X: Expression divergence between duplicategenes.  Trends Genet 2005, 21:602-607.4. Ohno S: Evolution by gene duplication.  New York, Springer;1970. 5. Force A, Lynch M, Pickett FB, Amores A, Yan Y, Postlethwait J: Pres-ervation of duplicate genes by complementary, degenerativemutations.  Genetics 1999, 151:1531-1545.6. Lynch M, Force A: The probability of duplicate gene preserva-tion by subfunctionalization.  Genetics 2000, 154:459-473.7. Moore RC, Purugganan MD: The early stages of duplicate geneevolution.  Proc Natl Acad Sci USA 2003, 100:15682-15687.8. Lawton-Rauh A: Evolutionary dynamics of duplicated genes inplants.  Mol Phylogenet Evol 2003, 29:396-409.9. Moore RC, Purugganan MD: The evolutionary dynamics of plantduplicate genes.  Curr Opin Plant Biol 2005, 8:122-128.10. Adams KL, Wendel JF: Polyploidy and genome evolution inplants.  Curr Opinion Plant Biol 2005, 8:135-141.11. Adams KL, Daley DO, Whelan J, Palmer JD: Genes for two mito-chondrial ribosomal proteins in flowering plants are derivedfrom their chloroplast or cytosolic counterparts.  Plant Cell2002, 14:931-943.12. Mollier P, Holffmann B, Debast C, Small I: The gene encoding Ara-bidopsis thaliana mitochondrial ribosomal protein S13 is arecent duplication of the gene encoding plastid S13.  CurrGenet 2002, 40:405-409.13. Adams KL, Qiu YL, Stoutemyer M, Palmer JD: Punctuated evolu-tion of mitochondrial gene content: high and variable ratesof mitochondrial gene Loss and transfer to the nucleus dur-ing angiosperm evolution.  Proc Natl Acad Sci USA 2002,99:9905-9912.14. Claros MG, Vincens P: Computational method to predict mito-chondrially imported proteins and their targetingsequences.  Eur J Biochem 1996, 241:779-786.15. Emanuelsson O, Nielsen H, Brunak S, von Heijne G: Predicting sub-cellular localization of proteins based on their N-terminalamino acid sequence.  J Mol Biol 2000, 300:1005-1016.16. Small I, Peeters N, Legeai F, Lurin C: Predotar: A tool for rapidlyscreening proteomes for N-terminal targeting sequences.Proteomics 2004, 4:1581-1590.17. Bannai H, Tamada Y, Maruyama O, Nakai K, Miyano S: Extensivefeature detection of N-terminal protein sorting signals.  Bio-informatics 2002, 18:298-305.18. Soltis DE, Soltis PE, Endress PK, Chase MW: Phylogeny and Evo-Additional file 1Mt rps13 distribution among rosids. The figure shows losses of mt rps13 among rosids in a phylogenetic context, and the inferred timing of numit rps13 formation by gene duplication.Click here for file[http://www.biomedcentral.com/content/supplementary/1471-2148-8-25-S1.PDF]Additional file 2Pairwise Ka and Ks analyses. The table shows the results of Ka analysis and Ks analysis for numit rps13 and nucp rps13 in seven rosid species.Click here for file[http://www.biomedcentral.com/content/supplementary/1471-2148-8-25-S2.PDF]Additional file 3Microarray data legend. The table lists the organs and developmental stages from Arabidopsis thaliana for which microarray data were ana-lyzed.Click here for file[http://www.biomedcentral.com/content/supplementary/1471-2148-8-25-S3.PDF]Additional file 4RNA editing of mt rps13 in Malus compared with other flowering plants. The figure shows an alignment of mt rps13 from several land plants, indicating the positions of RNA editing sites in Malus, and a list of mt rps13 RNA editing sites from several angiosperms.Click here for file[http://www.biomedcentral.com/content/supplementary/1471-2148-8-25-S4.PDF]Additional file 5Sequence alignment. The figure shows an alignment of numit rps13, nucp rps13, and cDNAs of mt rps13 for taxa analyzed in this study. The targeting sequences at the 5' end of the numit rps13 and nucp rps13 genes do not align with each other and were excluded from Ka/Ks analysis.Click here for file[http://www.biomedcentral.com/content/supplementary/1471-2148-8-25-S5.PDF]Additional file 6RT-PCR primers. The table lists the primers used for RT-PCR and qRT-PCR experiments.Click here for file[http://www.biomedcentral.com/content/supplementary/1471-2148-8-25-S6.PDF]Page 14 of 16(page number not for citation purposes)lution of Angiosperms.  Sunderland, Sinauer Associates, Inc., USA;2005:370. BMC Evolutionary Biology 2008, 8:25 http://www.biomedcentral.com/1471-2148/8/2519. Jansen RK, Kaittanis C, Saski C, Lee SB, Tomkins J, Alverson AJ, Dan-iell H: Phylogenetic analyses of Vitis (Vitaceae) based on com-plete chloroplast genome sequences: effects of taxonsampling and phylogenetic methods on resolving relation-ships among rosids.  BMC Evol Biol 2006, 6:32.20. Jaillon O, Aury JM, Noel B, Policriti A, Clepet C, Casagrande A,Choisne N, Aubourg S, Vitulo N, Jubin C, Vezzi A, Legeai F, HugueneyP, Dasilva C, Horner D, Mica E, Jublot D, Poulain J, Bruyere C, BillaultA, Segurens B, Gouyvenoux M, Ugarte E, Cattonaro F, Anthouard V,Vico V, Fabbro CD, Alaux M, Gaspero GD, Dumas V, Felice N, Pail-lard S, Juman I, Moroldo M, Scalabrin S, Canaguier A, Clainche IL,Malacrida G, Durand E, Pesole G, Laucou V, Chatelet P, MerdinogluD, Delledonne M, Pezzotti M, Lecharny A, Scarpelli C, Artiguenave F,Pe ME, Valle G, Morgante M, Caboche M, Adam-Blondon AF, Weis-senbach J, Quetier F, Wincker P: The grapevine genomesequence suggests ancestral hexaploidization in majorangiosperm phyla.  Nature 2007, 449:463-467.21. Kumar R, Drouaud J, Raynal M, Small I: Characterization of thenuclear gene encoding chloroplast ribosomal protein S13from Arabidopsis thaliana.  Curr Genet 1995, 28:346-352.22. Swanson WJ, Nielson R, Yang Q: Pervasive adaptive evolution inmammalian fertilization protein.  Mol Biol Evol 2003, 20:18-20.23. Yang Z: PAML: a program package for phylogenetic analysisby maximum likelihood.  Comput Appl Biosci 1997, 13:555-556.24. Yang Z Swanson, W.J.: Codon-substitution models to detectadaptive evolution that account for heterogeneous selectivepressures among site clesses.  Mol Biol Evol 2002, 19:49-57.25. Yang Z: PAML 4: Phylogenetic analysis by maximum likeli-hood.  Mol Biol Evol 2007, 24:1586-1591.26. Yang Z, Nielson R: Codon-substitution models for detectingmolecular adaptation at individual sites along specific line-ages.  Mol Biol Evol 2002, 19:908-917.27. Carter AP, Clemons WM, Broderson DE, Morgan-Warren RJ, Wim-berly BT, Ramakrishnan V: Functional insights from the struc-ture of the 30S ribosomal subunit and its interactions withantibiotics.  Nature 2000, 407:340-348.28. Yusupov MM, Yusupova GZ, Baucom A, Lieberman K, Earnest TN,Cate JHD, Noller HF: Crystal structure of the ribosome at 5.5Aresolusion.  Science 2001, 292:883-896.29. Broderson DE, Clemons Jr. WM, Carter AP, Wimberly BT, Ram-akrishnan V: Crystal structure of the 30S ribosomal subunitfrom Thermus thermophilus: structure of the proteins andtheir interactions with 16S RNA.  J Mol Biol 2002, 316:725-768.30. Schmid M, Davison TS, Henz SR, Pape UJ, Demar M, Vingron M,Scholkopf B, Weigel D, Lohmann JU: A gene expression map ofArabidopsis thaliana development.  Nature Genet 2005,37:501-506.31. Bonen L, Calixte S: Comparative analysis of bacterial-origingenes for plant mitochondrial ribosomal proteins.  Mol BiolEvol 2006, 23:701-712.32. Newcomb RD, Crowhurst RN, Gleave AP, Rikkerink EH, Allan AC,Beuning LL, Bowen JH, Gera E, Jamieson KR, Janssen BJ, Laing WA,McArtney S, Nain B, Ross GS, Snowden KC, Souleyre EJ, Walton EF,Yauk YK: Analyses of expressed sequence tags from apple.Plant Physiol 2006, 141:147-166.33. Shikanai T: RNA editing in plant organelles: machinery, physi-ological function and evolution.  Cell Mol Life Sci 2006,63:698-708.34. Conant GC, Wagner A: Asymmetric sequence divergence ofduplicate genes.  Genome Res 2003, 13:2052-2058.35. Byrne KP, Wolfe KH: Consistent patterns of rate asymmetryand gene loss indicate widespread neofunctionalization ofyeat genes after whole-genome duplication.  Genetics 2007,175:1341-1350.36. Thomas JH: Adaptive evolution in two large families of ubiqui-tin-ligase adapters in nematodes and plant.  Genome Res 2006,16:1017-1030.37. Sun X, Cao Y, Wang S: Point mutations with positive selectionwere a major force during the evolution of a receptor-kinaseresistance gene family of rice.  Plant Physiol 2006, 140:998-1008.38. Johnson DA, Thomas MA: The monosaccharide transportergene family in Arabidopsis and rice: A history of duplications,adaptive Evolution, andfunctional divergence.  Mol Biol Evol2007, 24:2412-2423.tors of flower development: adaptive evolution in protein-protein interaction domains after major gene duplicationevents.  Mol Biol Evol 2007, 24:465-481.40. Gu Z, Nicolae D, Lu HH, Li WH: Rapid divergence in expressionbetween duplicate genes inferred from microarray data.Trends Genet 2002, 18:609-613.41. Gu X, Zhang Z, Huang W: Rapid evolution of expression andregulatory divergences after yeast gene duplication.  Proc NatlAcad Sci USA 2005, 18:707-712.42. Chung WY, Albert R, Albert I, Nekrutenko A, Makova KD: Rapidand asymmetric divergence of duplicate genes in the humangene coexpression network.  BMC Bioinformatics 2006, 27:7:46.43. Blanc G, Wolfe KH: Functional divergence of duplicated genesformed by polyploidy during Arabidopsis evolution.  Plant Cell2004, 16:1679-1691.44. Casneuf T, De Bodt S, Raes J, Maere S, Van de Peer Y: Nonrandomdivergence of gene expression following gene and genomeduplications in the flowering plant Arabidopsis thaliana.Genome Biol 2006, 7:R13.45. Ganko EW, Meyers BC, Vision TJ: Divergence in Expressionbetween Duplicated Genes in Arabidopsis.  Mol Biol Evol 2007,24:2298-2309.46. Adams KL, Palmer JD: Evolution of mitochondrial gene content:gene loss and transfer to the nucleus.  Mol Phylogenet Evol 2003,29:380-395.47. Bittner-Eddy P, Monroy AF, Brambl R: Expression of mitochon-drial genes in the germinating conidia of Neurospora crassa.J Mol Biol 1994, 235:881-897.48. Adams KL, Song K, Roessler PG, Nugent JM, Doyle JL, Doyle JJ,Palmer JD: Intracellular gene transfer in action: Dual tran-scription and multiple silencings of nuclear and mitochon-drial cox2 genes in legumes.  Proc Natl Acad Sci USA 1999,96:13863-13868.49. Sandoval P, Leon G, Gomez I, Carmona R, Figueroa P, Holuigue L,Araya A, Jordana X: Transfer of RPS14 and RPL5 from themitochondrion to the nucleus in grasses.  Gene 2004,324:139-147.50. Choi C, Liu Z, Adams KL: Evolutionary transfers of mitochon-drial genes to the nucleus in the Populus lineage and coex-pression of nuclear and mitochondrial Sdh4 genes.  New Phytol2006, 172:429-439.51. Bininda-Emonds ORP: transAlign: using amino acids to facili-tate the multiple alignment of protein-coding DNAsequences.  BMC Bioinformatics 2005, 6:156.52. Hall TA: BioEdit: a user-friendly biological sequence align-ment editor and analysis program for Windows 95/98/NT.Nucleic Acids Symp Ser 1999, 41:95-98.53. Keane TM, Naughton TJ, McInerney JO: MultiPhyl: A high-throughput phylogenomics webserver using distributedcomputing.  Nucleic Acids Res 2007, 35:W33-W37.54. The statistical package R.   [http://www.r-project.org/]55. Yang Z, Nielson R, Goldman N, Pedersen AMK: Codon-substitu-tion models for heterogeneous selection pressure at aminoacid sites.  Genetics 2000, 155:431-449.56. Friedman R, Hughes AL: Likelihood-ratio tests for positiveselection of human and mouse duplicate genes reveal non-conservative and anomalous properties of widely used meth-ods.  Mol Phylogenet Evol 2006, 42:388-393.57. Kapralov MV, Filatov DA: Widespread positive selection in thephotosynthetic Rubisco enzyme.  BMC Evol Biol 2007, 7:73.58. Yang Z, Wong WS, Nielsen R: Bayes empirical bayes inferenceof amino acid sites under positive selection.  Mol Biol Evol 2005,22:1107-1118.59. Schwede T, Kopp J, Gues N, M.C. P: SWISS-MODEL: an auto-mated protein homology-modeling server.  Nucleic Acids Res2003, 31:3381-3385.60. Mitra K, Schaffitzel C, Fabiola F, Chapman MS, Ban N, Frank J: Elon-gation arrest by SecM via a cascade of ribosomal RNA rear-rangements.  Mol Cell 2006, 22:533-543.61. RCSB Protein Data Bank.   [http://www.rcsb.org/pdb]62. TAIR: The Arabidopsis Information Resource.   [http://www.arabidopsis.org/]63. Wu Z, Irizarry RA, Gentleman R, Murillo FM, Spencer F: A modelbased background adjustment for oligonucleotide expres-Page 15 of 16(page number not for citation purposes)39. Hernandez-Hernandez T, Martinez-Castilla LP, Alvarez-Buylla ER:Functional diversification of B MADS-box homeotic regula-sion arrays.  J Am Stat Assoc 2004, 99:909-917.Publish with BioMed Central   and  every scientist can read your work free of charge"BioMed Central will be the most significant development for disseminating the results of biomedical research in our lifetime."Sir Paul Nurse, Cancer Research UKYour research papers will be:available free of charge to the entire biomedical communitypeer reviewed and published immediately upon acceptancecited in PubMed and archived on PubMed Central BMC Evolutionary Biology 2008, 8:25 http://www.biomedcentral.com/1471-2148/8/2564. Adams KL, Cronn R, Percifield R, Wendel JF: Genes duplicated bypolyploidy show unequal contributions to the transcriptomeand organ-specific reciprocal silencing.  Proc Natl Acad Sci USA2003, 100:4649-4654.yours — you keep the copyrightSubmit your manuscript here:http://www.biomedcentral.com/info/publishing_adv.aspBioMedcentralPage 16 of 16(page number not for citation purposes)


Citation Scheme:


Citations by CSL (citeproc-js)

Usage Statistics



Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            async >
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:


Related Items