UBC Faculty Research and Publications

A complex and punctate distribution of three eukaryotic genes derived by lateral gene transfer Rogers, Matthew B; Watkins, Russell F; Harper, James T; Durnford, Dion G; Gray, Michael W; Keeling, Patrick J Jun 11, 2007

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

Item Metadata


52383-12862_2007_Article_383.pdf [ 1.02MB ]
JSON: 52383-1.0223227.json
JSON-LD: 52383-1.0223227-ld.json
RDF/XML (Pretty): 52383-1.0223227-rdf.xml
RDF/JSON: 52383-1.0223227-rdf.json
Turtle: 52383-1.0223227-turtle.txt
N-Triples: 52383-1.0223227-rdf-ntriples.txt
Original Record: 52383-1.0223227-source.json
Full Text

Full Text

ralssBioMed CentBMC Evolutionary BiologyOpen AcceResearch articleA complex and punctate distribution of three eukaryotic genes derived by lateral gene transferMatthew B Rogers1, Russell F Watkins2,4, James T Harper1, Dion G Durnford3, Michael W Gray2 and Patrick J Keeling*1Address: 1Department of Botany, University of British Columbia, 3529-6270 University Boulevard, Vancouver, British Columbia V6T 1Z4, Canada, 2Department of Biochemistry and Molecular Biology, Dalhousie University, Halifax, Nova Scotia, B3H 1X5, Canada, 3Department of Biology, University of New Brunswick, Fredericton, NB, E3B 5A3, Canada and 4Centre for Molecular Medicine and Therapeutics, 950 West 28th Avenue, Vancouver, B.C., V5Z 4H4, CanadaEmail: Matthew B Rogers - mbrogers@interchange.ubc.ca; Russell F Watkins - rwatkins@cmmt.ubc.ca; James T Harper - jtharper@interchange.ubc.ca; Dion G Durnford - durnford@unb.ca; Michael W Gray - M.W.Gray@Dal.Ca; Patrick J Keeling* - pkeeling@interchange.ubc.ca* Corresponding author    AbstractBackground: Lateral gene transfer is increasingly invoked to explain phylogenetic results thatconflict with our understanding of organismal relationships. In eukaryotes, the most commonobservation interpreted in this way is the appearance of a bacterial gene (one that is not clearlyderived from the mitochondrion or plastid) in a eukaryotic nuclear genome. Ideally such anobservation would involve a single eukaryote or a small group of related eukaryotes encoding agene from a specific bacterial lineage.Results: Here we show that several apparently simple cases of lateral transfer are actually morecomplex than they originally appeared: in these instances we find that two or more distantly relatedeukaryotic groups share the same bacterial gene, resulting in a punctate distribution. Specifically,we describe phylogenies of three core carbon metabolic enzymes: transketolase, glyceraldehyde-3-phosphate dehydrogenase and ribulose-5-phosphate-3-epimerase. Phylogenetic trees of each ofthese enzymes includes a strongly-supported clade consisting of several eukaryotes that aredistantly related at the organismal level, but whose enzymes are apparently all derived from thesame lateral transfer. With less sampling any one of these examples would appear to be a simplecase of bacterium-to-eukaryote lateral transfer; taken together, their evolutionary histories cannotbe so simple. The distributions of these genes may represent ancient paralogy events or genes thathave been transferred from bacteria to an ancient ancestor of the eukaryotes that retain them.They may alternatively have been transferred laterally from a bacterium to a single eukaryoticlineage and subsequently transferred between distantly related eukaryotes.Conclusion: Determining how complex the distribution of a transferred gene is depends on thesampling available. These results show that seemingly simple cases may be revealed to be morecomplex with greater sampling, suggesting many bacterial genes found in eukaryotic genomes mayhave a punctate distribution.Published: 11 June 2007BMC Evolutionary Biology 2007, 7:89 doi:10.1186/1471-2148-7-89Received: 9 January 2007Accepted: 11 June 2007This article is available from: http://www.biomedcentral.com/1471-2148/7/89© 2007 Rogers et al; 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 13(page number not for citation purposes)BMC Evolutionary Biology 2007, 7:89 http://www.biomedcentral.com/1471-2148/7/89BackgroundLateral gene transfer is the movement of genes betweendistantly related organisms, a phenomenon that hasbecome a major focus in the study of genome evolution[1-7]. The importance of gene transfers between prokary-otic genomes is now generally recognized due to the manysuch genomes now available for comparison, althoughthere is still controversy about how common prokaryote-to-prokaryote gene transfer is, and what long term effectsit has [8-10]. For eukaryotes there are far fewer sequencedgenomes to compare, and emerging evidence suggests thatlateral gene transfer may not be prevalent in many of thelineages where the most data are available, such as verte-brates [11,12]. Nevertheless, convincing examples ofprokaryote-to-eukaryote gene transfers have beendescribed [13-17], and transfers between eukaryotes arealso known [15,18-21]. While these studies make it clearthat lateral transfer has affected eukaryotic nucleargenomes, the frequency of such events and the extent oftheir evolutionary impact, particularly for eukaryote-to-eukaryote lateral transfers remains unknown. In particu-lar, most known cases involve genes moving from aprokaryote to eukaryotes, whereas comparatively little isknown about transfers between eukaryotes.Various approaches have been used to infer gene transfersbetween prokaryotic genomes [22]. By far the most com-mon method of detecting events involving eukaryotes isto observe incongruence between phylogenetic treesbased on a gene and the tree that is considered (on thebasis of other evidence) to reflect the evolution of theorganism in which the gene is encoded. Lateral gene trans-fer events are thus commonly invoked to explain trees thatdepart from an expected topology. Nevertheless, otherexplanations can account for these incongruent topolo-gies, including reconstructions that do not accuratelyreflect the history of the gene due to problems such as thefailure to account for rate-across-site variation, or covari-ance and biased amino acid composition across the tree[23]. Similarly, the history of the gene may be complex inways that erroneously suggest lateral transfer even whenthe phylogeny is accurately reconstructed. For example,gene duplication events and differential extinction of theresulting paralogues across different lineages can lead to atree that appears to reflect lateral transfer but that reallydescribes the history of a duplicated gene.Another emerging problem for the interpretation of lat-eral gene transfer is the observation of genes demonstrat-ing punctate distributions. For eukaryotes, the term'punctate distribution' has been used to refer to caseswhere two or more distantly related eukaryotic lineagespossess closely related genes that are either not found into 'patchy distribution', a term that has been used to referto genes with a limited distribution in both eukaryotesand prokaryotes [25]. Genes with a punctate distributionare significant because they have been interpreted in sev-eral different ways, each interpretation having its ownimportant implications. On one hand, such distributionshave been supposed to represent a single transfer to thecommon ancestor of two or more disparate lineages. Thereconstruction of several large-scale relationships amongeukaryotes (the so-called supergroups) has been partiallybased on the documentation of shared, rare characteristicssuch as gene fusions or indels in two or more lineages [26-28]. Shared lateral gene transfers have also been regardedin this way. Recent examples include the shared posses-sion of a nanoarchaeal prolyl-tRNA synthetase in tri-chomonad and diplomonad flagellates [29] and ahaloarchaeal tyrosyl-tRNA synthetase in opisthokonts[30]. In contrast, more complex distributions have beeninterpreted to represent multiple transfers or eukaryote-to-eukaryote transfers. These cases are also significantbecause detecting transfers between eukaryotes is madedifficult by poor sampling of many lineages and poor res-olution of many phylogenies, both of which impede thedistinction between horizontal and vertical descent. Forthis reason, some eukaryote-to-eukaryote transfers havebeen argued based on the fact that the gene bears somespecial feature, such as an insertion, an accelerated rate ofsubstitution, or even an origin from another lateral trans-fer event [18,24,29].For any of these genes the possibility of ancient paralogyfollowed by selective loss or retention in diverse eukaryo-tes must also be considered. Even in such cases however,the origin of the gene may still ultimately be due to a lat-eral transfer event, whereas its distribution is due to otherfactors. Deciding between these interpretations involvesbalancing a variety of observations, including how wide-spread the gene in question is, how closely related theorganisms that possess the gene are believed to be,whether other close relatives possess or lack the gene, andhow convincingly a source lineage for the gene can beidentified. In many cases we cannot answer these ques-tions because data are lacking from a sufficient diversity ofeukaryotes, such that it is impossible to conclude whatmight be the underlying cause. Moreover, it is impossibleto say whether the factors resulting in punctate clades arecommon or rare: in many cases bacterium-to-eukaryotetransfers have been inferred from data with seeminglysimple distributions, but it is possible that some of thesedistributions only appear simple because of samplingdeficiency.Here we use EST data to evaluate what are apparently sim-Page 2 of 13(page number not for citation purposes)other eukaryotes, or are clearly different from othereukaryotic homologues [24]. This situation is in contrastple cases of lateral transfer of genes involved in core car-bon metabolism. Surprisingly, we find in each instance aBMC Evolutionary Biology 2007, 7:89 http://www.biomedcentral.com/1471-2148/7/89more complex, punctate distribution than suggested bythe initial observations. In this study, we characterize pro-tist homologues of three genes: ribulose-5-phosphate-3-epimerase (RPE), glyceraldehyde-3-phosphate dehydro-genase (GAPDH), and transketolase (TK). In the case ofRPE, a transfer event has been described between a γ-pro-teobacterium related to Pseudomonas and a chlorarachnio-phyte [15]. Similarly, many lateral transfer eventsinvolving eukaryotic GAPDH have been described[15,21,31-34], including one isolated clade of diplone-mid genes inferred to be have been transferred from a pro-teobacterium [33]. In the case of transketolase (TK), thechlorarachniophyte TK is not closely related to othereukaryotic homologues, but is very similar to homologuesfrom Chlamydiales. For each of these genes, a single trans-fer from a bacterium to a eukaryote is evident from therelationship between the eukaryotic and a particular sub-group of bacteria, but we show here that the distributionwithin eukaryotes suggests a more complicated history. Inall three cases we find homologues in other eukaryotesthat are only distantly related to the organism in whichthe gene was first found. How the complex distribution ofthese genes arose is uncertain, but these examples indicatethat such a punctate pattern of presence is more commonthan previously thought, leading us to suggest that otherapparently simple cases of lateral transfer may be morecomplex than they appear.Results & DiscussionRibulose-5-Phosphate-3-EpimeraseRibulose-5-phosphate epimerase (RPE) catalyzes the bidi-rectional conversion of ribulose-5-phosphate to xylulose-5-phosphate both in the Calvin cycle in the plastid of pho-tosynthetic eukaryotes and in the cytosolic pentose-phos-phate pathway of both photosynthetic and non-photosynthetic eukaryotes. Phototrophic eukaryotestherefore have two forms of this enzyme: the plastid-tar-geted form in red algae, green algae and plants is cyano-bacterial, whereas the cytosolic form is related to that ofnon-photosynthetic eukaryotes, as expected (Figure 1).The relationships among cytosolic epimerases of eukaryo-tes are poorly resolved, with only recently diverginggroups such as vascular plants and metazoa recoveredwith strong bootstrap support.Previously, it has been shown that the plastid-targetedRPE of the chlorarachniophyte B. natans is not related toother plastid-targeted or even cyanobacterial genes, as onewould expect, but is instead closely related to the γ-pro-teobacterial genus Pseudomonas [15]. By increasing thesampling of RPE from EST and genomic data of other pro-tists, a similar proteobacterial RPE sequence was found insix different chromalveolate genomes. Phylogenetic anal-of the γ-proteobacterial type, as are those from the dia-toms Thalassiosira pseudonana and Phaeodactylum tricornu-tum (Figure 1). A fragment of a highly similar gene wasalso identified in the haptophyte Isochrysis galbana (Gen-Bank accession EC139053). It was too short to beincluded in the analysis, but preliminary trees confirmedit was closely related to the other haptophyte RPE genes(not shown). Relationships between the chromalveolateand B. natans genes (here designated RPE-γ) are unre-solved, but collectively they form a strongly-supportedgroup with γ-proteobacterial homologues, and more spe-cifically the relationship with pseudomonads andalteromonads remains well supported. The RPE-γ from B.natans was reported to have a truncated N-terminal leader,suggesting it is plastid-targeted [15]. Evidence for thisclade consisting of plastid-targeted proteins also comesfrom P. tricornutum and E. huxleyi RPE-γ sequences, thathave full-length leaders predicted to encode signal pep-tides at the N-terminus, which is a characteristic of plastid-targeted proteins in these organisms. The P. parvum, P.lutheri and T. pseudonana RPE-γ sequences also all havetruncated N-terminal leaders, further suggesting that thisentire clade of proteobacterium-derived RPE proteins isplastid-targeted. Unlike the RPE of B. natans which hasbeen demonstrated to be a bacterial gene in a eukaryoticgenome by the presence spliceosomal introns, and thediatoms RPEs which have been assembled in to a eukary-otic genome, the possibility of bacterial contamination inthe P. parvum, P. lutheri and E. huxleyi ESTs cannot be for-mally ruled out, but is very unlikely for several reasonsrelating to how the sequences were generated (see Meth-ods for details).The two diatom genomes encode another RPE thatbranches with the cytosolic homologues of other eukary-otes, but there is no evidence in any of the five chromal-veolates for a cyanobacterium-derived, red algal-type RPEgene that would be expected to operate in the plastids ofthese organisms. This search included the completegenome sequence of T. pseudonana and the nearly com-plete sequence of the P. tricornutum genome, as well as theextensive EST databases that have been generated for thethree haptophytes. It would appear that the ancestral,cyanobacterium-derived RPE is absent from all of thesechromalveolates. A second, bacterial-type RPE was foundin I. galbana (GenBank accession EC141129), but it wasnot found to be related to plastid or other eukaryotichomologues.The origin of the γ-proteobacteria-like RPE gene in eukary-otes and its distribution must be considered separately.The origin of the RPE-γ gene is addressed by the strongsupport for the eukaryotic genes being sister to a specificPage 3 of 13(page number not for citation purposes)ysis confirmed that RPE genes from the haptophytes Emil-iania huxleyi, Prymnesium parvum and Pavlova lutheri are alland taxonomically narrow group of bacteria, the pseu-domonads. This result argues for a relatively recent originBMC Evolutionary Biology 2007, 7:89 http://www.biomedcentral.com/1471-2148/7/89Page 4 of 13(page number not for citation purposes)Bayesian phylogenetic tree of ribulose-5-phosphate-3 epimerase (RPE)Figure 1Bayesian phylogenetic tree of ribulose-5-phosphate-3 epimerase (RPE). The tree was inferred from 183 amino acid characters with branch lengths estimated using PROML. Bootstrap values > 50% are shown. Values shown above a node corre-spond to PHYML bootstrap support, those below a node correspond to WEIGHBOR support. Eukaryotic sequences are enclosed in boxes where blue corresponds to the major clade of cytosolic proteins, green corresponds to plastid-targeted pro-teins, and red corresponds to bacterium-derived genes. Filled circles adjacent to taxon names indicate that a complete genome is available from this organism.0.1Treponema pallidumStreptococcus pyogenesLactococcus lactisRhodobacter capsulatusListeria innocuaCampylobacter jejuniRhodospirillum rubrumNeisseria meningitidisXanthobacter autotrophicusXylella fastidiosaMarinobacter aquaeoleiPseudomonas fluorescensPseudomonas syringidaePseudomonas putidaPseudomonas aeruginosaPavlova lutheriPrymnesium parvumEmiliana huxleyiBigelowiella natansThalassiosira pseudonanaPhaeodactylum tricornutumThalassiosira pseudonanaVibrio choleraeMannheimia succiniproducensHaemophilus influenzae RdPasteurella multocidaSerratia marcescensYersinia pestisBuchnera aphidicolaEscherichia coliSalmonella typhimuriumBacillus subtilisCyanidioschyzon merolaeSolanum tuberosumOryza sativaSpinacia oleraceaArabidopsis thalianaOstreococcus tauriChlamydomonas reinhardtiiProchlorococcus marinusSynechocystis PCC 6803Caulobacter crescentusAgrobacterium tumefaciensSinorhizobium melilotiMethanococcus jannaschiiThermotoga maritimaMycobacterium tuberculosisStreptomyces coelicolorThermobifida fuscaGiardia lambliaEntamoeba histolyticaTrichomonas vaginalisChlamydomonas reinhardtiiSchizosaccharomyces pombeCyanidioschyzon merolaeOstreococcus tauri fusionPlasmodium falciparumTrypanosoma bruceiToxoplasma gondiiSaccharomyces cerevisiaeLeishmania majorThalassiosira pseudonana X5P fusionPhaeodactylum tricornutum X5P fusionPhaeodactylum tricornutumTetrahymena theromophilaDrosophila melanogasterHomo sapiensDictyostelium discoideumArabidopsis thalianaGossypium hirsutumLycopersicon esculentumOryza sativaOstreococcus tauri759987615391 9080941005995828294998473681001005271971007362-9253-575190 981006087648660888110082535210062929870100100Plastid-targetedChromistans andChlorarachniophytesCyanobacteria andPlastid-targetedRed algae, Green algaeand PlantsCytosolic Eukaryoticand Xylulose-5-phosphatefusion Proteins83855055991001001004168BMC Evolutionary Biology 2007, 7:89 http://www.biomedcentral.com/1471-2148/7/89by lateral gene transfer from the pseudomonads toeukaryotes. The distribution of this gene, however, ismore complicated because the eukaryotes that containRPE-γ are not all closely related. Evidence exists that hap-tophytes and diatoms are both members of the super-group Chromalveolata, but chlorarachniophytes belongto a completely different supergroup, Rhizaria [35].Explaining this complex distribution by paralogy wouldbe relatively simple if the enzyme were cytosolic: onewould then propose that rhizaria and chromalveolateswere specifically related and that many of the constituentlineages of these two supergroups had lost the enzyme(considering only complete or nearly complete genomes,this would include apicomplexa, Perkinsus and ciliates).However, the fact that RPE-γ is targeted to the plastid sub-stantially complicates this interpretation because the B.natans plastid is derived from a green alga whereas thechromalveolate plastids are derived from a red alga. Theplastid-targeted RPEs of both green algae and red algae arecyanobacterial (Figure 1) and in neither group has theproteobacterial RPE-γ type been found in available com-plete genome sequences. Accordingly, if the plastid RPEsin chromalveolates and chlorarachniophytes were derivedfrom a common ancestor, the proteobacterial type wouldhave had to coexist with the cyanobacterial type in anancestor of red and green algae, with subsequent diversifi-cation involving a complex pattern of reciprocal losses notonly in rhizarians and chromalveolates, but also in redand green algae. Alternatively, if the enzyme was acytosolic RPE in a hypothetical common ancestor of chlo-rarachniophytes, haptophytes, and diatoms, then itwould have had to have taken over plastid function twiceindependently, in addition to reciprocal losses. Both ofthese explanations are very complicated and invokehigher-order relationships among eukaryotes that are notknown. In addition, the close specific relationshipbetween the pseudomonad and the eukaryotic RPE-γsequences is more suggestive of a recent origin by lateraltransfer than an ancient origin. Taken together, the sim-plest explanation for the current distribution is that RPE-γwas transferred from a pseudomonad to an undefinedeukaryotic lineage and then transferred between twoeukaryotic lineages (the direction cannot be inferred fromthe phylogeny because the topology of rhizarian andchromalveolate RPEs is not resolved).In the course of this study, we also observed an interestinggene fusion event involving the cytosolic RPE of the dia-toms T. pseudonana and P. tricornutum and the prasino-phyte green alga Ostreococcus tauri. In these threeorganisms, the cytosolic-type RPE is found as a xyluloki-nase-RPE fusion-protein. In P. tricornutum and O. tauri thetwo proteins are part of an uninterrupted ORF, whereas inframeshift is most likely due to the presence of an unan-notated intron. Xylulokinase catalyses the phosphoryla-tion of xylulose to xylulose-5-phosphate for entry into thepentose phosphate pathway. This reaction occurs imme-diately prior to the ribulose-5-phosphate epimerase reac-tion, raising the intriguing possibility that the fusionprotein may catalyze both reactions. Interestingly, the N-termini of the P. tricornutum and T. pseudonana RPE pro-teins also encode a predicted signal peptide, suggestingthat this fusion protein may be targeted to the plastid. P.tricornutum and O. tauri also encode canonical cytosolicRPEs related to other cytosolic isoforms. Gene fusionevents involving carbon metabolic enzymes have beenreported from other algae, notably those involvingGAPDH and enolase in dinoflagellates [36], and betweentriose phosphate isomerase and GAPDH in the mitochon-dria of heterokonts [37]. Whether the fusions had a com-mon origin or arose independently is not clear: O. tauri isa green alga whereas diatoms have plastids derived fromred algae. The fusion genes are not demonstrably related(Figure 1), suggesting perhaps that the fusion arose twiceindependently.Glyceraldehyde-3-Phosphate DehydrogenaseGlyceraldehyde-3-phosphate dehydrogenase (GAPDH)catalyzes the bi-directional conversion of glyceraldehyde-3-phosphate to 3-phosphoglycerate in both glycolysis andthe Calvin cycle. GAPDH has been extensively sampledfrom eukaryotes and bacteria, revealing many cases of lat-eral transfer and paralogy. Bacteria and plastids canoni-cally use a class of enzyme known as GapA/B, whereas thetypical eukaryotic cytosolic GAPDH is called GapC. Theancient evolution of this family is complex as several geneduplications have taken place and GapC as a whole hasbeen suggested to be derived by lateral transfer [38,39].These ancient events remain uncertain, but several eukary-otic groups also have genes that are clearly from the GapA/B class: these are not derived from the plastid endosymbi-ont but are considered to have originated by relativelyrecent lateral transfer [15,21,31,40]. One of these cases isthe divergent class of GapA/B (here designated GapA/B*)previously found only in diplonemids, heterotrophic rel-atives of kinetoplastids [33]. Once again, however, withincreased sampling we find the same class of GAPDH inhaptophytes (I. galbana) and diatoms (T. pseudonana andP. tricornutum) (Figure 2). EST data from the haptophyteE. huxleyi also include three transcripts encoding a similarGapA/B* gene (GenBank accessions CX777351,CX776621, and EG034112), but these sequences are tooshort to be included in the phylogeny. As noted above,heterokonts and haptophytes are thought to be membersof the same supergroup, chromalveolates, but they are notclosely related to diplonemids, which are members of thePage 5 of 13(page number not for citation purposes)the T. pseudonana genome, the xylulokinase and RPE arein different reading frames, although this apparentexcavates. Nevertheless, the GapA/B* sequences from allthree groups form a strongly supported clade, which inBMC Evolutionary Biology 2007, 7:89 http://www.biomedcentral.com/1471-2148/7/89turn branches within a eubacterial group consisting ofproteobacteria and cyanobacteria, as found previously fordiplonemids alone [33]. The chromalveolate GapA/B*sequences are paraphyletic in this analysis, as the diplone-mids share a robust sister relationship with I. galbana tothe exclusion of diatoms. The I. galbana GAPDH is not fulllength, but the N-termini of the two diatom sequences areof comparable length to GapA/B in bacteria and do notencode a predicted signal peptide, so these proteins arelikely cytosolic.The distribution of GapA/B* bears many similarities tothat of RPE-γ : the gene is found in distantly related clades(chromalveolates and diplonemids), each of which hasrelatives that lack it. Also in common with RPE-γ, the closerelationship between the eukaryotic GapA/B* genes andtheir bacterial sisters, together with their distant relation-ship to canonical eukaryotic GapC genes, indicates thatthe eukaryotic homologues originated by lateral transferfrom a bacterium. However, this conclusion does notaddress present-day distribution of GapA/B* in eukaryo-tes. If there was a single transfer to the ancestor of chroma-lveolates and diplonemids, then the gene must have beenlost in many of the relatives of these two groups (consid-ering only taxa where complete or nearly completegenomes are known, this includes apicomplexa, Perkinsus,ciliates, trypanosomatids, and perhaps Giardia and Tri-chomonas). However, given the large number of lateraltransfer events already known to have involved GAPDH,including one between two eukaryotes [21], the currentnarrow range of taxa possessing the GapA/B* gene sug-gests instead that it was transferred from a bacterium rela-tively recently and subsequently spread to othereukaryotes by eukaryote-to-eukaryote transfers. Thebranching order between the eukaryotic GapA/B* genes iswell supported, and at face value this result suggests thatthere were either multiple transfers between eukaryotes orthat the gene originated in chromalveolates and was trans-ferred to diplonemids.TransketolaseTransketolase (TK; glycoaldehydetransferase) catalyzesthe reversible transfer of a C2 unit between two 5-carbonsugars, producing a 3-carbon sugar and a 7-carbon sugar,or between a 4-carbon sugar and a 5-carbon sugar, pro-ducing a 6-carbon sugar and a 3-carbon sugar. TK func-tions in the cytosol of non-photosynthetic eukaryotes,where it is involved in reactions of the classic pentose-phosphate pathway. It is also found in the plastids ofalgae and plants, where it functions in the Calvin cycle aswell as the reversible branch of the pentose-phosphatepathway. In at least some plants, an additional cytosolicisoform of TK, related to the plastid form, exists [41].TK exhibits a complex phylogenetic distribution acrossdifferent groups of eukaryotes. Metazoa and ciliates havea highly divergent form of TK characterized by many gapsand deletions; we have not included these sequences inour data set because they are difficult to align with otherTKs. Most eukaryotic cytosolic TKs belong to a more con-served group that is widespread and constitutes a singlewell-supported clade (Figure 3). Similarly, most plastid-targeted TKs have robust cyanobacterial affinities, asexpected, although one with an unusually close relation-ship to the cytosolic clade described above. The phylog-eny within the plastid-targeted clade is generally not wellresolved; however, one interesting exception are the plas-tid-targeted TKs from the euglenid Euglena gracilis and thedinoflagellate Heterocapsa triquetra, which form a verystrongly supported branch in all analyses and which fall atthe base of the plastid-targeted clade with strong supportin analyses of the full protein (Figure 3, 'Plastid-targeted').Euglenids and dinoflagellates are not closely related andtheir plastids are derived from green and red algae, respec-tively. That these two sequences branch together is there-fore unusual and reminiscent of a proposed transfer of aplastid-targeted GAPDH between euglenids and dinoflag-ellates [36].The most interesting clade, however, comprises severaleukaryotic TKs that are unrelated to either the majorcytosolic clade or plastid-targeted clade, but instead areclosely related to the bacterial group Chlamydiales (Figure3). This clade (which we term TK-Ch) includes not two,but several distantly related groups of eukaryotes. Thechromalveolates are most heavily represented, includingthe diatoms T. pseudonana and P. tricornutum, the hapto-phyte I. galbana, and the dinoflagellate with a haptophyteplastid, Karlodinium micrum. Also included are the amoe-bozoans Dictyostelium discoideum and Physarum polycepha-lum, the excavate E. gracilis, and the rhizarian B. natans, anassemblage that accounts for four of the five eukaryoticsupergroups (the exception being plants). In addition,more highly truncated ESTs were found from several otherchromalveolates, specifically the haptophyte P. parvumand the cryptomonad Guillardia theta, as well as ESTs rep-resenting another copy of the gene from K. micrum. Thesesequences were too short to include in the analysis shownin Figure 3, but in phylogenies restricted to the 3' end ofthe gene they consistently branch within this clade withhigh support (not shown). Amoebozoans and E. gracilisconsistently form a strongly supported group, as do thechromalveolates and B. natans. Within the latter clade, I.galbana occupies a basal position in analyses based on thenearly complete amino acid sequence, and this result isstrongly supported both by bootstrap values and by thepresence of a unique conserved 4-amino acid insertion inPage 6 of 13(page number not for citation purposes)the K. micrum, B. natans and diatom TK-Ch sequences thatis absent from I. galbana TK-Ch (Figure 4).BMC Evolutionary Biology 2007, 7:89 http://www.biomedcentral.com/1471-2148/7/89Page 7 of 13(page number not for citation purposes)Bayesian phylogenetic tree of glyceraldehyde-3-phosphate dehydrogenase (GAPDH)Figure 2Bayesian phylogenetic tree of glyceraldehyde-3-phosphate dehydrogenase (GAPDH). The tree was inferred from 278 amino acid characters with branch lengths estimated using PROML. Bootstrap values > 50% are shown. Values shown above a node correspond to PHYML bootstrap support, those below a node correspond to WEIGHBOR support. Eukaryotic sequences are enclosed in boxes where blue corresponds to the major clade of cytosolic proteins, green corresponds to plas-tid-targeted proteins, and red corresponds to bacterium-derived genes (the B. natans plastid targeted GAPDH is also bacte-rium-derived and is coloured red). Filled circles adjacent to taxon names indicate that a complete genome is available for this organism.0.1Streptomyces coelicolorCorynebacterium glutamicumMycobacterium lepraeChlorobium tepidum TLSCytophaga hutchinsoniiClostridium pasteurianumEscherichia coliThermus aquaticusThermotoga maritimaAquifex aeolicusChlorobium phaeobacteroidesTrichomonas vaginalisPhaeosphaeria avenariaPhaeodactylum tricornutumIsochrysis galbanaPhaeodactylum tricornutum TPI/GAPDH fusion proteinDiplonema ATCC 50224Zea maysGlycine maxPichia pastorisCandida glabrataHeterocapsa triquetraGuillardia thetaPhaeodactylum tricornutum Odontella sinensisIsochrysis galbanaPavlova lutheriFlavobacterium MED217Vibrio choleraeSalmonella typhimuriumDesulfovibrio vulgarisSymbiobacterium thermophilumYersinia frederikseniiVibrio choleraeShewanella putrefaciensAzotobacter vinelandiiRhodobacter capsulatusXanthobacter autotrophicusProchlorococcus marinus str.Synechococcus 5701Isochrysis galbanaDiplonema ATCC 50225Diplonema ATCC 50224Rhynchopus ATCC 50231Thalassiosira pseudonanaPhaeodactylum tricornumSynechococcus elongatusNostoc 7120Synechococcus JA2Chloroflexus aurianticusBigelowiella natansShewanella oneidensisBrevibacterium linensArthrobacter FB24Corynebacterium efficiensAzotobacter vinelandiiPseudomonas syringidaePelotomaculum thermopropinicumSynechocccus sp. WH8102Gloeobacter violaceusSynechococcus elongatusPorphyra yezoensisCladophora rupestrisMesostigma virideMesostigma virideChara vulgarisArabidopsis thalianaOrzya sativaCyanophora paradoxaChlamydomonas reinhardtiiChlorella sp. JPScenedesmus vacuolatusChara vulgarisArabidopsis thalianaArabidopsis thalianaOryza sativaEuglena gracilisPyrocystis lunulaBacillus cereusClostridium tetaniMyxococcus xanthusIdiomarina balticaEscherichia coliPseudomonas aeruginosaNeisseria gonorrhoeaeBurkholderia vietnamiensisBrucella suisSinorhizobium melilotiGeobacter sulfurreducens74858310094547910010098507680501009596756398539471819210010097691001008010010099781008588100656958675253100100100100929710010097701009670829810062100-1008237689099435310059835860819579929758Plastid-TargetedGAP A/BCytosolicGAP C Diplonemidand ChromistGAP A/B939169884757534386918878Plastid-TargetedBMC Evolutionary Biology 2007, 7:89 http://www.biomedcentral.com/1471-2148/7/89Page 8 of 13(page number not for citation purposes)Bayesian phylogenetic tree of transketolase (TK)Figure 3Bayesian phylogenetic tree of transketolase (TK). The tree was inferred from 473 amino acid characters with branch lengths estimated using PROML. Bootstrap values > 50% are shown. Values shown above a node correspond to PHYML boot-strap support, those below a node correspond to WEIGHBOR support. Eukaryotic sequences are enclosed in boxes where blue corresponds to the major clade of cytosolic proteins, green corresponds to plastid-targeted proteins, and red corre-sponds to bacterium-derived genes. Filled circles adjacent to taxon names indicate that a complete genome is available from this organism. Multiple isoforms of TK are present in the diatom genomes, both a Chlamydia type TK as well as a eukaryotic form of this enzyme are present. Two types of TK are present in Euglena as well, a form related to the plastid TK of other eukaryotes as well as a Chlamydia type.Cytosolic 0.1Streptococcus pneumoniaeOceanobacillus iheyensisBacillus subtilisListeria monocytogenesLactococcus lactisStreptococcus pyogenesDeinococcus radioduransKarlodinium micrumEuglena gracilisDictyostelium discoideumPhysarum polycephalumIsochrysis galbanaBigelowiella natansThalassiosira pseudonanaPhaeodactylum tricornutumChlamydophila pneumoniaeChlamydia muridarumChlamydia trachomatisCandidatus Kuenenia stuttgartiensisAcidobacteria bacteriumGluconobacter oxydansMyxococcus xanthusNitrosospira multiformisKarlodinium micrumHartmannella vermiformisBurkholderia xenovoransBlastopirellula marinaAquifex aeolicusEntamoeba histolyticaSalinibacter ruberClostridium acetyobutylicumPolaromonas napthalenivoransBurkholderia cenocepaciaChlorobium tepidum TLSProsthecochlorosis aestuariiXanthobacter flavusNeisseria meningitidisPseudomonas aeruginosaVibrio choleraeBuchnera aphidicola APSEscherichia coliAgrobacterium tumefaciensSinorhizobium melilotiLeishmania mexicana mexicanaThalassiosira pseudonanaPhaeodactylum tricornutumPerkinsus marinusPlasmodium falciparumCandida albicansSaccharomyces cerevisiaeSaccharomyces cerevisiaeNeurospora crassaSchizosaccharomyces pombeUstilago maydisTrichomonas vaginalisGiardia lambliaHeterocapsa triquetraEuglena gracilisCyanidioschyzon merolaePorphyra yezoensisCyanophora paradoxaProchlorococcus marinusSynechococcus WH 8102Chlamydomonas reinhardtiiCraterostigma plantagineum (cytosolic1)Craterostigma plantagineum (cytosolic2)Spinacia oleraceaArabidopsis thaliana Oryza sativaPinus taedaCapsicum annuum8993 586671-1001001001001001009995989890100 91948583 74- 10010098100100997670999910010010099 5277 8287100100999998989790 65935059100998679999910010098979885100100607810010050-585479665352525260100100-195675Plastid-TargetedTK-Ch-BMC Evolutionary Biology 2007, 7:89 http://www.biomedcentral.com/1471-2148/7/89We cannot determine whether the TK-Ch proteins of B.natans and K. micrum are cytosolic or plastid-targeted asthey are N-terminally truncated. However, the E. gracilisand two diatom sequences are comparable in length attheir N-terminus to bacterial TKs and cytosolic TKs ofother eukaryotes, and neither is predicted to encode a sig-nal peptide. The transketolase of Isochrysis galbana encodesa long n-terminal leader predicted to encode a signal pep-tide, suggesting that this transketolase may be plastid-tar-geted in I. galbana. Altogether, the evidence suggests thatthese TK-Ch proteins are plastid-targeted in some organ-isms such as the haptophyte I. galbana, but cytosolic inother photosynthetic organisms such as the diatoms P. tri-cornutum and T. pseudonana and the euglenid E. gracilis.In addition to this group, there are also a few eukaryoticsequences that fall outside any of the TK groups character-ized so far. In particular, the amoebozoan Hartmannellavermiformis and the dinoflagellate K. micrum both haveEST-predicted TKs related to a proteobacteria-planctomyc-etes-CFB bacterial clade (a relationship that is further sup-ported by an insertion: Figure 4), and that constitute aweakly supported clade (Figure 3). The Entamoeba histolyt-Distinguishing between lateral transfer and paralogy inthe TK case is a more complex problem than in the RPEand GAPDH situations considered above, because thediversity of eukaryotes with TK-Ch is much greater. At thesame time, this breadth makes the TK case potentiallymuch more interesting, and significant. This wide distri-bution makes a stronger case for paralogy – that this generepresents an ancient eukaryotic paralogue present in thelast common ancestor of these groups. This argumentimplies that the gene was lost in close relatives of extantorganisms that contain TK-Ch, which is a substantialqualification because a large number of losses would berequired: considering only groups where genomesequences are complete or nearly so, this would includeanimals, fungi, Entamoeba, apicomplexa, ciliates, Perkin-sus, kineteoplastids, Giardia and Trichomonas. The specificrelationship between the eukaryotic TK-Ch and Chlamy-diales TK sequences is also difficult to reconcile with suchan ancient origin, suggesting instead that the correspond-ing gene may have originated more recently by lateralgene transfer from an ancestor of the Chlamydiales group.If, in fact, the TK-Ch gene was transferred to an ancientTransketolase alignment flanking a 4-amino acid insertion present in diatom, dinoflagellate and chlorarachniophyte TK-Ch geneFigure 4Transketolase alignment flanking a 4-amino acid insertion present in diatom, dinoflagellate and chlorarachnio-phyte TK-Ch genes. A nearby 11- to 15-amino acid insertion characterizes a mixed group of CFB bacteria and proteobacte-ria as well as the amoebozoan Hartmannella vermiformis and the dinoflagellate Karlodinium micrum. Eukaryotic TKs putatively derived from lateral transfer events are surrounded by black squares (TK-Ch being the lower box).Page 9 of 13(page number not for citation purposes)ica TK, on the other hand, branches outside any eukaryo-tic clade but does not show an affinity to any other group.ancestor of most or all eukaryotes, no eukaryote-to-eukaryote transfers need be invoked. However, this strictBMC Evolutionary Biology 2007, 7:89 http://www.biomedcentral.com/1471-2148/7/89interpretation runs into difficulties when the phylogenywithin the TK-Ch clade is considered. If no between-eukaryote transfer had occurred, then the supergroupsshould be monophyletic or at least unresolved. This is notthe case, since the rhizarian B. natans branches within thechromalveolates with strong support (Figure 3), and itsrelationship to K. micrum and the diatoms is further sup-ported by the shared insertion (Figure 4). To explain theseobservations without lateral transfer, it would be neces-sary to propose additional cases of paralogy arising sincethe gene originated in eukaryotes. The alternative explana-tion is that the B. natans gene is derived from lateral trans-fer from another eukaryote, which is consistent with theobservation that nearly a dozen other B. natans genes havebeen derived from other phototrophs by lateral transfer[15]. By extension, there is no reason to exclude the pos-sibility that there have been other transfers betweeneukaryotes (the present distribution could be achievedwith as few as three transfers), which would also explainthe punctate distribution without having to argue for lossin close relatives. This interpretation provides the simplestexplanation of the current data. If eukaryote-to-eukaryotegene transfer is the underlying mechanism by which TK-Ch came to exhibit its present distribution, then theresulting pattern is second in complexity only to that ofthe previously described case of a novel elongation factor-like GTPase, EFL [24].ConclusionWe describe three examples where the phylogeny of a car-bon metabolic enzyme at first appeared to indicate a sim-ple case of bacterium-to-eukaryote lateral gene transfer,but where greater sampling has shown the situation to beconsiderably more complex. In all three cases othereukaryotes with the same bacterial gene have been discov-ered, and in each case these eukaryotes are only distantlyrelated to one another at the organismal level. The firstimportant point to note here is that these observationsemerged only with an increased sampling of eukaryoticmolecular diversity, implying that the distributionsreported here are likely to change further as taxon sam-pling becomes even more comprehensive. However, thedistribution can only change in one direction – towardgreater complexity. It is possible that these genes will ulti-mately be found in such a large sample of eukaryotes wewill ultimately conclude that they represent ancient para-logues whose distribution is mostly due to gene loss; onthe other hand, considering the high frequency of theabsence of these particular sequences in current data, itseems more likely they will continue to be rare and toexhibit a punctate distribution. In the case of the EFLGTPase further sampling has revealed additional organ-isms that possess the corresponding gene [42,43]; never-A second noteworthy point is that we often contrast lateralgene transfer and lineage sorting as two contradictory pos-sibilities, whereas they work concurrently. If a new genearrives in a lineage by lateral transfer, some descendentsmay keep it and others may lose it, resulting in an appar-ently complex distribution. Distinguishing this patternfrom that resulting from serial transfers is difficult andmay well be impossible in certain circumstances; how-ever, we can still weigh the observations in favour of onepossibility over the other. In particular, lineage sorting isprobably more likely in cases where a complex distribu-tion of presence and absence is found in closely relatedspecies, whereas serial transfers are more likely when thetime frame is significantly longer. In the three casesdescribed here, the time frame is very long indeed, andthere is evidence from the internal phylogenies for eukary-ote-eukaryote transfer. The significance of this findingextends beyond these three genes, to the process of trans-fer between eukaryotes in general. Currently, few cases ofsuch transfers are known because they are difficult todetect in the absence of better sampling than is currentlyavailable for most genes. The results reported here use anunusual feature of the gene, its origin from bacteria, as aflag to draw attention to subsequent transfers; however,there is nothing to indicate that the same process couldnot be occurring in many other genes where it is not asevident because of the lack of such flags.MethodsCharacterization of new sequencesUsing TBestDB [44], clones corresponding to TK wereidentified in EST projects from Euglena gracilis (12 ESTs forthe chloroplast-targeted isoform and 3 ESTs for the TH-Chisoform), Bigelowiella natans (3 ESTs),Hartmannella vermi-formis (9 ESTs), Karlodinium micrum (4 ESTs for the TK-Chisoform and 1 EST for the CFB group isoform), Isochrysisgalbana (7 ESTs) and Physarum polycephalum (18 ESTs).Clones corresponding to RPE were identified from Pavlovalutherii (3 ESTs). ESTs were re-sequenced to obtain full-length assemblies wherever possible. Putative full-lengthtranscripts were assembled in this way for TK from E. gra-cilis, P. polycephalum and H. vermiformis. 5'-Truncatedassemblies were obtained from K. micrum and I. galbana.The 5' end of B. natans TK was obtained through PCRamplification using degenerate TK primers (CGCGACTA-CAGGCCCNYTNGGNCARGG and GCGCAAG-GCGAACWSNGGNCAYCCNGG) and specific primersbased on the EST sequences (CTCTCCAACACCGATA-GAATCATGAGTC and GCCTTGTACCGGGTGATGA-CATCCTCAG). A PCR product from I. galbana withsimilarity to bacterial GAPDHs was obtained throughPCR using degenerate primers (CCAAGGTCGGNATH-AAYGGNTTY and CGAGTAGCCCCAYT CRTTRTCRT-Page 10 of 13(page number not for citation purposes)theless, EFL remains far less common than its counterpart,EF-1α.ACCA). All PCR products were cloned using the TOPO TAvector (Invitrogen) and multiple copies were sequencedBMC Evolutionary Biology 2007, 7:89 http://www.biomedcentral.com/1471-2148/7/89on both strands. New sequences were deposited in Gen-Bank as accessions EF216678-EF216685, EF221881 andEF375722. RPE genes were also identified in EST projectsfrom Emiliania huxleyi (7 ESTs) and Prymnesium parvum (1EST). Homologues of all three genes were also identifiedin the completed genomes of Thalassiosira pseudonana [45]and Phaeodactylum tricornutum [46].Sequence analysesFor organisms possessing a complex plastid, the probableplastid localization of all full length sequences was evalu-ated using SIGNALP v. 3.0 [47] and leader sequences weremanually scanned for characteristics expected of transitpeptides for the group in question.New sequences were aligned to homologues from publicdatabases using CLUSTAL X, and manually edited usingMacClade 4.07. Publicly available sequences used inalignments were downloaded from nr GenBank, ESTdb,or from complete eukaryotic genome databases. Allsequences found to be closely related to the eukaryoticgenes in question were included, along with a representa-tive selection of other genes. Genes that were closelyrelated to some other gene but not the eukaryotic genes inquestion were generally excluded, as were some genes thatwere highly divergent (e.g. the animal TK). Positions thatwere not clearly homologous were excluded, resulting ina full TK alignment with 473 characters, and a reducedalignment with 212 characters that was also used toinclude those sequences with missing 5' sequence. TheRPE alignment consisted of 183 alignable characters andthe GAPDH alignment comprised 278 alignable charac-ters. Alignments of all three genes are available uponrequest from PJK.While the possibility that several of these sequences areartifacts from a bacterial genome exists, several lines ofevidence argue against this. ESTs encoding the proteobac-terial-like RPE from haptophytes branch together(weakly) to the exclusion of all bacterial sequences. Thissuggests that if these sequences arose through bacterialcontamination, the source of the contamination musthave been similar in each library. Also, multiple ESTsencoding RPE in P. lutheri and E. huxleyi are present inthese two libraries (two and seven respectively), indicat-ing that if this RPE is due to a bacterial contaminant, thecontamination must be highly represented in these librar-ies. Lastly, the RPE sequence of E. huxleyi encodes an N-terminal extension with a predicted signal peptide, a hall-mark of plastid-targeted proteins in eukaryotes with com-plex plastids. Similarly, a contamination artefact cannotbe precluded for the K. micrum and H. vermiformis TKs thatbranch within a clade consisting of CFB bacteria, proteo-contamination arose from a similar source since the twosequences branch together with moderate support in ouranalysis. The K. micrum TK is encoded by a single EST,though the H. vermiformis TK is encoded by nine ESTs sug-gesting that if this H. vermiformis sequence is a contami-nant, it is highly represented among the H. vermiformisESTs.Phylogenetic analyses were carried out using maximumlikelihood, distance and Bayesian methods. Maximumlikelihood phylogenies were performed using PHYML2.4.4 [48] with the WAG substitution matrix and site ratedistribution modeled on a discrete gamma distributionwith 4 rate categories and one category of invariable sites.The estimated alpha parameters were 1.154, 1.156, and1.419 and the estimated proportions of invariable siteswere 0.067, 0.050, and 0.101 for RPE, GAPDH, and TK,respectively. Bayesian analysis of trees was performedusing MrBayes 3.0b4 [49] run using the WAG substitutionmatrix and a gamma distribution with 4 rate categoriesand one category of invariable for 1,000,000 generationswith sampling every 1,000 generations. Distance analyseswere performed using TREE-PUZZLE 5.2 [50] with fourrate categories and 1 category of invariable sites. Alphaparameters inferred by TREE-PUZZLE were 0.89, 1.04 and1.21 and the estimated proportions of invariable siteswere 0.04, 0.05 and 0.10 for RPE, GAPDH and TK respec-tively. WEIGHBOR 1.0.1a was used to reconstruct dis-tance trees Bootstrapped distance matrices were generatedusing PUZZLEBOOT [51] with alpha parameter and pro-portion of invariable sites estimated using TREE-PUZZLE5.2.Authors' contributionsMBR identified and characterized new genes, performedphylogenetic analyses and drafted the manuscript. RFWand JTH characterized new genes. DGD and MWG con-tributed to interpretation and helped to draft the manu-script. PJK conceived of the study, and participated in itsdesign and coordination and helped to draft the manu-script. All authors read and approved the final manu-script.AcknowledgementsThis work was supported by a grant from the Natural Sciences and Engi-neering Research Council of Canada to PJK (227301). EST sequencing from E. gracilis, H. triquetra, H. vermiformis, I. galbana, K. micrum, P. lutheri and P. polycephalum was supported by the Protist EST Program of Genome Can-ada/Genome Atlantic, with additional funding from the Atlantic Innovation Fund. PJK and MWG are Fellows of the Canadian Institute for Advanced Research, MWG is Canada Research Chair in Genomics and Genome Evo-lution at Dalhousie University, and PJK is a Senior Scholar of the Michael Smith Foundation for Health Research.Page 11 of 13(page number not for citation purposes)bacteria and planctomycetes. Like the RPEs of the hapto-phytes, we would have to conclude that theReferences1. Doolittle WF: Lateral genomics.  Trends Cell Biol 1999, 9:M5-8.BMC Evolutionary Biology 2007, 7:89 http://www.biomedcentral.com/1471-2148/7/892. Eisen JA: Horizontal gene transfer among microbial genomes:new insights from complete genome analysis.  Curr Opin GenetDev 2000, 10:606-611.3. Coleman ML, Sullivan MB, Martiny AC, Steglich C, Barry K, Delong EF,Chisholm SW: Genomic islands and the ecology and evolutionof Prochlorococcus.  Science 2006, 311:1768-70.4. Koonin EV, Makarova KS, Aravind L: Horizontal gene transfer inprokaryotes: quantification and classification.  Annu Rev Micro-biol 2001, 55:709-742.5. Ochman H, Lawrence JG, Groisman EA: Lateral gene transfer andthe nature of bacterial innovation.  Nature 2000, 405:299-304.6. Beiko RG, Harlow TJ, Ragan MA: Highways of gene sharing inprokaryotes.  Proc Natl Acad Sci USA 2005, 102:14332-14337.7. Ragan MA, Harlow TJ, Beiko RG: Do different surrogate meth-ods detect lateral genetic transfer events of different relativeages?  Trends Microbiol 2006, 14:4-8.8. Lawrence JG, Hendrickson H: Lateral gene transfer: when willadolescence end?  Mol Microbiol 2003, 50:739-749.9. Kurland CG: What tangled web: barriers to rampant horizon-tal gene transfer.  Bioessays 2005, 27:741-747.10. Hao W, Golding GB: The fate of laterally transferred genes: lifein the fast lane to adaptation or death.  Genome Res 2006,16:636-643.11. Stanhope MJ, Lupas A, Italia MJ, Koretke KK, Volker C, Brown JR:Phylogenetic analyses do not support horizontal gene trans-fers from bacteria to vertebrates.  Nature 2001, 411:940-944.12. Salzberg SL, White O, Peterson J, Eisen JA: Microbial genes in thehuman genome: lateral transfer or gene loss?  Science 2001,292:1903-1906.13. Boucher Y, Doolittle WF: The role of lateral gene transfer in theevolution of isoprenoid biosynthesis pathways.  Mol Microbiol2000, 37:703-16.14. Nixon JE, Wang A, Field J, Morrison HG, McArthur AG, Sogin ML,Loftus BJ, Samuelson J: Evidence for lateral transfer of genesencoding ferredoxins, nitroreductases, NADH oxidase, andalcohol dehydrogenase 3 from anaerobic prokaryotes to Gia-rdia lamblia and Entamoeba histolytica.  Eukaryot Cell 2002,1:181-90.15. Archibald JM, Rogers MB, Toop M, Ishida K, Keeling PJ: Lateral genetransfer and the evolution of plastid-targeted proteins in thesecondary plastid-containing alga Bigelowiella natans.  ProcNatl Acad Sci USA 2003, 100:7678-7683.16. Andersson JO, Sjogren AM, Davis LA, Embley TM, Roger AJ: Phylo-genetic analyses of diplomonad genes reveal frequent lateralgene transfers affecting eukaryotes.  Curr Biol 2003, 13:94-104.17. Waller RF, Slamovits CH, Keeling PJ: Lateral gene transfer of amultigene region from cyanobacteria to dinoflagellatesresulting in a novel plastid-targeted fusion protein.  Mol BiolEvol 2006, 23:1437-1443.18. Keeling PJ, Palmer JD: Lateral transfer at the gene and subgeniclevels in the evolution of eukaryotic enolase.  Proc Natl Acad SciUSA 2001, 98:10745-10750.19. Andersson JO: Lateral gene transfer in eukaryotes.  Cell Mol LifeSci 2005, 62:1182-97.20. Bergthorsson U, Adams KL, Thomason B, Palmer JD: Widespreadhorizontal transfer of mitochondrial genes in floweringplants.  Nature 2003, 424:197-201.21. Takishita K, Ishida K, Maruyama T: An enigmatic GAPDH gene inthe symbiotic dinoflagellate genus Symbiodinium and itsrelated species (the order Suessiales): possible lateral genetransfer between two eukaryotic algae, dinoflagellate andeuglenophyte.  Protist 2003, 154:443-54.22. Ragan MA: On surrogate methods for detecting lateral genetransfer.  FEMS Microbiol Lett 2001, 201:187-91.23. Roger AJ, Hug LA: The origin and diversification of eukaryotes:problems with molecular phylogenetics and molecular clockestimation.  Philos Trans R Soc Lond B Biol Sci 2006, 361:1039-54.24. Keeling PJ, Inagaki Y: A class of eukaryotic GTPase with a punc-tate distribution suggesting multiple functional replace-ments of translation elongation factor 1alpha.  Proc Natl AcadSci USA 2004, 101:15380-15385.25. Andersson JO, Hirt RP, Foster PG, Roger AJ: Evolution of fourgene families with patchy phylogenetic distributions: influxof genes into protist genomes.  BMC Evol Biol 2006, 6:27.26. Baldauf SL, Palmer JD: Animals and fungi are each other's clos-est relatives: congruent evidence from multiple proteins.Proc Natl Acad Sci USA 1993, 90:11558-11562.27. Archibald JM, Longet D, Pawlowski J, Keeling PJ: A novel polyubiq-uitin structure in Cercozoa and Foraminifera: evidence for anew eukaryotic supergroup.  Mol Biol Evol 2002, 20:62-66.28. Stechmann A, Cavalier-Smith T: Rooting the eukaryote tree byusing a derived gene fusion.  Science 2002, 297:89-91.29. Andersson JO, Sarchfield SW, Roger AJ: Gene transfers fromnanoarchaeota to an ancestor of diplomonads and parabasa-lids.  Mol Biol Evol 2005, 22:85-90.30. Huang J, Xu Y, Gogarten JP: The presence of a haloarchaeal typetyrosyl-tRNA synthetase marks the opisthokonts as mono-phyletic.  Mol Biol Evol 2005, 22:2142-2146.31. Markos A, Miretsky A, Muller M: A glyceraldehyde-3-phosphatedehydrogenase with eubacterial features in the amitochon-driate eukaryote, Trichomonas vaginalis.  J Mol Evol 1993,37:631-43.32. Wiemer EA, Hannaert V, van den Ijssel PRLA, Van Roy J, OpperdoesFR, Michels PA: Molecular analysis of glyceraldehyde-3-phos-phate dehydrogenase in Trypanoplasma borelli: an evolution-ary scenario of subcellular compartmentation inkinetoplastida.  J Mol Evol 1995, 40:443-54.33. Qian Q, Keeling PJ: Diplonemid glyceraldehyde-3-phosphatedehydrogenase (GAPDH) and prokaryote-to-eukaryote lat-eral gene transfer.  Protist 2001, 152:193-201.34. Figge RM, Schubert M, Brinkmann H, Cerff R: Glyceraldehyde-3-phosphate dehydrogenase gene diversity in eubacteria andeukaryotes: evidence for intra- and inter-kingdom genetransfer.  Mol Biol Evol 1999, 16:429-440.35. Keeling PJ, Burger G, Durnford DG, Lang BF, Lee RW, Pearlman RE,Roger AJ, Gray MW: The tree of eukaryotes.  Trends Ecol Evol2005, 20:670-676.36. Takishita K, Patron NJ, Ishida K, Maruyama T, Keeling PJ: A tran-scriptional fusion of genes encoding glyceraldehyde-3-phos-phate dehydrogenase (GAPDH) and enolase indinoflagellates.  J Eukaryot Microbiol 2005, 52:343-8.37. Liaud MF, Lichtle C, Apt K, Martin W, Cerff R: Compartment-spe-cific isoforms of TPI and GAPDH are imported into diatommitochondria as a fusion protein: evidence in favor of a mito-chondrial origin of the eukaryotic glycolytic pathway.  Mol BiolEvol 2000, 17:213-223.38. Martin W, Brinkmann H, Savonna C, Cerff R: Evidence for a chi-meric nature of nuclear genomes: eubacterial origin ofeukaryotic glyceraldehyde-3-phosphate dehydrogenasegenes.  Proc Natl Acad Sci USA 1993, 90:8692-8696.39. Henze K, Badr A, Wettern M, Cerff R, Martin W: A nuclear geneof eubacterial origin in Euglena gracilis reflects cryptic endo-symbioses during protist evolution.  Proc Natl Acad Sci USA 1995,92:9122-9126.40. Viscogliosi E, Muller M: Phylogenetic relationships of the glyco-lytic enzyme, glyceraldehyde-3-phosphate dehydrogenase,from parabasalid flagellates.  J Mol Evol 1998, 47:190-199.41. Bernacchia G, Schwall G, Lottspeich F, Salamini F, Bartels D: Thetransketolase gene family of the resurrection plant Crater-ostigma plantagineum: differential expression during therehydration phase.  Embo J 1995, 14:610-8.42. Ruiz-Trillo I, Lane CE, Archibald JM, Roger AJ: Insights into theevolutionary origin and genome architecture of the unicellu-lar opisthokonts Capsaspora owczarzaki and Sphaeroformaarctica.  J Eukaryot Microbiol 2006, 53:379-384.43. Gile GH, Patron NJ, Keeling PJ: EFL GTPase in cryptomonadsand the distribution of EFL and EF-1alpha in chromalveo-lates.  Protist 2006, 157:435-444.44.  [http://amoebidia.bcm.umontreal.ca/pepdb/].45. Armbrust EV, Berges JA, Bowler C, Green BR, Martinez D, PutnamNH, Zhou S, Allen AE, Apt KE, Bechner M, et al.: The genome ofthe diatom Thalassiosira pseudonana: ecology, evolution, andmetabolism.  Science 2004, 306:79-86.46.  [http://genome.jgi-psf.org/Phatr1/Phatr1.home.html].47. Bendtsen JD, Nielsen H, von Heijne G, Brunak S: Improved predic-tion of signal peptides: SignalP 3.0.  J Mol Biol 2004, 340:783-795.48. Guindon S, Gascuel O: A simple, fast, and accurate algorithmto estimate large phylogenies by maximum likelihood.  SystPage 12 of 13(page number not for citation purposes)Biol 2003, 52:696-704.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 2007, 7:89 http://www.biomedcentral.com/1471-2148/7/8949. Ronquist F, Huelsenbeck JP: MrBayes 3: Bayesian phylogeneticinference under mixed models.  Bioinformatics 2003, 19:1572-4.50. Schmidt HA, Strimmer K, Vingron M, von Haeseler A: TREE-PUZ-ZLE: maximum likelihood phylogenetic analysis using quar-tets and parallel computing.  Bioinformatics 2002, 18:502-504.51.  [http://www.tree-puzzle.de].yours — you keep the copyrightSubmit your manuscript here:http://www.biomedcentral.com/info/publishing_adv.aspBioMedcentralPage 13 of 13(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