Previous Article |
Table of Contents
| Next Article 
From the Cover
EVOLUTION
Massive horizontal transfer of mitochondrial genes from diverse land plant donors to the basal angiosperm Amborella
Department of Biology, Indiana University, Bloomington, IN 47405-3700
Contributed by Jeffrey D. Palmer, November 9, 2004
 |
Abstract |
|---|
 Top Abstract Materials and MethodsResultsDiscussionReferences |
|---|
Several plants are known to have acquired a single mitochondrial gene by horizontal gene transfer (HGT), but whether these or any other plants have acquired many foreign genes is entirely unclear. To address this question, we focused on Amborella trichopoda, because it was already known to possess one horizontally acquired gene and because it was found in preliminary analyses to contain several more. We comprehensively sequenced the mitochondrial protein gene set of Amborella, sequenced a variable number of mitochondrial genes from 28 other diverse land plants, and conducted phylogenetic analyses of these sequences plus those already available, including the five sequenced mitochondrial genomes of angiosperms. Results indicate that Amborella has acquired one or more copies of 20 of its 31 known mitochondrial protein genes from other land plants, for a total of 26 foreign genes, whereas no evidence for HGT was found in the five sequenced genomes. Most of the Amborella transfers are from other angiosperms (especially eudicots), whereas others are from nonangiosperms, including six striking cases of transfer from (at least three different) moss donors. Most of the transferred genes are intact, consistent with functionality and/or recency of transfer. Amborella mtDNA has sustained proportionately more HGT than any other eukaryotic, or perhaps even prokaryotic, genome yet examined.
Recent studies indicate that plant mtDNAs are unusually active in HGT relative to all other organellar genomes and nuclear genomes of multicellular eukaryotes. Four papers (3–6) have reported a total of nine cases of mitochondrial HGT within seed plants. Three transfers involve parasitic angiosperms as putative donors or recipients and implicate direct, plant-to-plant transfer of DNA as one mechanism of HGT (5, 6). Each of the nine transfers involves a different set of recipient plants. For this reason, and because only a few mitochondrial genes have been scrutinized for potential HGT in these or any other plants, it is unclear whether these cases are singular exceptions in each genome or whether they are harbingers of perhaps massive mitochondrial HGT in certain plants.
To address this uncertainty, we have assessed the origin and history of the mitochondrial protein gene set of Amborella trichopoda and the five angiosperms whose mtDNAs have been sequenced. Amborella was chosen because it was already known to contain one foreign gene (3) and because preliminary studies suggested it might be unusually rich in HGT. We show that Amborella mtDNA has sustained remarkably massive HGT, whereas the five sequenced mtDNAs show no evidence of HGT.
|
Materials and Methods |
|---|
|
TopAbstractMaterials and MethodsResultsDiscussionReferences |
|---|
Sequences were aligned by using either BIOEDIT or SE-AL V2.0A11 (alignments available on request). Regions containing primers, poor alignment, or only a few taxa, as well as all sites subject to RNA editing in either Arabidopsis/Brassica or Oryza/Zea, were excluded from phylogenetic analyses. Analyses used PAUP* 4.0B10 within an automated script (courtesy of D. W. Rice, Indiana University). A starting topology was generated with maximum parsimony, from which the transition/transversion ratio and gamma shape parameter were estimated. A maximum-likelihood (ML) tree was built by using these parameters, the HKY85 model (7), four rate categories, and empirically determined base frequencies. If the ML and parsimony trees differed in topology, a new ML tree was built, using parameters from the preceding ML tree, and this process was repeated until a stable topology was obtained.
The Shimodaira–Hasegawa (SH) test (8) was used to assess whether phylogenetically anomalous gene placements suggestive of HGT are significantly favored over the hypothesis of strictly vertical transmission. This test assigns a P value to the difference in likelihood between the best ML tree found (as shown in all of our figures) and that ML tree, based on the same data set, in which the Amborella gene in question has been constrained to fit a vertical scenario of paralogy (duplication) by being placed as sister to its putatively vertically transmitted homolog.
All cases of suspected Amborella HGT from bryophyte donors and most cases from angiosperm donors were confirmed by obtaining the same sequence from multiple (3–5) independent preparations of Amborella DNA. These DNAs originated from material sent from four different sources. Two shipments of fresh leaves, received and DNA-extracted 18 months apart, came from the University of Santa Cruz Arboretum courtesy of Brett Hall. Silica-dried leaves were obtained from Doug Soltis (University of Florida, Gainesville), fresh leaves were obtained from the University of Massachusetts Greenhouse, Amherst, courtesy of Teddi Bloniarz, and Amborella DNA was received from Yin-Long Qiu (University of Michigan, Ann Arbor). Leaves were inspected carefully for any signs of epiphytic growth and other potential sources of biological contamination, in some cases under a dissecting microscope, and were thoroughly washed before DNA extraction. All attempts to confirm HGT by using alternative sources of Amborella DNA were successful, with PCR product ratios constant among DNA preps for those primers giving size-heterogeneous products. Further evidence against contamination or sample mix-up came from the pseudogene nature of nine cases of putative HGT (Table 1), i.e., contamination or mix-up is much more likely to result in artefactual isolation of intact, functional copies of a gene. Further verification was obtained for two cases of HGT by showing that cDNA sequences are identical to genomic sequences except for a few sites of RNA editing (ref. 3 and unpublished data).
|
|
Results |
|---|
|
TopAbstractMaterials and MethodsResultsDiscussionReferences |
|---|
100 pairs of primers was designed to PCR-amplify the entire set of 40 angiosperm mitochondrial protein genes (including introns) that were present in the last common ancestor of angiosperms (11). Pilot amplifications to assess primer efficacy involved three test plants. Plant mitochondrial genes are generally present once per genome, and rice and Arabidopsis routinely give a single PCR product of the expected size based on their known genome sequences (12, 13). But to our surprise, with many primer pairs Amborella gave either two distinct bands or a single broad band. Three of these mixed products were examined and found to consist of vertically and horizontally transmitted genes, similar to the atp1 case already described for Amborella (3). Finding so much HGT among so few examined Amborella genes led us to focus on Amborella. We amplified and sequenced all readily isolated Amborella mitochondrial protein genes, taking care to sequence multiple clones for each Amborella gene whose PCR products showed either size or sequence heterogeneity. For most genes, too few homologs were available to enable meaningful phylogenetic analysis. We therefore chose 13 diverse angiosperms and three gymnosperms (as outgroups) and sequenced their genes from PCR products, setting aside complicated cases (of potential HGT) involving size or sequence heterogeneity. Where appropriate, we also sequenced selected genes from a few nonseed plants. Phylogenetic analyses included all of these genes, all relevant genes from the five sequenced angiosperm mitochondrial genomes (12–16), and selected other available sequences. Of the 40 protein genes present in the ancestral angiosperm mitochondrial genome (11), 31 were recovered from Amborella (Table 1). Of these 31 genes, 20 showed what we interpret as reasonable to compelling evidence for one or more cases of HGT. The strongest evidence for HGT comes from seven genes for which Amborella possess a bryophyte-like copy (Table 1 and Fig. 1). Six of these seven bryophyte-like genes are far more similar in sequence to homologs from mosses than to angiosperm homologs, and in phylogenetic analyses these all group with mosses with convincing support (Fig. 1 and data not shown). No moss sequences are available for nad6, which is more similar to the only bryophyte sequence available (from the liverwort Marchantia) than to angiosperm homologs (Fig. 1). Three of the six moss-derived genes (cox2, nad5, and nad7) probably were acquired from different lineages of moss donors (Fig. 2). For the other three genes, there is insufficient sampling of mosses (Fig. 1) to address this issue.
|
|
A total of 13 Amborella genes show evidence of HGT from angiosperm donors only (Table 1). In each case, a pair of divergent gene copies was isolated, one of which is putatively vertical (see below) and the other putatively horizontal in origin. All but two of the 13 donors appear to be eudicots, albeit with varying levels of support. At one extreme are genes such as nad1, atp1, and ccmFN1, with strong bootstrap support for being of eudicot origin (Fig. 3) and which also pass the SH test for being the product of HGT (Table 1). Five other genes (ccmB, ccmC, atp8, atp4, and rps19) show moderately good bootstrap support (70–80%) for being derived from eudicots (Fig. 3) and/or pass the SH test (Table 1). The other five genes show weak or no support for being eudicot-derived and fail the SH test (Fig. 3 and Table 1).
|
Where do the 26 transferred genes reside within Amborella? The chloroplast genome can be ruled out because it has been sequenced in Amborella (23) and does not contain any of the transferred genes (data not shown). The nuclear genome can be ruled out and a mitochondrial location be assigned with confidence for those two transferred genes (atp1 and atp8) that are transcribed and subject to mitochondrial-characteristic RNA editing (ref. 3 and unpublished data). We favor a mitochondrial location for the 24 other horizontally transferred genes for three reasons. First, all six HGT cases (including the two above from Amborella) whose genomic provenance has been established are indeed located in mtDNA (ref. 3 and unpublished data). Second, nucleotide substitution rates are almost always far higher in the nucleus than in the mitochondrion in plants (9, 10), such that even relatively recent cases of functionally transferred mitochondrial genes present in the plant nucleus have extremely long branches in gene trees compared to their mitochondrial counterparts (e.g., refs. 10 and 24). Only the most divergent transferred genes in Amborella (e.g., atp6 and rpl16) even approach this level of divergence, and most transferred genes show conventional mitochondrial-like branch lengths (Figs. 1 and 3). Third, the transferred genes amplify by PCR to roughly the same abundance as vertically transmitted Amborella homologs. Because mtDNA is typically present in hundreds of copies per cell, this result, even though the PCR was not carried out in a quantitative manner, suggests that the putatively horizontal and vertical copies reside in the same genome, that of the mitochondrion [conuclear localization is highly improbable, given that survey of 280 diverse angiosperms (11) showed that genes corresponding to 23 of the 26 transfers have never, or in one case very rarely, been lost from mtDNA].
Outside of Amborella, there is no convincing evidence for HGT in the 31 mitochondrial gene trees. Between 24 and 30 of the 31 genes are present, as intact, single-copy genes (except for identical duplications; see Discussion), in the five sequenced angiosperm mtDNAs (12–16). None of these genomes shows any evidence of HGT. The two grasses (Zea and Oryza) always either pair together or form part of a larger clade of grasses, the grasses in turn always group with at least one to all three of the most commonly sampled other monocots (Eichhornia, Agave, and Philodendron), and monocots are almost always monophyletic. Likewise, the two crucifers (Arabidopsis and Brassica) always pair as sisters, and these, together with Beta, always tree within core eudicots, as expected. Core eudicots are almost always monophyletic. Although there are numerous phylogenetic anomalies involving basal eudicots (usually Platanus, Eschscholzia, and Mahonia) and the magnoliids (usually Piper, Laurus, Asarum, Calycanthus, and Liriodendron), these are only poorly supported and are best attributed to poor resolution of generally slowly diverging sequences in these parts of angiosperm phylogeny.
|
Discussion |
|---|
|
TopAbstractMaterials and MethodsResultsDiscussionReferences |
|---|
70,000 species of monocots and 175,000 eudicots) may reveal phylogenetically local cases invisible to our sampling. The 13 other angiosperms sampled extensively in this study also show little evidence for HGT, but with still further caveat. For these plants, we deliberately ignored complicated PCR results that might, as with Amborella, reflect a mixture of vertical and horizontal products, focusing instead on clean PCR sequences to boost phylogenetic coverage. At the same time, none of these plants showed nearly as many complex, heterogeneous PCR products as Amborella, and thus none could approach it in extent of HGT.
Clearly, then, the incidence of HGT varies markedly from one plant to another. One wonders how many other Amborella-type situations exist among the 255,000 species of flowering plants. Are the eight cases of thus-far singleton HGT identified in other seed plants (3–6) exceptional for these genomes, or are some of these genomes also replete with HGT?
Why is Amborella so extraordinarily rich in HGT? Amborella is a monotypic genus (and family) of shrubs endemic to New Caledonia, where it grows in midelevation (600–900 m high) montane tropical rain forests (25). Epiphytic and parasitic plants are common in this environment, and Amborella leaves and stems are often covered with diverse epiphytes, including mosses and other bryophytes (e.g., Fig. 4). This could readily promote direct, plant-to-plant HGT, especially given the potential for herbivory to introduce epiphytic tissue and exudates within wounded Amborella tissue. Evidence for direct plant-to-plant HGT has recently been reported in the context of parasitism, to account for three well supported cases of transfer of mitochondrial genes from parasitic angiosperms to their hosts (6) or vice versa (5). Epiphytism may offer similar opportunities for HGT.
|
3,400 vascular plants native to the island (26). Molecular examination of the flora growing on and in the general habitat of Amborella should prove crucial in efforts to (i) elucidate the factors promoting such extensive HGT, (ii) uncover other cases of extensive HGT, (iii) pinpoint donor identities, (iv) estimate the timing of transfer, and (v) estimate the number of transfers. This last issue relates to the fact that plant mitochondria frequently fuse (27), with their genomes recombining (28), which makes it easy to imagine multiple mitochondrial genes being acquired in a single event involving whole-mitochondrial transfer. Three of the moss transfers are evidently from different donor lineages (Fig. 2), as are two of the eudicot transfers (nad1 and ccmFN1; Fig. 3), but assessing the overall balance between fewer multigene transfers or more numerous single-gene transfers will require far more extensive sampling of genes and plants. Limits and Logical Bases of Inferring HGT in Plant Mitochondrial Genomes. The PCR approach we used to census the Amborella mitochondrial protein gene set will clearly miss important components of the mitochondrial genome. These include mitochondrial rRNA and tRNA genes (the latter are also too small for a meaningful PCR approach), chloroplast-derived sequences (which are commonly found in plant mtDNAs; refs. 12–16), and intergenic DNA (which makes up most of a typical plant mitochondrial genome; refs. 12–16). Furthermore, mitochondrial protein genes from nonland plants will be so divergent (land plant mtDNAs have exceptionally low rates of sequence evolution; refs. 9 and 10) as to be strongly disfavored by PCR when faced with competition from vertically retained homologs. Only by sequencing the Amborella mitochondrial genome can we census its population of horizontally acquired DNA in a comprehensive and phylogenetically unbiased manner.
A major limitation in our ability to detect HGT in plant mtDNA is the often poor resolution of individual gene trees (Figs. 1, 3, and 7), which is largely a consequence of the very low rate of nucleotide substitutions in most plant mtDNAs (9, 10) and the short length of most gene regions used in our phylogenetic analyses (Table 1). Importantly, though, some of the weakly supported conflicts between mitochondrial gene trees and organismal phylogeny are most likely, given the growing evidence for HGT as an ongoing and moderately frequent process in plant mitochondrial evolution, the residue of horizontal transfer occurring within poorly resolved portions of the gene trees. We are lucky that, of all angiosperms, Amborella happens to be so rich in HGT, because its distinctive position at the base of angiosperms makes it relatively easy to detect with reasonable confidence transfers from other angiosperms, even with the scanty taxon sampling of this study.
Even so, a number of the putative Amborella transfers are admittedly not well supported by purely phylogenetic criteria. The SH test is a stringent test, and those 14 transfers that passed it (Table 1) should therefore be regarded as well supported. Some of the 12 other angiosperm cases appear to be good candidates for HGT based solely on visual inspection of phylogenetic trees, but others are less compelling (Figs. 1 and 3). Importantly, there is a second, independent criterion that we hereby invoke, namely, the very existence of divergent copies of a gene within a plant mitochondrial genome. With one possible exception (29), we are unaware of any examples of divergent duplicate genes in plant mtDNAs that are paralogs, i.e., that trace back phylogenetically to duplication events within a mitochondrial lineage. Instead, all divergent duplicates behave phylogenetically as xenologs, as the products of horizontal evolution. Moreover, plant mitochondria possess evolutionary mechanisms that tend to prohibit paralogs from diverging with time: repeated elements larger than 500 bp in plant mtDNAs are subject to frequent concerted evolution such that they generally remain identical to one another (12–16). HGT may be the only mechanism plant mitochondria possess to establish divergent copies of a gene. Therefore, the presence of divergent duplicates in plant mtDNA (especially when they are distantly related by phylogeny, as here) can be taken as prima facie evidence for HGT.
Functionality of Transferred Genes in Amborella. Whereas all but one of the vertically transmitted genes in Amborella have intact ORFs, 8 of the 26 transferred genes are pseudogenes (Table 1). Whether any of the 18 intact transferred genes are functional and under selection is an open question. Both transferred genes (atp1 and atp8) whose expression has been assayed are transcribed and RNA-edited (ref. 8 and unpublished data); however, transcribed and RNA-edited pseudogenes are known to occur in plant mitochondria (30, 31). Although some of these transferred genes may be functional in Amborella mitochondria, we suspect that most are not, and that with time an increasing proportion will evolve into obvious pseudogenes. The time frame and dynamics of HGT in Amborella mitochondria may well be similar to those described in bacterial systems, where HGT regularly supplies the genome with foreign genes, most of which soon decay as pseudogenes (32).
HGT in Different Plant Genomes. These results highlight the disparity between plant mitochondrial and chloroplast genomes in their propensity to take up foreign DNA. Despite vastly more chloroplast than mitochondrial sequencing in plants, HGT is now well established for the latter but unknown for the former. Of greatest relevance, the sequenced chloroplast genome of Amborella (23) shows no evidence of HGT (D. W. Rice and J.D.P., unpublished work). This disparity in frequency of HGT is in keeping with other features that distinguish the two genomes. Plant mtDNAs contain much more noncoding DNA than compact chloroplast DNAs and are renowned for their frequent incorporation of chloroplast and nuclear DNA sequences, whereas chloroplasts show no evidence of intracellular gene transfer (12–16, 33). Plant nuclear genomes, on the other hand, have a loose, fluid organization (mostly noncoding DNA, many gene duplications) that would seem to accommodate HGT, are known to frequently take up DNA from organelle genomes via intracellular transfer (33), and offer one clear example of recent multigene HGT (from bacteria; ref. 34). Given this evidence and how rich it is in mitochondrial HGT, we predict that substantial levels of nuclear HGT will be found in Amborella.
|
Acknowledgements |
|---|
|
Footnotes |
|---|
Data deposition: The sequences reported in this paper have been deposited in the GenBank database (accession nos. AY831968–AY832318).
Present address: Department of Biology, University of New Mexico, Albuquerque, NM 87131. 
Present address: Department of Biological Sciences, Auburn University, Auburn, AL 36849.
To whom correspondence should be addressed. E-mail: jpalmer{at}bio.indiana.edu.
© 2004 by The National Academy of Sciences of the USA
|
References |
|---|
|
TopAbstractMaterials and MethodsResultsDiscussionReferences |
|---|
, S., Young, G. J. & Palmer, J. D. (2004) Nature 432, 165–166.[Medline]
, S., Rice, D. W. & Palmer, J. D. (2004) BMC Evol. Biol. 4, 35.[CrossRef][Medline]
CiteULike 
Complore 
Connotea 
Del.icio.us 
Digg What's this?
This article has been cited by other articles in HighWire Press-hosted journals:
|
|
 |
|
  S.-M. Chaw, A. Chun-Chieh Shih, D. Wang, Y.-W. Wu, S.-M. Liu, and T.-Y. Chou The Mitochondrial Genome of the Gymnosperm Cycas taitungensis Contains a Novel Family of Short Interspersed Elements, Bpu Sequences, and Abundant RNA Editing Sites Mol. Biol. Evol., March 1, 2008; 25(3): 603 - 615. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
H. Khan, N. Parks, C. Kozera, B. A. Curtis, B. J. Parsons, S. Bowman, and J. M. Archibald Plastid Genome Sequence of the Cryptophyte Alga Rhodomonas salina CCMP1319: Lateral Transfer of Putative DNA Replication Machinery and a Test of Chromist Plastid Phylogeny Mol. Biol. Evol., August 1, 2007; 24(8): 1832 - 1842. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 G. Jin, L. Nakhleh, S. Snir, and T. Tuller Efficient parsimony-based methods for phylogenetic network reconstruction Bioinformatics, January 15, 2007; 23(2): e123 - e128. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 A. O. Richardson and J. D. Palmer Horizontal gene transfer in plants J. Exp. Bot., January 1, 2007; 58(1): 1 - 9. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
G. Jin, L. Nakhleh, S. Snir, and T. Tuller Inferring Phylogenetic Networks by the Maximum Parsimony Criterion: A Case Study Mol. Biol. Evol., January 1, 2007; 24(1): 324 - 337. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
R. M. Bateman, J. Hilton, and P. J. Rudall Morphological and molecular phylogenetic context of the angiosperms: contrasting the 'top-down' and 'bottom-up' approaches used to infer the likely characteristics of the first flowers J. Exp. Bot., October 1, 2006; 57(13): 3471 - 3503. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
M. Turmel, C. Otis, and C. Lemieux The Chloroplast Genome Sequence of Chara vulgaris Sheds New Light into the Closest Green Algal Relatives of Land Plants Mol. Biol. Evol., June 1, 2006; 23(6): 1324 - 1338. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
A. G. B. Simpson, Y. Inagaki, and A. J. Roger Comprehensive Multigene Phylogenies of Excavate Protists Reveal the Evolutionary Positions of "Primitive" Eukaryotes Mol. Biol. Evol., March 1, 2006; 23(3): 615 - 625. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
S. Kim, P. S. Soltis, K. Wall, and D. E. Soltis Phylogeny and Domain Evolution in the APETALA2-like Gene Family Mol. Biol. Evol., January 1, 2006; 23(1): 107 - 120. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 Y. Cho, J. P. Mower, Y.-L. Qiu, and J. D. Palmer From the Cover: Mitochondrial substitution rates are extraordinarily elevated and variable in a genus of flowering plants PNAS, December 21, 2004; 101(51): 17741 - 17746. [Abstract] [Full Text] [PDF]
|
|
| |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
