The struggle for life of the genome's selfish architects

From incredulity to inescapability

The history of TEs is much shorter than the history of the theory of evolution: it is less than 70 years since Barbara McClintock first reported the existence of controlling elements. However, even though her discoveries were rigorously supported by experimental data it took much longer for McClintock's findings to gain acceptance than it had for Darwin's theory. Basics of Darwin's theory relied on common sense, and it was clearly its implications for evolution and the origin of the human species that aroused his virulent detractors. McClintock discovered transposable elements at a time dominated by the idea of genetic stability, which appeared to be essential for transmission to descendants, and for the conservation of species characteristics. The concept of genetic stability had emerged after Mendel's laws, and was later reinforced by discoveries such as the structure of DNA, and the regulation of bacterial genes. The established view of a static genome seemed to be unquestionable. Her work aroused a reception that may have been less hostile than that of Darwin's detractors, but her work was not understood, and gave rise to incredulity, rejection and sarcasm [4]. This lasted for several years and indeed decades, until the identification of similar elements in other genomes [5] began to win round the wider scientific community. Eventually the transposition of DNA fragments was demonstrated using the tools of molecular biology. In 1983 Barbara McClintock was finally awarded with the Nobel Prize for her discovery of transposable elements.

It is now no longer possible to ignore TEs. The impact of TE-derived sequences in regulating genes needs no further proof. Everyone accepts that genomes are quite flexible or plastic entities, that they are riddled with TEs, and that TEs affect both gene regulation and the composition and structure of the genome. The depiction of the genome as a linear succession of genes and the dogma of its stability have been replaced by a dominant view of a functional genome as a complex network of genetics, epigenetics, and cell interactions, in which TEs and other structural or functional elements are involved. 25 years after McClintock's Nobel Prize, have we fully embraced the full extent and diversity of the influence of TEs, notably in genome evolution?

The astonishing properties of jumping genes

TEs possess two main characteristics that distinguish them from other genomic components. They are mobile, so able to change their genetic environment, and by doing this they also change the genetic environment of the locus into which they insert. Since they have the intrinsic ability to multiply during the transposition process, they are almost inevitably repeated, with a virtually unlimited copy number, restricted only by the carrying capacity of their environment i.e. the genome. Hence they are simultaneously part of the genome and independent entities living their own life within the genome, in a way that reminds Dawkins' selfish gene [6].

How can TEs be integrated as a major evolutionary factor in Darwinian theory? How do TEs influence genome evolution, and how does genome evolution influence TEs? Do they exploit the genome? Are they exploited by the genome? Are they parasites of the genome or part of it? What would evolution and life have been like without them? The answers are complex, because the interactions between TEs and their host genomes are complex. In this review we attempt to propose some clues to the answers to these questions. Some of the properties described below show how TEs fit in with the most recent developments in evolutionary theory.

Structure and classification

Transposable elements exist in every known eukaryotic, bacterial or archaeal genome. They are defined as DNA sequences that are able to move from one chromosomal position to another within the same genome (i.e. within a single cell), which distinguishes them from phages and viruses, which move from cell to cell.

TEs usually encode the genes that promote their own transposition, but many non-autonomous elements use the transposition machinery of close relatives or unrelated elements instead. TEs are divided into two classes depending on their transposition mechanism, each class is further divided into subclasses, orders and superfamilies [7].

Class I elements transpose through an RNA intermediate, transcribed from DNA then reverse transcribed into double-stranded DNA (dsDNA) before or during their integration into a new position. They are replicative by nature. The key enzyme is a reverse transcriptase (RT), which is present in the telomerases of eukaryotes, but which is also an overall characteristic of mobile RNA entities (retroviruses, group II introns, and retrotransposons). RT is also present in bacteria, in elements such as retrons, group II introns and diversity-generating retroelements, although their mobility has been proven only for group II introns [8]. In Eukaryotes, four orders of autonomous retroelements are recognized [7], (i) Long Terminal Repeats (LTR) retroelements, similar in structure to retroviruses, (ii) Long INterspersed repeated Elements (LINEs), elements which have no LTRs but do have a polyA tail, (iii) DIRS (from DIRS-1, the first element identified in Dictyostelium) and (iv) PLEs (Penelope-like elements), these two last groups having somewhat unusual structures. In eukaryotes, several Class I non-autonomous elements have been identified. Short INterspersed repeated Elements (SINEs) are usually derived from tRNA and use LINEs to transpose. They may contain the 3' part of LINEs, probably fused to the tRNA at the time of retrotransposition [9]. All other non-autonomous retroelements possess typical structural features or are deletion derivatives of one of the four orders of autonomous retroelements (LTR, LINE, DIRS, PLE).

The diversity of retroelements reflects their complex origin. Indeed, phylogenies based on RT suggest that LINEs are related to group II introns, and that most retroviruses belong to one superfamily within the LTR order, despite several independent examples of infectious retroviruses originating from LTR-retroelements [10]. However, phylogenies based on other protein domains (endonuclease or RNAseH) display different topologies, suggesting that the various retroelements originated from independent fusions of different modules [10, 11].

Class II elements transpose directly with no RNA copy intermediate. They can excise from the donor site (they are known as cut-and-paste transposons, and the transposition is described as conservative) although this is not always the case, since several Class II elements are replicative (i.e. their transposition is coupled with replication). Hence, Class II has been divided into two subclasses depending on the number of DNA strand cuts at the donor site, which reflects these different transposition mechanisms. In the subclass I, the two strands are cut at both sites, and the element is fully excised [7]. This subclass comprises mainly those elements that are characterized by having two terminal inverted repeats (TIR) and at least one gene encoding the transposase (TIR elements Order). They are especially abundant in prokaryotes, where they are known as insertion sequences (IS), and are also widespread and diversified in eukaryotes. On the basis of transposase similarities, TIR elements can be divided into 12 to 17 superfamilies in eukaryotes [7, 12, 13], and more than 20 in prokaryotes [14, 15]. However, a number of prokaryotic and eukaryotic superfamilies are related and thus form trans-domain superfamilies, which suggests that these superfamilies are either old enough to have preceded the split into the three domains of life, or that horizontal transfers occurred in the distant past [16]. Subclass I also includes elements that do not possess a transposase, but instead have a recombinase that is able to recombine two DNAs without generating free ends. Recombinase-containing Class II elements are frequent in prokaryotes [14], and have also been found in eukaryotes, although so far only in some opisthokonts (crypton elements) [[17], and see RepBase http://www.girinst.org]. When only one strand is cut on each side, the transposition is said to be replicative. In eukaryotes, two recently discovered types of Class II elements (Polintons/Mavericks and Helitrons) are thought to transpose in such a way. Polintons are very large elements, bordered by TIRs and containing several genes, including an integrase (related to retroviral integrases and Class II transposases) and a polymerase [18]. Helitrons are moderately large, possess hairpin structures at the ends, and contain a helicase [19]. These characteristics are reminiscent of a rolling-circle mechanism, such as that involved in IS91. In bacteria, another recently identified family (IS608) is characterized by having a transposase related to the RCR protein of IS91, which recognizes specific secondary structures, such as hairpins, at the tips of the elements. However, the transposition mechanism seems to be different [20]. Finally, prokaryotes also carry more complex TEs-based structures that trap a large range of mobile genes, such as in composite transposons (Tn) or in Integrative and Conjugative Elements (ICEs) [21], illustrating that evolution can also occur by modularity [22].

Prokaryotic and eukaryotic TIR elements frequently generate considerably reduced non-autonomous elements known as Miniature Inverted repeat Transposable Elements (MITEs) that use a transposase encoded in trans to transpose. MITEs are either deletion derivatives of full-length elements, or only share TIRs with their autonomous partner. Helitrons are also often found as non-autonomous copies (derived from an internal deletion).

Abundance and distribution

The abundance of TEs in each eukaryotic and prokaryotic lineage is highly variable (Figure 1 and Additional file 1). TEs are more ubiquitous in eukaryotes (most genomes contain TEs) than in prokaryotes, in which more that 20% of the genomes so far sequenced lack both remnants and complete TEs [23]. Furthermore, TEs are far more abundant in eukaryotic genomes (comprising up to 80% of the genome) than in prokaryotes (up to 10% of the genome, averaging only 1-5%). However, in both prokaryotes and eukaryotes, there seems to be a positive correlation between genome size and TE abundance [23, 24]. Retroelements (that have intrinsic replicative properties and may be large in size) are often the main provider of TE DNA in eukaryotes, such as several mammals, yeasts, Drosophila, and plants with large genomes [25–27]. In some cases, however, (e.g. Trichomonas vaginalis, Caenorhabditis elegans), Class II elements dominate, at least in terms of copy number [16]. In contrast, small eukaryotic genomes (parasitic apicomplexa for example) are usually devoid of TEs, perhaps because of a general tendency towards genome size reduction.

In many genomes, a few elements dominate, but this does not preclude an extraordinary diversity, which is usually in the range of hundreds of element families. For example, LINEs of the L1 type and SINEs Alu are predominant in the human genome, and diversified as a few subfamilies of different ages. In contrast, LTR retroelements (endogenous retroviruses) are found in relatively low copy numbers, and belong to a few dozen different families [26]. So TE diversity and abundance is highly variable from one species to another, and reflects their specific genome-TE history. In addition, TE distribution within a genome is usually neither random nor uniform. First some (rare) elements are site-specific, such as LINE R2 elements, which exclusively insert into a single site in the rDNA, or some IS elements in prokaryotes. Secondly, TEs are frequently found in chromosomal regions where their potentially deleterious impact is reduced. Thus, TEs are common in the heterochromatin and in pericentromeric or telomeric regions, in low gene density regions, or within other elements. However, they can also be found in open chromatin regions, near tRNA genes, and near promoters and genes. Hence, in some plants DNA transposons are preferentially found around genes.

This non-random distribution, including an apparent preference either to insert near to genes or to avoid them, may result either from a true insertion preference or just from selection. The study of recent insertions obtained in the lab or in the wild may help distinguish between these two hypotheses [28].

Our view of TE landscape is biased, because genome sequencing efforts have mainly focused on bacterial genomes or on the "higher eukaryotes", and so are not representative of the full diversity of life. Moreover, TE contents can differ greatly between closely related species. For example, the Archaea Sulfolobus solfataricus carries at least 130 full IS copies and at least 200 partial IS copies, whereas the related species Sulfolobus acidocaldarius carries only a few partial IS copies [14, 15]. Furthermore, differences are often observed between strains of the same species in terms of copy number [28, 29]. As the number of genome research projects increases and technology progresses, we can hope that the representations of the tree of life and of intra-species diversity will improve. This may well reveal that TE history has as many versions as there are populations bearing them.

Consequences of TEs at the DNA level

The presence of TEs (as dispersed, mobile, and repeated elements) has two major mutational consequences at the DNA level: insertion within a locus, and ectopic recombination leading to different types of rearrangements. First, TEs can insert near or within genes, and by doing so alter or destroy the activity of the gene in a variety of ways, ranging from total inactivation to spatio-temporal changes in expression, alternative splicing, or changes in expression level or protein activity. Modifications of gene expression can be a direct consequence of adding extra nucleotides to the original sequence or an indirect consequence of the epigenetic marks on the element. Furthermore, in addition to promoter and terminator sequences, TEs sometimes carry silencers or insulators that are able to modify expression over distances of several kb, or binding sites for different proteins (i.e. heterochromatin protein) [30–32]. Second, the possibility of recombination between two copies at different loci can also have a more or less dramatic effect, ranging from small-scale inversions to major chromosomal rearrangements, including deletions, translocation or duplications [33, 34].

Although TEs are defined as intracellular parasites/entities, they are prone to being transferred from cell to cell, notably in prokaryotes through conjugation of the element or of the plasmids carrying it. A major consequence is lateral gene transfer (LGT), also known as horizontal transfer (HT), which is quite common in bacteria. In eukaryotes, numerous cases of TE HT have been reported, although the vector involved remains elusive. Interestingly, several TEs have been found in eukaryotic viruses, such as TED, piggyBac, or Tc1-like in baculoviruses [35], DIRS elements in a Polydnavirus genome [36] or TEs related to the IS605/IS607 family in Phycodnavirus and Mimivirus genomes [37]. Thus, these viruses could be used by TEs as a potential source of horizontal dissemination in eukaryotes.

TEs and the genome: an evolutionary point of view

The way we imagine genome evolution today has not departed from the Darwinian theory. Any new gene, or new function, which confers advantages in the host in a given environment, will be selected as long as the host is in this environment. This is true for the selfish genes of Dawkins: any gene that is able to propagate successfully (by vertical transmission) in a given environment (the genome, including other genes) will successfully disseminate in this environment [6]. This also holds on for TEs, which can be viewed as selfish DNA: the ultimate parasite [38], which is able to propagate itself through vertical transmission, through intra-genomic transposition, and through horizontal transfer.

TEs are able to replicate more rapidly than the genome, and so constitute a kind of genomic cancer. They are basically parasitic, i.e. selfish, and deleterious entities, conferring no benefits on the genomes they inhabit. For these reasons, they were long considered to be "junk DNA", part of the genome that by definition it would be better to get rid of, because it has no role, no function, and is just a kind of genetic burden for the host genome. This simplistic view must now be tempered. First of all, TEs and the rest of the genome have lived side by side for a very long time and such prolonged co-habitation almost inevitably leads to various kinds of interaction. Second, having no known role does not necessarily imply having no impact: day after day, portions of the genome that were previously thought to be useless have been shown to have important regulatory or structural roles. The same could be true for TEs. Hence, when considering genome evolution, TEs are far from being just parasitic sequences [39]. Starting from the simple assumption that DNA is separated into two compartments, the genome, and the TEs, we review below the relationships between them in all their diversity. We will focus first on how the genome deals with the sea of TEs that surrounds it, and then on how TEs deal with the host genome in which they are embedded.

For the genome, TEs are a major source of genetic variation

From mutations to polymorphism, genetic variants reflect the diversity within a population, and DNA alterations or changes constitute the basis of evolution. At the DNA level, two molecular mechanisms are responsible for generating diversity: mutation and recombination. Classically, mutations (changes in nucleotide sequences) arise either through uncorrected replication errors or after DNA lesions; whereas recombination is a normal process during the meiotic phase. However, both processes can also result from the activity or mere presence of a transposable element. Transposition does not result from fortuitous errors during replication or lesion repair, but can be considered to be an active mutagenic process, resulting in mutations that are different from SNPs (Single Nucleotide Polymorphisms). In contrast, TE-induced ectopic recombination can be viewed as an erroneous (albeit easy-to-produce) process in contrast to normal meiotic crossing-over.

Such DNA alterations may affect the function (of genes) and the structure (of genomes), the worst outcome being the immediate death of the cell, or its inability to complete meiosis. However, mildly detrimental or neutral effects are also to be expected, and insertions that produce such effects may survive and contribute to the genetic variation of the host genome. There are many diverse ways in which TEs can alter the genome, ranging from small sequence modifications to gross rearrangements. Finally, the frequency of such events is not negligible, which means that TEs are major actors in diversity [40, 41].

For the genome, TEs are disturbing invaders but can also be useful helpers

Epigenetic control is widely used by multicellular organisms, such as higher metazoans or plants, to implement cell lineage-specific gene regulation, and more generally for any developmental process, including X inactivation, parental imprinting, cell cycle, germ line development, and early embryogenesis [108–112]. Epigenetic mechanisms are also used to silence transposable elements, thus avoiding the detrimental effects of transposition. The present-day view is that epigenetics was first used to defend the genome against invading DNA (including TEs) before being exploited at a larger scale for gene regulation. The relationship between TEs, epigenetics, and gene regulation is in fact far more complex than this. TEs may have acted primarily as evolution drivers that led the genome to evolve defense mechanisms, and then gene expression control systems. Although present-day epigenetic gene regulation appears at first sight to be free of TE intervention, silenced TEs can nevertheless directly interfere with the expression of adjacent genes [68]. Furthermore, it has recently been proposed that TEs could ultimately have been exapted for regulation purposes [113]. Finally, occasional disruption of epigenetic control may offer an opportunity to enhance the evolvability.

For the genome, TEs carry useful sequences and functions that can be exploited

Data accumulated over several years have indicated that the contribution of TEs to the evolution and function of host genes is far from negligible. The direct participation of TEs in genome functional evolution can occur in different ways (Figure 2). First they can carry sequences into regulating, coding, or intronic regions. These sequences may trigger useful functional changes (expression pattern, alternative splicing, transcription initiation and termination) as a result of the presence of particular motifs or their physico-chemical properties [see [162] as a recent example]. Second, they can provide a function normally encoded by the element, which is then recruited to implement a cellular function. In this case, either an entire domain or the full protein is recruited, i.e. domesticated by the genome. The molecular domestication of transposable elements has long been known to occur, even if the role of the domesticated copy in the cell is not always obvious [163, 164]. It concerns both classes of TEs. The roles assumed by TEs in the cell are far from anecdotal, and can lead to important evolutionary innovation.

There are several criteria that may indicate that a domestication event has occurred: the loss of mobility, presence at only one locus, fixation in the population, presence of an intact open reading frame, presence at orthologous sites in several species, or traces of positive selection on these orthologous sequences [165]. Obviously, none of these criteria is sufficient in isolation, because each of them occurs in a normal TE life cycle. For example, traces of selection are visible in some cases of suspected horizontal transfer [166, 167]. Fixation of an immobile copy in one or several species can be achieved simply as a result of demographic history and genetic drift. When a TE family is on its way to being eliminated, it loses its members one by one, until only one copy remains.

Parasitic TEs and host fitness

At first sight, most of the DNA changes described above might have some deleterious consequences for the cell and for the organism. And indeed, in a worst case scenario, inactivating genes or rearranging chromosomes can have immediately lethal consequences. In other cases, the alteration of genes or their rearrangement may be less harmful (only slightly deleterious). However it looks as if a TE insertion usually has no dramatic impact if it occurs in dispensable/non-genic DNA for example (for instance in other TEs), which often corresponds to most of the genome. In rare cases, an insertion can even have beneficial effect, and so will ultimately become fixed.

However, transposable elements have tended to be known solely for their harmful mutagenic effects, which once raised the question of how they manage to survive despite natural selection. This implied that a genome with high fitness would be one with few TEs. But in fact this is rarely the case. First of all, we have to remember that transposable elements are the archetype of selfishness. Their only raison d'être is to amplify and perpetuate themselves in the genome. Encoding the ability to self-propagate within the genome is a simple but very powerful aspect of their selfishness. The "selfish genes" of Dawkins work in a much more complicated manner to propagate themselves in the population by exploiting sophisticated organismal "survival machines". When faced by threatening natural selection, it is far easier for a TE to duplicate itself than for a gene to do so. Second, when the genetic burden caused by TEs becomes too great, individuals or an entire population may become extinct. This may explain why many TEs are only found in moderate numbers of copies. Third, most TE insertions are in themselves probably neutral, as are most mutations, with some deleterious insertions that are ultimately eliminated, and occasional beneficial insertions that are eventually fixed. So, on average, the fitness cost of carrying TEs may be relatively limited.

So far we have only glimpsed the potential enormous positive impact of TEs on long-term genome evolution. However, this long-term benefit cannot be set against the short-term deleterious effects. Such a consideration resembles the sex paradox, where the benefits of sex (which generates genetic diversity) are visible in the long-term, but cannot offset the short-term, two-fold cost of sex compared to asexuality [200]. In both cases, the discrepancy in time scale is reinforced by a difference between the levels at which the effects act, at the individual level for short-term effect, and at the population or species level for long-term effect.

Genome-TE interactions are often viewed as an arms race in which each opponent successively devises fresh tricks to overcome the opponent's latest displays, resulting in tight co-evolution. TEs are genomic parasites, subjected to the fire of natural selection that may act directly on any insertion, or indirectly by favoring on the one hand a genome with good defenses, and on the other hand TEs that are able to tame themselves. The ultimate weapon developed by the genome is the impressive epigenetic defense system that does not destroy TEs but efficiently silences them, as can be observed in Arabidopsis [67]. Arms race is visible in the rapidity with which proteins involved in the defense system evolve, but in contrast, the rate of evolution of an element is difficult to infer. However the huge diversity of TEs suggests that this arms race does indeed exist. Of course, each time a TE escapes from epigenetic control, amplification bursts can occur (and indeed do occur from time to time). TEs can also escape control as a result from their ability to colonize new hosts after horizontal transfers.

TE dynamics are influenced by several parameters

The TE lifecycle

The emergence of TEs in a naive genome may have two origins. The first, and perhaps the most frequent origin, is the horizontal transmission (HT) of an active copy into the germ line. This phenomenon is frequent in prokaryotes, and the mechanisms of transfer are known (conjugation, transformation, and transfection). In eukaryotes, such transfers seem to occur far less frequently, and their mechanisms remain unknown. It is quite possible that one or several intermediates, including bacteria, viruses, or parasites, could be required [229]. However, whatever the mechanism, the TEs must reach the germline.

Comparisons of HT frequency show clearly that significant differences exist between the main superfamilies. Recently, Loreto et al. [229] estimated that among the 98 HTs described in Drosophila, 51% involve DNA transposons, 44% LTR retrotranposons and 5% non-LTR retrotransposons. Quantitative estimations cannot be provided for other species, but several cases of HT have been reported in mammals and tetrapods [230], in bdelloid rotifers [231], and in plants [232].

The alternative hypothesis of TE origin is the de novo emergence or re-emergence of autonomous sequences as a result of recombination between inactive copies. While there is less supporting evidence for this, it has been demonstrated that ectopic recombination between different copies of the same family or copies from different subfamilies can occur. For instance, it has been shown that some of the TEs described in yeast as Ty1/2 elements are in fact hybrids between Ty1 and Ty2 [233]. More recently, Sharma et al. [234] reported new elements resulting from repeated recombinations that may occur during the hybridization of sympatric species and polyploidization. Similarly, Marco and Marin [235] showed the emergence of a new Athila lineage as a result of recombination between distantly-related copies. In all these cases, this is not a de novo emergence; it is rather a re-emergence of autonomous copies from non-autonomous or dead copies.

As soon as a new element appears in a naive genome, it has to face a new challenge since there is generally a single copy, in a single individual, in a single population. To avoid being lost, this copy must invade the population and the genome. The transposition rate estimated from several natural populations, laboratory strains and for several types of elements is about 10^-4 transpositions/copy/generation. If we apply this rate to a newly arriving copy, this copy would almost systematically be lost. Therefore, two scenarios for a successful invasion have been proposed. First, either a high rate of new elements arriving by HT or recombination, or a high transposition rate of the initial copy i.e. close to 10^-1 or 1, according to the model prediction [236]. Of course, such a transposition rate cannot be maintained for long without risk to the population. Therefore, regulation of the transposition rate can be expected to occur rapidly. This could result from self-regulation by the TE or host regulation [203].

After the successful initial invasion of the genome and of the population, it becomes difficult to lose an element. Thus, it is important to follow the TE dynamics both in the genome and in the species. In most of the models published in the 1980s and 1990s, it was assumed that the copy number of an element had to reach an equilibrium (see [201]). However, most of these models failed to take the impact of mutation on TE activity into consideration. When this effect was included in the model, it could be shown that it is almost impossible to reach a long-term equilibrium, and several dynamic outcomes can be observed, including the loss of the active or trans-mobilizable copies, or the domestication of a copy.

TE competition and the ecology of the genome

With the exception of a few species, a genome does not normally contain only one type of TE. For a given family, several types of copies with differing levels of activity can be detected, including inactive copies. This is clearly demonstrated by analyses of a large number of genomes involved in sequencing projects. Since this situation is observed for almost all TEs, several questions arise: Is there any competition between different families, or between different types within the same family? Can an equilibrium resembling an Evolutionary Stable Strategy (ESS) be reached by these TEs in a genome? Can we apply models of population biology to the dynamics of TEs in a genome?

In the last few years, it has been assumed that the genome can be viewed as an ecosystem in which TE copies are considered as individual members of a species [202, 237, 238]. In such an analogy, autonomous and non-autonomous copies of the same family are competing entities rather than belonging to the same "species". In any case, the resources are produced by autonomous or truncated copies that have kept an intact ORF. These resources correspond to the transposition machinery like the transposase for the Class II elements, and can be used both by autonomous copies and by trans-mobilizable non-autonomous ones. Simulations and analytic models both provided TE cyclic dynamics due to the competition between active and non-autonomous copies, which are similar to the prey-predator dynamics described by Lotka and Volterra in population biology [[238], see Figure 3 and Additional file 1]. In such a context, non-autonomous copies can be viewed as parasites of autonomous copies, providing a nice illustration of how a genome can be viewed as an ecosystem. This cyclical pattern may be disrupted by changing any of the parameters of the system, leading occasionally to the loss of one or other of the elements. Hence TE interactions within a family should probably not be considered to constitute an ESS (Evolutionary Stable Strategy). Furthermore, it must be stressed that transposition bursts may occur [239–242], and that these sometimes reflect perturbations that can lead to long-term changes in TE content.

In addition, several families may coexist in a single genome. TE interactions have been poorly studied in cases where trans-family mobilization is not possible. However, it looks likely that if the genome is considered as an ecosystem, its "biotic capacity" is probably restricted. In other words, the expansion of a given family could have an impact on the dynamics of another family, reflecting a struggle for survival between TE families similar to that which occurs between species sharing the same ecological niche.

Reviewer 1

Jerzy Jurka, Genetic Information Research Institute

This reviewer provided no comments for publication

Reviewer 2

Jürgen Brosius, University of Muenster

This is a review on the impact of TEs on the evolution of genomes and genes including their regulation and epigenetic phenomena. Unlike many previous reviews, parts of the present one explore the (potential) significance of TEs from different angles and hence make it a worthwhile read. At the same time, the manuscript still carries the burden of past and present misunderstandings or ambiguities concerning TEs.

1) One of the difficulties to write a general review on all classes of TEs, is the fact that they are very different from each other, especially when comparing DNA transposons (class II) that usually operate in the cut and paste mode and class I TEs that operate in a copy and paste mode via an RNA transcript. Although this is discussed beginning on page 7, perhaps for the reader it would be easier to mention it at the onset. As an aside, few class I elements are transposABLE elements as, after integration into the genome most copies are not able to produce additional copies because they are dead on arrival. Especially, SINEs are rarely being transcribed because they lack the necessary flanking regions for autonomous transcription while LINEs are mostly truncated. An interesting exception is a small nucleolar RNA (snoRNA)-derived class of SINEs in Platypus that often maintains the ability to be co-expressed and processed into distinct RNAs when retroposed into introns of RNA polymerase II genes [1]. Nevertheless, in most cases a better description would be transposED elements for class I TEs.

Authors' response: In this manuscript, we used a classical terminology to describe transposable elements. "Transposable elements" are generally defined as sequences able to promote their own mobility and/or duplication in genomes, but in practice this definition largely extends to all TE-derived sequences, even if they have lost this autonomy. Consequently, non-autonomous copies (such as SINEs or MITEs) are generally considered as transposable elements (even if they are phylogenetically unrelated to the corresponding autonomous element), as well as totally inactive TE-derived pseudogenes, even if these are not actually "transposable". On the opposite, retroprocessed sequences that are not repeated (but just happened to be reverse-transcribed accidentally) are not considered as transposable (perhaps "transposed" here would fit). Although non-homologous, "transposable elements" and their derived sequences thus constitute an unambiguous group of sequences, characterized by their distribution in genomes (they are the middle repetitive fraction of the genome DNA), their evolutionary properties, and their "ability to transpose", where "transpose" stands for both "copy and paste" and "cut and paste" mechanisms.

Reviewer's response: Good point! Use "classical" for a term that is imprecise and not quite correct and you are off the hook. An admittedly extreme would be the "classical view" of the sun revolving around the earth. Reviews often address readers outside the field. Imprecise terminology leads to misconceptions that are difficult to purge. Hopefully readers will get as far as this section. Were I not familiar with this research topic, I would have been confused about much of the content concerning class I elements including the following statement from the background section (despite the attempt for clarification in the last sentence):

"1 - TEs are a major factor in evolution because they are an important source of variability. Mutations caused by TEs are diverse, ranging from small-scale nucleotide changes to large chromosome rearrangements, including epigenetic modifications. Although TEs are mobile, the nucleotide (or epigenetic) changes resulting from their transposition can persist, being transmitted through generations and through populations."

With respect to class I TEs, it is not the DNA that is moving but the RNA. RNA is transcribed from the DNA and then reverse transcribed and integrated. Neither master copy(ies) from which the RNA is originating nor the numerous integrated cDNAs do not move once integrated into their respective loci (except by genomic rearrangement as for any piece of DNA). If these class I TEs weren't absolutely sedentary, they could not be used as reliable phylogenetic markers as mentioned in this review under the heading "From mutations and (epi)-genetic variation to genetic novelty and adaptation". Actually, the first publication on the phylogenetic potential of retroposed elements (Alus) came from A. Dugaiczyk's laboratory: Ryan, S. C., and A. Dugaiczyk. 1989. Newly arisen DNA repeats in primate phylogeny. Proceedings of the National Academy of Sciences USA 86:9360-9364. N. Okada's laboratory perfected the approach to solve many interesting phylogenetic questions. This effort is summarized in the following review: Shedlock, A., K. Takahashi, and N. Okada. 2004. SINEs of speciation: tracking lineages with retroposons. Trends in Ecology and Evolution 19:545-553. Perhaps, part of the discrepancy stems from the DNA-centric majority view versus my RNA- centric view (see also comments 3 and 11).

Another point: I do not see how TE insertions cause small-scale nucleotide changes. Those changes are at least as large as the TE, i.e., at least 70 nt plus direct repeats.

Authors' response: If one considers mobility or transposability as the ability to excise, then we agree that Class I are neither mobile nor transposable. But if one considers mobility or transposability as the ability for a DNA segment to be inserted at a new position, then, Class I element are TEs. It seems that this last definition is more universally used. On what, in our opinion, should be considered as TEs, see comment 11.

Small-scale nucleotide changes occur when Class II elements excise. Usually the initial sequence is not exactly restored after double strand break repair. This is referred to as the excision footprint (added in the text).

2) The title of the manuscript already contains two terms that require qualification. First of all, non-autonomous TEs are not necessarily selfish. In theory, any RNA (including messenger RNAs) can be retroposed [2]. Some, however, are reverse transcribed and retroposed in a highly efficient manner by the machinery of autonomous TEs (e.g., LINEs) and hence give rise to thousands, up to a million copies in a genome. Of course, one could argue that some of the frequently retroposed RNAs do not have cellular functions any longer, but evolved structures that tricks the machinery of autonomous TEs into using them as templates [1]. Even though the RNAs might have lost (original) function, genes encoding such RNAs survive, not necessarily at its original genomic locus, but fortuitously due to the sheer number of copies generated. As a consequence, a few of them are bound to integrate into a genomic locus that permits autonomous transcription [3]. This is probably the case with Alu elements that survived in some form or another from the beginning of mammalian radiation up to now. Such a mode of persistence is documented, for example, by the presence of Alu subfamilies that were active at different times in primates. Of course, one cannot rule out a cellular function of some Alu RNAs at this juncture. The second problematic term is "architects" which implies foresight and planning. Perhaps, the term "agents" would be less ambiguous. In a similar vein, perhaps one should stay clear of the term "create" or similar (used elsewhere in the text).

Authors' response: In theory, there is indeed a stage where a non-selfish sequence (as e.g. the ancestor of Alu elements) starts, for some reason, to be amplified by an autonomous copy. However, genome-level selection will take place very rapidly, and among all amplified copies, the one that have a slight "superparasitism" advantage will become more frequent than the others, and the original ancestors will be outnumbered. Since there is no reason why the mutations favoring transposition would maintain the original cellular function, the probability for a sequence to be both selfish-DNA and "altruistic DNA" at the same time remains infinitesimal. The same infinitesimal (at the evolutionary scale) transition period exists when a functional copy inserts in a site where it brings some selective advantage, the copy being both potentially selfish and useful. A similar unstable stage probably also exists when species evolve from e.g. parasitism to symbiosis, but does not preclude an operational classification between two non-exclusive categories. As in remark #1, the issue is probably linked to the fact that "selfish", in the same way as "transposable", generally also qualifies derived sequences that are not by themselves "selfish" or "transposable", but exist because their direct ancestors were selfish and/or transposable.

Although the reviewer's remark about the use of the term "architect" is formally exact, we note that similar stylistic effects are common in the literature (Mattick (2001) "Non-coding RNAs: the architects of eukaryotic complexity" EMBO reports 2, 11, 986-991), and our feeling was that our "selfish architects" could not be understood in a different way than e.g. Dawkins' blind watchmaker. Potentially misleading occurrences of "create" were removed from the text, and we believe that this comment published along with the article will prevent misinterpretation of the title.

Reviewer's response: Concerning the infinitesimal probability for a sequence to continue to be both selfish-DNA and "altruistic DNA" at the same time, BC1 RNA is a counter example. It arose in a common ancestor of rodents via retroposition of a tRNA, has a function in the central nervous system and is the master copy of thousands of ID repetitive elements generated over long time periods. However, as the authors stated above, a few rare integrated copies that happened not to be transcriptionally silent, became master copies of additional sub-families of ID repeats [reviewed in ref. [2], given at the end of this section]. Once more, for class I TEs, it would be the RNA that is selfish, not the bookkeeping DNA [7], just as the RNA of an RNA virus would be selfish and not the integrated genomic DNA copy. On the other hand, DNA transposons (class II) might be considered selfish DNA.

Authors' response: This is indeed a nice counter example. Retroposition ability and cellular function may be both present because the gene is in a transitional stage before retrotransposition ability be lost, or because both reside on the same sequence in the gene (the non-tRNA part).

3) For most investigators, evolutionary considerations begin with the last common ancestor (LUCA) with RNA, protein and DNA already in place. A look at the RNA world and major evolutionary transitions [4], especially those from the RNP world to modern cells with DNA as bookkeeper, provides some scenarios to questions [5–8] such as: "Are we able to understand why they [TEs] are here, and why they are still here?". This also should qualify the statement, "the Central Dogma could not be questionable". See ref. [2], Figure 3.

Authors' response: the reference to the Central Dogma was indeed unnecessary here, and we have reformulated this sentence. In order to address a remark from reviewer 1, most open questions were reformulated, so that we could not directly refer here to the origin of DNA.

4) "TEs possess two main characteristics that distinguish them from classical genes...". One should remind the reader that some TEs are not genes. LINEs and LTR elements are more like small operons and thus harbour at least two genes. Furthermore, most SINEs or mRNA-derived retrocopies are not true genes but inactive pseudogenes (SINEs with extremely high copy numbers).

Authors' response: Of course, we wanted to refer to non-TE sequences. This sentence was changed into "... from other genomic components".

5) The sentence " the core of Darwin's theory was never really questioned" needs qualification. Perhaps, the authors mean that it was never questioned in the scientific community. Even that would be inaccurate, see refs. [9–12].

Authors' response: This sentence was indeed misleading, we meant that it was never successfully challenged. This was fixed in the revision.

6) There are earlier references (in addition to refs. [34, 35] concerning "TE as major actors of diversity" [13–15].

Authors' response: The Kidwell and Lisch (1997) reference seems well adapted here since they review the effects of all classes of TEs in both animals and plants. The second reference illustrates through several examples the involvement of epigenetics in TE-induced phenotypic variations.

7) "homologues of the three proteins involved in RNAi (ARG family, DICER and RdRP) can be found in all supergroups"

What is meant by "supergroups" major clades perhaps?

Authors' response: Eukaryotes are divided in 6 clades called supergroups (Rhizaria, Chromalveolates, Archaeplastidae, Opisthokonts. Amoebozoa and Excavates. The very same term is used in the cited reference, and elsewhere to refer to these 6 clades.

8) When discussing the CRISPR elements, it should be mentioned that the small RNAs were acquired from invaders, such as phages. The acquisition of these elements even resembles something akin to Lamarckism [16].

Authors' response: We gave a little more details on these very interesting CRISPR elements. Contrary to Koonin and Wolf 2009, we are however a bit reluctant to qualify this process as "Lamarckism" (Lamarck's theory, which was a general framework to explain evolution, cannot be validated by rare observations in which Darwinian evolution has led to a system superficially similar to Lamarck's wrong model of evolutionary change).

9) "First they [TEs] can bring sequences in regulating, coding or intronic regions. Those sequences may trigger useful functional changes (expression pattern, alternative splicing, transcription initiation and termination), by the presence of particular motifs or their physico-chemical properties [see [144] as a recent example]. Second, they can bring coding sequences, which modify the initial sequence and create new genes. Concerning "coding sequences" I do not see much difference between "first" and "second", once you bring TEs into coding sequences, they usually have to be coding or they would destroy the ORF.

Authors' response: Indeed, the second point was already included in the first, and has been removed.

10) "The full domestication is the most extreme case in which the totality of the coding region is used to ensure the new function." True, but one should find smaller contributions of fragments of TE-derived genes (this is mentioned only in the legend to Figure 2). For example as novel (alternative) exons oder contributing new termini to existing proteins, just as mRNA-derived retrocopies do [17].

Authors' response: This is what we meant by "extreme case". We refer to the less extreme cases two paragraphs later, "In numerous other cases, the domestication concerns only a part of the TE protein [...]". Here, the domain function is exapted. Smaller contributions are mentioned in the next part, which depicts exaptation of TE sequences (and not TE protein function). In some cases, exapted sequences become part of a coding region.

11) "While first examples of TE domestication and cooptation appeared as the exception (although of prime importance in regard to the function), the recent and numerous data prove that this is actually a recurrent phenomenon in genome history. Since the beginning, genomes regularly feed on TEs," and "TE and genome have been in constant contact since probably the beginning of life and such promiscuity has had repercussions on the evolution on both partners." Genomes ARE transposed (RNA) elements [5–8, 18].

Authors' response: In this paragraph, we refer to transposable elements, and not to other sequences retroprocessed accidentally. We think that "transposition" is too specific to be applied to any kind of reverse-transcription event.

Reviewer's response: There is not much difference between class I transposable elements (retroposons) and other retroprocessed sequences. Once more, the key to the difference lies in the properties of the RNA: some are more others less efficient templates for retroposition. Where do you draw the line: One hundred retrocopies of a tRNA are retropseudogenes and one thousand copies of a tRNA or tRNA-like RNA are SINEs?

Authors' response: The copy number is clearly not the good criterion to decide whether a sequence is a TE or not. The property to be efficiently retroposed is crucial, and must not depend on the environment, meaning that basically, RNA produced from any intact retroposed copy must keep the ability to be reinserted.

12) page 25, Exploiting TE sequences

A discussion about the persistence of exapted TEs in short evolutionary branches (gain and loss of exapted TEs e.g., in primates) [19] and long evolutionary branches (e.g., constitutive expression of exapted TEs in deep mammalian branches) [20] should be added.

Authors' response: This discussion on the long-term persistence of domesticated sequences is indeed interesting, and is now mentioned in the manuscript. However, it is also important to consider that there is no strong evidence that TE-derived exons behave in a different way than new coding sequences from different origins, and that this could simply reflect the "average" fate of genetic novelties in genomes.

Reviewer's response: Agreed, there should be no difference between TE-derived novel exons and those from anonymous genomic sequences [8], because even the latter are ancient TEs who are not discernible anymore, due to mutations over long time periods [6, 18]. Actually, most if not all genomic DNA is TE-derived, which would return us to evolutionary transitions following the RNA and RNP-worlds [5, 6].

13) Page 32, TE competition and ecology of the genome

For marsupials, Nilsson et al. could show an overlapping activity of RTE and LINE mobilized SINE elements along a single phylogenetic marsupial branch. The parallel activity of the two different retropositional systems was further supported by detecting frequent nested insertions of RTE in LINE mobilized elements and vice versa [21].

Authors' response: There is indeed no doubt that several TE families can be active simultaneously in genomes. Reciprocal transpositions in inserted copies is a strong piece of evidence that this was the case in the marsupial lineage, and such a coexpression is regularly observed in modern insect species. The missing information, however, remains the degree of interaction between these families: do they use the same resources, do they fight the same regulation mechanisms? So far, it is not clear whether co-invading TE families are competitors, commensals, or mutualists.

14) An additional earlier reference for the role of TE-derived genes in placenta formation should be cited [22].

Authors' response: The literature has been updated.

Reference:

[1] Schmitz J, Zemann A, Churakov G, Kuhl H, Grützner F, Reinhardt R, Brosius J Retroposed SNOfall--a mammalian-wide comparison of platypus snoRNAs. Genome Res 18:1005-10.

[2] Brosius J (1999) RNAs from all categories generate retrosequences that may be exapted as novel genes or regulatory elements. Gene 238:115-34.

[3] Brosius J (2003) The contribution of RNAs and retroposition to evolutionary novelties. Genetica 118:99-116.

[4] Szathmáry E, Smith JM (1995). The major evolutionary transitions. Nature 374: 227-32.

[5] Brosius J (2003) Gene duplication and other evolutionary strategies: from the RNA world to the future. J Struct Funct Genomics 3:1-17.

[6] Brosius J (1999) Transmutation of tRNA over time. Nat Genet 22:8-9.

[7] Brosius J (2005) Disparity, adaptation, exaptation, bookkeeping, and contingency at the genome level. Paleobiology 31(2 Suppl):1-16.

[8] Brosius J (2005) Echoes from the past--are we still in an RNP world? Cytogenetic Genome Res. 110: 8-24.

[9] Kellogg, Vernon L. (1907) Darwinism To-Day. A Discussion of Present-Day Criticism of the Darwinian Selection Theories, Together with a Brief Account of the Principal Other Proposed Auxiliary and Alternative Theories of Species-Forming" Henry Holt and Company, New York.

[10] Hull, David L. (1983) Darwin and his Critics. The Reception of Darwin's Theory of Evolution by the Scientific Community. University of Chicago Press, Chicago, ISBN 0-226-36046-6

[11] Mayr, Ernst (1991) One Long Argument. Charles Darwin and the Genesis of Modern Evolutionary Thought. Harvard Univiversity Press, Cambridge, ISBN 0-674-63905-7

[12] Woese CR (2004) A new biology for a new century. Microbiol Mol Biol Rev 68:173-86.

[13] Brosius J (2001) Retroposons--seeds of evolution. Science 251:753.

[14] Brosius J, Gould SJ (1992) On "genomenclature": a comprehensive (and respectful) taxonomy for pseudogenes and other "junk DNA". Proc Natl Acad Sci USA 89:10706-10.

[15] Brosius J, Tiedge H (1995) Reverse transcriptase: mediator of genomic plasticity. Virus Genes 11:163-79.

[16] Koonin EV, Wolf YI (2009) Is evolution Darwinian or/and Lamarckian? Biol Direct 4:42.

[17] Baertsch R, Diekhans M, Kent WJ, Haussler D, Brosius J (2008) Retrocopy contributions to the evolution of the human genome. BMC Genomics 9:466.

[18] Brosius J (2009) The fragmented gene. Ann NY Acad Sci 1178:186-93.

[19] Krull M, Brosius J, Schmitz J (2005) Alu-SINE exonization: en route to protein-coding function. Mol Biol Evol 22:1702-11.

[20] Krull M, Petrusma M, Makalowski W, Brosius J, Schmitz J (2007) Functional persistence of exonized mammalian-wide interspersed repeat elements (MIRs). Genome Res 17:1139-45.

[21] Nilsson MA, Churakov G, Sommer M, Van Tran N, Brosius J, Schmitz J (2010) Tracking marsupial evolution using archaic genomic retroposon insertions. PloS Biol 8:e1000436).

[22] Mi S, Lee X, Li X, Veldman GM, Finnerty H, Racie L, LaVallie E, Tang XY, Edouard P, Howes S, Keith JC Jr, McCoy JM. Syncytin is a captive retroviral envelope protein involved in human placental morphogenesis. Nature. 2000 Feb 17;403(6771):785-9.

Reviewer 3

I. King Jordan, School of Biology, Georgia Institute of Technology

In this manuscript, Hua-Van et al. present a fairly extensive review of the interactions between transposable elements and their host genomes. The review emphasizes the numerous ways that transposable element derived sequences have influenced the structure, function and evolution of genomes and tries to reconcile these influences with classical (neo-)Darwinian evolutionary theory. The review is distinguished by the fact that it deals with two perspectives on transposable elements that are usually treated separately: the impact of the transposable elements on their host genomes and the function and evolution of the elements themselves. This paper makes a nice contribution to the field of transposable element biology and also fits well with the recent series of papers that Biology Direct has published dealing with current perspectives on Darwinian evolutionary theory.

Much of what is covered in this review has been treated elsewhere previously. Nevertheless, it is both timely and useful to have much of this material presented together in an evolutionary framework. Some of the newest and most relevant material is on the relationship between transposable elements, RNA interference and epigenetic phenomena. From my admittedly biased perspective, this represents the single most important contribution of this review. But this is an area of investigation that is changing rapidly, and I would urge the authors to consult some of the most recent literature on transposable elements and epigenetics to deepen this part of the manuscript. With all apologies for being self-serving, our own lab has recently published a couple of reviews on these topics: (Jordan IK and Miller WJ 2009 Genome defense against transposable elements and the origins of regulatory RNA in Genome Dynamics and Stability Lankenau and Volff (Eds) 4: 77-94 and Huda A and Jordan IK 2009 Epigenetic regulation of mammalian genomes by transposable elements in Ann NY Acad Sci 1178: 276-284). In addition, we have also recently shown that transposable element mediated epigenetic effects on host genomes may not be confined to repressive epigenetic modifications, as emphasized in this review, but also by activating modifications that are recruited to transposable elements in the vicinities of host genes (Huda A et al 2010 Epigenetic histone modifications of human transposable elements: genome defense versus exaptation in Mobile DNA 1:2). There are a couple of other recent papers that are directly related to this topic - and this list is by no means exhaustive - that the authors may wish to have a look at (Rebollo R et al. 2010 Jumping genes and epigenetics: towards new species in Gene 454: 1-7 and Lisch D 2009 Epigenetic regulation of transposable elements in plants Annu Rev Plant Biol. 2009;60:43-66).

Authors' response: This part has been reorganized to integrate this aspect and update citations.

I agree strongly with the authors' sentiment that transposable elements play critical roles in genome structure, function and evolution. However, some caution is warranted in order to avoid overstating the case. For example, the statement in the abstract that "...since Darwin's theory, transposable elements are maybe the discovery that has changed the most our vision of (genome) evolution." is somewhat overwrought considering that Darwin lacked even the most basic concept of the molecular mechanisms of heredity or any notion whatsoever of what constituted a genome. Indeed, the authors point this very fact out in several places in the manuscript. Thus, they may wish to be more circumspect when placing the impact of transposable elements into the context of evolutionary theory and genome evolution as a whole.

Authors' response: The concerned sentences have been reformulated

The statement in the introduction that "the core of Darwin's theory was never really questioned." (page 4) is factually inaccurate. The core of this theory has been, and continues to be, continually questioned at a fundamental level. It may be more accurate to state that the core of theory has never been successfully challenged or over-turned.

Authors' response: We agree with this remark and changed the sentence accordingly.

The authors' imply that biologists were reluctant to accept McClintock's discovery of transposable elements because it did not fit with the 'Central Dogma' (Introduction page 4). But the Central Dogma is a concept from molecular biology that came later, and while the discovery of mobile genetic elements made by McClintock clearly challenged prevailing ideas about how static the genome was, it did not directly address or contradict the Central Dogma. Further on in the same section the Central Dogma is referred to as depicting 'the genome as a linear succession of genes'. Again, the linear 'beads-on-a-string' concept of a static genome is distinct from the Central Dogma.

Authors' response: The confusion between the central dogma and the static genome dogma has been clarified.

The authors point out an important concept that the evolutionary dynamics of transposable elements occur at two levels: intra-populational, based on the competition between individual organisms, as is the case for static host genes, and intra-genomic based on the competition between individual element copies. This is indeed a critical aspect of transposable element evolution that impacts how the elements affect their host genomes. However, they then go on to posit a third conceptually distinct level based on horizontal transfer. It is well known that elements may be particularly prone to horizontal transfer between species, but it is not clear how and whether this phenomenon entails a third distinct level of transposable element evolutionary dynamics.

Authors' response: This third level become apparent only when an analogy with an ecological concept is considered, which was not clearly stated. The intra-genomic competition may be compared to competition between individuals for the same resource in a unique ecological niche. A TE family in one genome corresponds then to a population. The intra-populational level represents a metapopulation in which TE populations mix by a kind of migration process triggered by sex. By analogy, horizontal escape toward a new genome can be viewed as new ecological niche colonization and represents the extreme case of migration with foundation of a new isolated population and ultimately allopatric speciation. (In comparison a static host gene (allele) will not use the intra-genomic level to expand). We agree that the ability to transfer horizontally does not impact the TE dynamics at the species level, but provides only new seeds for TE expansion in the living world as a whole. The idea has been reformulated, hopefully with more clarity.

The authors often refer to the conflicting, and seemingly dichotomous, notions of transposable elements as genomic parasites versus the creative or adaptive contributions that the elements make to their host genomes throughout the manuscript. However, these two concepts are not mutually exclusive. In a very nice review on this topic (Kidwell MG and Lisch D 2001 Perspective: transposable elements, parasitic DNA, and genome evolution in Evolution 55:1-24), the authors nicely lay out the idea that transposable elements do not exclusively occupy extreme positions on either end of this dichotomy. They hold, rather, that transposable elements can best be considered as occupying a variety of positions on a dynamic continuum from extreme parasitism to obligate mutualism. This kind of more nuanced perspective would add nicely to the evolutionary role of transposable elements presented here.

Authors' response: We agree with the view of Kidwell and Lisch. However we cannot deny that TE are intrinsically selfish, which was our starting point to further pinpoint facts that actually support other TE-host relationships. The concerned parts have been modified to erase the impression of too clear-cut views.

Apparently the manuscript was written by a non-native English speaker and it has numerous grammatical errors. The authors should proof the manuscript closely for these and other language related issues or enlist a native English speaking colleague to help with this. For instance, the last sentence in the abstract that reads 'The review attaches to explore ...' does not make sense. The presence of many errors of this kind has the unfortunate effect of obscuring the important message of the manuscript as well as the authors' unique perspective on transposable elements.

Authors' response: The revised text was corrected by a native English speaker.