Supporting Information

This page provides information/entry points to these topics:

Selfish Transposons

The selfish paradigm for transposons was originated by Doolittle and Sapienza (1980) and Orgel and Crick (1980). The idea is that the increase in copy number resulting from transposition can maintain the transposon in the genome instead of (or perhaps regardless of) the selective pressure for function that maintains ordinary genes. In its purest form, the selfish transposon has no function except to replicate itself, and is under no selective pressure except to propagate itself. Propagation for a transposon is limited to passage through the germline. Although this is a slow process, newly inserted copies of the transposon can eventually spread through a population over many generations of the host.

Selfish Transposons and Host Function

An obvious example of transposons that do benefit their host are the bacterial transposons carrying antibiotic resistance genes. In this case the relationship among host, transposons, and conjugating plasmids is so intimate that the system can be considered symbiotic. Another example is that some LINE-like transposons in Drosophila, HeT-A and TART, have become involved in extending the host's telomeres (Pardue et al, 1994).

So if a transposon provides a benefit to the host, does that invalidate the concept of it being selfish?

At the most basic level of description, "selfish" is a kind of selection acting on the transposon, rather than a description of the transposon itself. Selection for host benefit is a different pressure that can be acting on the transposon at the same time. Other properties attributed to a "selfish" transposon tend to emerge out of the host transposon interaction. See emergent properties of transposon-host systems.

A key to understanding the balance of selective forces on a transposon is to know its expected lifetime in the absence of providing a host function, and its transposition frequency. If each transposon reproduces itself (transposes) within its expected lifetime, additional selection for host function will not be required to maintain it. If it cannot reproduce itself within its expected lifetime, then it can only persist by extending its lifetime in some way. One such way might be to provide a benefit to the host and have its lifetime extended by positive selection. However, there are some limitations to the application of this idea as follows:

Host Function and Transposition Tend to Dissociate

If a family of transposons does not transpose fast enough to survive as a purely selfish element, selfish selection will still operate to select more aggressively transposing varieties. Hence the need for alternative forms of conservative selection will tend to dissipate. If nontransposing variants that still carry out the host function can arise, then they will become normal host genes and eliminate host dependence on the transposon. When transposons make multiple copies, selection is relieved on individual copies to carry out the host function. So normally one should expect symbiotic systems to devolve into a stable host gene and transposons that do not carry out host function. For example, the Het-A and TART transposons mentioned above no longer jump to sites other than telomeres, so they have lost the general properties of a transposon and are becoming specialized as host genes. Another example is the Fv1 retroviral resistance gene that was derived from the gag gene of a retroviral-like sequence family (Benit et al. 1997), and has subsequently lost its mobility.

So I'd claim that there is a limit to how much stability a transposon can derive from providing a host benefit. If the beneficial function is actually required often enough for the host to conserve a gene for that function alone, then such a gene will usually be derived from the transposon and the functions will dissociate. A transposon might, however, derive some stability from providing functions that are needed less often by the host, such that it can not conserve its own gene for the purpose.

None-the-less bacterial transposons and resistance genes provide a powerful example that symbiotic relationships can exist. An important aspect of the bacterial system is the three-way interaction between bacterial genomes, transposons, and horizontally transferring plasmids.

Transposons as a source of genetic variation

It has been argued that transposons provide a selective advantage to the host by disrupting the occasional host gene that has become a liability to the host and by promoting other useful rearrangements (Chao et al., 1983; Nevers & Saedler, 1977; Syvanen, 1984). Starting with the classic paper by Britten and Davidson (1969) and extending to this day (Britten, 1998), it is often argued that transposons (in particular, including LINE-1) provide a benefit to the host by distributing promotor, enhancer, or other cis-acting sequences around the genome. These are more properly called "effects" rather than "functions", at least if we reserve the word "function" for a positive action that is sufficient to conserve the sequence in the absence of any other positive selective forces.

Consideration of transposon activity as a source of beneficial genetic variation is complicated by the effect of isogenization on transposon activity. Ie. inbreeding causes activation of transposons. If it is true that founder effects in the establishment of new species are sufficiently powerful to have the same effect, then transposons will exhibit an upswing in activity right at the time when selective pressure is highest to modify or even destroy prior functions, at right at a crucial time for group selection between species that have transposons and species that do not. So the impact of transposons on fitness may not be adequately modelled if we only consider transposons in large extant populations.

Transposon activation by isogenization

Inbreeding leads to transposon activation, at least in Drosophila and maize. I am writing this as a summary of presentations at the 18th International Congress of Genetics by Sergey Nuzhdin, Peter A. Peterson, and V.A. Ratner. I will be adding literature references in the near future. It is apparently a long-standing observation that inbreeding of maize causes activation of transposons. I am reminded that someone once told me that McClintok's discovery of the transposon in corn came about because she inbred her strains. Her assignment of the term "genomic stress" was tied to a peculiar chromosomal instability (breakage bridge fusion) that was under investigation in this strain. She introduced the idea that the corn was under genomic stress, and that genetic variability was induced as a means to increase the rate of adaptation to this situation. Apparently, in long-standing strains of maize, transposons have become the source of regenerated variation which is otherwise declining due to the inbreeding effect.

Sergey Nuzhdin expressed this phenomena in the following terms. He believes that the host strains are polymorphic at many loci that restrict activity of specific transposons. Inbreeding fixes the permissive alleles of some of these as a founder effect leading to the activation of specific transposon families in different inbred strains. In both maize and Drosophila, impressive activations are seen (up to 10-1 - 10-2 transpositions per host generation.

Incorporation of this phenomenon raises a number of points:

  • Someone should screen inbred mouse strains for LINE-1 activation.
  • The question of whether natural founder effects are typically high enough to cause transposon activation is raised.
  • Risk of insertional mutagenesis would be unevenly distributed within a population according to genetic susceptibility.
  • The question of why the resistant alleles do not fix is raised. In particular, is there balancing selection? Ie. is the permissive allele favored in the absence of pressure from transposons?
  • Can the transposition-selection model be reformulated as a transposition-host resistance gene balance, and should such a balance be an equilibrium or a fluctuating system. Does the existence of a shifting balance of host resistance allow for mechanisms of negative selection to bring about an equilibrium that could not meet the criterion of a concave decreasing function?
  • How does this influence containment of transposon variants. Is it true that genomes have the means to crush any transposon if all resistant alleles were brought into play at once. In other words, is this like a QTL system where strong artificial selection will nearly always uncover a large capacity of a population to adapt by combining existing polymorphic alleles, even though the traits are such that one might have expected natural selection to have optimized them in the first place.
  • How many of the resistance genes are defective interfering transposons, as opposed to intrinsic host functions?

    Run-away Transposition

    Intuitively, it seems that introduction of an aggressively replicating transposon would lead to an uncontrolled exponential increase in transposon copy number until the host species was in danger of being overcome by the accumulated negative effects. Is this a realistic scenario? Transposons that are well characterized generally are relatively tame, and examples of a species under serious stress from its transposons are not available. So is there a flaw in the intuitive concept?

    As an example of a hypothetical run-away transposon, consider what will happen if a variant rogue transposon arises that makes new copies at 10-1 in a diploid genome. Its original frequency in the population is 1/2N where N is the census population size. If the population is not changing in size, then the frequency of the original rogue locus will stay at about this level in the short term (say 100 generations) with some stochastic likelihood of loss and some likelihood of drifting slightly upwards. Ignoring those cases where the locus is lost early, the output of new copies will increase the total density in the population by 1.1 per generation. The amplification factor per number of generations is:

    Gen. #Ampl. factor
    1001.4x104
    5005x1020
    10002.5x1041

    The exponential buildup quickly outstrips N such that there are now more than one rogue transposon per genome. Then, if not somehow checked, the number continues to rise exponentially to ridiculous levels. Once the rogue reaches a density in the genome such that essentially all progeny have one new insertion per generation, then there is no negative selection that can stop the process. That is because there will be no individuals with low copy number to select over those with increased copy numbers. Clearly organisms will quickly reach a state where they can not handle the load of transposons. If insertional mutagenesis does not overwhelm them, then the pure mass of excess DNA must eventually become unmanageable. The entire host species would be placed under stress at once.

    The equilibrium theory of transposon population genetic does not allow for this scenario. The equilibrium theory starts with the description:

    delta_n = n @ ln wn /@n + n (u-v)

    u = insertion rate per element, n = equilibrium copy number v = excision rate per element wn = is a function describing a loss of fitness with n that is presumed to become more severe as n increases. A more aggressive variant can be modelled by plugging in a higher value of u. What happens is that a more aggressive variant elicits a sufficiently severe loss of fitness that it just comes to equilibrium at a higher n. Secondly, in the presence of both tame and aggressive transposons, the concave decreasing fitness function will actually provide a positive selective coefficient for the tame transposon to accumulate at the expense of the aggressive one. In other words, the same form of fitness function that is hypothesized to contain n for any given u, will also eliminate more aggressive variants above some limit in u.

    There are a number of holes in the above:

    So whether or not run-away transposition puts constraints on the host-transposon relationship beyond what can be investigated around the equilibrium point is not clear to me. A line of investigation that suggests that thinking about high levels of transposition is not just a distraction is the observation of transposon activation by isogenization. These experiments imply that transposons are intrinsically capable of jumping at high frequencies and that host genomes already contain alleles at resistance loci that are capable of containing the transposons.

    Normal Transposition Frequencies.

    The frequency of transposition from natural transposons is remarkably low. A survey of Drosophila transposons of all descriptions shows that individual transposons typically jump at a rate of about 10-4 to 10-5 per host generation (reviewed Charlesworth and Langley, 1989). The average transposon family has about 10 members per genome, thus causing about 10-3 to 10-4 new inserts per fly. This is thought to represent an equilibrium value. By comparison, mouse LINE-1 makes about 1 new insert per 50 new born mice. However, 90% of these are defective due to truncation, and another half or so should be defective due to reverse transcription errors. So the total output of active copies is similarly about 10-3 per mouse. The total number of active copies per genome is also thought to be on the order or 3 to 10 active copies per genome, so the core of active LINE-1 elements may bear quantitative similarities to Drosophila transposons.

    The Transposition-Selection Equilibrium Theory

    In order for there to be an equilibrium, there would have to be an opposing negative selection against more aggressively transposing elements. The way the system is usually described is to contain the copy number of a transposon with a fixed transposition frequency, however the same treatment also extends to containing more aggressive elements. Charlesworth and Charlesworth (1983) describe an equilibrium that may pertain to transposons in Drosophila. Drosophila transposons differ from mammalian LINE-1 in that the allele frequency of any particular insert is vanishingly small. These authors describe how a very small level of negative selection against the presence of each new insert in combination with the observed insertion and excision frequencies would put the system into an equilibrium. This negative selection was variously attributed to alteration of gene function for genes near the insertion sites or to a process called ectopic exchange.

    Mathematical description of Drosophila equilibrium

    For the longest time I thought that the transposition-selection equilibrium theory couldn't apply to LINE-1 because LINE-1 has high allele frequencies. But now I think that it does apply and that the high allele frequencies are mainly a natural consequence of the relatively small effective population size of humans. See presentation to the 18th ICG on application of transposition-selection equilibrium theory to LINE-1.

    Models that are more general than equilibrium

    In a larger sense, an equilibrium point should be an attractor in the space defined all possible variations of the transposon and the host. In other words, small variations in the activity of the transposon should be countered by the host either automatically or through adaptive evolution to restore equilibrium. Similarly, small increases in the pressure of the host brought against the transposon should be countered by the transposon either automatically or through adaptive variations. In this broader sense there may be fluctuations in the transposon copy number from time to time, but always tending towards a stable plateau.

    The expectation for the system to be at an equilibrium is derived from the simplest tenant of complexity theory. That is that systems not at an attractor tend to move towards an attractor; whereas systems at an attractor tend to stay at an attractor. Hence most of the time when you look at a system, it should be at an attractor.

    There are only a few general forms of attractors available in principle for the transposon/host systems:

    A consideration of long term evolution of host-parasite systems would suggest that cyclical expansion and contraction of transposon number should be the natural form of the interaction. "Cyclical" in this case is not meant to imply regularity, but rather fluctuations where increased replication is always favored by selfish selection and this is opposed by selection for host resistance whenever the transposons are getting out of hand. The exponential form of run-away transposition would tend to favor explosive fluctuations, whereas the incidences of failure to adapt leading to loss of the transposon or even of the host would tend to sift out systems that damped the oscillations over the long term. In this view, equilibrium and near-equilibrium would appear often in systems actually undergoing mild fluctuations over time. See emergent properties in host-transposon systems.

    In addition, if different transposon families interact, then there can be chaotic periodic shifts in the fortune of one family at the expense of others, with the overall envelope of all transposons taken together obeying one of the dynamics listed above.

    Lifetime of active LINE-1 loci

    The minimum transposition rate for a stable system requires that on average each active transposon produces an active insert before it is lost from the genome. In the full blown transposition-selection balance theory, active allele lifetime is not usually calculated, and if it were it would be a function of drift and negative selection. We have used the concept of a lifetime which corresponds to a pure drift model to characterize LINE-1. In the context of the larger theory, this corresponds to calculating a maximum limit based on using only one of the processes (base substitution) leading to loss of active loci. However, I propose that base substitution is a sufficiently major component of the loss of active LINE-1 alleles that this limit is also a reasonable estimator of the overall average. See Application of transposition-selection equilibrium theory to LINE-1.

    For LINE-1, we estimate the lifetime as the time a full length LINE-1 can be expected to survive damage from damaging base substitutions under a neutral drift scenario. The fraction of base substitutions that damage function of LINE-1 is estimated from the ratio of nonsynonymous:synonymous differences between known active sequences. A number of studies have set this ratio at about 1:1 (Loeb et al. 1986; Hardies et al., 1986). This contrasts with a ratio of available sites for nonsynonymous and synonymous mutations of 3.2:1. In combination, this means that natural selection culls out about 1/2 of all base substitutions that affect the reading frames of LINE-1, or about 2/3 of all amino acid replacements.

    So then, out of a total of about 5000 bp of reading frame, L1 has a vulnerable target equivalent to about 2500 bp. The drift rate in mouse is about 1%/Myr, so the expected lifetime of a mouse L1 locus is about 40,000 years before it will become damaged. The drift rate in humans is about 0.1%/Myr so the estimated lifetime of a human L1 locus is about 400,000 years.

    This already sets up an interesting contrast between mouse and human LINE-1 families. In human it is reasonable to expect a single active LINE-1 locus to have persisted for the entire duration of the modern species, and to have reached fixation in the genome. In mouse the same statements are not reasonable, at least not without some large influence not represented in the minimal model. Activity should be expected to transfer from one locus to another repeatedly during the evolution of the mouse. Further, each active murine L1 locus should not be expected to reach fixation (at least not in an active form). So one might expect geographical variation in which loci (and hence which sequence variations) are most actively amplifying. Finally, factors that influence the allele frequency reached by active LINE-1 alleles at each locus may have a more direct effect on the total output of that locus across the species in mouse than in humans.

    Mechanisms of negative selection

    The transposition-selection balance equilibrium model proposes a concave decreasing fitness function of total copy number. This was explained as meaning "that the adverse effect of an additional copy on fitness increases with the number of copies already present". In early papers by Charlesworth and/or Langley, these authors presented as an alternative that the transposition rate might be a decreasing function of the copy number. However, later they take the view that the fitness function must have this form as a causative agent of the equilibrium, and that the transposon might regulate itself to lower activity at higher copy number if it improved its survival in the face of the prevailing fitness function.

    The negative selection is therefore attributed to a small likelihood of mischief created by the presence of the elements. The mechanistic example given is that of DNA transposons in Drosophila that cause rearrangements in trans with other copies of the same family. Hence the loss of fitness (fraction of seriously damaged genomes) is increased both by the intrinsic activity of the transposon and the number of other copies in the genome with which it can interact. A number of points about this are worth considering.

    1. First, simple insertional mutagenesis could be the end result that delivers the negative selection. To get the concave curve, it would be necessary that each new insertion not only made insertions itself, but also increased the mutagenic potential from other templates in trans. The selective coefficient in a diploid genome would be small, consisting of the fraction of insertion sites that were exposed in the hemizygous male and vulnerable to generating sterile/lethal phenotypes, and the fraction of similar autosomal sites, that already carried one defective allele. However, insertional mutagenesis is not favored as the relevant mechanism in Drosophila, because the enrichment for transposons on autosomes that would be expected in not generally observed. There is however an excess of inserts on autosomes for some specific transposon families. (Montgomery and Langley, 1983).

      Without a trans-mobilization factor, simple insertional mutagenesis doesn't have the prerequisite concave curve shape described above but is more like a threshold function. See insertional mutagenesis below. Another issue is whether the mutagenic effect of LINE-1 is augmented by a series of mechanisms that produce non productive inserts (truncation, trans-mobilization of defective LINE-1 templates, and trans-mobilization of SINEs and other retroposons).

    2. Another issue to note is that the trans interaction of the DNA transposons doesn't transfer easily to the retrotransposons (since they use RNA as template instead of DNA). None-the-less, the retrotransposons of Drosophila seem to follow roughly the same dynamics as do the DNA transposons. One could propose that the RNA concentration in the relevant cells changes in proportion to the copy number, and try to regenerate a similar model in which the different RNAs are interacted with in trans by the transposition proteins. In this case, the relevant outcome is just to make more insertions, although there is the prospect of creating inserts from RNAs that otherwise would be unable to produce functional enzymes.

    3. Looking at the evolutionary history of LINE-1 has yet to reveal any substantial families of defective sequence. I've tended to take this to mean that there isn't much trans-mobilization and that L1 has adapted to avoid this process. However, significant amounts of transmobilization could hide if it is distributed among many templates, and we really need a way to measure this process. There are now experiments that observe LINE-1 trans-mobilization in an in culture system at a very low rate (Moran et al., 1996). These experiments weren't specifically controlled to determine the rate of trans-mobilization to cis-mobilization, and such a measurement could provide an estimate to work into the developing theory.

    4. There is a hypothesis that LINE-1 enzymes trans mobilize the SINEs, and other pseudogenes. It is unclear why these enzymes would seem cis-acting with respect to LINE-1 RNA, but trans-mobilize other RNAs. Perhaps it just works out in the numbers that a very low rate of trans-mobilization is amplified by the large amount of these other RNAs. However, a more intriguing idea is that the SINEs are generated from decoy templates flooded into the germline cells by the host as a means of competing with and inhibiting LINE-1 replication. I would note that if LINE-1 has adapted to acquire a cis-acting replicative mechanism, it will have to associate its protein with the RNA before it gets away from the ribosome. The templates for SINE amplification are also RNAs that go to the ribosome, so this might support a decoy mechanism.

    5. Another example of a mechanism that could provide the prerequisite selection is ectopic exchange (Langley et al., 1988). In this case the repetitive sequences are proposed to cause loss of fitness through a small frequency of unequal exchange producing gametes with inviable deletions. This, of course, would work much the same for DNA and RNA transposons, since it has nothing to do with the nature of their transpositional mechanism. This brings us to the idea of a composite selective mechanism. DNA transposons would presumably experience both forms of selection simultaneously. Also it bring us to the idea that the components of the negative selection function needn't be the same for different transposons, even in the same host.

    6. Thinking in terms of different mechanisms contributing to the restraint of the transposon, we have to face up to what is happening to the simple equilibrium brought about by fitness being a decreasing concave function of n. For each mechanism discussed above n is really something different. For the trans interaction of DNA transposons, defective copies count as long as they can be a target for the activity of the transposase. The math that has been done considers all the transposons to be of the same functional quality, but in reality there are defective copies that are only mobilizable in trans with a fully active element. Presumably, a similar equilibrium can be worked out with an accounting for the two different classes. For ectopic exchange, n counts any elements that can recombine, whether they have any activity or not. For insertional mutagenesis, fully active and transmobilizable defective elements must be treated differently, and I'm suggesting that active elements should be subdivided as a function of activity.

    7. So for LINE-1, we have a major subdivision problem to sort out. There are over 100,000 elements that might in principle participate in ectopic exchange, although there is a suggestion from Drosophila that only unfixed copies are a target for ectopic exchange. Most of the 100,000 are not able to make proteins, and many are not transcribed. So the 100,000 copy number is probably not a relevant number. There is a copy number of active elements that is somewhere between 3 and 50/genome. Then one can imagine that there are more aggressive elements that are there in low allele frequency ala Drosophila. There is currently no information to judge that, except that novel inserts that have been attributed to actual parental elements have identified active LINE-1s that are at high allele frequency, not ones that might belong to this hypothetical low frequency class. So if rogue LINE-1s exist at low allele frequency, they can not be responsible for the majority of LINE-1 inserts. And then finally the presence of amplifying non LINE-1 templates may be a factor. In many cases, it would be the amount of RNA that mattered, and so variation in expression from different DNA loci would come into play.

      Auto-elimination

      In consideration of restraint against extreme rogue elements, I note that protection of the parental locus by segregation from its progeny breaks down at transposition frequencies greater than about 1 new insert per generation. At these rates, the number of new inserts is high enough that the parental locus ends up nearly always co-segregating with some of them. That leads to a form of negative selection for which I would like to introduce the term "auto-elimination". The point of this is that whatever mechanism(s) create the fitness function, it doesn't have to extend to infinitely high transposition rates. There is an upper limit above which it is simply not feasible for the transposon to go.

      Insertional Mutagenesis

      A treatment incorporating adding the incidence of producing sterile and lethal defects to the other selective forces thought to be acting in Drosophila suggests that such defects would begin to exert a restraining effect at a level of 3.7% (Charlesworth and Langley, 1986). That treatment doesn't consider what will happen if the transposon makes large numbers of nonproductive inserts. Non productive inserts will tend to exclude transposons when the number of sterile/lethal mutations produced exceeds the number of active transposons produced.

      In diploid organisms, most genes are protected from immediate lethal effects of insertional mutagenesis by complementation. Many higher eucaryotes in addition have very non-compact genomes with lots of safe places for transposon insertion. None-the-less there should be a negative selection of somewhere between 0.1% and 1% against the activity of a transposon locus based on disruption of X-linked genes and production of dominant and semi-dominant effects. Mechanisms that force the transposon to make nonproductive inserts (such as defective copies of itself, or copies of extraneous templates) would amplify the effect of insertional mutagenesis. Problematically, insertional mutagenesis scales with the transposon activity such that at any activity it is either enough to cut off run-away transposition or it is not. Hence there is a threshold of intensity of insertional mutagenesis below which it fails to contain transposons of any activity, and above which it would appear to condemn transposons of any activity. To reach that threshold in mammalian genomes would require an amplifying effect in the order of hundreds-fold to thousands-fold. If there is significant insertional mutagenesis, but it is under that threshold, then it becomes a problem to understand if and how it might interact with other modes of containment such that the combination would be effective containment.

      Directly eliminating or inactivating Transposons

      Yoder et al. (1997) and Bird (1997) have debated whether or not inactivation by methylation may be an important part of transposon containment. Excision is thought to be an important component of confinement of transposons to low allele frequency. However, either of these acting in isolation simply shorten the lifetime of the transposon and hence just reset its equilibrium point at a higher transposition rate. Neither acting alone particularly does anything to eliminate transposon variants that exceed the equilibrium transposition rate.

      However, inactivation or excision of transposons has a useful interaction with insertional mutagenesis. Imagine that 90% of a transposon's progeny were inactivated by methylation. That would increase the number of insertions it must make 10x in order to make a single active progeny. That in turn would increase the negative selection derived from insertional mutagenesis 10x. Excision would have a similar effect as long as it was imprecise and left the affected gene damaged.

      Mammalian LINE-1 makes 10 truncated inserts for each full length one, and another fraction of the full length inserts are probably initially defective due to reverse transcription errors. So LINE-1 apparently experiences an increased insertional mutagenesis by a factor greater than 10x (out best estimate would be 20x from truncation and reverse transcription error). Although it has usually been considered that LINE-1 truncation is due to its own insufficiencies, we are now led to suggest that it may be due to host interference in the transposition mechanism.

      Host Resistance Genes

      Transposons that repress themselves in trans are vulnerable to the generation of defective copies that become host resistance genes. The P element has generated an accumulation of defective copies throughout the Drosophila population world-wide that repress the transposition of active elements (reviewed by Engels, 1989). The Fv1 retroviral host resistance gene is derived from the gag gene of a defective retrovirus-like repetitive sequence (Benit et al. 1997).

      Selfishness by Extortion

      Another kind of selfish gene has been described where the gene product is deleterious when expressed in heterozygotes but not in homozygotes (Medea; see Wade and Beeman, 1994). This kind of selfishness (called "maternal-effect") creates a selective pressure for the gene by killing off the progeny of individuals that are homozygous for the null allele.

      Maternal-effect at first seems unrelated to the selfishness of a transposon, but is it? Consider a P-element-like method of trans-repression (Engels, 1989; Misra and Rio, 1990; Engels; 1992; Misra et al., 1993; ). In addition to the usual advantages attributed to this form of self-regulation, the maternal effect preferentially inflicts the mutagenic burden of transposition disproportionately on the progeny of flies that didn't have P element (assuming a population that mostly has P elements to start with). In principle, any mechanism of transposon regulation that exhibits trans repression at high copy number and activation of transposition at a lower copy number is capable of generating a selfish selective pressure even if it isn't arranged as a maternal effect. I would like to introduce the term "extortion" to describe this phenomenon. Although the examination of transposon regulation by Langley and Charlesworth does incorporate something like this in the way they provide for negative selection against rising copy number, it is not clear to me if this effect is adequately modelled. Intuitively, regulated transposons after achieving a high degree of penetrance into a population might exhibit a loss of selective restraint on the high end of their dynamic range of activity. This might distort the equilibrium description in significant ways.

      Endogenous retroviruses may constitute a useful comparison to traditional transposons. The mouse immune system is known to suppress the replication of endogenous retroviruses within the germline by 350 fold from 1/10 to 1/3500 (reviewed by Coffin, 1996). [!? Summed over how many progenitor proviruses?] This is presumably triggered by the blood borne stage of the life cycle and so the specific mechanism wouldn't apply to ordinary transposons. However, the numerology is comparable to the run-away transposon problem. The lifetime of the provirus should be on the order of 64,000 generations [Similar amount of coding sequence as for LINE-1, but the selective pressure is higher. Whereas, LINE-1 allows 1/3 of nonsynonymous changes, retroviruses allow only about 5% of them (at least in the major conserved regions). So 80,000/1.25= 64,000. This uses 2 generations/yr.] so the inhibited condition is not far away from the transposon equilibrium point [20x]. There is also known to be an inhibitory contribution from methylation. [If methylation permanently shuts down 95% of proviruses, then this would be an equilibrium.] Without the immune system the provirus is in the run-away zone. So here is a system that appears to have evolved towards its equilibrium point after taking into account the specific inhibitory system arrayed against it.

      | Introduction| | Projects| |Publications | | Members|

      | LINE-1 Theory| | Datasets| | References| | Other sites|

      To Home(for no frame version)