selection on transposable element families

Tue Sep 19 09:27:14 EST 1995

David Witherspoon writes:

> 	I am studying a family of transposable elements (TBE1s) in a 
> couple of species of ciliates (Oxytricha fallax and trifallax.)  These
> are 
> DNA "Cut and Paste" transposons.  The most striking feature of this 
> family is the degree of conservation between elements: ...

> While I'm asking, if anyone knows of papers which 
> demonstrate selection operating during the evolution of other sorts of 
> repeated sequence families, I'd appreciate their sending me the 
> references.

I've published a body of work on selective pressure on mouse LINE-1

Hardies et al., Mol. Biol. Evol., 3, 109-125 (1986)
Hardies and Rikke, UCLA symposia on Molecular and Cellular Biology,
Vol 122 (1989)
Casavant and Hardies. J. Mol. Biol 241:290-397 (1994)

>... virtually all of 
> the several thousand elements in any given host genome are intact, 
> meaning that no stop codons, no deletions or insertions, and few 
> nonsynonymous changes (relative to synonymous changes) have 
> occurred during their divergence from each other.  I take this to be 
> evidence that some force of purifying selection has been acting on these 
> elements over a long period within the host population.

The upshot of the LINE-1 worrk is that, even though most mouse LINE-1s
are truncated, pairwise comparison comes out much as you have
indicated.  That is, few premature terminators, frameshifts (other
than the truncation itself); and a shortage of replacment changes in
relation to synonymous changes.  However, when we made a phylogenetic
tree and mapped the base changes to the branches, it became clear that
only the backbone of the tree deviated from a random mutation pattern. 
Since all the pairwise comparisons involve some segment of the
backbone of the tree, they all show some degree of selection, even
though the individual branches leading to the extant sequences do not.

More exactly, I think that a few lineages of selected elements exist
in each era.  Most of these peter out, but one gives rise to the
functional lineages in the next era.  If you didn't know they were
truncated, and if you didn't do the tree analysis, you might conclude
that the whole 100,000/genome was under selection.

> 	I would like to contrast this pattern with that of other DNA cut 
> and paste transposons, such as mariner elements:  when a 'population' 
> of mariners is found in a host population, virtually all of the element
> are 
> defective, meaning that their ORFs are interrupted by stop codons and 
> deletions (Hugh Robertson's recent papers.)  

It's important to realize that only a small fraction of deleterious
mutations result in frameshifts and terminators.  By the time these
obvious defects accumulate, the sequences have long been dead.
By our calculation, the mean time to pick up deleterious mutations by
mouse LINE-1 is < 100,000 yrs.; whereas the mean age of the most
recognizable mouse LINE-1 population is 2 Myr.  However, the mean time
to picking up an obvious defect is even longer, so you get this large
population that looks fine, but really isn't.
It's becoming apparent that the mouse also has a large collection of
old chewed up LINE-1s; (see work by Anthony Furano). 
The large number of fairly young mouse LINE-1s just kind
of blinded us to the older ones.

Unlike mouse LINE-1s, human LINE-1s are mostly chewed up by
terminators, frameshifts, etc.  A comparable young active lineage is
still there however.  See work by Beth Dombroski and Susan Holms
working with Haig Kazazian.  So the same family can either look like
your TBE1 distribution, or look like Mariner depending on whether the
bulk of the elements (which are mostly dead in either case) were put
out relatively recently, or a long time ago.

> This pattern [observed for Mariner] apparently 
> results from the fact that, in eukaryotes, the transposase encoded by 
> one copy of a transposon can be used to replicate other (possibly 
> defective) elements in the same genome;  there is therefore no selection 
> acting to conserve the transposase ORF, and the elements 'crumble' as 
> mutations accumulate (Kaplan, Darden & Langley, 1985, in Genetics).
I'm fond of using the Kaplan argument to claim that the LINE-1
protein is cis-acting on its own RNA.  (LINE-1 is a retrotransposon,
so at the point at which LINE-1 RNA produces the reverse
transcriptase, you can imagine that the mRNA is also the genomic RNA
for the new insertion.)  But to be fair, I think that
trans-mobilization has been observed for virtually all transposons
(including LINE-1); and they all manage to persist in their hosts
(with the possible exception of mariner, & I'm not sure I really
believe that).  In fact, it's pretty hard to figure a mechanism to
make a DNA transposase cis-acting in a nuclear organism, but P element
and many others clearly get along just fine.  So there must be a hole
in the Kaplan et al argument somewhere.

Steve Hardies, Assoc. Prof. of Biochem., Univ. of Texas HSC at San Antonio
Hardies at uthscsa.edu

For information on our graduate program in Biochemistry:

More information about the Mol-evol mailing list

Send comments to us at biosci-help [At] net.bio.net