From: Preston Garrison (garrisonp@uthscsa.edu)
Date: Wed Apr 02 2003 - 19:20:11 EST
>Preston:
>"What seems to be going on here, at least as interpreted by the
>researchers, is not site specific insertion, but selection for cells
>in which the virus inserted near a particular gene. I'm not a gene
>therapy expert, but I think it works like this. They isolate some
>pre-lymphocyte cells of a particular type from the patient and
>transform them in mass with the engineered retrovirus vector. They
>may then select for cells that have the vector incorporated, based
>on some selectable marker in the vector. I'm not sure that they
>would do that or not. They would do it if the transformation is
>inefficient. But they didn't apparently clone from a single
>transformant cell, since they assume in their analysis that the
>cells they injected back into the patient had the vector inserted at
>various sites. After the cells are in the patient, some of them
>behave well and go to the marrow and start producing the desired
>corrected lymphocytes. However, it turns out that some element(s) of
>this virus can interact "in cis" (if the virus inserts next to or
>within) with this LMO1 gene to give a cell that is unregulated for
>growth and/or programmed cell death. That rare cell grows out to
>give the leukemia. The reason it happened in two patients is that
>they are transforming and injecting enough cells to make it fairly
>likely that in each patient at least one cell will have the
>insertion somewhere near the LMO1 gene. Of course, it would be
>interesting to know if the two insertions occurred at exactly the
>same site or only in the same vicinity, which is much more likely.
>There are such things as hotspots for virus insertion, probably
>based on chromatin structure or sites related to normal
>recombination. It's possible something like that contributed here,
>but I believe that in general retroviruses are not very site
>specific."
Josh:
>-This is exactly why I've been pondering this piece of evidence for
>a while. Since viruses supposedly jump into the genome completely
>at random, the only way to get viruses integrated into the same site
>in the genome is to have an ancestor obtain the viral incorporation,
>then pass it on to the progeny. This requires incorporation into
>the genome of Gametes, but within one infected individual who knows
>the number of possible viral incorporation into gametes, and then
>there's the issue of that gamete being passed on...
P:
Not sure what your point is here. We know that there are retroviral
proviruses in the genome, so they must infect germ cells at a high
enough rate that some of those germ cells form people. They also must
get fixed in the population. There doesn't have to be positive
selection - population genetic theory predicts that some neutral
mutations will get fixed.
P:
>"As to your larger question on endogenous retroviruses at the same
>site in different species, a mechanism corresponding to that above
>would require that there be a selective advantage of the insertion
>at the same exact location for both species."
J:
>-Why is this so?
P:
My point was just that the leukemia result is explained by positive
selection, not a site specific mechanism of insertion. If you wanted
to invoke the same explanation for proviruses at the same site in
different species, you would have to say that there was positive
selection for incorporation of the same virus at the exact same site
in both species.
J:
> Two requirements for similar site insertions across species come to
>mind: gametes must be infected, and similar variables must lead to
>similar insertions (either virus has some unknown specificity for
>something like non-methylated DNA or degree of chromatin
>condensation/silencing...) If the insertions are neutral, they will
>be retained in the genome, because they will simply piggyback with
>the chromosome when DNA duplication occurs. I don't know of any
>virus element editing mechanisms during genome duplication that
>would get rid of any inserted virus.
P:
Retroviruses have LTRs - long terminal repeats, a sequence of several
hundred nucleotides that is present at each end of the provirus
after it inserts. The proximity of 2 long identical sequences tends
to induce recombination, which deletes the intervening sequence and
leaves one copy of the LTR. There are thousands of these single LTRs
in the genome.
J:
> Also, are known viral insertions demonstrated to be conferring
>selective advantage to the host?
P:
There are a handful of specific insertions that appear to have taken
on some function, but again, for the vast majority it would dubious.
Some cellular regulatory elements appear to have incorporated LTRs or
fragments of them.
P:
>There may be a few sites where there was a selective advantage, but
>that would be dubious for the vast majority. Even in that case, you
>wouldn't expect the insertion to occur at exactly the same site.
J:
>-Exactly why I find this 2/10 case for leukemia like symptoms in
>this gene therapy trial case interesting. Also, this has only taken
>3 years to develop, whatever events trigger these symptoms may occur
>later in more of the treated patients. This would appear to argue
>that the virus insertion has a site preference.
P:
It doesn't require a site preference, as I explained above, if one
out of 100,000 insertions is expected to be close enough to the gene
to activate it (that was the researchers' estimate) and they injected
a million cells. Frequent occurrence of the leukemia is expected even
in the absence of a site preference.
>P:
>Insertion at exactly the same site in humans and other primates is I
>think known to be the case for some retrovirus insertions where the
>region has been sequenced in more than one species. Other
>transposable elements, particularly L1 elements and Alu elements are
>present in mammalian genomes in much larger numbers than
>retroviruses (L1 > 10^5; Alu > 10^6) and thus provide thousands more
>examples of the same thing. These elements are so frequent in the
>genome that almost any segment over a few kb that is sequenced in
>both human and chimp or gorilla will have examples of this. Some
>particular insertions can be shown to be present only after a
>certain point in primate evolution. That is, they may be present
>only in chimps, gorilla and human and not in more primitive primates
>or other mammals.
J:
>-Elaborating on the question I am raising, this may indicate the
>range of hosts for a particular virus infection instead of an
>evolutionary relationship. The question is, are there any other
>ways to interpret the data in general. I am trying to think of
>other considerations.
>
>Consider the following from Sverdlov, "Retroviruses and Primate
>Evolution" Bioessays 22:161-171:
>
>"..since retroviruses can cause cross-species infections, and there
>are data suggesting multiple ancient cross-species transmissions of
>retroviruses in primates."
>Referenced articles: j. virol 1995 69:7877-7887, virology 1997 238 212-220.
>
>
>Additionally, since viruses mutate much more rapidly when not
>incorporated into a host, it is possible that events of viral
>incoporation in gametes occur within the same wave of "retroviral
>plague" if you will, and that the resulting incorporations would
>look different due to the differences in rapidly mutating viral
>strains being incorporated, not their divergence due to evolution
>after integration (at the same time alot of the mutations observed
>seem to compromise viral function....)
P:
This still doesn't account for the insertions being in the exact same
site. Retroviruses just don't have the absolute site specificity that
it would take to hit the same place twice with 3 x 10^9 places to
chose from. Another point is that other insertion elements that show
this same phenomenon of being in the same place in different species,
retrotransposons like L1 and Alu, have no extracellular transmissible
form. There are no cross-species epidemics of L1s or Alus because
they have no packaged, secreted form like viruses do. But there are
thousands of sites where they have been inserted at the same position
in related species.
--
This archive was generated by hypermail 2.1.4 : Wed Apr 02 2003 - 19:21:16 EST