From: Aubrey de Grey (ag24@gen.cam.ac.uk)
Date: Sat Apr 26 2003 - 09:01:17 MDT
Rafal Smigrodzki wrote:
> ### But in plants we have huge mitos with introns, 57 genes. Either
> there is another insurmountable barrier but a different one from
> hydrophobicity, or, more likely IMO, plants do not have a strong
> selective pressure against large mitochondrial genomes.
Quite right -- for most plants. The chlamydomonads, however, appear
to have experienced just as much selective pressure against large
mitochondrial genomes as animals, and the result is a genome with only
seven of our 13 protein-coding genes (plus one that we don't have).
(Plasmodium has only three, but it's done the Complex I trick so in
fact it has moved one less than Chlamy.) It seems that without the
code disparity evolution was able to get quite a bit further against
the hydrophobicity problem -- but still not nearly all the way.
> Also, there are many highly hydrophobic proteins made in the nucleus
> and imported into membranes, or through membranes.
Into membranes yes, through membranes no. The anion carriers are very
hydrophobic and yet nuclear-coded, but they get into the inner membrane
by a different route that doesn't involve an N-terminal leader and is
not preceded by complete import into the matrix. Ones that do go via
the matrix (eg cytochrome c1) are much less hydrophobic, at least by
what seems to be the appropriate measure, "mesohydrophobicity" (see
work by Claros in the mid-90s).
> The genetic code difference makes sense to me because it implies the
> need for at least three independent mutational events for mtDNA-nDNA
> gene transfer in vertebrates. You have to make a nuclear pseudogene
> from a mtDNA gene, then mutate it into a form which will produce a
> viable protein if read by the nuclear tRNA's, and then you have to
> couple it to a mitochondrial leader peptide. Each event happens all
> the time, but all three together would be quite uncommon.
The problem is that the "mutate it into a form which will produce a
viable protein" bit is actually not one mutational event but one for
each discrepant codon, and they all (all the UGA's and most of the rest,
anyway) have to occur in the absence of any other mutations that make
the protein not work. This seems to me to explain why (as far as we
can tell) the code disparity is in fact vastly harder for evolution to
solve than either of the other two hurdles.
> ### You lost me here. For the outcome of a xenohydrolase treatment it
> definitely matters whether the lysosomal aggregates are detrimental or
> not.
Sorry, all I meant here was that if there's no non-lysosomal problem
then by elimination the lysosomal aggregates must be detrimental.
> ### You are talking about the so called "fatty streaks", right? Fatty
> streaks can fully disappear.
This is news to me. Can you give references? This certainly flies in
the face of all I've ever come across on fatty streaks.
> The only way you can tell there is totally indigestible stuff in the
> plaque is if you cannot digest it in vitro in cell culture
True. As far as I'm aware, no one has done the simple experiment of
exposing macrophages to atheroma in culture and seeing what happens.
Or more likely, it was done 50 years ago and the result was boring...
> and if injected in vivo in a young animal, it doesn't get phagocytosed.
I think you meant "OR if injected ..." -- clearly both phagocytosis and
digestion have to happen, either without the other is useless.
> ### A different explanation is that there is a dynamic balance between
> the influx of digestible fatty material, and the ability of endothelium
> and macrophages to deal with it. As time goes by, the balance tilts
> towards buildup, with first only occasional macrophages dying, then as
> the quality of macrophages goes down with age (and it does), there
> isn't enough to deal with plaque even if it's still digestible, also
> because of scarring, and calcification.
I disagree that this is an explanation, because it doesn't explain why
fatty streaks appear in childhood. Any process of accumulating damage
(e.g. junk like this) that is already easily detectable in early life
is by definition primary, not brought on by the misfortune of inhabiting
the same body as a bunch of other aging tissues (and thus experiencing
a tilting of a dynamic balance).
> ### It is not clear that the mutant sequences, aside perhaps from some
> deletion mutants, have a selective advantage.
You're quite right that the evidence is best for large deletions. But
that's all we care about, because they're always (absolutely always)
homoplasmic or nearly so in COX-negative cells (muscle, anyway) -- see
Aiken's work.
> Chinnery and Turnbull did some calculations showing that mtDNA can
> widely fluctuate and fix mutations into homozygosity if only random
> processes are taken into account.
This is one body of work regarding which I tend to have trouble being
diplomatic. Patrick Chinnery is an excellent medic, but suffice to say
that we discussed his hypothesis for accumulation by drift five years
ago, before he'd published anything embarrassing, and he's still having
trouble understanding the fundamental inconsistency of that model with
the data, viz. that it predicts that mitochondrial hyperproliferation
will precede (hence be seen in the absence of) COX-negativity, whereas
with the sole exception of the MELAS mutation (which seems to have a
special way to win) the reverse is true, we see abundant COX-negative
cells that lack any mitochondrial hyperproliferation.
> A mutational ratchet mechanism tends to favor mutated DNA's in
> a 3 to 1 ratio, but this is not a selective advantage.
I don't know what underlies your arrival at this ratio, but one can get
any ratio one likes with suitable assumptions about rates of mutation
and turnover. The most general argument in favour of selection, though,
is that we know roughly the mitochondrial half-life (a week to a month
in rats) and thus we know how many mitochondrial generations it takes
for cells to be taken over -- and it's very, very few, probably only
a few times the theoretical minimum of log(2) of the number of mtDNAs
in the cell. Drift won't do that, whatever the assumptions.
> Actually, once you have mitofection, there are at least two feasible
> methods to select the mutated mtDNA's out of the cell. Previous
> attempts failed because of lack of the ability to introduce DNA into
> mitos, but this has now changed.
Your turn to lose me. What are you suggesting, in detail?
Aubrey de Grey
This archive was generated by hypermail 2.1.5 : Sat Apr 26 2003 - 09:11:01 MDT