Re: Why don't mitochondria have junk DNA?

John Harshman wrote:
> rev.goetz wrote:
>
> > John Harshman wrote:
> >
> >>rev.goetz wrote:
> >>
> >>
> >>>John Harshman wrote:
> >>>
> >>>
> >>>>rev.goetz wrote:
> >>>>
> >>>>
> >>>>
> >>>>>John Harshman wrote:
> >>>>>
> >>>>>
> >>>>>
> >>>>>>rev.goetz wrote:
> >>>>>>
> >>>>>>
> >>>>>>
> >>>>>>
> >>>>>>>John Harshman wrote:
> >>>>>>>
> >>>>>>>
> >>>>>>>
> >>>>>>>
> >>>>>>>>rev.goetz wrote:
> >>>>
> >>>>[snip]
> >>>>
> >>>>
> >>>>
> >>>>>>>And on my last post of this topic on Jan 8, I purposefully focused on
> >>>>>>>the evidence of natural selection in rps gene duplications that have
> >>>>>>>been preserved in the nuclear genome of several species for hundreds of
> >>>>>>>millions of years. The context for all of this should not confuse you.
> >>>>>>
> >>>>>>But it does. You are presenting evidence for conservation of a
> >>>>>>functional gene copy after all other copies have been lost, as if this
> >>>>>>is at all relevant to the fixation of that gene in the presence of
> >>>>>>another functional copy. But it isn't. And that's because you are
> >>>>>>yourself confused about the event you are trying to explain. Once more,
> >>>>>>you want to explain why, of the two genotypes below, the one on the left
> >>>>>>might have an advantage:
> >>>>>>
> >>>>>> Genotype A Genotype B
> >>>>>>n-genome gene X -
> >>>>>>mt-genome gene X gene X
> >>>>>>
> >>>>>>Notice that it is no explanation to say that gene X is important; both
> >>>>>>genotypes contain gene X. In fact, given that there are many
> >>>>>>mitochondria in the average cell but only one nucleus, the difference in
> >>>>>>copy number between genotypes must be negligible.
> >>>>>
> >>>>>Before I go further with this, I want to make sure that I understand
> >>>>>you on various points of the first major step of the gene transfer,
> >>>>>which is the fixation of the duplication in the nuclear genome. And
> >>>>>that you understand my definition of "allele." And I think that
> >>>>>your definition of "allele" is really should be used only for the
> >>>>>term "genotype."
> >>>>
> >>>>Whatever terms you like to use, I think that viewing this in terms of
> >>>>genotypes with different copy numbers is the clearest way to do it. The
> >>>>question you need to ask is why genotype A would be selected over
> >>>>genotype B.
> >>>>
> >>>>
> >>>>
> >>>>>I use a different definition of "allele" than you in the case of
> >>>>>gene transfer. I keep the concept of allele to apply to a single locus
> >>>>>while in the case of gene transfer there are two loci in two respective
> >>>>>genomes are involved. In the first step of the transfer, the new allele
> >>>>>is the duplication in nuclear genome. And in the case of rps1, the
> >>>>>first set of the transfer, the fixed duplication into the nuclear
> >>>>>genome occurred at least 33 times.
> >>>>>
> >>>>>And you are saying that the preservation by natural selection for the
> >>>>>nuclear genome copy may have only occurred after the transfer was
> >>>>>completed. Is that correct? If I understand you correctly, you are
> >>>>>saying that rps1 may have fixed in the nuclear genome by drift in 33
> >>>>>separate occasions. Is that correct? (At this point, let us refrain
> >>>>
> >>>>>from talking about the complete gene transfer that by definition
> >>>>
> >>>>>includes the loss in the mt genome. We will get back to that later.)
> >>>>
> >>>>I'm asking you to provide any suggestion of a reason why the fixation of
> >>>>a nuclear copy of rps1 might conceivably have been promoted by
> >>>>selection. I'm not claiming that drift was responsible for that
> >>>>fixation. I don't know. But at least drift is credible.
> >>>
> >>>
> >>>First, I will say that I doubt that drift is a credible explanation for
> >>>33 convergent duplications of a functioning gene, unless we see a
> >>>similar pattern of convergence with a gene sized chunk of
> >>>"nonconserved" DNA that duplicated from the mitochondrial genome to
> >>>the nuclear genome.
> >>
> >>There is no such DNA in the typical mitochondrion. Mitochondria tend to
> >>have almost no junk DNA.
> >>
> >>
> >>>And from what I have read, I do not here of such
> >>>convergence with nonconserved DNA. And I do not see how a statistical
> >>>model of neutral molecular evolution could possibly support such
> >>>convergence. Do you?
> >>
> >>Yes. The number 33 is not, of itself, any indication of selection. Are
> >>we agreed on that? What's at all interesting, even to a cursory glance,
> >>is a difference in number of transfers among genes. The obvious theory
> >>to explain that was suggested here by Larry Moran: that the rate of some
> >>particular mutations is higher than others. Either the gene in question
> >>is particularly likely to be duplicated or is particularly likely to be
> >>functional after duplication. Larry suggested a difference in the ease
> >>with which particular protein products can find their way into the
> >>mitochondrion.
> >>
> >>
> >>>And if there was such an intense mutational pressure that mutated many
> >>>duplicate copies of a functioning gene from the mitochondrial genome to
> >>>the nuclear genome, then it would likely happen multiple times in the
> >>>same lineage.
> >>
> >>Note that these gene duplications happen frequently, much more
> >>frequently than is required to explain any arbitrarily large number of
> >>transfers. Duplication isn't the problem. Transfer of function is the
> >>problem.
> >>
> >>
> >>>And at least in the case of plants, the rate mutation is
> >>>faster in the nuclear genome compared the mitochondrial genome, so if
> >>>everything in the gene transfer was neutral, then the greater amount of
> >>>mutations in the nuclear genome would more quickly silence the nuclear
> >>>copy, causing loss to the nuclear copy instead of the m copy .
> >>
> >>That made no sense. I think the problem is that "everything in the gene
> >>transfer was neutral" is such an ambiguous statement. The rates of
> >>mutation are irrelevant until such time as you can compare the absolute
> >>rate of duplication with the rate of gene transfer. You have no idea how
> >>many duplications were necessary to produce one gene transfer.
> >>
> >>
> >>>Likewise, there would have to be numerous neutral duplications into the
> >>>nuclear genome before drift would cause the loss of the mitochondrial
> >>>copy. And from what I have read, there are few examples of multiple
> >>>duplications of functioning genes from the mitochondrial genome to the
> >>>nuclear genome within the same lineage.
> >>
> >>You are wrong. Mitochondrial pseudogenes are plentiful, and duplications
> >>are frequent.
> >>
> >>
> >>>And I doubt that there are
> >>>enough such examples to defend a statistical model of neutral molecular
> >>>for duplications that occurred in over 30 lineages. Do you see such
> >>>evidence for a neutral model?
> >>
> >>No, but I haven't looked either. This is not getting you any closer to
> >>answering the question, though.
> >>
> >>
> >>>Concerning the advantage of what you call "genotype A" over
> >>>"genotype B," I saw a correlation for a proposed ranking of the
> >>>importance of the genes to the amount of times that it duplicated. This
> >>>may explain the advantage of genotype A over genotype B, but I now see
> >>>that it offers no explanation for the eventual loss of the m copy in
> >>>over 30 lineages. I clearly need to add something to my hypothesis.
> >>
> >>Quite so, because so far your hypothesis says nothing at all.
> >>
> >>
> >>>I
> >>>have more ideas on it after reading the article by Adams, K.L. and
> >>>Palmer, J.D. (2003) titles "Evolution of mitochondrial gene content:
> >>>gene loss and transfer to the nucleus," _Molecular_ _Phylogenetics_
> >>>_and_ _Evolution_ 29, 380-395. The article briefly reviews discusses
> >>>the potential buildup of deleterious mutations in the mitochondrial
> >>>genome. Now, such a buildup of deleterious mutations would not cause an
> >>>advantage for genotype A over genotype B, but such a buildup of
> >>>deleterious mutations would set up the nuclear copy for a rapid burst
> >>>of adaptive substitutions.
> >>
> >>What? That again made no sense. First, we're talking about plants, in
> >>which the mt evolution rate is slower than the nuclear rate. Second, how
> >>does a buildup of deleterious mutations (in which copy?) set up a burst
> >>of adaptive substitutions?
> >>
> >>
> >>>With all of this in mind, here is a possible scenario of gene transfers
> >>>that occurred in over 30 lineages. This scenario requires 3 major
> >>>steps: 1) the fixation of the duplication in the nuclear genome, 2) the
> >>>adaptive evolution of the nuclear copy, 3) the loss of the m copy to
> >>>drift. At first there was an advantage for more copies of the gene,
> >>>though that meant just two more copies in the nuclear genome, so this
> >>>led to the fixation of the duplication in nuclear genome.
> >>
> >>Not credible. Increasing the number of copies by two is just not a
> >>significant increase.
> >>
> >>
> >>>Then there
> >>>was a rapid burst of adaptive substitutions in the nuclear copy that
> >>>reversed a buildup of moderately deleterious substitutions that fixed
> >>>during the mitochondrial stage of the gene's ancestry.
> >>
> >>Muller's ratchet? Why would a nuclear copy release the ratchet?
> >>
> >>
> >>>And this burst
> >>>of adaptation made the nuclear copy superior to the m copy so that
> >>>there was no more selective pressure to preserve the m copy. And the m
> >>>copy was eventually lost by drift to a silencing mutation.
> >>
> >>At least this phase is credible, sort of.
> >
> >
> > I will focus on two points: 1) comparing rates of functioning
> > duplications versus rates of non-functioning duplications, 2) the
> > possibility of nuclear genome evolution reversing moderately
> > deleterious substitutions (subs) that built up in mitochondrial
> > ancestry.
> >
> > You mention that there is little junk DNA in mt genomes (the original
> > topic) so we will not see gene-size chunks of non-functioning
> > duplicated from the mt genome into the nuclear genome. This is a
> > misconception because mutational processes that generate duplications
> > have no conscience related to making sure that a complete gene is
> > duplicated.
>
> True, but most numts seem to be rather longer than any single
> mitochondrial gene, so the odds of a complete copy of at least one gene
> are quite high.
>
> (Are you familiar with the term "numt"? It means nuclear mitochondrial
> sequence, i.e. a nuclear copy of a mitochondrial sequence.)
>
> > So for hypothetically example, 50 sample lineages have 4
> > functional genes that duplicated from the mt genome into the nuclear
> > genome in at least 30 lineages while no similar size chunk of
> > non-functioning DNA duplicated from the mt genome into the nuclear
> > genome in more than 5 lineages. I clarify that I do not have the data
> > to support this, but such a difference of rates of functioning
> > duplications versus rates of non-functioning duplications would support
> > that natural selection was driving the fixation of the duplications
> > instead of random drift. Do you agree or disagree?
>
> Disagree. What you need is some idea of the frequency of numt events of
> different lengths. I contend (from what numts I have seen) that they
> generally are longer than the average gene, so would tend to contain at
> least one complete gene. If fragments containing full genes tend to be
> fixed at a higher frequency than suggested by their occurrence as
> mutations, then you have something.

I concede that perhaps the high frequency of numts could account for a
non-adaptive/neutral explanation for the fixation of the functioning
gene duplications in the nuclear genome. Or at least I cannot find the
data to refute a neutral explanation.

>
> > I am sure that if I had enough time, I could compile the data required
> > for such a test. And the results would either confirm or refute my
> > hypothesis. Unfortunately, I do not have enough time to compile this
> > data in the near future. (Or perhaps a researcher already as this data
> > compiled; do you know anybody that we could email about this?) Despite
> > my time constraints, it is possible to test if the functional gene
> > duplications from the mt genome into the nuclear genome are consistent
> > with either natural selection of random drift. On the other hand, if
> > the statistical testing suggests natural selection, this will not tell
> > us why there was natural selection driving the functioning gene
> > duplications. And you reject that an additive effect would select two
> > nuclear copies because there are several more mt copies. And I do not
> > know how we could test this.
>
> It's not a matter of testing, just logic. If there are dozens or
> hundreds of mitochondria in a cell, will one or two nuclear copies make
> a significant difference in expression? How could they?
>
> > Now I will address point 2. Other literature suggests that moderately
> > deleterious mutations may build up in mt genes.
>
> Not just mt genes. Nuclear genes too. Recombination of course can help
> dispose of these in nuclear genes but not (we think) in mt genes.
>
> > And in my last post I
> > proposed that selection in the nuclear copy could reverse moderately
> > deleterious subs that built up in the mt copy. For example, if there
> > was a deleterious sub with a 3% disadvantage in the mt gene (for
> > instance an A to G sub), upon duplication into the nuclear genome the
> > nuclear copy would need to generate only 100 G to A mutations before
> > there would be a nearly 1 probability that G would fix in the nuclear
> > locus and reverse the 3% disadvantage that originated in the mt genome.
> > And in most gene loci with G in nuclear genomes in a large population,
> > it does not take much evolutionary time to generate 100 G to A
> > mutations.
>
> Does it take a longer or shorter time to generate those mutations in a
> mt gene than a nuclear gene? Consider copy number, turnaround time, and
> mutation rate in your answer.
>
> > And if there were a few such deleterious loci in the mt
> > genes, then the nuclear copy would quickly become superior and the mt
> > copy may become negligible. Unfortunately, it is hard to test for a
> > burst of adaptive subs when the adaptations occurred over 80 million
> > years ago due to saturation by drift in the following years.
>
> This explanation, I will note, doesn't account for differences among
> genes in propensity to be transferred to the nucleus. We would expect it
> to affect all genes about the same. Wouldn't we?

I assume that it would take a shorter time to generate the same
mutations in the mt gene population. But many geneticists conjecture
that there is little or no recombination in the mt genome. If this is
true, then for the most part the entire mt genome is selected for or
against instead of selection acting on various mt loci. So a particular
sub at a nuclear locus may have a 3% advantage while the same sub at
the homologous locus in the mt copy would have a much smaller percent
of selective advantage in terms of the entire mt genome. And if the
percent of selective advantage for the respective mt genome is small
enough, then it would be nearly neutral in the mt genome as opposed to
the same mutation having a 3% advantage at the nuclear locus.

.

Relevant Pages

  • Re: Why dont mitochondria have junk DNA?
    ... >>>33 convergent duplications of a functioning gene, ... >>>the nuclear genome. ... >>>the nuclear genome, then it would likely happen multiple times in the ... >>Note that these gene duplications happen frequently, ...
    (talk.origins)
  • Re: Why dont mitochondria have junk DNA?
    ... >> 33 convergent duplications of a functioning gene, ... >> the nuclear genome. ... >> the nuclear genome, then it would likely happen multiple times in the ... > Note that these gene duplications happen frequently, ...
    (talk.origins)
  • Re: Why dont mitochondria have junk DNA?
    ... Either the gene in question ... >>>many duplications were necessary to produce one gene transfer. ... >>>which the mt evolution rate is slower than the nuclear rate. ... > hundreds of mitochondria in a cell, will one or two nuclear copies make ...
    (talk.origins)
  • Re: How old does the universe have to be for evolution to "work"?
    ... The interaction of proteinase inhibitors produced, in most cases, by ... "Role of selection in fixation of gene duplications."By Kondrashov ... Duplications of long DNA segments are constantly ... and are subject to divergent evolution of the ...
    (talk.origins)
  • Integrity
    ... Every gene in the genome has the same IBD relatedness ... one allele in one genome may not be ... Hamiltonian heuristic rb fitness ...
    (sci.bio.evolution)