In <79347319DB at mercury.uark.edu> DRHOADS at MERCURY.UARK.EDU ("Douglas
Rhoads") writes:
> From: JEThomas at ix.netcom.com (Jonah Thomas)
>> An organism that could change genes from dominant to recessive and back
>> could possibly evolve faster. A gene that's currently favorably
>> selected could spread faster if it was dominant. A gene that's common
>> but currently unfavorably selected would do less damage if it was
>> recessive. Also it would survive in the population longer, giving a
>> greater chance that it would still be present if sometime later it
>> became favorable.
>I am not sure this argument or hypothesis could ever hold water.
>Organisms don't change from dominant to recessive or back, traits do.
Yes, I misspoke.
>Clearly a phenotype can be modified by the environment and the
>penetrance can change with changing environmental pressure just as
>the effects on `fitness' can change. Normally, a gene that is
>unfavorably selected (what ever that means) will NOT be common unless
>the population has just come out of a situation where the gene was
>advantageous.
That's right. For an unfavorably selected gene to be common it must be
only recently unfavorable. This can happen due to an environmental
change that makes a previously-less-fit alternative allele more fit, or
it can happen when a new mutation happens to be more fit. Both of
these may happen fairly often.
>>Modifying genes that affect dominance would have to
>>evolve at each location, and the selection that would lead to their
>>evolution is weak. If they mutate at the same rates as the genes
>>whose dominance they modify, they have little effect on selection.
>>It doesn't work.
>Modifying genes that affect dominance do NOT have to be linked to the
>gene. Why should they?? And what point are YOU trying to make?
Ah, why should they be linked. OK, let's look carefully at that.
First, both Fisher and I suppose there is a general mechanism to affect
gene expression. A gene that's turned on, makes something that
eventually results in mRNA that produces a specific protein. A gene
that's turned off has that process interrupted so the protein is not
produced. Fisher and I both suppose that mechanisms exist which use
this mechanism to produce dominance. There is more here than the
simple model which says that recessive genes never produce a functional
protein. In this model a recessive gene may produce a protein when
it's homozygous, but that production gets shut off when the dominant
allele is present.
Here it comes: Could you select such a dominance-modifying gene based
on its effect on dominance? And the answer is, with reasonable
assumptions it's hard to select such a gene if it is unlinked to the
gene it affects. It is only selected when the version it makes
dominant is selected relative to the version it makes recessive. This
is a transient condition. When either allele is rare there isn't much
selection for the modifier. When the modifying gene is rare there
isn't much selection for the modifier. And after the dominant version
increases its frequency, the only way to get another pulse of selection
in favor of the modifier is to first select the other direction to get
the frequency back down. This is not an effective way to breed
modifying genes. It doesn't work.
But if the modifying gene is closely linked to the modified gene, then
whenever the allele it modifies is selected, the modifying gene
hitch-hikes. If it can make that gene dominant then they both increase
frequency faster. When the allele it modifies is unfavorable, they
both are selected against, and if it can then make that gene recessive then
they both lose slower. (For clarity -- I'm not suggesting that every copy
switches from dominant to recessive at the same time. I'm suggesting that
a few copies get switched at random, and selection increases their
numbers relative to the others.)
Also here's a handwaving molecular argument -- if there's a general
mechanism to modify a closely-linked gene, it can be applied to any
gene merely by translocation. If modifiers have to evolve separately for
each gene, they don't get so much flexibility and won't arise so often to
be selected.
>What if the modifying gene was pre-existent in the population. Multi
>enzyme complexes and receptor signaling process already exist and all
>upstream and downstream processes would have variants in the
>population.
Yes, but that isn't what I'm talking about. I don't want to propose a
specific molecular mechanism for dominance-change because 1st, I could
make a silly molecular mistake that would get it laughed at, and 2nd,
the reality could be some completely different molecular mechanism that
gave an equivalent result. But here's a start toward something like
that: Say you have an intron that has a part that can flip over at a
fairly high rate, say once per thousand individuals. When it is in
state A, during transcription it splices out a part that can interact with
the same intron in state B. Once the part is spliced out, the hnRNA the
intron is inserted into works normally.
The intron in state B works the same way exactly, except that when the
spliced-out RNA from a transcript of A (or a product of that RNA, etc)
interacts with it, it doesn't splice out the same way but instead puts
an ealy stop signal into the mRNA. Both versions result in functional
protein, but the state-B intron results in nonfunctional protein from
the state-B allele whenever the state-A version is present.
This intron could evolve once and then get spliced into a wide variety
of genes. Versions of it might get selected at whichever sites they do
a good job.
>> But if dominance-modifying genes could be transposed to different
>> locations at a relatively high rate, they would need to evolve only
>> once. And if they switch from dominant to recessive and back at a
>> relatively high rate, say an order of magnitude or two below the
>> selective rate, they could be selected.
>HUH?? (see above)
I hope this is clearer. The dominance-modifying gene needs to be
linked to the modified gene to be selected with it. It's plausible
that molecular mechanisms might be possible that would do direct
dominance-modification of closely-linked genes in a single standard
way.
What I haven't shown at all, is that there is selective advantage
in expressing only one of a pair of heterozygous alleles. If almost
every time organisms do as well or better expressing both, then the
selection that might drive my hypothetical mechanism won't be there and
the whole thing collapses. Never mind that it's possible, without that
selection, it probably won't happen.
So what's the truth? I think only molecular studies can tell. I have
a Just-So story that seems to me at least as plausible as the standard
one, the one which says that recessive genes simply produce
dysfunctional proteins. I can make mathematical models and computer
simulations which predict that under the right circumstances my
hypothetical dominance-modifiers would be selected. I can't tell whether
those circumstances arise, without real-world data which I think is not
available yet. There is likewise no real-world data available to test
the standard Just-So story.
