HISCOX (julian.hiscox at afrc.ac.uk) wrote:
: Brian Foley raises some interesting points, here are some possible ideas
: in answer to his hypothesis:
: Hypothesis 1: It is probably true that silent sites in the third position
: are not truely silent, hence the idea of codon preference! Although,
: structural requirements may necessitate the change in the third codon
: from the "most common" base in some cases.
I have seen several papers including [Proc. Natl. Acad. Sci. USA
85: pp 4378-4382 (1988) Codon Preference is but an Illusion Created by the
Construction Priciple of COding Sequences] that claim that codon "preference"
is not a cause of the codns we observe, but a refleciton of biased mutations
at silent sites and other factors. For example thermophilic organisms
prefer C and G in the silent sites not because they have more tRNAs for
those codons, but because a high total genomic G+C content is needed.
I am observing that mice are losing a lot of C's in silent sites
and I do not belive that all of the tRNAs that recognize codons with C in
the third position are becoming scarce. I believe that these C's are
being mutated to T by deamination of 5-methyl-C. I have a hard time believing
that selection could be so stringent as to select against silent codon
positions in most genes.
If there is selection in these positions, would we not see a greater
selection in important housekeeping genes than in nonimportant genes that are
only expressed in one cell type or only expressed at low levels? It seems
to me that if a protein must be cranked out at max level in order for the
cell to survive codon useage could be important, but if another protein is
pretty much optional in a cell anyway, codon usage should be not very
I know that different genes evolve at different rates (presumably
they all mutate at the same rate, selection keeps some from evolving), but
the variation in rate stems from an increase in the non-silent mutations
that do not get selected out, rather than an increase in silent and non-silent
mutations. [J. Mol. Evol. 28: 286-288 (1989) Nucleic Acid Composition, Codon
Usage, and the Rate of Synonymous Substitution in Protein-Coding Genes]
looked at 42 genes that had been sequenced in both rat and human. Although
they show a nice table on page 288, that shows that the genes with the
lowest rate of nonsynonymous do not have the lowest rate of synonymous
substitutions, I have a hard time understanding the data:
They say the table gives the nuber of substitutions/site and yet
the values go above 1.0. How can you measure more than one substitution
per site when comparing the gene from only two species? If the genes had
been nearly saturated with mutations, I guess a random distribution would
show us that if 90% of the sites have been changed, then something like
30% could be calculated to have been hit twice.
Anyway, if some genes have seen more than 1 mutation per silent
site that survived selection, there cannot be much selection on the basis of
codon bias in those genes.
: Hypothesis 2: It is also true that introns "mutate" faster than exons,
: because mutations within the codon region i.e exons that are deliterious
: to the organism are most likely to be selected against-i.e the organism dies
: and we do not sample the change!. Especially if the mutations are in
: a gene of
: functional importance. The histone genes are very conserved between species,
: and I imagine would not tolerate mutation because they are of extreme
Lets say introns "evolve" faster than exons then. They mutate at
the same rate, but selection against most mutations prevents evolution.
: Mutations within introns would be less deliterious because these are not
: as functionaly important as exons (except perhaps the 5' and 3' ends),
: hence would not be deliterious to the organism. I don't think the polymerase
: differentiates between exon and intron, although this would be an
: interesting hypothesis to test.
This is what I am trying to do.
: One could examine the replication of a defined piece of exon compared to an
: intron within an in vitro system (for example construct YACS with bits of
: exons and introns). One could introduce mutations within each region and
: if the polymerase preferentialy corrects/repairs exon DNA over intron DNA.
Why do we have to do benchwork? We already have organisms that have
been out there mutating and replicating for a few billion years. Why do
we always asume that some arteficial system is the only way to get the
First of all, it is most likely not the polymerase that does the
repair. There might be 100 proteins involved in repair of both damaged
DNA and DNA mismatches. Secondly, there is much evidence that mice genes
evolve faster than human genes (and we are both mammals!) so I do not want
to assume that some strain of yeast in a test tube is an accurate model of
what happens in either humans or mice.
The statistics are not trivial, but I think it is possible to sort
out some of the factors by looking at real genes in real wild organisms in
the real world. I'm sure we'll find that some regions of the genome
mutate faster than others. We'll find that some regions that mutate at
equal rates will evolve at different rates. We'll find that some organisms
evolve faster than others. So we may need to limit ourselves to comparing
one gene, located on the same chromosome, in relatively closely related
species; but we can still look at the few million years separate rat from
mouse, or human from chimpanzee and see a lot more than looking at a few
decades of test-tube evolution.
: However, I think mutations within the exon are selected against either by
: no translation of the protein or the protein that is produced is defective
: and the organism cannot reproduce to pass on this change to the next
I think everyone will agree with you there.
: Julian A Hiscox.
: I.A.H. Compton.
: E-mail: HISCOX at AFRC.AC.UK
* Brian Foley * If we knew what we were doing *
* Molecular Genetics Dept. * it wouldn't be called research *
* University of Vermont * *