On 27 Aug 1998, Anders Sandberg wrote:
> Damien Broderick <email@example.com> writes:
> > Everyone knows that DNA is prey to mutations, most of them quickly edited
> > out by vigilant repair mechanisms, most of the rest neutral or bad news,
> > only a few improving the genes they happen to occur within. But by what
> > chemical pathway do those mutations come about? I'd always assumed that
> > most of them involved gross *deletions* of one or more nucleotide bases,
> > the implication being that the least dangerous deletion would knock out
> > three adjacent bases in one go (otherwise *all* the subsequent three-letter
> > codons are fouled). However, many mutations occur at a single point. The
> > classic instance is the change creating sickle cell anaemia, in which
> > position 6 in the 146 amino acid chain has valine instead of glutamic acid.
> > Presumably this means that an adenine has been changed into a uracil (or,
> > in the DNA chain, into a thymine).
> > But how the hell does the equivalent of a random impact do that?
> Radiation or free radicals mess up single nucleotides, changing them
> into other molecules, but I think the major contribution is errors in
> replication. If a single base is damaged (or just unlucky), the enzyme
> doing the replication might make a wild guess and insert a random
> nucleotide in the copy.
> Just a guess from a computer scientists playing molecular biologist :-)
Well, I'm on the road to being both, and I can verify your wild guess. Many mutagens work by altering a nucleotide so that it binds better with a noncomplementary nitrogenous base, giving you a weird base-pair combination like Adenine-Guanine. If the change made by the mutagen is permanent, the error in pairing can be missed by "proof-reading" enzymes, and you will end up with half the progeny being mutants, the rest being normal.
If my memory of biochem serves me, mutations can also occur without a mutagen from the random and rare ketone to -enol molecular change in a nitrogenous base. This would also create a weird base pairing, but I think it is usually weeded out by proof-reading enzymes.