Tautomers--evil twins of the bases!

Molecular identity is a malleable thing-particularly in the case of nucleotides. While we tend to draw the bases as static molecules with only one form, in water solutions this is simply not the case. Indeed, at the time when Watson and Crick were guessing at the structure of DNA, the 'evil twin' structures of the bases were thought to predominate. This, of course, threw a major wrench into the model building, and it was only when this error was realized that the structure came forth.
The different structures arise from the alteration of single and double bonds in the ring systems of purines and pyrimidines. The disposition of binds is very important to the hydrogen bond presentation; for example, N₃ is a hydrogen bond Acceptor in cytosine and a Donor in uracil/thymine. The reason? in cytosine, 3 bonds form between the nitrogen and the other members of the ring, 'satisfying' its desires for partners and leaving only so the remaining free electron pair available. In uracil/thymine, only 2 bonds are formed to neighbors in the ring, the third thus forms to a donatable hydrogen. These are the high probability structures, but alternatives are possible, and in the molecular world, possibility ALWAYS manifests itself as something that is occasionally true. Given the millions of bases in the genome of even a single cell, it is INEVITABLE that the tautomers will pop up periodically.
So what? The key fact is that tautomerization alters the basepairing properties of a nucleotide. Since the genetic code IS the binary code of hydrogen bonding, this means that the tautomers intermittently 'masquerade' as other members of the code. Shown below are tautomeric forms of the Big 4 (again, tautomerization is a result of interaction with water; that's where the moving Hydrogens go to and come from). Note how the tautomeric forms lose the ability to make correct pairings and, worse still, create viable pairings with incorrect partners!

Dr. Base and Mr. Tautomer

Note also that the GT pairing shown does fit gracefully into B-form double helix, whereas the GT pairing between 'normal' T and G requires a rotation and does not fit into the helix.
Since the tautomerization of the bases is inevitable, and since key players (such as the copy machine DNA Polymerase!) choose DNA partners on the basis of 'fit' alone, mistakes will be incorporated into a new copy of DNA, regardless of the care taken by the machines doing the copying! However, since the cell's machinery has evolved in the presence of nucleotide tautomerization, several processes have arisen to minimize its effects.