As we all know, the genetic code is redundant. Within protein coding regions, substitutions at silent sites do not affect the amino acid sequence of the encoded protein. Because of this property, these synonymous substitutions (so-called because they result in the same amino acid) are often used to estimate the neutral rate of evolution — they should not be under selection. But there is evidence for natural selection operating on silent sites. That’s because, even though different codons encode the same amino acid, the tRNAs for the synonymous codons are found at different frequencies within an organism’s cell.

When we looked at the codon usage in bacteria, we see that the most commonly used codon for a particular amino acid is also the one with the most abundant tRNAs. Whether codon usage is determined by the tRNA pool or the tRNA pool is determined by codon usage, there is obviously selection for coadaptation between these different loci. That’s because the incorporation of amino acids into a polypeptide during translation isn’t as clean a process as presented in an introductory biology course. The incorrect tRNA may enter a ribosome and be incorporated, which must be corrected. That takes energy, and depends on the frequency of a particular tRNA in the pool — common tRNAs have a lower probability of misincorporation.

Additionally, the more common the tRNA, the more quickly it will find its way to the ribosome to be added to the polypeptide chain. Which leads us to this research (doi: 10.1126/science.1135308, but the article is not online yet). The researchers found that a synonymous mutation affects the structure of a protein pump. The phenotype could not be due to the amino acid sequence because the wild type and mutant encode the same polypeptide chain. But the mutant allele contains a codon for a less common tRNA than the wild type allele. The less common tRNA leads to slower incorporation of the amino acid. This means that the mutant allele is translated slower than the wild type allele, which affects the folding of the protein.

The mutant allele has the same amino acid sequence as the mutant, but encodes a different protein. Sort of. That’s because the structure of a protein consists of more than just its sequence. Additionally, this is another reason why it’s so damn hard to predict protein folding.