Ways of Introducing Variability into Genes

There are many ways of introducing mutations into genes of interest. The most commonly used is error prone polymerase chain reaction (EP-PCR) that exploits the low proofreading fidelity of Taq polymerase (Cadwell and Joyce, 1992). Thus, by varying the concentration of dNTPs and the divalent cation Mn²⁺, it is possible to obtain a range of introduced mutations typically from 0.1% to ~1% of the bases of the target DNA. Random point mutagenesis, that is, a base change at one of the three locations in the triplet that encodes a single amino acid, has an inherent limitation related to the structure of the genetic code itself. That is, depending on the degeneracy of the amino acid encoded by a particular triplet, one can only reach between three and seven amino acid substitutions per site. Compounding this problem, EP-PCR has been shown to exhibit considerable base change bias in that >70% of changes are seen from A and T, and <20% from C and G (Shafikhani et al., 1997). This further reduces the number of possible amino acid changes to less than the 4–7 that would occur for random substitutions. A DNA polymerase (Mutazyme, Stratagene, La Jolla, CA) was developed in which the base change proclivity is inverted from that of Taq, such that the two enzymes can be used in concert to minimize bias among variants.

Another powerful method of introducing changes into genes is to perform a partial digest with DNase followed by reassembly of the fragments in an autopriming PCR reaction and amplification of reassembled product with the addition of terminal primers (Stemmer, 1994a). This method exploits lack of fidelity in the reassembly reaction in which mutations are introduced at the borders of overlap extension reactions. Because DNase cuts randomly, the positions of introduced mutations occur randomly along the length of the target DNA. This method has been successfully used to generate a population of variants starting from a single parental gene. A limited analysis of the base changes introduced by this method suggests that it is less biased than EP-PCR. All of these methods suffer from the limited range of amino acids that can be reached by point mutagenesis as described above. To circumvent this limitation, a method called gene site saturation mutagenesis was devised in which oligonucleotides encoding all possible 19 amino acid substitutions at a particular site are used to make a library of variants that can be assayed for desired related activities (Desantis et al., 2003). Given sufficient resources, all possible substitutions can be made at every position along the amino acid chain to identify improved variants.