Biology Reference
In-Depth Information
of combinatorial crossover libraries.
39
41
Here, efforts are made to swap distinct structural
elements and to minimize disruptions to the overall protein architecture. PCR primers
corresponding to the selected crossover regions are then designed and used to direct the desired
recombination events. This method is particularly useful for chimeragenesis from distantly
related family members, where structure is largely conserved though sequence similarity is low.
Site-Directed Diversification
Above, site-directed mutagenesis in which amino acid positions are intentionally
substituted with alternate, specified residues is considered
protein design.
The classification becomes a hybrid between rational and combinatorial design when
targeted positions are instead randomized. With this approach, information from
structural and/or biochemical analyses is used to predict which positions in a protein
should be mutated in order to improve a desired property. However, rather than
constructing a single new variant, a focused gene library is instead constructed, in which
the codon positions to be mutated are diversified to a desired extent. Traditionally this
involves synthesizing a series of oligonucleotides corresponding to the regions to be
mutated, but where the bases at the codons of interest are replaced by mixtures of
nucleotides. The diversified oligonucleotides are used as primers to amplify the gene, and
the fragments are assembled into a gene library using PCR. The mixture of nucleotides at
each base position can be specified, such as using an equimolar combination of A, T, G,
andC(referredtoas
'
rational
'
).
The choice of codon diversification depends on how many and which amino acids
one wants to be represented in the protein library.
'
N
'
), or using a mixture of only G and C (referred to as
'
S
'
Site-saturation mutagenesis refers to a codon set that includes all 20 amino acids. There are 64
possible codons and three possible stop codons in an NNN set, while the set NNS still encodes
all 20 amino acids and only contains one stop codon (UAG). Still, 32
n
possible combinations
from
n
simultaneously randomized positions quickly becomes an intractable number of library
members to screen as
n
grows, depending on the screening method used. One can instead
iteratively perform saturation mutagenesis at a small number of positions (e.g. residues lining
an enzyme
30
s substrate binding pocket randomized in pairs or in triplicate), again depending on
the number of variants that can be screened.
42,43
Alternately, individual positions throughout
the protein can be subjected to saturation mutagenesis and screened. After screening each
library, mutations showing improvements can be combined and tested for additivity, further
mutagenesis can be performed, etc. Other approaches involve PCR-based gene assembly using
primers or gene fragments which have been doped with randomized codons.
44
In this manner,
any one position targeted for mutagenesis still has a high probability of being coded for the
wild-type amino acid, depending on the level of mutant doping. This technique is therefore
similar to randommutagenesis, except only selected positions are mutated, and the mutations
are specified (on oligonucleotides) and can therefore be entire codons or even multiple codons
on a single oligonucleotide. Many variations on these types of codon mutagenesis techniques
have been reported.
'
In order to examine a greater number of simultaneously diversified residue positions
without creating intractable library sizes, one may consider ways of restricting the codon
alphabet to limit the set of amino acids encoded at each position. A variety of such
methods have been developed and applied, with the chosen codon usage resulting from
experimental observations, chemistry of the corresponding amino acids, and/or predictions
from molecular modeling.
45
Researchers have explored protein libraries where
n
positions
are diversified to as few as two possible amino acids.
46
In a noteworthy example,
Reetz and coworkers compared the use of fully randomized codons (NNK, where
'
K
'
represents G or T) to NDT codons (
represents A, G, or T), which corresponds to
12 codons that code for 12 possible amino acids.
47
Given the same number of codon
'
D
'
