The recombination strategy, called iterative truncation for the creation of hybrid enzymes (ITCHY), is based on generation of N- or C-terminal fragment libraries of two genes by progressive truncation of the coding sequences with exonuclease III followed by ligation of the products to make a single-crossover hybrid library (Fig. 1B). ITCHY was used to obtain fusions of glycinamide ribonucleotide transformylases encoded by Escherichia coli (PurN) and human (GART) genes sharing only 50% DNA sequence identity. N-terminal fragments of PurN were fused to C-terminal fragments of GART, and active clones were selected using a survival-selection in an E. coli auxotroph lacking either of two GAR transformylase activities. For comparison, Ostermeier et al. also performed a DNA shuffling experiment in which the same N-terminal (PurN) and C-terminal (GART) fragments were used as templates for reassembly of a complete shuffled hybrid.
Comparison of active clones produced by the two methods showed that ITCHY crossovers occurred over a wider range of residues than with shuffling, in which all the crossovers were found in the coding sequences of just four residues in a region of identical amino acid and virtually identical DNA sequence. With ITCHY, crossovers occurred throughout the coding region. And, since all sizes of the PurN and GART fragments were generated, fusions of diverse lengths were also possible. However, the functional proteins were fusion products in which the PurN N-terminal sequence was followed exactly by the residue in GART predicted from the sequence alignment. Thus, it seems that there is value to maintaining the "proper" gene length and finding the small subset of fusions where the crossovers correspond to sequence alignments (which is an integral part of the DNA shuffling strategy, but not of ITCHY). Furthermore, the crossovers occurred in core regions of the protein, in particular near the active sites of the enzymes. One might have expected to see crossovers in unstructured loops, where, in principle, substitutions, deletions, or insertions should be less disruptive. Clearly, this was not the case, and Ostermeier et al. offer the explanation that the structural complementarity of the PurN and GART protein fragments are highest in these highly conserved regions.
This is a preview of subscription content, access via your institution