Affinity Maturation of Ligand-Receptor Binding through Epitope Expansion Using a Cystine Knot Scaffold

Interactions between cell-surface receptors and their activating ligands mediate a host of biological processes governed by specific molecular recognition events. Engineering molecular recognition to manipulate these processes has long been an approach taken by the biotechnology field to develop therapeutic proteins. This has been accomplished by affinity maturing native interactions for enhanced complementarily or by randomizing entire regions of scaffold proteins, such as antibody CDRs, to select variants with desired binding properties. In nature, however, there are many examples of binding interactions occurring at the junction of multiple domains such that the binding epitope is formed by separate domains proximal in the complex’s tertiary structure.  One example can be found in platelet derived growth factor, a homodimeric ligand which can bind to its receptor only when the binding epitope is formed upon the association of the two monomers.  This illustrates that one approach nature uses to enhance molecular recognition is adding and evolving domains to native binding interactions, thus expanding the original binding epitope. Using this strategy, we affinity-matured an existing protein-protein interaction through the inclusion and evolution of a synthetic binding domain based on a cystine knot scaffold. We chose the Gas6-Axl system to evaluate our approach as the native interaction is very strong (K_D ~1nM) and will consequently provide a stringent test of this method. In addition, the Axl receptor tyrosine kinase is an emerging target in the treatment of many forms of cancer and the development of a soluble, high-affinity Axl variant capable of antagonizing the natural Gas6 ligand would have potential as a therapeutic agent.

The 28 amino acid cystine-knot peptide, Ecballium elaterium trypsin inhibitor II (EETI) has been shown to be a viable scaffold for engineering novel molecular recognition through the randomization of its disulfide constrained loops. We created a library of yeast-displayed EETI-Axl fusion proteins, in which the length and amino acid composition of two of the three loops of EETI-II were concurrently randomized. Sequencing of the initial library confirmed that wild-type Axl was not mutated, and the starting library retained high binding to Gas6 suggesting the fusion does not alter the native Gas6-Axl interaction. Yeast-displayed EETI-Axl (EAxl) variants with enhanced binding to Gas6 were then isolated by fluorescence-activated cell sorting.  We identified three unique EAxl clones which possess ~3.5 fold improvements in affinity to Gas6 over wild-type Axl. Off-rate experiments, in which equilibrium binding is followed by an irreversible unbinding step, indicate that this enhanced affinity is due to the mutants’ improved off-rates which are ~3.5 fold slower than wild-type Axl. Furthermore, the unbinding of the Gas6-EAxl complex is fit best by a double-exponential decay suggesting a complex mechanism of binding in which the EETI makes a significant contribution. We also show that the synthetic binding surface is comprised of residues from both engineered loops through individually reverting the loops back to the wild-type EETI loop sequence and comparing the binding affinities to wild-type and the original engineered EETI-Axl variants.

In summary, we describe an approach to affinity maturing the interaction of Gas6 ligand to the Axl receptor through epitope expansion using a cystine knot scaffold. Since the native Gas6-Axl interaction is strong (K_D ~1nM), this work provides a stringent test for affinity maturation by epitope expansion evolution. Also, to our knowledge this is the first example in which a binding face has been evolved on a knottin scaffold through simultaneous mutation of adjacent loops. We envision this approach as a broadly applicable strategy for enhancing ligand-receptor interactions.