From molecules to organisms: Molecular recognition using proteins, peptides and peptide nucleic acids

This thesis describes experimental work I performed since 2004. It intends to give the reader an insight into molecular recognition processes that occur in biologically relevant systems. It approaches these processes in two different ways: rational design and non-rational selection; in both cases the main goal is to obtain strong and selective ligands for a variety of targets.;The first chapter provides the reader an introduction to biological systems, and the molecular recognition processes which govern their interactions. Toward the end of this chapter, random approaches for the selection of biomolecules with specific binding properties are presented. The use of G-rich PNA recognition of G-quadruplex DNA forming sequences is described in Chapter 2 of this thesis work. In this case we used rational design in order to improve the selectivity of our PNA molecules for a homologous DNA sequence over a complementary DNA sequence. In order to achieve our goal we incorporated modifications into the PNA backbone as well as abasic sites in the sequence of the PNA. These modifications resulted in selectivity that could be tuned depending on the goal of the project.;Recognition of PNA:DNA hybrid structures using engineered antibodies is described in Chapters 3 and 4. In these two chapters I describe the use of randomized libraries of proteins, in both cases derived from human antibodies, to find ligands that can bind strongly and selectively to PNA:DNA hybrid quadruplexes similar to the ones analyzed in Chapter 2. The selection of engineered antibodies took place using two different display formats: yeast display (Chapter 3) and phage display (Chapter 4). In addition, two different types of engineered antibodies were employed: heavy and light chain scFvs and heavy chain only Dabs in the two different display formats, yeast and phage respectively. These selections did not result in an antibody that meets our requirements: high affinity and good selectivity. Instead using scFvs in the yeast display context we found proteins that seem to bind to the hybrid quadruplexes but the affinity as well as the selectivity were below our expectations. Interestingly some of the analyzed scFvs were able to bind to the target on yeast surface; however this binding was not reproducible when the proteins were free in solution. On the other hand, when using phage display, no good binders were obtained.;The last chapter describes the rational optimization of a peptide that was previously selected using phage display to bind spores. Here we demonstrate that is possible to combine these two strategies, random selection and rational design, in order to enhance the affinity of binding. We created a multivalent ligand that binds to the spores several orders of magnitude better than the monovalent peptide. In addition, this peptide we created is readily modified and can be used in different contexts. While the rational design and use of randomized libraries to find ligands for different targets are completely different approaches, both are valid and offer advantages and disadvantages. This thesis intends to highlight the beauty of these strategies and how they complement each other.