Combinatorial protein libraries based on a protein template offer a vast potential for deriving protein variants harboring new receptor specificity while retaining other tem-plate functions to serve as library search-engines, cell-routing sequences and therapeutic domains. This concept was tested with the design and synthesis of protein libraries where short random peptide motifs were embedded directly within the catalytic A subunit of the bacterial ribosome-inactivating protein (RIP) known as Shiga-like toxin 1 (SLT-1). More precisely, a seven amino acid peptide epitope (PDTRPAP) was inserted between residues 245-246 of its A subunit (SLT-1APDTRPAP) and shown to preserve catalytic function while exposing the epitope. SLT-1 A chain libraries harboring tripep-tide and heptapeptide random elements were subsequently constructed, screened and shown to express more than 90% of expected cytotoxic A chain variants. Finally, more than 9,000 purified SLT-1 A chain variants were screened using their ribosome-inactivating function in a cell-based assay to identify mutants that are able to kill human melanoma 518-A2 cells. This search led to the striking discovery of a single chain RIP that displays selectivity for a panel of human melanoma cell lines as well as minimal immunogenicity when injected repeatedly into mice. This directed evolution of a RIP template provides a broad platform for identifying cell type specific cytotoxic agents.

Combinatorial protein libraries based on a protein template offer a vast potential for deriving protein variants harboring new receptor specificity while retaining other tem-plate functions to serve as library search-engines, cell-routing sequences and therapeutic domains. This concept was tested with the design and synthesis of protein libraries where short random peptide motifs were embedded directly within the catalytic A subunit of the bacterial ribosome-inactivating protein (RIP) known as Shiga-like toxin 1 (SLT-1). More precisely, a seven amino acid peptide epitope (PDTRPAP) was inserted between residues 245-246 of its A subunit (SLT-1APDTRPAP) and shown to preserve catalytic function while exposing the epitope. SLT-1 A chain libraries harboring tripep-tide and heptapeptide random elements were subsequently constructed, screened and shown to express more than 90% of expected cytotoxic A chain variants. Finally, more than 9,000 purified SLT-1 A chain variants were screened using their ribosome-inactivating function in a cell-based assay to identify mutants that are able to kill human melanoma 518-A2 cells. This search led to the striking discovery of a single chain RIP that displays selectivity for a panel of human melanoma cell lines as well as minimal immunogenicity when injected repeatedly into mice. This directed evolution of a RIP template provides a broad platform for identifying cell type specific cytotoxic agents.

This study is part of a project which aims to understand evolution of genetic code together with structural and functional analysis of prebiotic proteins. The repertoire of amino acids in the first proteins was probably developing in time and it influenced the development of structure and function of today's proteins. First amino acid alphabet was apparently only half of the size of present alphabet, which contains twenty amino acids. These ten amino acids were probably prebiotically available from endogenous and exogenous sources. This work includes cell-free expression and purification of two randomized protein libraries (containing approximately 1011 variants) with various amino acid composition and following comparison of their propensity to form secondary (using circular dichroism) and tertiary (using proteolytical analysis of sequences) structures. First library contains only ten probably prebiotically available amino acids; second library contains all twenty amino acids in today's genetic code. This project could help us understand benefits of genetic code expansion in terms of developing structure in protein sequences. The whole research could theoretically contribute a few basic questions not only in the fields of protein evolution but also in areas of synthetic biology or protein...

A wide range of screening strategies have been employed to isolate antibodies and other proteins with specific attributes, including binding affinity, specificity, stability and improved expression. However, there remains no high-throughput system to screen for target-binding proteins in a mammalian, intracellular environment. Such a system would allow binding reagents to be isolated against intracellular clinical targets such as cell signalling proteins associated with tumour formation (p53, ras, cyclin E), proteins associated with neurodegenerative disorders (huntingtin, betaamyloid precursor protein), and various proteins crucial to viral replication (e.g. HIV-1 proteins such as Tat, Rev and Vif-1), which are difficult to screen by phage, ribosome or cell-surface display. This study used the â-lactamase protein complementation assay (PCA) as the display and selection component of a system for screening a protein library in the cytoplasm of HEK 293T cells. The colicin E7 (ColE7) and Immunity protein 7 (Imm7) Escherichia coli proteins were used as model interaction partners for developing the system. These proteins drove effective â-lactamase complementation, resulting in a signal-to-noise ratio (9:1 – 13:1) comparable to that of other â-lactamase PCAs described in the literature. The model Imm7-ColE7 interaction was then used to validate protocols for library screening. Single positive cells that harboured the Imm7 and ColE7 binding partners were identified and isolated using flow cytometric cell sorting in combination with the fluorescent â-lactamase substrate, CCF2/AM. A single-cell PCR was then used to amplify the Imm7 coding sequence directly from each sorted cell. With the screening system validated, it was then used to screen a protein library based the Imm7 scaffold against a proof-of-principle target. The wildtype Imm7 sequence, as well as mutants with wild-type residues in the ColE7- binding loop were enriched from the library after a single round of selection, which is consistent with other eukaryotic screening systems such as yeast and mammalian cell-surface display. In summary, this thesis describes a new technology for screening protein libraries in a mammalian, intracellular environment. This system has the potential to complement existing screening technologies by allowing access to intracellular proteins and expanding the range of targets available to the pharmaceutical industry.