Phage display technology has emerged as the dominant approach in antibody engineering. Here I describe my work in developing a high-throughput method of reliably generating intracellular antibodies. In my first data chapter, I present the first known high-throughput pipeline for antibody-phage display libraries of synthetic diversity and I demonstrate how increasing the scale of both target production and library selection still results in the capture of antibodies to over 50% of targets. In my second data chapter, I present the construction and validation of a novel scFv-phage library that will serve as the first step in my proposed intrabody pipeline. Antibodies obtained from this library will be screened for functionality using a novel yeast-two-hybrid approach and have numerous downstream applications. This high-throughput pipeline is amenable to automation and can be scaled up to thousands of domains, resulting in the potential generation of many novel therapeutic reagents.

Peptide recognition domains (PRD) bind to short linear motifs on their biological partners and are found in several cellular pathways including those found to be critical in tumorigenesis. In this study, I aimed to generate peptide probes against PRDs present on proteins involved in ovarian cancer. Using bioinformatics, I identified 66 potential PRDs present on these proteins. I then used peptide phage display to successfully generate peptides against 27 of the 66 domains. To validate my results, I performed an extensive literature review and structural analysis. For several cases, the phage-display derived binding preferences are similar to previously reported studies. However, for a subset of domains, I identified non-canonical binding preferences that have not been reported previously in literature. The binding preferences obtained in this study can be used to design intracellular probes for studying the role of these PRDs in biological pathways important in ovarian cancer.

The availability of cost-effective high throughput screening assays combined with an enhanced understanding of oncogenesis has driven the development of more potent, specific, and less toxic anti-cancer agents. At the forefront of these advances are immunoglobulin molecules and their fragments. However, difficulties in producing antibodies in sufficient quantity and quality for commercial application have driven the development of alternative systems that can produce antibodies efficiently and cost-effectively. This thesis focuses on the engineering of an antibody fragment referred to as a single chain variable fragment (scFv), which consists of antibody light and heavy chain variable domains fused together by a peptide linker. Although the use of scFvs circumvents many of the issue of full-length antibody production, they still possess their own unique set of difficulties, including stability. In this thesis, we explored the following strategies to increase scFv stability. First, we increased the number of linkers used to join the variable light and heavy domains. We constructed two linear and two cyclic permutated scFvs that contained additional peptide linkers. Two linear permutated scFvs, named Model 1 and Model 3, showed increased stability with calculated melting temperatures (Tms) exceeding that of the unpermutated scFv. The two cyclic scFvs were less stable with Tms less than that of the unpermutated scFv. Second, we mutated light and heavy variable domains by introducing prolines or mutating glycine to alanine in the variable domain framework regions. Sites for proline mutations and glycine to alanine mutations were identified and scFvs containing the mutations were purified and their thermal stability tested. Unfortunately, there were no discernible differences between purified scFv mutants and the control scFv. Third, we designed a new selection/screening strategy using phage display and yeast two-hybrid assays to identify complementarity determining regions on scFvs that increased intracellular stability. We used this strategy to isolate anti-Abl-SH3 scFvs. Transient expression of scFvs in K562 cells indicated that two anti-Abl-SH3 scFv decreased viability.

<p>Heparan sulfates (HSs) are a class of anionic carbohydrate chains found at cell surfaces and in the extracellular matrix where they interact with a number of proteins. HS is characterized by extreme structural heterogeneity, and has been implicated in a number of biological phenomenon like embryogenesis, morphogen gradient formation and signalling of growth factors such as FGF, PDGF etc. Despite the characteristic structural heterogeneity, evidence from compositional studies show that the HS structure is expressed in a tightly regulated manner, implying a functional significance, which is most likely in the modulation of cell behaviour through HS-protein interactions. The lack of molecular tools has, however, hampered the understanding of HS structures with functional significance. This work therefore aims at characterizing the structural requirements on HS involved in the interaction with the anti-HS phage display antibodies HS4C3, AO4B08 and HS4E4 and a selected growth factor PDGF-BB. The characterization was done with the help of tailored oligosaccharide libraries generated from sources bearing structural resemblance to HS.</p><p>The work has thus made available tools that preferentially recognize certain structural features on the HS chain and will aid in the further study of HS structure and its regulation. Evidence is also provided to support the notion that HS protein interactions can occur in multiple manners, utilizing any of the structural features on the HS chain.</p>

Les DD-peptidases sont des enzymes impliquées dans la synthèse de la paroi bactérienne. Elles sont aussi appelées "Penicillin Binding Proteins" (PBPs) car elles forment des acyl-enzymes stables avec des antibiotiques de type beta-lactames comme la pénicilline, la stabilité de ces complexes étant à l'origine de l'effet antibiotique. Certaines bactéries sont résistantes aux b-lactames grâce à la production de beta-lactamases, capables d'hydrolyser ces antibiotiques jusqu'à 10E8 fois plus rapidement que les PBPs selon un mécanisme impliquant deux étapes, d'acylation puis de désacylation. Il est généralement accepté que les beta-lactamases ont évolué à partir d'une PBP ancestrale en intégrant au site actif un mécanisme catalytique efficace pour la réaction de déacylation. L'objectif de cette thèse s'inscrit dans la compréhension des mécanismes d'évolution de la catalyse enzymatique au niveau moléculaire. Nous tenterons de reproduire un mécanisme évolutif en créant in vitro une activité b-lactamase à partir d'une DD-peptidase. La protéine de départ pour ce travail est la PBP-A de Thermosynechococcus elongatus, appartenant à une nouvelle famille de PBPs homologue aux beta-lactamases de classe A. L'analyse biochimique de cette protéine suggère que c'est une DD-peptidase, et une approche rationnelle par substitution d'un résidu dans le site actif de PBP-A a permis d'augmenter la vitesse de désacylation de l'acyl-enzyme formé avec la pénicilline d'un facteur 90. L'analyse des structures 3-D de PBP-A sauvage et mutante laisse ouverte la question de savoir comment évoluer cette PBP en beta-lactamase. Ainsi, pour l'évolution dirigée de cette protéine, deux banques ont été construites par approches aléatoire et semi-rationnelle. Il a été possible de sélectionner par "phage display" des enzymes améliorées pour la réaction d'acylation. L'obtention de mutants améliorés pour l'acylation ou la désacylation constitue des premiers pas vers l'évolution de PBP-A en beta-lactamase.

This thesis describes the development and use ofnon-immunoglobulin affinity proteins denoted affibodies asalternatives to antibodies in different immunotechnologyapplications. A 58 aa IgG Fc binding three-helix bundle domainZ, derived from staphylococcal protein A has been used asframework for library constructions, in which the face of themolecule involved in the native binding activity has beenengineered by combinatorial protein engineering. Recruting 13surface-located positions for simultanenous substitutionmutagenesis, using degenerated oligonucleotides for libraryassembly at the genetic level, two libraries differing in thechoice of codons were constructed to serve as general sourcesof novel affinity proteins. The libraries were adapted fordisplay onE. colifilamentous phage particles allowingin vitroselection of desired variants capable ofbinding a given target molecule. In selections using human IgAas target, several new IgA specific affibodies could beidentified. One variant ZIgA1, was further investigated and showed binding toboth IgA1 and IgA2 human subclasses as well as to secretoryIgA. This variant was further demonstrated uesful as ligand inaffinity chromatography purification for recovery of IgA fromdifferent samples including unconditioned human plasma.Affibodies of different specificities were also fused to otherprotein domains to construct fusion proteins of relevance forimmunotechnology applications. Using Fc of human IgG as genefusion partner, "artificial antbodies" could be produced inE. colias homodimeic proteins, where the antigenbinding was confered by N-terminally positioned affibodymoieties of different valencies. One area of application forthis type of constructs was demonstrated through specificdetection of the target protein by Western blotting. Exploitingthe uncomplicated structure of affibody affinity proteins, genefusions between affibodies and the homotetrameric reporterenzyme β-galactosidase were constructed, which could beproduced as soluble proteins intracellularly inE. coli. The potential use of such recombinantimmunoconjugates in immunotechnology was demonstrated in ELISAdot-blot and immunohistochemistry, where in the latter case IgAdepositions in the glomeruli of a human kidney biopsy could bespecfically detected with low background staining ofsurrounding tissues. In a novel format for sandwich ELISA, thepossible advantage of the bacterial origin of the affibodyclass of affinity proteins was investigated. As a means tocircumvent problems associated with the presence of humanheterophilic antibodies in serum, causing bakground signals dueto analyte-independent crosslinking of standard capture anddetection antibody reagents, assay formats based oncombinations of antibody and affibody reagents for capture anddetection were investigated and found to be of potentialuse. <b>Keywords:</b>phage display, combinatorial, affinity, IgAligand, immunohistochemistry, affibody-fusions

Heparan sulfates (HSs) are a class of anionic carbohydrate chains found at cell surfaces and in the extracellular matrix where they interact with a number of proteins. HS is characterized by extreme structural heterogeneity, and has been implicated in a number of biological phenomenon like embryogenesis, morphogen gradient formation and signalling of growth factors such as FGF, PDGF etc. Despite the characteristic structural heterogeneity, evidence from compositional studies show that the HS structure is expressed in a tightly regulated manner, implying a functional significance, which is most likely in the modulation of cell behaviour through HS-protein interactions. The lack of molecular tools has, however, hampered the understanding of HS structures with functional significance. This work therefore aims at characterizing the structural requirements on HS involved in the interaction with the anti-HS phage display antibodies HS4C3, AO4B08 and HS4E4 and a selected growth factor PDGF-BB. The characterization was done with the help of tailored oligosaccharide libraries generated from sources bearing structural resemblance to HS. The work has thus made available tools that preferentially recognize certain structural features on the HS chain and will aid in the further study of HS structure and its regulation. Evidence is also provided to support the notion that HS protein interactions can occur in multiple manners, utilizing any of the structural features on the HS chain.

Serine proteases are a family of enzymes with a wide array of functions across both eukaryotes and prokaryotes. Here we have attempted to produce the serine proteases rat mast cell protease 2 and mouse mast cell protease 5 in a culture of HEK 293 cells; and mouse mast cell protease 4, platypus granzyme B-like protease and platypus hypothetical protease in a baculovirus expression system. Following production we wanted to analyse these serine proteases using a phage display assay and a battery of chromogenic substrates.

<p>This thesis describes the development and use ofnon-immunoglobulin affinity proteins denoted affibodies asalternatives to antibodies in different immunotechnologyapplications. A 58 aa IgG Fc binding three-helix bundle domainZ, derived from staphylococcal protein A has been used asframework for library constructions, in which the face of themolecule involved in the native binding activity has beenengineered by combinatorial protein engineering. Recruting 13surface-located positions for simultanenous substitutionmutagenesis, using degenerated oligonucleotides for libraryassembly at the genetic level, two libraries differing in thechoice of codons were constructed to serve as general sourcesof novel affinity proteins. The libraries were adapted fordisplay on<i>E. coli</i>filamentous phage particles allowing<i>in vitro</i>selection of desired variants capable ofbinding a given target molecule. In selections using human IgAas target, several new IgA specific affibodies could beidentified. One variant Z_IgA1, was further investigated and showed binding toboth IgA1 and IgA2 human subclasses as well as to secretoryIgA. This variant was further demonstrated uesful as ligand inaffinity chromatography purification for recovery of IgA fromdifferent samples including unconditioned human plasma.Affibodies of different specificities were also fused to otherprotein domains to construct fusion proteins of relevance forimmunotechnology applications. Using Fc of human IgG as genefusion partner, "artificial antbodies" could be produced in<i>E. coli</i>as homodimeic proteins, where the antigenbinding was confered by N-terminally positioned affibodymoieties of different valencies. One area of application forthis type of constructs was demonstrated through specificdetection of the target protein by Western blotting. Exploitingthe uncomplicated structure of affibody affinity proteins, genefusions between affibodies and the homotetrameric reporterenzyme β-galactosidase were constructed, which could beproduced as soluble proteins intracellularly in<i>E. coli</i>. The potential use of such recombinantimmunoconjugates in immunotechnology was demonstrated in ELISAdot-blot and immunohistochemistry, where in the latter case IgAdepositions in the glomeruli of a human kidney biopsy could bespecfically detected with low background staining ofsurrounding tissues. In a novel format for sandwich ELISA, thepossible advantage of the bacterial origin of the affibodyclass of affinity proteins was investigated. As a means tocircumvent problems associated with the presence of humanheterophilic antibodies in serum, causing bakground signals dueto analyte-independent crosslinking of standard capture anddetection antibody reagents, assay formats based oncombinations of antibody and affibody reagents for capture anddetection were investigated and found to be of potentialuse.</p><p><b>Keywords:</b>phage display, combinatorial, affinity, IgAligand, immunohistochemistry, affibody-fusions</p>

Plasmodium vivax Duffy binding protein (DBP) is an essential ligand for reticulocyte invasion making it a premier asexual blood stage vaccine candidate. However, strain-specific immunity due to DBPII allelic variation may complicate vaccine efficacy, suggesting that an effective DBPII vaccine needs to target immune responses to conserved epitopes that are potential targets of strain-transcending neutralizing immunity. Anti DBPII monoclonal antibodies, which were previously characterized by COS7 cell binding assay as inhibitory and non-inhibitory to DBPII-erythrocyte binding, were mapped to DBPII gene fragment libraries using phage display. Inhibitory mAb 3C9 binds to a conserved conformation-dependent epitope in subdomain 3 while non-inhibitory mAb 3D10 binds to a linear epitope in subdomain 1 of DBPII. More definitive epitope mapping of mAb 3D10 was achieved using a random peptide library displayed on phage. Since DBP region II is mostly made up of alpha-helices, we used a randomized helical scaffold library, the Affibody library, displayed on phage, to determine epitope of conformation-dependent antibodies. The immunogenicity of the identified epitopes was evaluated in mice and the immune sera evaluated for binding to DBPII by ELISA and inhibition of DBPII-erythrocyte binding by the COS7 cell assay. Immune serum from the mAb3C9 epitope blocked DBPII-erythrocyte, suggesting this epitope could be a good subunit vaccine target.