Oligonucleotides enable many biotechnological applications; however they are easily degraded by nucleases. Many nucleotides modified at the 2’ position are degraded at decreased rates which improves oligonucleotide utility. Most applications of oligonucleotides rely on enzymatic synthesis. Unfortunately, native DNA polymerases do not recognize most useful modified nucleotide substrates. Directed evolution has been used to identify mutants of Taq DNA polymerase I (Taq) that recognize substrates with 2’ modifications. While mutant enzymes capable of modified nucleotide addition have been identified, to date, all of these enzymes are limited by their inability to synthesize full length modified DNA. Despite considerable efforts to evolve new activity there has been little work done to quantitatively characterize these evolved enzymes. This thesis work presents efforts to synthesize modified primers that will help comparatively and quantitatively characterize three enzymes previously evolved to recognize 2’ modified substrates. Using the methods developed in this thesis project, our lab will be able to characterize the relationship between the number of modified nucleotides in the primer terminus and the rate of modified and unmodified nucleotide addition. Future work will identify key enzymatic steps that limit extension in these enzymes with implications for the future design of Taq mutants capable of synthesizing long 2’ modified oligonucleotides.

Recent years have seen tremendous developments in methods for computational modeling of (bio-) molecular systems. Ever larger reactive systems are being studied with high accuracy approaches, and high-level QM/MM calculations are being routinely performed. However, applying high-accuracy methods to large biological systems is computationally expensive and becomes problematic when conformational sampling is needed. To address this challenge, classical force field based approaches such as free energy perturbation (FEP) and empirical valence bond calculations (EVB) have been employed in this work. Specifically: Force-field independent metal parameters have been developed for a range of alkaline earth and transition metal ions, which successfully reproduce experimental solvation free energies, metal-oxygen distances, and coordination numbers. These are valuable for the computational study of biological systems. Experimental studies have shown that the epoxide hydrolase from Solanum tuberosum (StEH1) is not only an enantioselective enzyme, but for smaller substrates, displays enantioconvergent behavior. For StEH1, two detailed studies, involving combined experimental and computational efforts have been performed: We first used trans-stilbene oxide to establish the basic reaction mechanism of this enzyme. Importantly, a highly conserved and earlier ignored histidine was identified to be important for catalysis. Following from this, EVB and experiment have been used to investigate the enantioconvergence of the StEH1-catalyzed hydrolysis of styrene oxide. This combined approach involved wildtype StEH1 and an engineered enzyme variant, and established a molecular understanding of enantioconvergent behavior of StEH1. A novel framework was developed for the Computer-Aided Directed Evolution of Enzymes (CADEE), in order to be able to quickly prepare, simulate, and analyze hundreds of enzyme variants. CADEE’s easy applicability is demonstrated in the form of an educational example. In conclusion, classical approaches are a computationally economical means to achieve extensive conformational sampling. Using the EVB approach has enabled me to obtain a molecular understanding of complex enzymatic systems. I have also increased the reach of the EVB approach, through the implementation of CADEE, which enables efficient and highly parallel in silico testing of hundreds-to-thousands of individual enzyme variants.

Enzymes are attractive catalysts in organic synthesis since they are efficient, selective and environmentally friendly. A large number of enzyme-catalyzed transformations have been described in the literature. If no natural enzyme can carry out a desirable reaction, one possibility is to modify an existing enzyme by protein engineering and thereby obtain a catalyst with the desired properties. In this thesis, the development of enantioselective enzymes and their use in synthetic applications is described.  In the first part of this thesis, enantioselective variants of Candida antarctica lipase A (CALA) towards α-substituted p-nitrophenyl esters were developed by directed evolution. A highly selective variant of CALA towards p-nitrophenyl 2-phenylpropanoate was developed by pairwise randomization of amino acid residues close to the active site. The E value of this variant was 276 compared to 3 for the wild type. An approach where nine residues were altered simultaneously was used to discover another highly enantioselective CALA variant (E = 100) towards an ibuprofen ester. The sterical demands of this substrate made it necessary to vary several residues at the same time in order to reach a variant with improved properties. In the second part of the thesis, a designed variant of Candida antarctica lipase B (CALB) was employed in kinetic resolution (KR) and dynamic kinetic resolution (DKR) of secondary alcohols. The designed CALB variant (W104A) accepts larger substrates compared to the wild type, and by the application of CALB W104A, the scope of these resolutions was extended. First, a DKR of phenylalkanols was developed using CALB W104A. An enzymatic resolution was combined with in situ racemization of the substrate, to yield the products in up to 97% ee. Secondly, the KR of diarylmethanols with CALB W104A was developed. By the use of diarylmethanols with two different aryl groups, highly enantioselective transformations were achieved. / At the time of the doctoral defense, the following paper was unpublished and had a status as follows:<strong>  </strong>Paper 5: Submitted.

Determining the properties of mutations is fundamental to understanding the mechanisms of adaptive evolution. The major goal of this thesis is to investigate the mechanisms of bacterial adaptation to new environments using experimental evolution. Different types of mutations were under investigations with a particular focus on genome rearrangements. Adaptive evolution experiments were focused on the development of bacterial resistance to antibiotics. In paper I, we performed stochastic simulations to examine the role of gene amplification in promoting the establishment of new gene functions. The results show that gene amplification can contribute to creation of new gene functions in nature. In paper II, the evolution of β-lactam resistance was studied by evolving S. typhimurium carrying a β-lactamase gene towards increased resistance against cephalosporins. Our results suggest that gene amplification is likely to provide an immediate solution at the early stage of adaptive evolution and subsequently facilitate further stable adaptation. In paper III, we isolated spontaneous deletion mutants with increased competitive fitness, which indicated that genome reduction could be driven by selection. To test this hypothesis, independent lineages of wild type S. typhimurium were serially passaged for 1000 generations and we observed fixation of deletions that significantly increased bacterial fitness when reconstructed in wild type genetic background. In paper IV, we developed a new strategy combining 454 pyrosequencing technology and a ‘split mapping’ computational method to identify unique junction sequences formed by spontaneous genome rearrangements. A high steady-state frequency of rearrangements in unselected bacterial populations was suggested from our results. In paper V, the rates, mechanisms and fitness effects of colistin resistance in S. typhimurium were determined. The high mutation rate and low fitness costs suggest that colistin resistance could develop in clinical settings. In paper VI, a novel Metallo-β-lactamase (MBL) with low resistance against β-lactam antibiotics was employed as the ancestral protein in a directed evolution experiment to examine how an enzyme evolves towards increased resistance. For most isolated mutants, in spite of their significantly increased resistance, both mRNA and protein levels were decreased as compared with the parental protein, suggesting that the catalytic activity had increased.

Enzymes are gaining increasing importance as catalysts for selective transformations in organic synthetic chemistry. The engineering and design of enzymes is a developing, growing research field that is employed in biocatalysis. In the present thesis, combinatorial protein engineering methods are applied for the development of Candida antarctica lipase A (CALA) variants with broader substrate scope and increased enantioselectivity. Initially, the structure of CALA was deduced by manual modelling and later the structure was established by X-ray crystallography. The elucidation of the structure of CALA revealed several biocatalytically interesting features. With the knowledge derived from the enzyme structure, enzyme variants were produced via iterative saturation mutagenesis (ISM), a powerful protein engineering approach. Several of these variants were highly active and enantioselective towards bulky esters. Furthermore, an extensively combinatorial protein engineering approach was developed and investigated. A CALA variant with a spacious substrate binding pocket that can accommodate an unusually bulky substrate, an ester derivate of the non-steroidal anti-inflammatory drug (S)-ibuprofen, was obtained with this approach. / At the time of the doctoral defence the following paper was unpublished and had a status as follows: Paper nr. 5: Manuscript

Cyanide is acutely toxic to the environment. However, this simple nitrile is used in several industrial applications especially the mining industry. Due to its high affinity to metals, cyanide has been used for years to extract gold and other precious metals from the ore. Cyanide nitrilases are considered for the detoxification of the industrial wastewaters contaminated with cyanide. Their application in cyanide remediation promises cheaper and safer processes compared to chemical detoxification. However, application of these enzymes in industry requires improving their characteristics. The goal of this dissertation is to better understand cyanide nitrilases, in particular the cyanide dihydratase from of Bacillus pumilus and Pseudomonas stutzeri and to improve their activity and stability. The lack of any high resolution structure of these enzymes calls for isolating or screening for mutants showing enhancement in enzyme properties. Described first is a simple and efficient method utilizing in vivo recombination to create recombinant libraries incorporating the products of PCR amplification. This method is useful for generating large pools of randomly mutagenized clones after error-prone PCR mutagenesis. Several parameters were investigated to optimize this technique; length of homology region, vector treatment, induction time and ratio of fragment to vector. Using error-prone PCR for random mutagenesis, several CynDpum mutants were isolated for higher catalysis at pH 7.7. Three point mutations, K93R, D172N and E327K increased the enzyme’s thermostability. The D172N mutation also increased the affinity of the enzyme for its substrate at pH 7.7 suggesting an effect on the active site. However, the A202T mutation located in the dimerization or the A surface rendered the enzyme inactive by destabilizing it. No significant effect on activity at alkaline pH was observed for any of the purified mutants. Lastly, an important region for CynDstut activity was identified in the C-terminus. This same region increased the stability of CynDpum compared to the wild-type enzyme. Also, CynDpum-stut hybrid was found to be highly more stable than CynDpum. This same hybrid exhibited 100% activity at pH9, a pH where the parent enzyme is inactive, and retained 40% of its activity at pH 9.5 making it a true pH tolerant mutant.

A new approach to the process of Directed Evolution is proposed, which utilizes different machine learning algorithms. Directed Evolution is a process of improving a protein for catalytic purposes by introducing random mutations in its sequence to create variants. Through these mutations, Directed Evolution explores the sequence space, which is defined as all the possible sequences for a given number of amino acids. Each variant sequence is divided into one of two classes, positive or negative, according to their activity or stability. By employing machine learning algorithms for feature selection on the sequence of these variants of the protein, attributes or amino acids in its sequence important for the classification into positive or negative, can be identified. Support Vector Machines (SVMs) were utilized to identify the important individual amino acids for any protein, which have to be preserved to maintain its activity. The results for the case of beta-lactamase show that such residues can be identified with high accuracy while using a small number of variant sequences. Another class of machine learning problems, Boolean Learning, was used to extend this approach to identifying interactions between the different amino acids in a proteins sequence using the variant sequences. It was shown through simulations that such interactions can be identified for any protein with a reasonable number of variant sequences. For experimental verification of this approach, two fluorescent proteins, mRFP and DsRed, were used to generate variants, which were screened for fluorescence. Using Boolean Learning, an interacting pair was identified, which was shown to be important for the fluorescence. It was also shown through experiments and simulations that knowing such pairs can increase the fraction active variants in the library. A Boolean Learning algorithm was also developed for this application, which can learn Boolean functions from data in the presence of classification noise.

In this study, the novel idea of in-vivo directed evolution was applied in order to achieve variants of the enzyme galactose oxidase with increased activity. This procedure was done under starvation conditions in Escherichia coli BL21 Star (DE3). Previous studies have been carried out in order to improve the activity of this enzyme using directed evolution methods. In this study, the same idea was used in-vivo, during stationary phase adaptive mutations inside the host organism, hence called in-vivo directed evolution. This method gave variants with improved enzyme activity as compared with the wild-type enzyme, and some variants showed activities that were even higher than the variants of previous directed evolution studies, hence making this method a promising approach for the random mutagenesis of genes of interest. The above mentioned mutations are carried out by a special group of polymerases, the error-prone polymerases. Phylogenetic analysis of these error-prone polymerases was also carried out in order to investigate the relationship between the number of error-prone polymerases and the level of complexity of organisms, and both the number of error-prone polymerases and the ratio of error-prone polymerases to total DNA polymerases of six organisms were studied. It was found that as the organism gets more complex, the number of error-prone polymerases and their ratio to the total polymerases increase.

B and T cells comprise the two major arms of the adaptive immune response tasked with clearing and preventing infection; molecular recognition in these cells occurs through antibodies and T cell receptors (TCRs), respectively. Highly successful therapeutics, clinical diagnostics and laboratory tools have been engineered from fragments of these parent molecules. The binding specificity, affinity and biophysical characteristics of these fragments determine their potential applications and resulting efficacies. Thus engineering desired properties into antibody and TCR fragments is a major concern of the multi-billion dollar biopharmaceutical industry. Toward this goal, we (1) designed antibody specificity using a novel computational method, (2) engineered thermoresistant Fabs by phage-based selection and (3) modulated binding kinetics for a single-chain TCR. In the first study, de novo modeling was used to generate libraries of FLAG peptide-binding single-chain antibodies. Phage-based screening identified a dominant design, and activity was confirmed after conversion to soluble Fab format. Bioinformatics analysis revealed potential areas for design process improvement. We present the first experimental validation of this in silico design method, which can be used to guide future antibody specificity engineering efforts. In the second study, the variable heavy chain of a moderately stable EE peptide-binding Fab was subjected to random mutagenesis, and variants were selected for resistance to heat inactivation. Thermoresistant clones where biophysically characterized, and structural analysis of selected mutations suggested general mechanisms of stabilization. Framework mutations conferring thermoresistance can be grafted to other antibodies in future Fab stabilization work. In the third study, TCR fragment binding kinetics for a clonotypic antibody were modulated by varying valence during phage-based selection. Binding affinity and kinetics for representative variants depended on the display format used during selection, and all TCR fragments retained binding to native pMHC antigen. This work demonstrates a general engineering platform for tuning protein-protein interactions. Using a combination of computational design and phage-based screening, we have identified antibodies and TCR fragments with improved binding properties or biophysical characteristics. The optimized variants possess a wider range of potential applications compared to their parent molecules, and we detail engineering methods likely to be useful in the engineering of many other protein-based therapeutics. / text

The conflict between HIV-1 and the host immune system plays out over a time-scale of months and years, and on a grander scale in the co-evolution of lentiviruses and the immune systems of their host species. Directed evolution of HIV-1 entry inhibitors using controlled randomization together with a display system offers a means of recapitulating one side of this conflict in vitro on an accelerated time-scale. To address limitations in existing display systems, we constructed a vector (pDQ1) integrating phage-display and mammalian-expression systems. This vector displays on phage when expressed in bacteria, and as an Fc-fusion when expressed in tissue culture, thus accelerating the iterative process of randomization, display, and characterization. We demonstrated the utility of this vector in the evolution of a CD4-mimetic peptide.