Search

B-cell activation factor (BAFF) and A proliferation-inducing ligand (APRIL), two cytokines regulating B-cell homeostasis, are known to play important roles in the pathogenesis of B-cell malignancies. In B-cell malignancies, significant up-regulation and receptor-mediated signalling of BAFF and APRIL have been shown to facilitate tumour proliferation, treatment-resistance, and contribute to immune suppression in the tumour microenvironment (TME). Multiple myeloma (MM) and diffuse large B-cell lymphoma (DLBCL) are two of the most prevalent types of B-cell malignancies. Despite recent approvals of novel therapeutics such as CAR T-cell therapy (CAR T), disease relapse and treatment resistance continue to pose clinical challenges, often resulting in patient mortality. Here, we found elevated expression levels of BAFF, APRIL and their receptors, and observed the resulting activation of critical survival signalling pathways in patient specimens of MM and DLBCL. To effectively neutralize BAFF and APRIL-mediated signalling, we adopted a ligand trap approach, and developed a decoy receptor by fusing the extracellular cysteine rich domain (ECD) of the BCMA (B-cell maturation antigen) receptor (soluble BCMA, or sBCMA) with a human IgG1 Fc domain, to initially test our hypothesis. Next, to enhance the therapeutic efficacy of the approach, we engineered a mutant sBCMA, and developed a second-generation Fc fusion protein with significantly increased binding affinities toward BAFF and APRIL. In vitro and in vivo studies demonstrated that our sBCMA-Fc ligand trap approach is efficacious at inhibiting tumour growth with minimal off-target normal tissue toxicities, providing preclinical evidence to support the use of our molecules as a treatment for MM and DLBCL.

In eukaryotes, DNA is packaged into repeating units called nucleosomes, which consist of ~150 base pairs of DNA wrapped around eight proteins called histones. These histones are often laden with post-translational modifications (PTMs), and while many PTMs have been shown to have profound effects on gene expression and chromatin organization, most remain uncharacterized. This thesis strives to elucidate the functions of PTM containing histones. Chapter 1 describes a light-driven reaction that site-specifically inserts sidechains of interest into histones via radical addition to previously installed dehydroalanine residues, with a scope encompassing many native and PTM containing sidechains, as well as several reactive functional groups. Chapter 2 utilizes the reaction from Chapter 1, as well as others from our group, in several independent applications: The enzymatic removal of installed histone PTMs is tested, histone methylation and phosphorylation are assessed in the structural context of a nucleosome, PTM processing on fluorinated histone sidechains is tracked by 19F-NMR, and interaction partners are crosslinked to histones bearing electrophilic sidechains. Chapter 3 studies histones in more complex, living systems. The understudied ability of histones to penetrate cell membranes is investigated, revealing the incorporation of extracellular histones into the chromatin of nearby cells, provoking a strong inflammatory and immune response. Furthermore, several light-driven histone modification reactions are attempted in the translucent model organism, the zebrafish. In conclusion, this thesis describes powerful tools for modifying histones and applies them to relevant problems in epigenetics, revealing unknown roles and functions of histones and their PTMs.

Compartmentalisation is a universal feature of life, allowing organisms to separate their functions from each other and the environment. It enables greater control and complexity for cellular processes. Taking a cue from nature, engineered compartmentalisation has become one of the top priorities of synthetic biology. While many alternative strategies have emerged, building a synthetic membranous organelle de novo remains prohibitively difficult, despite its potential as a tool for metabolic engineering. This lack of an orthogonal synthetic organelle especially hampers the development of the nascent plant molecular farming industry. In this thesis, I describe the design and characterisation of a novel synthetic membranous compartment derived from the plant endoplasmic reticulum (ER), and proof-of-principle experiments to showcase its potential as a bioproduction platform in plants. The formation of the compartment is triggered by overexpression of a single recombinant scaffold protein, and I investigated the relationship between the structure of the scaffold protein and the morphology of the resultant compartment. I also compared the compartment to a conceptually similar ER rearrangement called organised smooth endoplasmic reticulum to determine their relation. Notably, the formation of the compartment had very little impact on the overall health of the host plant, and successful pilot experiments for the purification of the scaffold and other proteins anchored to it, as well as running enzymatic reactions on the surface of the compartment demonstrate the great potential of this system to serve as a plant synthetic biology tool. With further development, this platform could be a valuable asset to the plant biosynthesis and recombinant protein production industry.

Self-assembling protein tools are highly desirable in the field of synthetic biology. They enable the creation of new macromolecular structures for novel applications, by providing users with the ability to 'build' with proteins post-translationally: both in vivo and in vitro. In this thesis, I investigate and describe applications for a range of self-assembling protein tools. These include the BslA protein, SpyTag/SpyCatcher and SnoopTag/SnoopCatcher peptide/protein pairs, mVirD2 protein, and split inteins. For each of these tools I outline their capabilities, and then demonstrate a new application, either individually or in tandem. BslA is an amphiphilic protein that has the remarkable ability to self-assemble into monolayers on hydrophobic glass surfaces. Using a BslA-SpyTag or BslA-SnoopTag fusion protein, I show that BslA monolayers can be formed and subsequently functionalised via these tags with proteins attached to SpyCatcher or SnoopCatcher respectively. I show proof of principle with the attachment of fluorescent proteins to the glass surface via these specific tags. Fluorescent labelling of the BslA monolayer with these proteins provides a better understanding of the monolayer properties, specifically with the movement of proteins within the monolayer. Further expansion of the toolbox for functionalising BslA surfaces is achieved using the mVirD2 protein, which forms a covalent bond with DNA at a specific recognition site. This enables DNA to bind to the glass surface through a series of specific and covalent interactions: first to mVirD2-SpyCatcher fusion protein, and then to BslA-SpyTag. These tools are utilised to test different applications with biotechnological relevance. The applications proposed and tested in this thesis include: screening for new protein/peptide interactions, nanopillars for anti-glint purposes, and real-time cell-free analyte monitoring. Within the cell-free experiments, I focus on the capabilities of a BslA surface to capture proteins expressed in situ, and test cell-free protein expression from DNA covalently-bound to the surface. Following this, I attach protein biosensors to the BslA surface via SpyTag/SpyCatcher peptide/protein pairs and also use them to monitor fluctuations in phosphate levels during a cell-free reaction as it proceeds. I also help with characterisation of an expanded library of split inteins for protein trans-splicing, by testing the efficiency of splicing and comparing the orthogonality of the library in vitro. Subsequently, I demonstrate the use of split inteins in the development of a protocol for the solid-phase assembly of large, extremely repetitive proteins using split intein trans-splicing. I provide two protocols, that enable the ligation of six protein units, using either five orthogonal split inteins simultaneously, or using just two split inteins sequentially. This enables the creation of repetitive proteins to a length that would be difficult to achieve in vivo. In summary, this thesis presents new applications for self-assembling protein systems and further characterises the tools that underpin them.

Botulinum toxins (BoNTs) comprise a family of extremely potent neurotoxins, which have been harnessed as muscle relaxants for the treatment of a wide variety of debilitating diseases, particularly for the treatment of movement disorders. BoNT entry into neurons leads to destructive cleavage of cellular proteins critical to vesicle fusion and neurotransmitter release at the neuromuscular junction. A dual receptor model has been proposed for BoNT binding to target neurons, comprising a low affinity ganglioside interaction followed by a higher affinity interaction with a protein receptor. A deeper understanding of the molecular nature of these interactions will facilitate the generation of modified BoNT proteins with novel characteristics and significant therapeutic potential. This project was aimed principally at molecular characterisation of the interaction of BoNT/A with gangliosides, using a combination of computational, biochemical, and biophysical methods. Specific achievements included: 1) The development and optimisation of two biophysical protocols for measuring Botulinum neurotoxin binding to gangliosides. 2) The preparation of a well curated carbohydrate database that contained all known structures of protein-carbohydrate. 3) The generation of a comprehensive and meaningful benchmark for algorithms that are designed to predict affinity values for protein-small molecule interactions. 4) The training of a state-of-the-art machine learning algorithm that can predict affinity values for protein-small molecule interactions. 5) The use of computational and biophysical approaches to explore the specificity and selectivity of ganglioside binding pockets. The objective was to make contributions to the development of an engineered BoNT with novel binding properties and therapeutic potential. In addition, a set of novel tools and methodologies that can be applied across most types of structural data was developed. In addition to the main project this thesis describes a series of collaborative efforts, not directly related to BoNTs, that were undertaken during my PhD. This section focuses mainly on projects related to the COVID-19 pandemic, which heavily disrupted ordinary research work, as well as a parallel project modelling the proteome of Mycobacterium abscessus.

How do new enzymatic motifs emerge in evolution? Charting new mechanistic solutions for biocatalytic challenges is a core obstacle in the constant evolutionary adaptation of living organisms. If understood and harnessed, these processes hold the potential to revolutionise industrial processes in the chemical and pharmaceutical industry. The recent massive release of new, man-made substance classes into the environment in the form of pesticides provides a dynamic testing ground to observe protein evolution in fast-forward mode. In this global experiment, we can observe the emergence of new enzymatic activities and compare the adaptability of different catalytic motifs. Among the xenobiotic substances used, phosphotriester pesticides range among the most toxic. Phosphotriester pesticides were designed to irreversibly inhibit the catalytic triad of synaptic acetylcholinesterase. Numerous enzymes from different evolutionary origins have rapidly evolved to hydrolyse phosphotriesters since. Interestingly, all known enzymes have convergently reached the same mechanistic solution by recruiting divalent cations as a cofactor for metal-ion catalysis. The starting point of this study is the α/β hydrolase P91. This enzyme was recently identified in a functional metagenomic screening and displays weak promiscuous phosphotriesterase (PTE) activity. P91's PTE activity is remarkable because, in contrast to other currently known PTEs, it utilises a fundamentally different, metal-independent mechanism. At the same time, it has the same fold and a very similar active site configuration as acetylcholinesterase, the biochemical target of most organophosphate pesticides and nerve gases. In this dissertation, I subject P91 to directed evolution in order to explore the evolvability of its catalytic motif to this new enzymatic activity (chapter 2). To be able to cover otherwise inaccessible sequence space, I use droplet microfluidics to screen large libraries of enzyme variants in high throughput. Within only two rounds of evolution, P91's phosphotriesterase activity could be increased by 400-fold. The improved enzyme displays a catalytic efficiency close to 10^6 M^(−1) s^(−1) and a turnover rate of > 10 s^(−1). This activity is comparable to the catalytic efficiencies of many metal-dependent 'conventional' phosphotriesterases. These result highlight that enzymes can be evolved to hydrolyse phosphotriester pesticides using a catalytic Cys-His-Asp triad, independent of metal cofactors. Detailed mechanistic analysis reveals that P91 is pre-disposed for the turnover of organophosphates due to the intrinsic suitability of its cysteine nucleophile (chapter 3). This residue is able to quickly break down the otherwise unresolvable intermediate providing the initial foothold for rounds of directed evolution to improve upon the rate-limiting initial formation of the covalent intermediate. Further phylogenetic exploration of P91 homologues reveals that this promiscuous activity is widespread within the dienelactone hydrolase protein family, a cluster of proteins in the α/β hydrolase superfamily, which possess a Cys-His-Asp catalytic triad (chapter 4). By detailed structural analysis and targeted engineering I show that the determinant for the level of displayed PTE activity is not a specific arrangement of specificity-determining residues in the active site (chapter 5), but rather the length and flexibility of loops surrounding the active site (chapter 6). In summary, these results highlight the enormous potential for synergy by combining focussed, combinatorial libraries with high-throughput droplet microfluidics. In combination, these techniques can establish new mechanistic tracks and find efficient biocatalytic solutions that - despite strong natural selective pressure - were previously unseen in Nature.

Proteins are a valuable class of therapeutics because proteins, like antibodies or non-antibody scaffolds, can bind their targets with high selectivity and affinity. The upper gastrointestinal tract, however, rapidly inactivates protein-based targeting reagents. Compared to injections, oral administration of protein therapies is favourable for treating gastrointestinal disease. In this thesis, I have developed a protein scaffold termed a gastrobody, that is suited to applications in the gastrointestinal tract. Gastrobodies are derived from Kunitz-type soybean trypsin inhibitor (SBTI). SBTI is resilient to digestion with stomach-like concentrations of pepsin and pH 2. Furthermore, SBTI is stable in the presence of intestinal proteases and bile acids, as well as being resilient to high temperatures. I used the protein modelling software Rosetta to assist design of a binding surface on SBTI. I also established phage display of SBTI to enable directed evolution. As proof of concept, gastrobody binders to domains of two toxins of the enteric pathogen Clostridium difficile were developed. After optimisation of the binding surface and affinity maturation, I identified gastrobodies that bound their target with nanomolar affinity. Gastrobodies bind their target after incubation with physiological concentrations of gastrointestinal proteases, including pepsin at pH 2. Lastly, I explored alternative gastrobody backbones using structural homologues of SBTI. The work presented in this thesis may provide a foundation for generating binding reagents for therapeutic and diagnostic use in the gastrointestinal tract.

Incorporation of fluorine into biomolecules provides versatile opportunities for clinical and pre-clinical imaging, with ¹⁸F-PET and ¹⁹F-MRI and as reporter tags for "zero-background" ¹⁹F-NMR. Considerable effort has been made towards achieving direct fluorination of biomolecules, however nucleophilic fluorination methods typically require anhydrous conditions and organic solvents, while electrophilic methods suffer from high reactivity of fluorine gas and low specific activity, conditions which are protein incompatible. There are currently no known methods for direct C-F bond formation on large biomolecules, however the fluorinase enzyme has been shown to allow for direct fluorination of peptides in aqueous media using inorganic fluoride. Enzymatic catalysis provides an alternative to standard chemical methods, allowing for reactivity under mild conditions. Herein, a method for direct nucleophilic site-selective incorporation of fluoride into a protein containing an engineered recognition motif in aqueous media is described. Initial work was performed on a small-molecule model to allow investigation of critical parameters pertaining to the reactivity of this enzyme. Substrate promiscuity of the enzyme was evaluated alongside alterations to the co-factor/additive of the enzyme to improve reactivity. In addition, a side-product of the enzymatic reaction was observed and its role was investigated. Optimised conditions from the small-molecule reactions were applied for on-protein work and allowed for the first example of direct C-F bond formation on a protein in aqueous media. The compatibility of this methodology was assessed with a range of different proteins and further assessed for the ¹⁸F-fluorination of a protein. An alternative approach to ¹⁸F-labeling of biomolecules was also explored in this work, where three enzymes were combined in a reactor to allow for transformation of the radiotracer, [¹⁸F]FDG to another radiotracer of interest, [¹⁸F]FDT. The enzymes were immobilized and introduced into a cartridge, and flow-chemistry was adopted to allow for easy purification, facile handling, removal of endotoxins, and possible automation of this process in the future. We anticipate that the methodologies described in this work will contribute towards ready labelling of complex biomolecules under diverse mild conditions for potential clinical applications.

There is an ever-growing demand for diagnostic relevant biological reagents. However, the high cost of these biomaterials limits their accessibility to burdened low- and middle-income countries. This thesis explores a synthetic biology approach to deliver a recombinant Geobacillus stearothermophilus (BST) DNA polymerase capable of low-cost purification, ease-of-use, and compatible with sensitive and specific loop mediated isothermal amplification (LAMP) assays. Fusing the enzyme with a silaffin tag (R5) and a fluorescent protein (mCherry) enabled rapid and cost-effective purification through unmodified silica immobilisation and built-in visual tracking from protein production to diagnostic assay, respectively. With the immobilised enzyme compatible with standard LAMP assays, additional elution steps were not necessary simplifying the purification from more traditional methods. The activity of the synthetic polymerase was highly dependent on the primer set for the target assay but had comparable outputs to commercial LAMP assays. In all 3 trialled sites (Cambridge, Ghana, Malaysia) the construct could be produced from a single flask culture (0.5L E. coli culture), at a rate of 2500 assays per 24 hours and achieve a limit of detection as low as 10 copies of target DNA. The synthetic polymerase was further evaluated with a Plasmodium falciparum diagnostics trial of 500 clinical venous blood samples, in collaboration with the West African Centre for Cell Biology of Infectious Pathogens. Achieving a maximum sensitivity and specificity of 69% and 95%, revealed some protocol discrepancies, but the recombinant protein showed promise with a faster turnaround time and better tolerance to sample contamination (blood) when compared to more standard nucleic acid amplification diagnostic methods. In addition, this thesis initiates consideration of an extremely thermostable red fluorescent protein through structural analysis of mCherry and thermal green protein (TGP) for future integration with other polymerases.

The surprising discovery of intramolecular isopeptide bonds within Gram-positive bacteria pili and surface adhesins gave rise to the development of a new type of protein conjugation method, namely the Tag/Catcher system. Among the different kinds of Tag/Catcher platforms, SpyTag/SpyCatcher is the leading system with more than 200 articles utilising the technology due to the rapid and buffer-independent nature of the reaction. In this thesis, I have developed a purification method called Spy&Go to complement the SpyTag/SpyCatcher platform. Spy&Go enables the purification of proteins using SpyTag as the affinity peptide tag. Spy&Go thereby eliminates the dependence on extraneous tags for purification of proteins intended for SpyCatcher-based protein conjugation reactions. Spy&Go was developed by engineering SpyCatcher2.1, a fast reacting variant of SpyCatcher, to become SpyDock by removing reactivity whilst maintaining tight binding to SpyTag. The purity of proteins purified through Spy&Go was comparable to the purity of proteins purified by Ni-NTA using a polyhistidine tag. Subsequently, a different route for elution of SpyTag-proteins was explored by conferring pH-sensitivity to a SpyDock variant through the introduction of histidine residues at the SpyTag003-binding interface. With the combined effects of three histidine mutations, SpySwitch was developed, which can bind SpyTag003 at pH 8.0 and release at pH 5.0. SpyTag/SpyCatcher has been used to modularly multimerise proteins for vaccination, nanomedicine and making enzymatic nanoreactors. To extend the scope of SpyCatcher-based protein multimerisation platforms, I developed a Spy oligomerisation toolbox composed of SpyCatcher002 fused to coiled coils, to multimerise SpyTag-proteins into dimer, trimer, tetramer, pentamer, hexamer and heptamer assemblies. The SpyCatcher002-oligomers enable a rapid and scalable route to multimerise different proteins into defined assemblies. An agonistic nanobody against Death Receptor 5 was multimerised through SpyCatcher002-oligomers and showed a stoichiometry-dependent effect on breast cancer cell line apoptosis upon receptor activation.

Protein phosphatases are regulatory enzymes with unique substrate specificity for serine/threonine/tyrosine/histidine phosphorylated proteins and phosphoinositides. They mediate critical cell processes such as cell metabolism, cell cycle, growth and differentiation. For example, phosphoinositide phosphatases act as regulatory enzymes that mediate critical cellular processes such as cell signalling, membrane trafficking, cell cycle and growth. Loss-of-function mutations in these enzymes are linked to a plethora of disease conditions. Tumour suppressor PTEN is the most frequently mutated gene in various forms of human cancer; Myotubularins are mutated in X-linked myotubular myopathy (XLMTM) and Charcot-Marie-Tooth disease type 4B (CMT4B); Synaptojanin I is mutated in early-onset progressive Parkinsonism. Engineered PTPs are therefore seen as useful tools to enhance our understanding of signalling pathways involving phosphorylation events. Here, we used a combination of structure-guided rational and computational design to alter the substrate specificity of Low Molecular Weight Protein Tyrosine Phosphatase (LMW-PTP), from tyrosine-specific to PI(3,5)P2-, PI(3)P-, PI(4)P-, and PI(5)P-specific phosphatase (Low Molecular Weight Phosphoinositide Phosphatase - LMWPIP). Four LMW-PIP mutants dephosphorylate phosphoinositides with single or broad substrate specificity. We combined Rosetta Enzyme designs containing frequently introduced mutations with molecular docking studies to assess binding affinity to the ligand substrate. This approach facilitates the design selection process by reducing the number of candidate mutants to be validated experimentally. We also designed mutants to be tested for specificity towards serine/threonine phospho-peptide. Finally, we set up a system in yeast that can be used for in vivo characterisation of LMWPTPB mutants, by cloning and confirming the expression of LMWPTP in yeast using western blot and RT-PCR. Engineered LMW-PIP mutants can be a useful tool in investigating lipid-regulated cell signalling pathways, and in the long term, could serve as biotherapeutics in enzyme replacement therapy.

Tandem-repeat proteins differ from globular proteins, both in their biophysical characteristics and in how they interact with their respective partners, yet they comprise nearly one third of the human proteome and are central to many cellular processes and disease phenotypes. Repeat proteins have been shown to behave like nano-sized biological springs: they are flexible, dynamic and elastic. Using coarse-grained models, I discuss how intrinsic flexibility may arise in repeat proteins and how it could be crucial for the biological function of two systems: PR65, the scaffold protein of the protein phosphatase 2A, and Rap proteins, which are involved in quorum sensing. To interrogate α-solenoids at physiologically relevant forces, I performed force spectroscopy experiments using a dumbbell optical tweezers set up for which it is necessary to attach the relevant protein to DNA. As PR65 is not amenable to current DNA-protein attachment methods, I developed a protocol that allows the cross-linking of DNA oligos to proteins using bio-orthogonal chemistry. I then explored the mechanics of the natural repeat protein, PR65, and a series of designed TPR proteins. I find that these proteins respond to forces in a novel manner which is significantly different to what has been previously reported. TPRs unfold and refold in quasi-equilibrium at constant force without energy loss. In contrast, PR65 unfolds in separate domains and refolds along an entirely different pathway. In conclusion, my doctoral studies explore the physical characteristics of repeat proteins in more detail. Using both experimental and computational techniques, I provide unique perspectives on different aspects of their mechanical and dynamic capabilities. This work provides the basis for future investigations of how such interesting mechanical behaviour relates to biological function. Are repeat proteins simply a molecular recognition platforms for their multitude of binding partners, or do their mechanics matter in a biological context?

There is a clear need for affordable point-of-care diagnostics, especially in vulnerable low- and middle-income countries where access to laboratory-based testing is limited. This thesis explores whether recombinant protein technology in combination with a low-cost support matrix could provide a basis for an inexpensive, simple, and robust production process for the bio-sensing element of a diagnostic that would be amenable to manufacture in resource constrained settings. Silica was selected as the solid support for this work, given its biocompatibility and wide-availability (including extraction from natural sources, like sand, as demonstrated here). By employing an affinity binding sequence for silica in fusion with the central assay reagent protein targeting the analyte, simultaneous isolation and immobilisation onto silica carrier particles was achieved directly from lysate. In addition, the incorporation of a coloured fluorescent protein in the fusion enabled the protein production and immobilisation to be followed visually without significant laboratory equipment. Diagnostic sensing activity was retained in the immobilised fusion proteins, even over two months at 20-22 ⁰C in a dried state. A comparable limit of detection was achieved with immobilised reagents as with the soluble form. Taken together with the reduced downstream processing attained by a one-step purification and immobilisation approach, this supports the use of particle-bound reagents in the development of point-of-care tests. To make use of the particle-bound reagents, a novel falling particle biosensor design was explored in this thesis, where the sedimentation of the silica particles was used to drive mixing in an otherwise stationary fluid compartment. The performance of this design was compared against two other formats commonly employed with bio-functionalised particles - (A) a simple suspension in a microcentrifuge tube and (B) a packed bed in a microfluidic channel. The falling particle device outperformed both formats. Overall, this work has demonstrated that the integrated functionality of the fusion proteins could facilitate a production pathway from raw material to end diagnostic, highlighting the use of silica as a protein carrier and presenting a novel biosensor format for utilizing particle-bound enzymes.

P450 BM3 is a natural fusion protein containing a heme catalytic domain fused to a CPR-like domain that binds two flavin cofactors (FAD and FMN). This allows this enzyme to have one of the highest catalytic rates of any P450. Such characteristics make P450 BM3 an attractive enzyme for mutagenesis to produce non-natural products. Unfortunately, no full-length structure revealing the natural domain interactions has been elucidated using X-ray crystallography or other means. HDX-MS is a relatively new technique to explore the surface of a protein and any solvent accessible channels, such as active sites and allosteric binding sites. HDX-MS and other MS techniques were utilized to probe the surface of WT P450 BM3 to gain new insights into the structure of the enzyme including potential dimeric interfaces. The observation of many shielded areas has given insights into conformational changes during ligand binding and on the dimeric interface of the protein, aiding in piecing together the full-length structure of this enzyme. WT BM3 binds specifically to fatty acids and typically produces hydroxylated products. Previously, two mutations were discovered that greatly affect the substrate and product profile of the double mutant (DM, A82F/F87V) enzyme. The substrate/ligand-binding profile of the DM enzyme was investigated using an FDA-approved library. From the library, 59% of the drugs caused significant shifts in the UV-Vis spectrum. These compounds have a variety of targets, masses and structures. Around 80 compounds were chosen due to their structural and binding properties for further analysis, including binding affinity determination, EPR, HPLC, LC-MS, LC-MS/MS, NMR and X-ray crystallography. From these analyzes, the binding of 18 inhibitors was identified, including a novel binding mode involving a substituted pyrimidine ring. Many substrates were also identified, which were found to be metabolized to known human metabolites during turnover.

Cytochrome P450 monooxygenases (P450s) are a widespread class of enzymes involved in biosynthetic pathways and drug metabolism. They can catalyse a diverse range of challenging reactions, including C-H activations. Since these reactions often occur in a regio- and stereoselective fashion, harnessing the oxidative power of these enzymes for biocatalysis has attracted pharmaceutical and fine chemical industries. Despite their immense potential, the implementation of these enzymes as industrial biocatalysts is limited by some issues such as low expression levels, stability and substrate scope. Catalytically self-sufficient P450s, in which the heme and reductase domains are fused in a single polypeptide, are of particular interest for the development of chemical routes to high value-added compounds, as the need for identification and expression of separate redox partners is negated. In an effort to enrich the number of P450s with valuable biocatalytic properties and encourage their utilization, this thesis describes the discovery, characterization, engineering and application of self-sufficient P450s. Enzymes from thermophilic organisms were selected as starting point of our investigation, as thermostable proteins often possess increased stability under process conditions and mutational robustness. This initial search led to a panel of new and diverse class VII P450s displaying not only higher expression levels and thermal stability compared to their mesophilic counterpart, but also a remarkable substrate promiscuity. From this expanded panel of characterized enzymes, the first crystal structure of a class VII P450 was also determined, providing a valuable framework for future protein engineering. Concurrently, P450-enabled biocatalytic cascades were also developed. Complications associated with the application of crude enzyme preparations and biocatalysts compatibility were addressed to demonstrate the applicability of these multi-enzyme systems on a preparative scale.

Synthetic biology is the engineering of life to imbue non-natural functionality. As such, synthetic biology has considerable commercial potential, where synthetic metabolic pathways are utilised to convert low value substrates into high value products. High temperature biocatalysis offers several system-level benefits to synthetic biology, including increased dilution of substrate, increased reaction rates and decreased contamination risk. However, the current gamut of tools available for the engineering of thermostable proteins are either expensive, unreliable, or poorly understood, meaning their adoption into synthetic biology workflows is treacherous. This thesis focuses on the development of an accessible tool for the engineering of protein thermostability, based on the evolutionary biology tool ancestral sequence reconstruction (ASR). ASR allows researchers to walk back in time along the branches of a phylogeny and predict the most likely representation of a protein family's ancestral state. It also has simple input requirements, and its output proteins are often observed to be thermostable, making ASR tractable to protein engineering. Chapter 2 explores the applicability of multiple ASR methods to the engineering of a carboxylic acid reductase (CAR) biocatalyst. Despite the family emerging only 500 million years ago, ancestors presented considerable improvements in thermostability over their modern counterparts. We proceed to thoroughly characterise the ancestral enzymes for their inclusion into the CAR biocatalytic toolbox. Chapter 3 explores why ASR derived proteins may be thermostable despite a mesophilic history. An in silico toolbox for tracking models of protein stability over simulated evolutionary time at the sequence, protein and population level is built. We provide considerable evidence that the sequence alignments of simulated protein families that evolved at marginal stability are saturated with stabilising residues. ASR therefore derives sequences from a dataset biased toward stabilisation. Importantly, while ASR is accessible, it still requires a steep learning curve based on its requirements of phylogenetic expertise. In chapter 4, we utilise the evolutionary model produced in chapter 3 to develop a highly simplified and accessible ASR protocol. This protocol was then applied to engineer CAR enzymes that displayed dramatic increases in thermostability compared to both modern CARs and the thermostable AncCARs presented in chapter 2.

By applying the principles of Darwinian natural selection in the laboratory, directed evolution has become a powerful practical approach to study enzymes and optimize them to catalyze industrially relevant transformations. In this thesis, I applied this strategy to the engineering of amino acid dehydrogenases for biocatalysis of chiral amines, focusing on two crucial features for successful directed evolution experiments. A first key aspect is the development of technologies allowing the screening of large libraries of enzyme variants to explore sequence space efficiently. Massive scale-down of assay volumes by compartmentalization of library members in water-in-oil emulsions has recently led to the development of ultrahigh-throughput screening platforms that allow sorting of more than 106 variants per hour. So far, these microfluidic droplet sorters have relied exclusively on fluorescent readouts. To further extend the range of applications toward enzymes for which no fluorescent assays are available, I successfully developed a sorting module based on absorbance detection. Using this new module, microdroplets could be sorted based on an absorbance readout at rates of up to 1 million droplets per hour. To demonstrate the utility of this module for protein engineering, three rounds of directed evolution were performed to improve a poorly stable NAD+ dependent phenylalanine dehydrogenase (PheDH) toward its native substrate. Five hits showed increased activity (improved up to 10-fold in lysate; kcat increased >3.5-fold), soluble protein expression levels (>2.5-fold) and thermostability (Tm, 8 °C higher). To increase the sensitivity of the device (3–4 orders of magnitude lower than fluorescence assays) for detection of enzymes with limited stability and low turnovers, an extra step of growth in droplets from single cell encapsulation, followed by piconinection of substrates and lysis agents was implemented. As a result, a fivefold signal enhancement over background was achieved, for an amine dehydrogenase (AmDH) reaction shown to be undetectable in a droplet single cell assay. Second, I investigated how mutational robustness may correlate with protein stability and lead to successful hits after mutagenesis and screening. To examine this issue, I initially investigated various approaches (including ancestral resurrection and computational design) to identify stabilized PheDH variants. One such variant (dubbed Pross 4) showed increased expression levels (>3.3-fold) and thermostability (Tm, 13 °C higher) compared to the wild-type PheDH. I further compared the mutational tolerance and the hit rate between PheDH and Pross 4 by generating variant libraries focused on key active site residues and screening them for improved AmDH activity. The Pross 4 background generated 6.4 times more active variants than the PheDH background, the best hits displaying increased activity (up to 2.5-fold in lysate; kcat/KM increased up to 8-fold) compared to previously engineered AmDHs with the PheDH scaffold. In conclusion, this work highlights how directed evolution experiments could be designed for increased success rates, by combining reliable high-throughput screens with careful choice of evolutionary robust starting points.

Protein-based materials are an important area of research for various reasons. Natural protein materials such as spider silk have mechanical properties which compare favourably to artificial or inorganic materials, and in addition are biodegradable and can be produced from easily available feedstocks. It is also possible to produce materials that incorporate the functionality of a natural protein, such as ligand-binding or catalysis of reactions, thus allowing this functionality to be used in the solid rather than solution phase. Two particularly interesting components for protein-based materials are amyloid fibrils and tandem repeat proteins. Amyloid fibrils are exceptionally strong, tough, highly-ordered structures that self-assemble from a wide range of simple building blocks. Meanwhile, tandem repeat proteins are a class of proteins that act as scaffolds to mediate protein-protein interactions and are known to act as elastic springs. Unlike globular proteins, tandem repeat proteins can be designed to bind specific ligands, and their ligand-binding properties and stability can be tuned separately. This work details the synthesis and characterisation of repeat protein and amyloid fibril components for a “smart” hydrogel, the production of these gels, and their characterisation using a microfluidic method that I developed. Although amyloid fibrils have previously been decorated with functional proteins, hitherto, this has usually been done by assembling the fibrils from already-functionalised components. This approach limits the functionality to species that can survive the harsh conditions of amyloid aggregation and do not disturb fibril assembly. Therefore, a method was developed to produce amyloid fibrils that displayed an alkyne functionality on their surface to allow functional proteins or other species to be attached after assembly. This involved the design and synthesis (using solid-phase peptide chemistry) of a peptide based on the previously known TTR105-115 peptide (derived from the amyloidogenic Transthyretin protein). These fibrils were characterised by AFM and TEM and it was then shown that the assembled fibrils could be functionalised using an azide-alkyne “click” reaction. The reaction was shown to work with a variety of ligands including proteins, which were found to retain their structure and function after crosslinking to the fibril. The fibrils with ligands attached were characterised by a variety of methods including LCMS (liquid chromatography-mass spectrometry) and super-resolution optical microscopy. Next, repeat proteins were produced recombinantly containing non-natural azido amino acids at their termini. Incorporation of non-natural amino acids was carried out using a number of different methods including amber codon suppression and methionine replacement. Micron-sized hydrogels were then formed from microfluidic-generated droplets by covalently crosslinking the alkyne-functionalised fibrils with the azide-functionalised repeat proteins. The initial experiments to show proof of principle were carried out with consensus-designed repeat proteins, but repeat proteins based on natural sequences were also used to make hydrogels that could later be tested for potential uptake of peptides known to bind these proteins. These hydrogels could potentially be used for drug delivery or other applications in which a chemical response to a mechanical stimulus is desired. The mechanical properties of the hydrogels were measured using novel microfluidic devices, which were designed and fabricated using standard PDMS-based soft lithography.

Biocatalysis represents an attractive route to the production of various compounds which are difficult or impossible to synthesise and isolate using traditional chemical synthesis. In particular, the production of chiral molecules is a function ideally suited to biocatalysis, due to the natural stereospecificity of enzymes. The synthesis of such chiral molecules is essential in the production of pharmaceuticals, additives for the food and drinks industry and the creation of specialist polymers. CYP105AS1, isolated from Amycolatopsis orientalis, is a cytochrome P450 enzyme which produces the inactive 6-epi-pravastatin of the blockbuster anti-cholesterol drug pravastatin. Previous directed evolution efforts have engineered this enzyme to produce a five-point mutant, known as P450prava, which partially reversed the stereospecificity of the enzyme to produce a majority pravastatin product mixture. This thesis details work to use structure-led engineering approaches to redesign the active site of P450prava to introduce stringent stereospecificity. A combinatorial approach of manual and computational rational design was pursued, leading to the creation of a novel T95F/V180M double mutant of P450prava. This double mutant was found to have successfully eliminated the unwanted 6-epi-pravastatin enantiomer from the product mix, leaving a pure pravastatin product. P450prava was also shown to bind and hydroxylate other statin substrate molecules, demonstrating its versatility in the production of drug metabolites and other high-value oxyfunctionalised molecules. This property, along with its proven tolerance of significant active site engineering efforts, demonstrates the viability of the P450prava as a platform for the creation of novel biocatalysts for the production of various hydroxylated products from diverse substrate molecules.

There is a growing demand for rapid, large-scale production of recombinant proteins, either as biotherapeutics or as research tools. The use of mammalian expression systems for the production of recombinant targets has significantly increased, due to their ability to perform complex post-translational modifications. However, certain recombinant proteins are deemed 'difficult to express' in mammalian expression systems and the unpredictability of production has significant consequences on drug development processes. To this effect, published examples of 'difficult to express' proteins have reported 'bottlenecks' at different stages along the protein expression pathway. As a result, separate protein-specific cell engineering strategies have been employed to overcome such limitations. No universal approach has been defined to overcome poor recombinant protein production. This Thesis has investigated the molecular mechanisms that restrict expression of recombinant proteins of extensive sequence similarity. The limiting step(s) that prevent successful production of a panel of recombinant targets have been defined by systematic screening of cellular processes along the protein expression pathway. It was found that the processes that restrict efficient protein production were of a protein-specific nature. Further, a protein engineering strategy was developed that allowed successful secretion of 'difficult to express' targets limited in their post-translational processing after initial processing in the endoplasmic reticulum. Screening of amino acid sequences using a computational approach reported on the predictability of protein secretion and identified problematic sequence features that may limit secretion. Computational analyses revealed that increased abundance of positively-charged or hydrophobic surface regions was associated with poor or no protein secretion. Engineering of these problematic sequence features in a model 'difficult' recombinant target resulted in successful secretion. Therefore, application of computational screening and re-engineering of unfavourable sequence features prior to expression in mammalian cells offers the potential to overcome challenges in the production of 'difficult to express' recombinant targets. A further theme focused on identification of cell engineering for creation of robust host cell lines for the production of recombinant proteins. Transcriptomic profiling revealed gene expression response distinctions between Chinese Hamster Ovary cells that expressed retained and secreted proteins. That study also highlighted significant gene expression changes between cells in early and later cell growth phase. Transcripts that were significantly differentially expressed were linked to endoplasmic reticulum stress and/or unfolded protein response signalling and cell cycle canonical pathways. Targeting of upstream regulators within these pathways acts as a potential cell engineering approach to enhance recombinant protein production.

Novel biocatalysts for the enantioselective reduction of imines and reductive amination of a broad range of carbonyl compounds have been developed. Unlike other imine reductases (IREDs), the IRED from Amycolaptosis orientalis (AoIRED) features an aprotic "catalytic" residue Asn171 and as such became an interesting candidate for detailed mechanistic, specificity and stereoselectivity studies. AoIRED has been shown to be an efficient catalyst for the enantioselective reduction of imines and iminium ions to yield the corresponding chiral amines in high conversions and good to excellent enantioselectivity. The enzyme exhibits unusual stereoselective properties, displaying a selectivity switch for structurally similar substrates and in certain cases for the same substrate depending on the age of the enzyme. Mutagenesis studies have highlighted important residues that may play key roles in the substrate specificity and stereoselectivity of the enzyme. The reductive aminase from Aspergillus oryzae (AspRedAm) is a multifunctional catalyst that efficiently catalyses i) the reductive coupling of carbonyl compounds and amine nucleophiles, ii) the enantioselective reduction of prochiral cyclic and preformed imines or iii) the oxidative deamination of amines towards kinetic resolution of racemic amines. Detailed kinetic studies have led to the construction of a kinetic model/mechanism and based on structure guided investigation of conserved active site residues, a putative catalytic mechanism has been proposed. It has also been possible to engineer wild-type AspRedAm for improved stereoselectivity as well as to invert the enzyme's enantioselectivity towards a range of substrates. Using AspRedAm as a catalyst, efficient systems have been developed that allow the kinetic resolution of several racemic amines. This thesis has been organised into separate chapters each addressing a specific theme. Chapter 1 gives an overview of recent advances in the field of amine biocatalysis with emphasis on biocatalytic imine reduction and reductive amination; it also outlines the objectives of this project. Chapter 2 describes methods and materials used in these studies while Chapters 3-7 present and discuss results from different projects that constitute the work in this thesis. Initial discovery and characterisation studies of IREDs are described in Chapter 3. Chapter 4 describes detailed characterisation of AoIRED with particular emphasis on stereoselectivity and synthetic applicability while Chapter 5 presents and discusses results from the study of the reductive aminase (AspRedAm) from Aspergillus oryzae. Chapters 6 and 7 respectively describe the engineering of AoIRED and AspRedAm, and the application of AspRedAm in kinetic resolution of racemic amines. The results from these chapters have been summarised and discussed in Chapter 8 and recommendations for future directions in this field have been offered.

Human drug metabolites are frequently biologically active, with many implications for human health. Pharmaceutical companies have become increasingly aware of the need to identify and test these metabolites. The P450 BM3 enzyme from Bacillus megaterium offers substantial advantages to the current methods of metabolite synthesis, as its soluble, catalytically self-sufficient nature, coupled with its high catalytic activity, make P450 BM3 ideal for engineering towards specificity for human drugs. The highly-active I401P BM3 mutant was characterized for its reactivity towards human drugs and for the development of a human P450-like metabolite profile. The I401P mutant exhibits binding to molecules including alkaloids, steroids, and azole drugs, along with many other compounds. I401P binds/oxidizes human CYP substrates, including alosetron, phenacetin, caffeine, nicotine and diclofenac. LC-MS product identification shows that I401P BM3 forms 4OH-diclofenac, the major human metabolite for diclofenac. I401P BM3 also produces nornicotine, the second major human metabolite of nicotine. I401P BM3 also forms theophylline, theobromine and paraxanthine, the three major human metabolites of caffeine. Thermostability (DSC) data show that the I401P mutation destabilizes the BM3 heme domain in both its substrate-free and substrate-bound forms. The I401P heme domain X-ray crystal structure reinforces previous structural observations that the Pro401 mutation causes the BM3 protein to adopt a high-spin, "substrate-bound" state, with a displaced heme iron axial water, producing a "catalytically primed" mutant with greater diversity in substrate selectivity. The destabilisation of the BM3 heme domain structure due to the Pro401 mutation increases conformational plasticity in this mutant, allowing it to function as a platform for future mutagenesis aimed at improved binding and metabolite yield from specific drug substrates. Further proline mutations (A330P, A330P/I401P and A82F/F87V/I401P) were examined for increased affinity for drug substrates. The A330P mutant shows no novel drug substrate specificity, despite its reported affinity for small molecules. The A330P/I401P double mutant demonstrates weak binding to WT BM3 and I401P substrates, but no synergistic effects were obtained by combining the two mutations. The double mutant exhibits very low solvent tolerance and significant structural destabilisation. DSC data confirms this, with the double mutant destabilising the BM3 heme domain by up to 20 °C. Initial work with the A82F/F87V/I401P mutant showed increased affinity for A82F/F87V- and I401P-type substrates, including diclofenac. LC-MS product analysis confirms that the A82F/F87V/I401P mutant oxidises diclofenac into its major human metabolite 4OH-diclofenac. These data indicate that human-like oxidation reactions are feasible with BM3 mutants. In this work, proline insertion mutants were generated that introduced novel affinity for biotechnologically relevant substrates. In particular the I401P mutant offers an excellent platform for future biotechnological engineering.

Vaccines have saved more lives than any other medical intervention. Recombinant vaccines provide unmatched safety profiles, but at the expense of reduced immunogenicity. Virus-like particles (VLPs) resemble viruses in size, shape and repetitive arrangement but are devoid of pathogenic genetic material and therefore safe. Poor immunogens can be rendered immunogenic by display on VLPs. Successfully decorating VLPs is still a major challenge. Genetic fusion or chemical modification is often time-consuming and can lead to misassembly or misfolding, which obstructs generation of the desired immune response. SpyCatcher is a genetically encodable protein, previously engineered to form a covalent isopeptide bond to its peptide-partner SpyTag. Presented in this thesis are SpyCatcher-VLPs, based on the fusion of SpyCatcher to the bacteriophage VLP AP205. SpyCatcher- VLPs can be conveniently conjugated with SpyTag fused antigens, simply by mixing. I demonstrate the modularity of this approach by covalently linking several complex, cysteine-rich malarial antigens to SpyCatcher-VLPs, such as the transmission-blocking antigen Pfs25 and the blood-stage antigen CIDR. A single administration of Pfs25-SpyTag conjugated to SpyCatcher-VLPs induced potent antibody generation against Pfs25, even in the absence of adjuvant. Anti-Pfs25 antibodies induced by this platform conveyed potent transmission-blocking activity in the mosquito vector. The thesis further demonstrates the feasibility of more complex Catcher-nanoparticle architectures. The previously engineered SnoopCatcher covalently reacts with SnoopTag peptide and is orthogonal to the SpyCatcher / SpyTag pair. IMX313 is an engineered chimera of the multimerization domain of chicken complement inhibitor C4-binding protein. This work describes fusion of SnoopCatcher and SpyCatcher to IMX313, which yields independently addressable Catcher-moieties on a single IMX313 nanoparticle. Display of two antigens on one particle may enable single-particle, multi-disease vaccines as well as multi-stage vaccines to tackle immune evasion of parasites. The platforms presented should accelerate and enhance vaccine development and may create opportunities for imaging and metabolic engineering.

The increased demand for an alternative form of fuel has raised a great interest towards exploring various metabolic pathways and enzymes in several microbial species for hydrocarbon production. In recent years, cyanobacteria have emerged as an attractive microbial host and cyanobacterial metabolic pathways were targeted for engineering to produce "drop in" fuels such as propane and butane. Whilst appealing, practicalities for producing biofuels in cyanobacteria remain challenging, requiring the identification and engineering of natural biocatalysts and their integration into metabolic processes. Cyanobacterial hydrocarbon biosynthesis arises from fatty acid metabolism involving a potential enzyme, aldehyde deformylating oxygenase (ADO), which catalyses the decarbonylation of long-chain fatty aldehydes to alkanes, mainly in the conversion of octadecanal (C17H35CHO) to heptadecane (C17H36) and formate. The substrate specificity and preferences for long-chain aldehyde by ADO necessitates a detailed kinetic and structural characterisation in order to optimise/engineer this enzyme for future biotechnological applications. Thus, the main objective was to identify a potential ADO enzyme that can be optimised for shorter chain alkane production. By studying the substrate specificity and reaction kinetics of different ADO enzymes, it was found that ADO from Prochlorococcus marinus MIT 9313 (PmADO) is a potential target for short chain alkane production. The crystal structural of PmADO was solved and further GC-MS analysis was carried out to identify the chemical origin of a mixture of long-chain fatty acid in the active site, originated from E. coli cells during recombinant over-expression and purification. It was suggested that the structure-guided protein engineering for short-chain alkane production should be carried out along with the removal of this adventitious ligand from the active site in order to increase the alkane production. Four important residues present at the entrance of the ligand-binding cavity were targeted and saturated mutagenesis was performed on PmADO to identify variants that excluded the long fatty acid ligands from the active site but have specificity and higher conversion rates for shorter chain aldehydes. This identified two variants, V41Y and A134F, with the A134F variant that not only exhibiting an improved activity and turnover value of PmADO by four-fold but also improved binding affinity for butyraldehyde by 2 times. Finally the improved variants were incorporated in a host organism (E. coli) and the possibilities for the development of a microbial platform for renewable propane synthesis based on a fermentative clostridial butanol pathway were explored. Four pathways were designed namely atoB-adhE2, atoB-TPC7, nphT7-adhE2 and nphT7-TPC7 routes, which utilise CoA intermediates selected to incorporate ADO as the terminal enzyme. When PmADO was co-expressed with these pathways, the engineered E. coli host produced propane. The atoB-TPC7-ADO pathway was the most effective in producing propane (220 ± 3 μg/L). By (i) deleting competing pathways, (ii) including a previously designed A134F variant ofPmADO with an enhanced specificity towards short-chain substrates, and (iii) including a ferredoxin-based electron supply system, the propane titre was increased up to 3.40 ± 0.19 mg/L. It was also shown that the best propane producing pathways are scalable in a 250 mL flask and in a large-scale (up to 30 L) fermentor setup. This thesis focuses on the detailed kinetics and structure-guided characterisation and engineering studies on the ADO enzyme for the development of a renewable microbial biofuel platform.

The research carried out within this thesis was aimed at the development and implementation of a biocatalytic route towards Ticagrelor, a platelet-aggregation inhibitor. A bio-retrosynthetic consideration of the target compound yielded different possible strategies, which were analysed in terms of enantioselectivity and efficiency. The ultimate goal was to generate a biocatalyst specifically tailored to the starting material to yield the target compound in high optical purity and conversion. Different approaches to the chemoenzymatic generation of the cyclopropyl subunit (cf figure) in enantiomerically pure form were proposed and tested. The lipase from Thermomyces lanuginosus proved to be the most selective and active enzyme tested and was used as a model enzyme, initially yielding an E of 76 at a conversion of 50% after 48h. Through both reaction engineering and rational protein design approaches the time to attain 50% conversion could be reduced to 24 h while the enantioselectivity of the process increased to 100. Moreover, in a rational protein design approach different residues in the lid of the lipase were identified through analysis of the resolved crystal structures and subsequently mutated in order to investigate the influence of these residues on the overall performance of the lipase towards the target biotransformation. Mutations on Asn88 resulted in the inactivation of the enzyme while an Asp57Asn mutation resulted in a more active enzyme. Ultimately, this research has contributed to making the synthetic route towards Ticagrelor more environmentally sustainable, diminishing the need for the use of toxic, unsustainable and sterically demanding auxiliaries, as well as the amount of waste produced. The principles of green chemistry have been applied to the case studied. The synthetic route towards a key fragment of Ticagrelor has been significantly shortened using a biotransformation with an enzyme that can be recycled and employed in catalytic quantities.

Specific recognition of pathogenic molecules by the immune system is mediated by proteins known as antigen receptors. One such component is the antibody. Binding properties of natural and engineered antibodies can be understood by studying the structure of their variable domains, VH and VL. In this thesis we investigate how the two variable domains orientate with respect to one another and therefore influence the geometry of the antigen binding site which is formed between them. We describe a method which fully characterises the VH-VL orientation in a consistent and absolute sense using five angles and a distance. The ABangle method is used to investigate variable domain orientation in structures collected by our database SAbDab. Using the ABangle method we compare VH-VL orientation to the corresponding property in a different component of the immune system, the T-cell receptor (TCR). Despite having similar individual domain structures the variable domain orientations of antibodies and TCRs are found to be distinct. This is found to affect an antibody’s ability to mimic TCR specificity. ABangle's characterisation is used to find determinants of the VH-VL orientation. We identify sequence and structural properties that influence the variable domain pose. A feature based method for predicting VH-VL orientation is presented and assessed. Future directions of this research and its application to the development of antibody therapeutics are described.

Haptens are small to be recognised by the immune system unless conjugated to larger carrier molecules. Characterisation of the binding sensitivity and specificity of anti-hapten antibodies, in relation to their targets, were the main focus of various previous studies. However, an understanding of the ability of antibodies to recognise haptens, at the molecular level, is still unclear. This is even factual for sheep antibodies which have generated the best anti-hapten binders. Haptens were conjugated to large carrier proteins, and were used in sheep immunisation. An antibody library was constructed, after successive rounds of antigen boosts, as representative of the extracted antibodies' genes taken from isolated sheep lymphocytes. Antibodies were selected against the hapten antigens (SQA, POR, 5BS, and HSL) using a bio-panning technique, and expressed as soluble scAb proteins. All the unique clones, that demonstrated promising recognition of the free hapten antigens, were extensively characterised. The isolated anti-SQA antibodies showed the structural naivety of SQA difficult to be recognised with high specificity and sensitivity. Although both POR and COP possess relative structural rigidity, but the polar groups of POR were necessary to generate more sensitive and specific antibodies when compared to the hydrophobic COP. Whereas the "optimum hapten structural model" was demonstrated by HSL, which has combined relative backbone firmness, with the presence of polar groups. In order to specifically understand the antibody-antigen interaction process, it was essential to investigate the effect of these haptens on the amino acid distribution, complementarity determining regions (CDRs) length, and the canonical classification of these antibodies. The canonical classification was affected by the CDR length and specific amino acids at exact positions. Sheep genetic diversity and age have influenced both the CDR length and canonical classification of the examined CDRs. In addition, the highly stringent panning protocols have necessitated the antibodies to conserve specific positions to enable the development of the required binding pockets. Anti-hapten antibodies recognition of haptens was by establishing complementary binding pockets to which the haptens can conveniently fit into and bind. These complementarities were examined in term of pockets' size, shape, and electrostatics potential. In addition, the computational docking affinity and experimental binding sensitivity can be correlated, when factors that might influence the antibody behaviour are taken into consideration, such as aggregation, adsorption, protein expression, solvent solubility and stability.

Lanthanides are critical elements used in numerous modern technologies, but their mining and purification pose significant environmental challenges. This thesis describes the development of synthetic biology tools to enable more sustainable extraction of lanthanides using engineered microbes. In Chapter 1, I engineer LanTERN, a first-of-its-kind fluorescent biosensor that directly transduces lanthanide binding into a fluorescent signal. LanTERN enables direct fluorescent detection of lanthanide binding, overcoming the limitations of previous sensors. Chapter 2 details early work on generating small, stable lanthanide-binding peptides through de novo design. Finally, Chapter 3 places this work in the broader context of synthetic microbiology for sustainability applications. I present a framework categorizing sustainability efforts based on their level of environmental containment and discuss research priorities for responsibly deploying engineered microbes. Together, the tools and perspectives provided help lay a foundation for engineering microbes to extract, concentrate, and separate lanthanides in a more environmentally friendly manner.

Cell separation based on antibody-targeted magnetic beads has been widely used in a number of applications in immunology, microbiology, oncology and more recently, in the isolation of circulating tumour cells (CTCs) in cancer patients. Although other cell separation techniques such as size based cell filtration and Fluorescence Activated Cell Sorting have also been in popular use, immunomagnetic cell isolation possesses the advantages of high throughput, good specificity and reduced cell stress. However, certain fundamental features of the cell-bead interface are still unknown. In this study, some of the key features of the cell-bead synapse were investigated in an effort to improve the efficiency of immunomagnetic cell isolation and reduce its dependence on high expressing cell surface markers. A clinically relevant antibody fragment (Fab) against tyrosine kinase receptor HER2 was applied to study the immunomagnetic isolation of HER2 expressing cancer cells. First, the minimum number of target proteins required on a cell for it to be isolated was determined. Second, the importance of the primary antibody affinity was investigated, using a series of Fab mutants with known kinetics and it was shown that despite starting with sub-nanomolar affinity, improving Fab affinity increased cell isolation. Third, the influence of the connection between the primary antibody and the bead was studied by comparing Fab bridged to the magnetic bead via a secondary antibody, Protein L or streptavidin; the high affinity biotin-streptavidin linkage increased isolation sensitivity by an order of magnitude. Fourth, the effect of manipulating cytoskeletal polymerization and cell membrane fluidity using small molecules was tested; cholesterol depletion decreased isolation and cholesterol loading increased cell isolation. The insights from these observations were then applied to isolate a panel of cell lines expressing a wide range of surface HER2. While the standard approach isolated less than 10% of low HER2 expressing cancer cells from spiked rabbit and human blood, our enhanced approach with the optimized cholesterol level, antibody affinity and antibody-bead linkage could specifically isolate more than 80% of such cells. The final part of this work focussed on developing an antibody clamp that could physically restrict the antigen within its binding site on the Fab and prevent antigen dissociation, using the HER2-Fab complex and the anti-myc peptide antibody 9E10. Work from this thesis provides useful insights into the molecular and cellular parameters guiding immunomagnetic cell isolation and can be used to extend the range of target receptors and biomarkers for tumour cell isolation and other types of cell separation, thereby enhancing the power and capacity of this approach.

Engineered protein nanopores can be used to investigate a wide range of dynamic processes in real time and at the single-molecule level, for example covalent bond making and breaking or the interaction of ligands with their cognate binding sites. The detection of such processes is accomplished by monitoring the current carried by ions through the pore in an applied potential, which is modulated as molecules of interest interact with engineered binding sites within the pore. In contrast to ensemble measurements, where the behaviour of individual molecules is obscured by averaging, single-channel recordings can identify short-lived intermediates and rare reaction pathways, thereby adding to our understanding of fundamental processes in chemistry and biology. The goal of my thesis work was to engineer alpha-hemolysin (αHL) pores to gain insight into such processes. Chapter 1 provides an overview of common techniques used to study single- molecule processes, in particular single channel recordings. General techniques to engineer ion channels and pores are presented, followed by examples of how the alpha-HL pore has been engineered to monitor dynamic processes at the single- molecule level. Chapter 2 describes how alpha-HL pores can be chemically modifeed with a tridentate "half-chelator" ligand. Single channel recordings show that this modifeed pore can be used to determine rates of chelation and the stability of divalent metal ion complexes. The modifeed pore can also be used as a stochastic sensor for the detection of different divalent metal ions in solution. Chapter 3 investigates the chelate-cooperativity between two half-chelator ligands installed in close proximity in the alpha-HL pore, as they form a full complex with a single Zn2+ ion. The single channel recordings reveal a two step process, in which the Zn2+ ion must fiferst bind to one of the two half-chelators, before the second one completes the complex. The rate constants for all the major steps of the process are determined and the extent of cooperativity between the half-chelators is quantifeed. Chapter 4 demonstrates that genetically encoded subunit dimers of alpha-HL can be used to control the subunit arrangement in the heptameric pore. Although techniques exist to prepare heteroheptameric pores, pores containing more than one type of modifeed subunit are not commonly used because it is impossible to distinguish between the permutations of the pore. By using subunit dimers, heptamers in which two defefined subunits are adjacent to each other can be formed, which increases the range of structures that can be obtained from engineered protein nanopores. Chapter 5 explores the possibility of following the nuclease activity of a metal complex in the alpha-HL pore at the single-molecule level. The Rh(III) complex [Rh(bpy)2phzi]2+ binds strongly to CC mismatches in dsDNA, and on activation with UV light promotes the cleavage of one of the two strands. To follow this reaction by single channel recording, a piece of dsDNA with the bound Rh-complex was immobilised in the HL pore and the single current changes under UV irradiation were monitored. The preliminary data indicate that the rate of the photocleavage reaction can be measured.

Experiments facilitating the engineering of the PETNR active site to accommodate a range of non natural enone substrates with substituents localised on the α and β carbons of the unsaturated bond are described. In order to facilitate the high throughput purification of PETNR libraries poly histidine (PETNRHis) and biotin (PETNRBio) tagged PETNR variants were generated. High throughput protocols were developed for the automated generation, purification and screening of libraries in a 96 well format. Protocols were optimised and trialled using blocks consisting of PETNRHis WT only and characterised in terms of intra block variation. A range of single site saturation mutagenic libraries were generated at positions in the active site consisting of T26, Y68, W102, H181, H184, Y186, Q241 and Y351. Sequencing results indicated randomised libraries with the occasional instance of bias evident. Expression and purification in a 96 well format was monitored by SDS PAGE and protein quantitation. Library activity was quantified and demonstrated to retain varying degrees of activity with the model substrate 2-cyclohexenone. Following this verification of the experimental protocol libraries were screened against a range of substrates analogous to substrates demonstrated to be active with PETNRWT but incorporating substituents at the α and β carbons. 'Hits' generated from these screening reactions were studied further by the determination of the specific activity and quantitation of substrate/product from biotransformation reactions. From these screening experiments totalling 3,600 individual reactions, 35 were identified as potential hits, of these 8 proved to be genuinely improved variants. Substituents at the β carbon were demonstrated to compromise the activity of the WT enzyme most severely. Positions 68, 102, and 351 were demonstrated to play an important role in the accommodation of substituents at the α carbon whilst residues 26 and 351 are important for the β carbon. The best variants demonstrated up to 9 fold improvements in poor substrates which represented rates in excess of those observed for model substrates.

With advances in mass spectrometry, there is increasing demand for more effective and adaptive approaches to systematically extract biological insights from glycomic and lipidomic data. This thesis explores the application of Markov modeling in understanding two key cellular processes, N-glycosylation and lipid biosynthesis.In Chapter 2, we demonstrate that a Markov model of N-glycosylation network successfully captures intricate glycosyltransferase interactions by reproducing a set of glycoprofiles from glycoengineered CHO cells producing erythropoietin (EPO). We further validate the model parsimony and accuracy by predicting the glycoprofiles of other glycoengineered drugs from the trained models and their wildtype glycoprofiles. The results attest to the model’s ability to learn glycosyltranferase activities and substrate specificities. To increase the impact of this approach, we also develop GlycoMME to allow broader access to this modeling pipeline, presenting a promising direction for rational glycoengineering. Chapter 3 extends the modeling approach to lipid biosynthesis, introducing the Lipid Synthesis Investigative Markov Model (LipidSIM). As a low-parameter, biologically interpretable framework, LipidSIM proves powerful in leveraging the interdependency in lipidomic data and extracts and quantifies perturbations to lipid biosynthesis reactions, generating hypotheses directly testable by transcriptomic data. This method is showcased in 5 different scenarios with 3 different lipidomic datasets, and the results substantiate LipidSIM as a valuable tool for extracting insights from high-dimensional lipidomic data of different types. In Chapter 4, a Taguchi design is used to systematically characterize the impact of 15 media supplements on potential cellular phenotypes for CHO cells, especially N-glycosylation. The analytical pipeline allows disentangling the impact of individual supplements at different concentration levels with minimal numbers of experimental configurations. This approach answers the demand to find more flexible strategies for controlling N-glycosylation beyond genetic engineering. When applied in conjunction with the Taguchi design, our modeling framework has the potential to facilitate rapid customization of media for the growing market of glycoprotein drugs. This thesis encapsulates the innovative applications of probabilistic modeling in accounting for the biological dependency underlying omics data, offering insights into intricate cellular processes and motivating further exploration in actionable modeling of biological systems.

Genetically encoded DNA recorders noninvasively convert transient biological events into durable mutations in a cell’s genome, thereby allowing the hidden forces that guide tissue development to be chronicled. After the developmental process of interest is complete, the cells’ experiences can be read out via high-throughput DNA sequencing. Two types of transient events are captured with DNA recorders—the existence of progenitor cells that give rise to daughter cells (which is recorded as lineage relationships among cells by generating lineage-specific DNA barcodes), and stimuli that trigger a molecular response in the cells (which is recorded in an analog fashion by linking the accumulation of mutations to the presence of inducers of interest). This dissertation focuses on the development of two distinct, but complementary DNA recorders. First, Cell History Recording by Ordered iNsertion (CHYRON), which uses a homing guide RNA (hgRNA), CRISPR-Cas9, and a template-independent polymerase (TdT) to generate ordered insertion mutations, was developed. A large portion of this technology was already established prior to my joining the project, so this dissertation focuses only on the developments that overlapped with my efforts. Specifically, alternative nuclease components (instead of Cas9) were explored for CHYRON recording, and dose- and duration-dependent hypoxia recording was accomplished. Also, the nucleotide insertion preferences of TdT were engineered in a novel, massively parallel screen, to achieve greater lineage barcode diversity and enable recording of two different stimuli with differently-biased TdTs. The other DNA recorder covered in this work is prime editing CHYRON (peCHYRON), which uses prime editor to generate iterative insertion mutations. Prime editor requires its edits to be precisely pre-defined in prime editing guide RNAs (pegRNAs), so pegRNA sequences that accomplish continuous insertion accumulation were carefully engineered via a combination of rational design and high-throughput screening. Once continuous editing was embodied by peCHYRON, we used the system for in vitro cell lineage tracing and molecular stimulus recording. Efforts to expand peCHYRON’s ability to record a massive collection of stimuli (i.e., any stimulus that can induce transcription with RNAP II) are discussed.

Cytokine signaling a core mechanism by which immune activity is regulated in both health anddisease. Cytokine-mediated signaling regulates the proliferation, differentiation, and activity of cells in both the innate and adaptive immune systems. Due to their powerful regulatory capacity, cytokines have been leveraged as immunotherapies in a wide range of disease indications; for example, interleukin-2 (IL-2) has been explored as a potential immunostimulant for the treatment for cancer, as well as an immunosuppressant for the treatment of autoimmune diseases. However, in many such cases, the pleiotropic nature of cytokine signaling has stymied the development of efficacious and safe therapies due to the induction of signaling in off-target populations. To overcome this limitation and bias cytokines towards signaling in target populations, engineered cytokines with a variety of alterations, such as mutations affecting their binding interactions with their cognate receptors, fusion to antibody fragments, or co-formulation with antibodies to that cytokine have been developed. However, without a quantitative model of signaling the effects of such mutations and alterations are often difficult to anticipate, leading to inefficient cytokine engineering efforts. To address this lack of quantitative understanding, we conducted a battery of computational studies. First, using a mechanistic binding model, we developed a general, quantitative understanding of the landscape of cell-selective cytokine signaling, and found that affinity, valency, and multi-specificity must be simultaneously optimized to engineer optimally selective cytokines. We then specifically studied the IL-2 signaling pathway, and used both ordinary differential equation models and our mechanistic binding model to study the signaling characteristics of wild-type and engineered IL-2 mutants. Leveraging our newfound quantitative of how affinity and valency interact to determine a cytokine’s selectivity profile both generally and in the specific context of IL-2, we developed affinity-optimized tetravalent IL-2 mutants with superior regulatory cell selectivity. Using these models of IL-2 signaling, we also elucidated the mechanism by which engineered antibody-IL-2 fusions induced regulatory cell-selective signaling and conferred protection against autoimmunity. In total, this body of work demonstrates the critical role that computational modeling plays in potentiating the engineering of superior cytokine-based immunotherapies.

Cell surface proteolysis plays a crucial role in regulating various biological and cellular processes, including but not limited to cell signaling, protein maturation, and extracellular matrix remodeling. Dysregulation of these proteolytic events is often a hallmark of numerous diseases. Understanding these changes in proteolysis may present opportunities for therapeutic intervention as neo-epitopes are created that may be preferentially displayed in diseased states. Under hypoxic conditions, Pancreatic Dunctal Adenocarcimoma (PDAC) experiences global changes to its cell surface proteolytic state. Additionally, it is important to understand the mechanism of action for protein-protein interactions in disease, particularly in the case of viral infection, as these interactions may be therapeutic targets for disruption. The following work describes the development of novel protein tools to understand how cell surface proteolysis changes when various oncogenes are overexpressed in breast cancer and how SARS-CoV-2 Spike protein interacts with the human ACE2 receptor. In Chapter 1, I describe the work done to develop a method utilizing subtiligase to study changes to the proteolytic landscape of MCF10A cells under the influence of KRAS G12V and Her2 overexpression. Chapter 2 describes the work done in studying the effect of hypoxia on cell surface proteolysis in Pancreatic Ductal Adenocarcinoma (PDAC). In Chapter 3, I describe the develop of various protein-Fc fusions to study the protein-protein interactions of Spike protein and ACE2.

Deep learning has become widely used in biological sciences. More specifically, the development of protein deep learning models has leveraged the evergrowing collection of biological data to learn the patterns that govern protein biochemistry. Here, we focus on the assessment of different protein deep learning models to better understand each of their capabilities, benefits and drawbacks. Our work aims to provide insights for future protein engineering efforts and for the discovery of protein homologs. In Chapter 2 we assessed a structure-based protein ML model in its ability to make biochemically meaningful predictions and tested weather or not the model can predict specific allowed amino acids in a protein. We compared the performance of models trained on different input sizes and correlated model predictions with natural variation in order to better understand how these models learn protein structure and biochemistry. In Chapter 3, we compared the predictions of two structure models and two language models to determine if different protein representations affect what information each model type learns and their performance. Finally, in Chapter 4, we apply a sequence-based protein model to searching for antibacterial microcin peptides in bacterial genomes.

SCR1-3, SCR3 and SCR3₂ are soluble recombinant proteins, consisting of the first three, third and two consecutive third N-terminal domains of CR1 respectively and provide the basis for the work described in this thesis. SCR1-3 has four tryptophan residues, one conserved in each domain at positions 51, 113 and 184, and an additional non conserved residue located at position 7 in SCR1. Individual site-directed replacement of these residues by phenylalanine residues determined that the fluorescence properties of SCR1-3, observed at 338nm, are due to the contribution of the non-conserved tryptophan located in SCR1, whilst the fluorescence of the remaining three conserved tryptophan residues is quenched by the N-terminal disulphide bond of each SCR unit. Fluorescence and circular dichroism techniques were utilised to determine the conformational stabilities and folding properties of SCR1-3, SCR3 and SCR3₂. SCR1-3 is significantly more stable than SCR3 and SCR3₂. Kinetic analysis of the folding and unfolding processes of the triple domain SCR1-3 indicates the presence of folding intermediates, not detectable at equilibrium, which are possibly attributable to the formation of interdomain contacts. The single domain, SCR3, demonstrates a simple two state transition between the folded and unfolded forms, with no detectable folding intermediates. Kinetic analysis of the refolding process of the W7F mutant protein confirmed that the rate limiting intermediate step was not due to spectral perturbations caused by the refolding of the non-conserved tryptophan and was therefore most probably due to interdomain interactions. Chemical modification and site directed mutagenesis studies have identified groups of residues and regions within SCR1-3 involved in the interaction with complement proteins.