Search

In the post genomic and high throughput era we have a wealth of sequence and expression data, yet we are still learning to understand the punctuation and syntax of the genome. This thesis presents case studies in gene regulation for a range of organisms, addressing whole transcriptome pattern changes in Plasmodium falciparum and Bartonella quintana, and then more focused descriptions of the binding sites of transcription factors from Mycobacterium tuberculosis, yeast, chimps, and humans. Early on in my thesis, I studied the immunology of Plasmodium infections. Later I profiled the transcriptomes of Plasmodium parasites after exposure to artemisinin, currently the front-line standard of care, and uncovered evidence for a developmental stall. The parasite was not previously thought to undergo any cell cycle arrest, and this may explain observed clinical recrudescence of infection after artemisinin monotherapy treatment and could represent a means for the parasite to adapt to drug pressure and select for resistance. After these studies I turned to a technique, called MITOMI2.0 (for mechanical trapping of molecular interactions), a means of measuring the specificity and energetic affinities of DNA binding proteins. The primary data produced with a lab-on-a-chip is the amount of DNA binding to a given protein, over a library of DNA sequences. We improved made the technique more robust and were able to apply it to several systems. We found that the yeast unfolded protein response factor Hac1 bound two DNA sites with distinct sequences. For another yeast stress regulator, Msn2, we measured its absolute affinity for regulated sites in target promoters and confirmed that it was a low affinity DNA binder, capable of a linear induction of its targets, as opposed to a frequently observed more binary induction response. ChIP-seq and MITOMI analysis revealed that FOXP2 the best-studied example of a protein involved in the development of human language, has had a conserved binding site preference, yet the complement of available binding sites has changed in humans. We confirmed the sequence specificity of a Mycobacterium tuberculosis factor that controls virulence in macrophage infection, and studied the way the protein interacts with DNA both as a dimer and a monomer.

The research described within this thesis is primarily motivated by two fields of science with reversed, yet complementary, approaches to addressing the same essential objective: manipulating and understanding the complex program of life. Whereas synthetic biology has to do with the bottom-up construction of living systems from a library of ‘parts’, systems biology takes a top-down, reverse-engineering analysis of the same processes within functioning cells to elucidate the crucial elements and interactions. Both perspectives have led to important contributions in learning how biological systems operate. Discoveries from these studies can have far-reaching implications, notably in medicine, manufacturing, energy, and the environment. One exciting outcome of the research efforts within synthetic biology is the ability to design genetic constructs encoding proteins with novel, optimized functions. Although advances have been made in predicting how a protein’s amino acid sequence relates to its higher order structural form and behavior, experimental methods for protein engineering remain invaluable for evaluating computationally derived designs. Unfortunately, the design-build-test cycle for rational protein development is commonly a time, labor and resource intensive endeavor. This is primarily due to limitations in the ability to generate libraries of gene variants and bottlenecks in cell-based expression and quantitative screening of protein products. This thesis describes the development of a solid-phase gene assembly technique and its integration with microfluidic protein analysis to accelerate the prototyping of novel protein designs. The gene assembly method utilizes commodity-scale, chemically manufactured DNA, and operates without the need for ligase or restriction enzymes. As a bench top process amenable to scale up, the modularity of the assembly process allows the efficient and economical generation of expression-ready gene variants. We directly coupled this gene assembly pipeline to a microfluidic device to enable high-throughput, on-chip, cell-free protein expression, purification, and characterization. By circumventing molecular cloning and cell-based steps, the lag time between protein design and quantitative analysis was dramatically reduced. As a proof of concept, this protein engineering platform was applied towards the construction of over 400 artificial, engineered variants of C2H2 zinc finger (ZF) proteins. ZF protein domains are the most prevalent form of transcription factor (TF) in humans, and are found throughout the tree of life. They provide a convenient structure for refactoring due to their relatively small size and composability, and are an ideal model for exploring the biophysics of TF-DNA specificity. The ability to engineer ZF proteins makes them useful as programmable, DNA-targeting units for applications in biotechnology and synthetic biology. We demonstrate that although ZFPs can be readily engineered to recognize a particular DNA target, engineering the precise binding energy landscape remains a challenge. Additionally, we show that ZF-DNA binding affinity can be tuned independently of sequence specificity. Together, these results demonstrate the versatility of the coupled gene assembly and microfluidic analysis platform as a new tool for rational protein engineering.

The goal of biology is to understand how complex systems such as cells and entire organisms function. Systems Biology attempts to quantitatively characterize all components comprising these systems. A considerable task. Microfluidics provides a powerful tool for undertaking this endeavor. This thesis describes the development of Microfluidic Large Scale Integration (MLSI) using devices fabricated by Multilayer Soft Lithography (MSL). MLSI and fluidic components, such as multiplexers and free-standing membranes, serve as the infrastructure for performing large-scale biophysical measurements of biological systems. Transcription factor binding energy landscapes were determined using MLSI and MITOMI, a novel method for measuring molecular interactions. The biophysical characterization of transcription factors described herein were the first comprehensive measurements of its kind, and answered basic questions regarding how transcription factors recognize DNA. Furthermore, it was possible to predict the in vivo function of transcription factors using only the measured binding topographies and a genome sequence, indicating that biological processes can be predicted with high accuracy. More generally, the methods described in this thesis are generally applicable to understanding the properties of any biological system and should find broad usage in the field of Systems Biology.

Measuring binding affinities and association/dissociation rates of molecular interactions is important for a quantitative understanding of cellular mechanisms. Many low-throughput methods have been developed throughout the years to obtain these parameters. Acquiring data with higher accuracy and throughput is, however, necessary to characterize complex biological networks. Here, we provide an overview of a high-throughput microfluidic method based on mechanically induced trapping of molecular interactions (MITOMI). MITOMI can be used to obtain affinity constants and kinetic rates of hundreds of protein-ligand interactions in parallel. It has been used in dozens of studies to measure binding affinities of transcription factors, map protein interaction networks, identify pharmacological inhibitors, and perform high-throughput, low-cost molecular diagnostics. This article covers the technological aspects of MITOMI and its applications.

Transcription factor binding to a single binding site on DNA and its functional consequence in a promoter context are beginning to be relatively well understood. However, binding to clusters of sites has yet to be characterized in depth, and the functional relevance of binding site clusters remains uncertain. We employed a high-throughput biochemical method to characterize transcription factor binding to clusters varying across a range of affinities and configurations. We found that transcription factors can bind concurrently to overlapping sites, challenging the notion of binding exclusivity. Furthermore, compared to an individual high-affinity binding site, small clusters with binding sites an order of magnitude lower in affinity give rise to higher mean occupancies at physiologically-relevant transcription factor concentrations in vitro. To assess whether the observed in vitro occupancies translate to transcriptional activation in vivo, we tested low-affinity binding site clusters by inserting them into a synthetic minimal CYC1 and the native PHO5 S. cerevisiae promoter. In the minCYC1 promoter, clusters of low-affinity binding sites can generate transcriptional output comparable to a promoter containing three consensus binding sites. In the PHO5 promoter, replacing the native Pho4 binding sites with clusters of low-affinity binding sites recovered activation of these promoters as well. This systematic characterization demonstrates that clusters of low-affinity binding sites achieve substantial occupancies, and that this occupancy can drive expression in eukaryotic promoters. In the antibody discovery phase, before advancing a particular antibody sequence to clinical trials, a large number of candidates are characterized for properties of therapeutic interest. Above all, this includes their ability to bind to the target antigen. Standard processes currently in use to characterize antibody binding are either only semi-quantitative (ELISA), or low-throughput (SPR). We developed an experimental process to first retrieve DNA encoding promising synthetic antibody variants directly from the output pools of phage and ribosome display methods, followed by expressing these variants on a microfluidic chip for characterization in a fluorescence-based binding assay. We applied our method to characterize the binding of retrieved Sybody variants against the Spike trimer and RBD domain of the SARS-CoV-2 variant. After an initial proof-of-concept, we improved our method by removing bottlenecks, reducing costs, and incorporating a large degree of process automation to enable measuring binding affinities at scale.

Gene regulatory networks (GRNs) determine cellular behaviour, and ultimately the functioning of single- and multicellular organisms. Transcription factors regulate gene expression by binding to DNA or via remodelling chromatin. Recent advances in biotechnological methods have made it possible to precisely characterise the binding of transcription factors to single consensus binding sites. Advanced sequencing methods allow the efficient discovery of de novo binding consensus sequences, and thus provide insight into the evolution of GRNs. However, clusters of transcription factor binding sites have been poorly studied despite their prevalence in eukaryotic genomes. Widely-used methods of GRN interrogation lack the sensitivity for transient binding events with low-affinity binding sites. In my PhD, using the MITOMI platform I showed that two orthologs of the transcription factor Pho4 from S. cerevisiae and C. glabrata that have evolved distinct regulation have similar 3D structure and DNA binding preferences over a wide affinity range. I have developed the iMITOMI assay that preserves the unique advantages of the original MITOMI platform, and thus too has the ability to rapidly characterise transient binding events with high-throughput. In addition, iMITOMI enables the study of transcription factor binding site clusters. We have studied in detail the interactions of Pho4 and Zif268 with clusters of one to six binding sites. The individual binding sites had affinities one to two orders of magnitude below the affinity of the corresponding consensus sequence. We discovered that clusters of low-affinity binding sites could give rise to transcription factor occupancies comparable to or higher than those of single-consensus sequences. Our results demonstrate that low-affinity binding site clusters give rise to high occupancy levels. These have significant implications for the biophysical characterisation of transcription factor binding, the de novo evolution of enhancers and promoters, and the development of synthetic promoters.