The drug development process is notoriously long and expensive. During preclinical studies, inaccurate prediction of pharmacokinetic properties such as the ability of a drug candidate to passively permeate cell plasma membranes contributes to the high failure rate of drug candidates during clinical trials. Passive drug permeability is currently predicted using in vitro techniques such as parallel artificial membrane permeability assays, or PAMPA. In PAMPA, drug transport is predicted between aqueous compartments via a synthetic filter filled with a phospholipid solution in an organic solvent. The lack of translatability of preclinical predictions to humans can be attributed, in part, to lack of biological similarly between models used for permeability prediction and cell plasma membranes in vivo. Here, I demonstrate a new method for pharmacokinetic prediction, built by using droplet interface bilayers (DIBs) as human-mimetic artificial cell membranes. DIBs are bilayer sections created at the interface of two aqueous droplets. In the literature, DIBs have been used as artificial cell plasma membranes to study, for example, electrophysiological properties, protein insertion, water permeability, and molecular transport. DIBs can be formed between droplets of differing composition such that one droplet can be used as a donor compartment and the other as an acceptor compartment for the quantification of molecular transport across the artificial cell membrane. DIBs have previously been used to measure the passive permeability of numerous fluorophores as well as the drugs caffeine and doxorubicin. However, the extent to which DIBs have been tuned to mimic human cell plasma membranes and transport across them is limited. I present here the use of microfluidic platforms for bespoke DIB formation, where variables such as temperature, bilayer composition, and droplet contents are customized to create biomimetic cells-on-a-chip. These artificial cells are then used to measure molecular transport with the aim of predicting permeability. In Chapter 2, I investigate the effectiveness of literature methods for the modification of polydimethylsiloxane (PDMS) microfluidic device channels for aqueous droplet formation and storage. While numerous techniques have been presented as mitigation strategies for common challenges in droplet microfluidics, it is not clear from the literature if any of these methods would be effective or necessary for the formation and analysis of DIBs. With the aim of facilitating aqueous droplet formation, I tested the effect of PDMS silanization using trichloro(1H,1H,2H,2H-perfluorooctyl)silane (PFOS) on surface hydrophobicity and oleophobicity. To assess their effect on reducing the rate of aqueous droplet evaporation, I tested surface treatment of PDMS with Teflon AF or Aquapel. I also tested modifications to the device fabrication process by bonding a glass coverslip to the surface of the device and soaking the device in water overnight. To quantify changes in PDMS surface chemistry, I performed contact angle measurements, aqueous droplet formation experiments, and measurements of droplet size during on-chip storage. I determined that baking PDMS microfluidic devices at 65 C overnight produced channel surfaces which allowed for aqueous droplet formation and storage. In Chapter 3 I present a systematic study on the role of temperature in DIB formation using naturally derived phospholipids. The use of increased temperature to form DIBs using total lipid extracts has previously been demonstrated, but has never before been investigated systematically using naturally derived phospholipids and bespoke formulations thereof. I hypothesized that, in order to form complete phospholipid monolayers and DIBs, the microfluidic device must be held at the phase transition temperature of the phospholipids. Using a custom-built heating platform, I tested DIB formation over a range of temperatures to determine conditions which allowed DIB formation rather than droplet coalescence. I show that temperature is a key parameter for DIB formation using naturally derived phospholipids in a microfluidic device. In Chapter 4, I demonstrate the use of DIBs as a new type of pharmacokinetic compartment model for intestinal absorption. Using three-droplet networks, the components of which were designed to mimic the intestinal space, the enterocyte cytosol, and the blood, I measured fluorescein permeability across intestine-mimetic DIBs. The model was able to predict the transport of fluorescein more accurately than the current state-of-the-art technique, PAMPA. Chapter 5 describes the development of complex DIB models for pharmacologically relevant membranes as well as an investigation into novel methods of drug transport detection on-chip. I created a new DIB model for the small intestine, incorporating more components of the enterocyte plasma membrane such as cholesterol. Measurement of calcein permeability served as a control experiment, as calcein does not cross cell plasma membranes. Measurement of fluorescein permeability yielded a significantly shorter permeation half-life than was determined in Chapter 4, indicating an increase in permeability with the more complex, biomimetic phospholipid formulation. I also developed sex-specific models for intestinal absorption to investigate the effect of sex-based membrane differences on permeability. This relationship has never before been explored in the literature. In comparison to the initial intestinal phospholipid formulation, the sex-specific formulations contained acyl chain tail groups which have been found in different ratios in male and female cells. A significantly longer half-life for fluorescein permeability was found in female intestine-mimetic DIBs, mirroring the slower drug absorption observed in female patients. I also used DIBs to model blood-brain barrier permeability. I demonstrate this application using two different brain lipid extracts, polar and total brain lipids. Polar brain lipids have previously been used in PAMPA to predict blood-brain barrier permeability, but have been found to overpredict the permeation of charged molecules in comparison to custom lipid formulations which mimic the composition of human brain endothelial cells. Permeability measurements in DIBs formed using polar brain lipids gave results which agree with PAMPA, as DIBs formed using polar brain lipids were permeable to fluorescein, but those formed using total brain lipids were not. Blood-brain barrier-mimetic DIBs formed using either lipid extract are impermeable to calcein and FITC-dextrans (both 40 and 500 kDa). I also show the formation of the first DIBs to be created using a total lipid extract from human cells as well as their impermeability to calcein. The extract tested was prepared from testicular Sertoli cells, which exhibit properties similar to the blood-brain barrier, but future work will focus on extracts prepared from human brain endothelial cells. Finally, I explore new options for the on-chip detection of the transport of nonfluorescent molecules. To move away from reliance on fluorescent molecules for permeability measurements, I selected three fluorogenic molecular recognition agents (fluorescamine, Chromeo P540, and DimerDye 4) whose fluorescence signal is activated by amine groups. None of the tested methods proved to be viable in DIBs, potentially due to slow permeation, low quantum yield, and side reactions with phospholipids. Overall, I demonstrate here the microfluidic formation and application of several novel types of biomimetic DIBs to permeability prediction. My work shows that DIBs can be used to predict permeability and mirror effects observed in vivo. Future work will focus on the development of new methods for the detection of drug transport and the application of the pharmacokinetic compartment models presented to predicting drug permeability. Further work using total lipid extracts prepared from human cells will also be vital to enhancing the use of biomimetic DIBs as pharmacokinetic permeability prediction tools. As their biological similarity and capacity to accurately predict transport increase, so will the potential of DIBs to improve the accuracy and translatablity of preclinical drug development. / Graduate / 2023-08-09

Modern drug development is an astronomically expensive and time consuming undertaking. Because of this, studying the pharmacokinetic properties of drugs in vitro has become an integral step early in the process of drug development, with the goal of preventing costly failures late in the process, and dangerous side effects. Artificial phospholipid bilayers known as droplet interface bilayers (DIBs) have the potential to be used for these pharmacokinetics assays, combining the low cost of cell-free assays with the ability to more closely mimic structures found in life than current cell-free in vitro techniques. Combined with the reproducibility, ease of use, and low reagent consumption found with microfluidic methods, disruptive new low cost techniques for assessing pharmacokinetics in drug development may be possible using DIBs as an artificial cell membrane model. In this work, I establish the potential of DIBs to be used as a pharmacokinetics modelling platform, and advance the use of microfluidic methods for carrying out pharmacokinetics assays in drug discovery. I first developed a new microfluidic platform for the formation of DIBs, which sought to solve some of the shortcomings of current microfluidic methods for DIB formation (Chapter 2). This device is the first that can be used to form DIB networks from dissimilar droplets in parallel, without use of active controls, and with droplet contact gentle enough to enable use of biomimetic lipid mixtures. I examine for the first time the behaviour of phospholipids on microfluidic devices, and characterise the interaction that they have with a common material used to construct microfluidic devices (Chapter 3). Not only has this interaction never been studied before, but my unexpected findings indicate a new area requiring further study in order to advance the adoption of DIBs on microfluidic devices. In collaboration with my colleague Jaime Korner, I use my newly developed microfluidic platform to carry out an on-chip permeation assay for the first time using biomimetic lipid formulations and bespoke compartments modelled after the human intestine. We demonstrate that this on-chip assay has predictive accuracy greater than that of a current widely used cell-free technique (Chapter 4). Finally, I demonstrate that a DIB based microfluidic platform enables, and is critical for, characterising the effect of structural features such as membrane asymmetry on drug permeation. With this, I find measurable, previously unknown effects of membrane asymmetry on the absorption of the chemotherapy drug doxorubicin, highlighting a possible contributing factor to chemoresistance in some cancers (Chapter 5). I find, and demonstrate throughout the body of this work that microfluidic methods and DIBs can not only provide alternatives to current cell-free in vitro pharmacokinetics assays, but that they can exceed the performance of existing assays, and be used for entirely new ways of examining pharmacokinetics. Through building bespoke artificial cell membranes from the ground up, I hope to demonstrate herein the great potential of these powerful new cell-free methods. / Graduate / 2022-09-12

Droplet Interface Bilayers (DIBs) have a number of attributes that distinguish them from conventional artificial lipid bilayers. In particular, the ability to manipulate bilayers mechanically is explored in this thesis. Directed bilayer area changes are used to make precise measurements of the specific capacitance of DIBs and to control the two dimensional concentration of a membrane protein reconstituted in the bilayer. Chapter 1 provides a general introduction to the role of the lipid membrane en- vironment in the function of biological membranes and their integral proteins. An overview of model lipid bilayer systems is given. Chapter 2 introduces work carried out in this laboratory previously and illustrates the experimental setup of DIBs. Some important bilayer biophysical concepts are covered to provide the theoretical background to experiments in this and in later chapters. Results from the characterisation of DIBs are reported, and an account of the development of methods to manipulate the bilayer by mechanical means is given. Chapter 3 describes experiments that apply bilayer area manipulation in DIBs to achieve precise measurement of specific capacitance in a range of lipid systems. Chapter 4 reports results from experiments investigating the response of bilayer specific capacitance to an applied potential. Chapter 5 covers the background and experimental setup for total internal fluo- rescence microscopy experiments in DIBs and describes the expression, purification and characterisation of the bacterial β-barrel membrane protein pore α-Hemolysin. Chapter 6 describes experiments that apply the mechanical manipulation of bilayer area in DIBs to the corralling and control of the surface density of α-Hemolysin.