<<

The Role of Domains and Chemistry in Nanoparticle-

Interactions

A thesis presented to

the faculty of

the Russ College of Engineering and Technology of Ohio University

In partial fulfillment

of the requirements for the degree

Master of Science

Andrew B. Fuhrer

August 2020

© 2020 Andrew B. Fuhrer. All Rights Reserved. 2

This thesis titled

The Role of Lipid Domains and Sterol Chemistry in Nanoparticle-Cell Membrane

Interactions

by

ANDREW B. FUHRER

has been approved for

the Department of Chemical and Biomolecular Engineering

and the Russ College of Engineering and Technology by

Amir M. Farnoud

Assistant Professor of Chemical and Biomolecular Engineering

Mei Wei

Dean, Russ College of Engineering and Technology 3

Abstract

FUHRER, ANDREW B., M.S., August 2020, Biomedical Engineering

The Role of Lipid Domains and Sterol Chemistry in Nanoparticle-Cell Membrane

Interactions

Director of Thesis: Amir M. Farnoud

There is a growing interest in the scientific research community to develop nanoparticles for use in novel commercial and biomedical applications, fueled by recent advances in nanotechnology and nanoparticle synthesis. Potential applications for nanoparticles include use as catalysts during chemical manufacturing processes, use as drug delivery vehicles and imaging agents for biomedical applications, and as surfaces for adsorption during removal of environmental pollutants. The use of nanoparticles in such applications has raised questions concerning their safety and impact on human health. Answers to these questions require a greater understanding of the interactions between nanoparticles and living cells. Models of the cell membrane have been employed to investigate how nanoparticles may adsorb to, fuse with, or penetrate the cell membrane, however careful consideration of the membrane model for such mechanistic studies is necessary. This thesis investigates the role of domains, which are lipid phase segregations comprised of saturated and , in modulating nanoparticle-membrane interactions and further explores how sterol chemistry impacts said interaction. Model membranes were synthesized with an equimolar ratio of sphingomyelin, 1,2-dioleoyl-sn-glycero-3-phosphocholine, and varied sterol composition to yield vesicles with varied lipid domain properties. Fluorescence anisotropy and Förster 4 resonance energy transfer of fluorescent probes was measured to quantify the degree of ordered domain formation in model vesicles. Additionally, confocal microscopy was performed to visualize lipid domains. Following lipid domain characterization, vesicles in which a self-quenching fluorescent dye was encapsulated were exposed to plain silica nanoparticles (diameter 37.5 ± 1.8 nm) and leakage of dye was measured to determine the degree of membrane disruption. By analyzing the results of vesicle leakage assays alongside the results from domain characterization, it was concluded that the lipid domain profile of the membrane alone is not an ideal predictor of nanoparticle-membrane interactions. By expanding the range of membrane models to include vesicles containing a more varied selection of sterol, it is shown that the structure of the sterol present in the membrane can impact nanoparticle-membrane interactions in a manner not predicted on the basis of the sterol’s impact on lipid domain formation. These findings help better elucidate the disruptive effects of nanomaterials on biological membranes, depending on membrane lipid chemistry and biophysical properties.

5

Dedication

To my parents, Brad and Wendy, and my brother, Aaron

6

Acknowledgments

I would like to acknowledge my research and thesis advisor, Dr. Amir M.

Farnoud, for the mentorship he has provided to me throughout my academic career.

Many of my academic goals I would not have met were it not for his guidance. Dr.

Farnoud sets an excellent example for all academic mentors to strive to emulate. His enthusiasm for teaching, both in the classroom and in the lab, and his passion to always learn more inspires his peers and tutors. His charisma and genuine interest in the growth and progress of his mentees cannot be understated. My first encounter with Dr. Farnoud was as an undergraduate student taking one of several courses under his instruction. It was through knowing him in this capacity that I expressed interest to him in attending graduate school. Dr. Farnoud generously offered to accept me into his lab as a master’s student, where I have grown and learned much under his assistance. I must also acknowledge the rest of the Farnoud lab group for being supportive of each other and may we continue to support each other as we embark on whatever journey we may take in the future. Lastly, I would like to thank my committee members, Drs. Sumit Sharma,

Shiyong Wu, and Douglas Goetz, who additionally is the head of the Biomedical

Engineering program at Ohio University.

7

Table of Contents

Page

Abstract ...... 3 Dedication ...... 5 Acknowledgments ...... 6 List of Figures ...... 8 Chapter 1: Introduction* ...... 10 The Cell Membrane ...... 11 Use of Model Membranes to Study Nanoparticle-Cell Membrane Interactions ...... 15 Objectives ...... 18 Chapter 2: Lipid Domain Characterization in Model Membranes ...... 20 Introduction ...... 20 Materials and Methods ...... 25 Commercial Reagents ...... 25 Vesicle Synthesis ...... 26 Fluorescence Anisotropy ...... 26 Förster Resonance Energy Transfer ...... 27 Confocal Microscopy ...... 27 Results and Discussion ...... 28 Conclusion ...... 33 Chapter 3: Role of Lipid Domains and Sterol Structure in Silica Nanoparticle-Membrane Interactions ...... 35 Introduction ...... 35 Materials and Methods ...... 37 Commercial Reagents ...... 37 Fluorescence Anisotropy ...... 37 Leakage Assays ...... 37 Results and Discussion ...... 40 Conclusion ...... 46 Chapter 4: Conclusions and Future Work ...... 47 Future Work ...... 49 References ...... 50 8

List of Figures Page

Figure 1-1. Example of structures (left) with hydrophilic head groups shown in green and acyl chains shown in black; a saturated and unsaturated lipid are shown. Structure of (middle). Diagram of showing cholesterol packing (right)...... 13 Figure 1-2. Commonly used cell membrane models. Lipid monolayer (left), supported lipid bilayer (middle), multilamellar and unilamellar vesicles in suspension (right)...... 17 Figure 2-1. Schematic depicting fluorescence anisotropy. Light first passes the monochromator, where the proper wavelength is selected, passes through a polarizer, and strikes the sample exciting fluorophore. Light emitted from the sample passes through a set of changing horizontal and vertical polarizers, to measure the contribution of light in each plane. The emitted light passes through a second monochromator before reaching the detector...... 22 Figure 2-2. FRET donor is attached to saturated lipid and FRET acceptor is attached to unsaturated lipid. FRET efficiency is high when lipids are dispersed randomly thought the bilayer and low when segregated lipid domains are formed. Reprinted from ‘Lipid Rafts - Methods and Protocols’...... 24 Figure 2-3. Selected sterol structures, drawn using Chemspace. Cholesterol and are expected to induce formation of LO domains whereas and androstenol are not...... 29 Figure 2-4. Anisotropy values for vesicles containing 1:1:1 SM/DOPC/Sterol. Each line corresponds to a vesicle containing a unique sterol (A). The lipophilic probe DPH was dissolved in acetone and incorporated into preformed vesicles at 0.2 mol% concentration and anisotropy values were recorded (n = 3) across a range of temperatures. High anisotropy corresponds to high lipid packing order. Anisotropy values at selected temperatures for side by side comparison of sterol effect on anisotropy values (B). Asterisks denote statistical difference. Only for coprostanol and androstenol vesicles was there a statistical difference in anisotropy at increased temperature, suggesting these vesicles are more susceptible to perturbations in lipid order based on temperature...... 30 Figure 2-5. Temperature dependent FRET between the saturated lipid probe NBD-DPPE (donor) and the unsaturated lipid probe rho-DOPE (acceptor). NBD was inserted into SM/DOPC/sterol vesicles at 0.5 mol% while rho-DOPE was inserted at 2 mol%, n = 1 trial...... 32 Figure 2-6. GUV images of cholesterol containing vesicles (left), coprostanol containing vesicles (middle), and ergosterol containing vesicles (right). Dark regions indicate ordered domains where the unsaturated lipid probe rho-DOPE is excluded...... 33 Figure 3-1. The leakage assay detects nanoparticle induced disruption of the vesicle lipid bilayer. Self-quenching fluorescent dye is released upon disruption, becomes dilute, and fluoresces...... 38 9

Figure 3-2. Per cent leakage of carboxyfluorescein dye from vesicle models following silica nanoparticle exposure. A high per cent leakage indicates greater disruption of the lipid bilayer by nanoparticles. Asterisks denote statistical difference as determined by ANOVA with post-hoc Tukey analysis (n=3)...... 41 Figure 3-3. Selected sterols used as constituents in vesicle models, including two additional sterols: epicholesterol and 5-androsten-3β-ol (drawn using Chemspace)...... 42 Figure 3-4. Anisotropy values for ternary vesicle models, including those containing epicholesterol and 5-androsten-3β-ol...... 43 Figure 3-5. Per cent leakage of carboxyfluorescein dye from vesicle models following silica nanoparticle exposure, including epicholesterol and 5-androsten-3β-ol vesicles. For each sterol, the bar on the left represents data taken at 25 °C, the bar on the right represents data taken at 37 °C...... 44

10

Chapter 1: Introduction*1

In the modern age, the use of nanotechnology has allowed scientists and engineers to deliberately manipulate nanoscale objects to enhance their material properties, but the history of nanotechnology extends far back into the ancient world. As early as the fourteenth and thirteenth centuries BCE, the ancient Egyptians and Mesopotamians may have unknowingly incorporated metallic nanoparticles in their glasswork to provide it with previously unobtainable optical properties.1 Damascus steel blades, forged in India during the seventeenth century, have been analyzed and found to contain nanotubes and cementite nanowires.2 It is not until 1959, however, that anyone is credited with exploring the idea of using modern tools and technological methods to intentionally create nanosized objects with unique properties.3 In December of that year, then a professor at the California Institute of Technology, Richard Feynman delivered his famous lecture “There’s Plenty of Room at the Bottom” during a meeting of the

American Physical Society, urging scientists to devise methods to manipulate individual atoms and to create nanoscale machines.4 Later, in 1974, the term nanotechnology was first used by Norio Taniguchi at an international industrial production conference in

Tokyo.3,5

Since the early days of its conception in the 20th century, nanotechnology has progressed at a staggering rate. More unifying definitions of nanoscale materials have

* Part of this work has been published as a chapter (authored by Andrew Fuhrer and co-authored by Amir M. Farnoud) in the book titled ‘Lipid Rafts – Methods and Protocols’, published by Springer and edited by Erhard Bieberich. Part of this work will also be submitted to the peer reviewed journal, Physical Biology, in a paper authored by Andrew Fuhrer and co-authored by Amir M. Farnoud. 11 been established to include materials that contain only one, two, or three dimensions at the nanoscale. Foundations such as the U.S. National Nanotechnology Initiative (NNI) have been established to provide research and development funding and a long-term vision for nanotechnology.6 Since its establishment in 2000, the NNI has invested almost

$29 billion in nanotechnology research and is supported by such federal organizations as the NIH, NSF, DOE, DOD, and NIST.7 Today, nanoparticles are used in a variety of applications, ranging from food packaging, microelectronics, and environmental remediation.8 Promising applications for nanomaterials in biology and medicine include use as biological labels, drug and gene therapy delivery agents, scaffolds for tissue engineering, use in bio separations, and use as imaging contrast agents.9 Of special interest to the biomedical field are silica nanoparticles, which exhibit high payload capacity and surface modification potential when compared to other nanomaterials.10 In order to make effective use of such nanoparticles in biomedical applications and reduce cytotoxicity, greater understanding of how these particles interact with living cells is required.

The Cell Membrane

Human and other mammalian cells are surrounded by a biological membrane, called the cell (or plasma) membrane, that separates the interior of the cell from the external environment, protecting it from foreign substances. The plasma membrane is composed of a lipid bilayer (Figure 1-1). The lipids, which are amphiphilic containing a hydrophilic headgroup and a hydrophobic tail region, arrange themselves in the bilayer to form two leaflets, such that the headgroups in the interior leaflet are 12 exposed to the aqueous cytosolic interior of the cell and the headgroups of the outer leaflet face the external aqueous environment. The hydrophobic tails, comprised of acyl chains, are tucked into the bilayer to minimize contact with water. Such a bilayer is formed spontaneously based on the principles of thermodynamics.11 Associated with the cell membrane are various integral proteins spanning the bilayer and peripheral proteins loosely attached at one leaflet. These proteins may act as ion channels, enzymes, or cell signal receptors and provide several other functions to the cell membrane aside from merely protecting the cell from harm.12 Many membrane lipids and proteins are glycosylated, providing greater diversity of structure and function.

Lipids are a diverse class of molecules, but the cell membrane is primarily composed of , glycolipids, and sterols.13 Phospholipids are characterized by a headgroup containing a phosphate . This class of lipids includes sphingomyelin, which consists of a phosphocholine headgroup, a sphingosine backbone, and a tail. Sterols are composed of a basic four ring structure, with a hydroxyl group at the number 3 carbon of the A ring. The most well-known sterol is cholesterol, and it is a major constituent of mammalian cell membranes. The exact lipid composition of the cell membrane varies by cell type; furthermore, for an individual cell, lipid composition varies over time.13,14 13

Figure 1-1. Example of phospholipid structures (left) with hydrophilic head groups shown in green and acyl chains shown in black; a saturated and unsaturated lipid are shown. Structure of cholesterol (middle). Diagram of lipid bilayer showing cholesterol packing (right).

The cell membrane is dynamic, in that its lipid constituents can laterally diffuse quickly throughout the bilayer. On average, a lipid molecule can laterally diffuse a distance of 2 μm per second in the cell membrane.15 On the other hand, movement of phospholipids from one leaflet to the other is slow; electron spin resonance has been used to show a phospholipid transversely diffuses across the membrane only once in several hours.15 Thus, transverse diffusion of phospholipids across the membrane is usually facilitated by the ATP-dependent transporter proteins flippase, floppase, and scramblase, owing to the thermodynamically unfavorable interactions a lipid faces when rotating through the membrane.16 14

Membrane fluidity is a measure of the ease with which lipids and other small molecules diffuse laterally throughout the membrane. Fluidity is impacted by how closely lipids pack in the bilayer. Lipids with long saturated acyl chains pack closely together due to hydrophobic interactions. Unsaturated lipids disrupt close packing because of the steric hindrances imposed by the double bond kinks in their acyl chains. This effect is especially prominent because the double bonds found naturally in the cell membrane almost always adopt the cis conformation, with extending from the same side of the double bond.

Cholesterol also profoundly impacts the fluidity of the cell membrane.

Cholesterol is a much smaller molecule than the surrounding lipids and tucks into the membrane such that its hydrophobic tail and ring structure pack with the acyl chains of the bulk lipids and the hydrophilic OH group interacts with the lipid head groups. The

Umbrella Model has been proposed to suggest that the phospholipid head groups shield the cholesterol molecule from the aqueous phase.17 Regardless, the cholesterol molecule is very stiff and rigid due to its ring structure and the double bond between carbons 5 and

6. Cholesterol is able to impart its stiffness to the surrounding membrane, impacting its fluidity. In fact, membrane fluidity is naturally regulated when cells alter their membrane cholesterol content. The cell regulates its phospholipid and cholesterol content in response to temperature changes, which also impact membrane fluidity. High temperature, high energy states provide greater mobility to lipids and increase membrane fluidity, so the cell will often increase the saturated lipid and cholesterol content of the membrane in such cases to counteract this effect. 15

When discussing membrane fluidity, it is useful to describe lipid phases. A loosely packed membrane exhibiting high fluidity is said to be in the liquid disordered

(LD) phase and a tightly packed membrane with more restricted fluidity is said to be in

18 the liquid ordered (LO) phase. A third phase is described which appears only in synthetic model membranes or under conditions of extremely low temperature that would kill the cell. This is the gel phase that is characterized by extremely tight lipid packing

19 and little to no membrane fluidity. Very often, spatially segregated regions of LO phase and LD phase lipids arise in the membrane. A phase boundary of transitional lipid packing is found between, separating the regions. Such intermingled regions are often referred to as lipid domains (i.e. LO and LD domains).

A special case of lipid domains, called the , is often distinguished. Lipid rafts are small nano-sized LO clusters especially rich in saturated phospholipids and

20 cholesterol, surrounded by a bed of LD lipids. The fact that these lipid rafts are rich in cholesterol suggests that cholesterol has the ability to stabilize saturated lipid interactions, and indeed it has been demonstrated that single phase LD model membranes lacking cholesterol can be induced to form phase segregated LD and LO lipid domains upon the addition of cholesterol.21 Other sterols, such as epicholesterol and ergosterol, have been observed to support domain formation.22 It has further been shown that some sterols, such as coprostanol and androstenol, do not support the ability to form ordered domains.22,23

Use of Model Membranes to Study Nanoparticle-Cell Membrane Interactions

As previously described, silica nanoparticles show great potential for application in the biomedical field. However, research points to their potential cytotoxic effects, such 16 as membrane damage, reactive species generation, and depletion of the antioxidant glutathione.24,25 The nanoparticle-cell membrane interaction must be better understood in order to design safer and more efficacious nanoparticles for biomedical applications. However, due to the complex and dynamic nature of the natural cell membrane, it is challenging to study and understand the mechanistic detail of the factors governing these nanoparticle-membrane interactions.

In order to facilitate nanoparticle-cell membrane interaction research, several cell membrane models have historically been applied. Models are advantageous for their ease of preparation and the methods of investigation that they lend themselves well to. These models include lipid monolayers, supported lipid bilayers, and liposomal (vesicle) suspensions (Figure 1-2).26,27 Lipid monolayers are assembled in a trough by injection of lipids at the air-water interface. The lipids spontaneously assemble to form a film, minimizing surface tension by pointing polar heads down into the water and exposing hydrophobic tails to the air.28 Supported lipid bilayer models are formed when a film of lipids interacts with a solid hydrophilic support to form one leaflet of a lipid bilayer as other lipids associate to form the second leaflet, limiting acyl chain exposure to the aqueous environment.29 Lastly, liposomes, or vesicles, are spontaneously self-assembled at sufficient lipid concentration in aqueous solution. Vesicle size and lamellarity (a lipid bilayer may form in the aqueous interior of another) varies, thus classification of vesicles is necessary. There are multilamellar vesicles (MLVs), small unilamellar vesicles

(SUVs), which are generally agreed to range in size from 15-100 nm, large unilamellar vesicles (LUVs) which range from 100-200 nm, and giant unilamellar vesicles (GUVs), 17 which are considered greater than 1 micron in size.30,31 Size exclusion methods exist to isolate unilamellar vesicles of a uniform size distribution for study.32

Figure 1-2. Commonly used cell membrane models. Lipid monolayer (left), supported lipid bilayer (middle), multilamellar and unilamellar vesicles in suspension (right).

Each of the membrane models has its merits, for instance, supported lipid bilayers are particularly suited to surface topographical techniques such as surface plasmon resonance and atomic force microscopy and have been used to study membrane electrochemical properties.29 However, consolidation of nanoparticle-membrane interaction research findings remains an issue. For example, it is not clear that the principles governing nanoparticle interactions with supported lipid bilayers have the same impact as in lipid vesicles. It has already been shown that silica nanoparticles interact differently with lipid monolayers than lipid bilayers.33 This demonstrates that careful selection of a membrane model is required and that researches should be cautious to apply their findings broadly to all cell membranes.

Use of vesicles as membrane models has become ubiquitous in research owing to their biological relevance, but there is still much that is unaccounted for. Many vesicle models lack intricate complexities such as the presence of lipid domains. Even amongst those studies that account for lipid domains, results are often conflicting. In 2016, one 18 group found that cationic gold nanoparticles experienced greater attachment to model membranes exhibiting liquid ordered domains when compared to those lacking such domains, though the model was a supported lipid bilayer.34 Another study, published in

2018, found that gold nanoparticles coated with amphiphilic organic ligands penetrated vesicles exhibiting a fluid phase but not vesicles exhibiting the gel phase.35 Yet another team used molecular dynamics simulations to propose that cationic nanoparticles adsorb most strongly to multicomponent phase segregated bilayers, less strongly to single phase

36 LD bilayers, and poorly to single phase LO bilayers. These inconsistencies in the literature clearly point to the need for a better understanding of how membrane lipid domains regulate the interactions of nanomaterials and lipid membranes.

Objectives

To better understand the interactions of nanoparticles with the cell membrane and to develop nanoparticles effectively for biomedical applications, the impact of lipid domains must be better understood. By taking advantage of the previously described ability of certain sterols to promote or hinder domain formation in membranes, vesicle models can be synthesized to display a range of domain characteristics and in turn silica nanoparticles can be introduced to study their interactions with said domains. The strategy is to synthesize vesicle models composed of a saturated lipid, an unsaturated lipid, and a sterol. By changing the sterol and not the mole balance of saturated vs. unsaturated lipids, the impact of domain formation can be observed and minimal changes in membrane curvature and surface charge are made. In the current study, vesicles are synthesized with a 1:1:1 molar ratio of the saturated lipid sphingomyelin (SM), the 19 unsaturated lipid 1,2-dioleoyl-sn-glycero-3-phosphocholine (DOPC), and a choice of sterol, ranging from cholesterol and ergosterol (expected to promote ordered domain formation) to coprostanol and androstanol (expected to hinder ordered domain formation). Interactions with silica nanoparticles (nominal diameter 30 nm) are observed.

Our overall hypothesis is that presence of ordered lipid domains will impact nanoparticle-membrane interactions and that membranes exhibiting a greater quantity of ordered lipid domains (rafts) will experience less disruption by silica nanoparticles. This is tested through the following aims:

Aim 1. Characterize lipid domain formation in vesicles containing varied sterol compositions.

Aim 2. Investigate the interactions of silica nanoparticles with vesicles that display different degrees of lipid domain formation. 20

Chapter 2: Lipid Domain Characterization in Model Membranes

Introduction

Advances in analytical techniques have allowed for greater understanding of lipid rafts. Now, it has been proposed that lipid rafts play several important roles in . It has been suggested that lipid rafts may be able to recruit specific proteins

(such as glycosylphosphatidylinositol anchored proteins), leading to their sequestration and localization into certain regions of the cell membrane.37 This would allow the cell membrane to spatially regulate certain membrane functions, such as signal transduction and vesicular transport.38 It has even been suggested that raft-association may alter protein conformation and regulate protein activity.37 Additionally, lipid rafts are thought to be highly conserved among various species. Raft like microdomains have been observed in fungi (where domains are rich in ergosterol rather than cholesterol), , and even in intracellular membranes.39–41

To more accurately mimic natural cell membranes, it is desired to synthesize vesicle models that exhibit lipid microdomains (rafts). The strategy selected is to choose a vesicle model with a saturated lipid, an unsaturated lipid, and a sterol, and to alter the sterol composition to induce varied degree of domain formation in the bilayer.

Given the small size of lipid domains, researchers have focused on developing novel methods to examine their presence in cells and/or model membranes. Fluorescence- based models have received special attention. Methods commonly used to characterize lipid domains in model membranes include the measurement of fluorescence anisotropy, 21 measurement of Förster resonance energy transfer (FRET), and visualization of lipid domains using confocal microscopy. Each model is described in more detail below.

Fluorescence anisotropy: Fluorescence anisotropy provides a measure of overall membrane fluidity. Using this method, a small lipophilic fluorescent probe is inserted into the membrane in trace quantities.42 Polarized light strikes and excites the fluorophore. Fluorophore molecules which have their electric dipole moments of transition randomly aligned in the direction of the electric field of the polarized light are preferentially excited.43 If fluorophore Brownian motion is restricted, fluorophore is not free to spatially reorient and thus the emitted light retains its polarization. In cases where fluorophore reorientation occurs at time scales comparable to the absorption and emission of light, polarization of emitted light is diminished. The Brownian motion of a fluorophore in a lipid bilayer membrane is partially dependent on the fluidity of the membrane. Anisotropy is the ratio of light intensity in the polarized plane to total light intensity,

퐼 −퐼 푟 = ∥ ⊥ (2-1) 퐼푇 where r is the anisotropy, I is light intensity, and the subscripts ∥ and ⊥ denote light in the parallel and perpendicular planes, respectively. Anisotropy is a value between 0 and 1, the more restricted fluorophore motion is, the greater the value. Membranes with a large portion of ordered domains restrict fluorophore motion to a greater extent, thus polarization is retained, and anisotropy is high. In the opposite case, when the membrane 22 exhibits a small portion of ordered domains, fluidity is high, and anisotropy is low.

Measurement of fluorescence anisotropy is depicted in Figure 2-1.

Figure 2-1. Schematic depicting fluorescence anisotropy. Light first passes the monochromator, where the proper wavelength is selected, passes through a polarizer, and strikes the sample exciting fluorophore. Light emitted from the sample passes through a set of changing horizontal and vertical polarizers, to measure the contribution of light in each plane. The emitted light passes through a second monochromator before reaching the detector.

While fluorescence anisotropy provides an overall measurement of membrane fluidity, it does not necessarily prove the existence of segregated lipid domains. High anisotropy may correspond to a single-phase LO membrane and low anisotropy may correspond to a single-phase LD membrane. Intermediate values may or may not correspond to a membrane with intermingled LO and LD domains. To infer the existence of domains, additional techniques are required. One such technique is the measurement of

Förster resonance energy transfer (FRET) between two fluorophores. 23

FRET: As stated, FRET is a measurement of energy transfer between two fluorophores. The technique requires that the emission spectrum of the first fluorophore, designated the ‘donor’, overlaps the excitation of the second fluorophore, the

‘acceptor’.44 If such a condition is met, upon excitation, the donor molecule may transfer energy to the acceptor via a non-radiative process that depends on transfer of energy through dipole-dipole coupling.45 The efficiency of the energy transfer process is inversely proportional to the distance between the FRET pair raised to the power of six, making FRET very sensitive to measurements of distance in the 1-10 nm range. FRET can be adapted to identify membrane lipid domains by conjugating a donor fluorophore to the headgroup of an unsaturated lipid and an acceptor molecule to the headgroup of a saturated lipid, or vice versa. If segregated lipid domains form in the membrane, the saturated/unsaturated lipids are separated by space and thus the FRET pair is separated, and energy transfer is low. For a visual depiction of FRET used as described, see Figure

2-2. 24

Figure 2-2. FRET donor is attached to saturated lipid and FRET acceptor is attached to unsaturated lipid. FRET efficiency is high when lipids are dispersed randomly thought the bilayer and low when segregated lipid domains are formed. Reprinted from ‘Lipid Rafts - Methods and Protocols’.46

Confocal Imaging: Finally, lipid domains can be visualized in micron sized giant unilamellar vesicles (GUVs). GUVs can be synthesized using several methods, but the electroformation method is common. Lipids are smeared on a conductive coverslip, placed in a chamber with DI water, and electrodes are attached to provide an external electric field.47 This causes the lipids, with charged headgroups, to swell to a large size. A small fluorescent molecule can be conjugated to the headgroup of an unsaturated lipid and incorporated into GUVs for visualization under confocal microscope. If homogeneous fluorescence is observed throughout the vesicle, then no lipid domains form. If dark regions are observed, they correspond to regions where the unsaturated 25 lipids were excluded, indicating formation of ordered domains rich in saturated lipids and disordered domains rich in unsaturated lipids.

Materials and Methods

Commercial Reagents

Porcine brain sphingomyelin (SM), 1,2-dioleoyl-sn-glycero-3-phosphocholine

(DOPC), cholesterol (chol), and the fluorescently labeled lipids 1,2-dioleoyl-sn-glycero-

3-phosphoethanolamine-N-(lissamine rhodamine B sulfonyl) (rho-DOPE) and 1,2- dipalmitoyl-sn-glycero-3-phosphoethanolamine-N-(7-nitro-2-1,3-benzoxadiazol-4-yl)

(NBD-DPPE) were purchased from Avanti Polar Lipids, Inc. (Alabaster, AL). Ergosterol

(erg) was purchased from Matreya, LLC (State College, PA). Coprostanol, and androstenol were purchased from Steraloids, Inc. (Newport, RI). Lipid stocks were prepared by dissolving dry powder lipids in HPLC grade chloroform from Fisher

Scientific (Hampton, NH). The lipophilic fluorescent probe 1,6-diphenyl-1,3,5-hexatriene

(DPH) was purchased from Sigma Aldrich (St. Louis, MO). Acetone and ethanol for cleaning were purchased from Fisher Scientific and Decon Labs, Inc. (King of Prussia,

PA), respectively. Purified water, with resistivity 18.2 mΩ·cm, was obtained from an

ELGA PURELAB classic (High Wycombe, UK). Phosphate buffered saline (PBS) solution was prepared with powder packs from Fisher BioReagents (Fair Lawn, NJ).

Silica nanoparticles were purchased from Mikromod Partikel Technologie GMBH

(Rostock, Germany).

26

Vesicle Synthesis

LUVs were prepared using the protocols of Zhang et al.48 Stock lipids were mixed in chloroform at a 1:1:1 molar ratio of SM/DOPC/sterol. Choice of sterol was either cholesterol, ergosterol, coprostanol, or androstenol. Lipid solution was placed in a

SpeedVac™ vacuum concentrator from Thermo Fisher Scientific (Waltham, MA) for 15 minutes to remove chloroform. Dried lipid film was heated to 70 °C and hydrated in 1x strength PBS, at concentration of 5 mM total lipids. Lipid suspension was vortexed to assemble MLVs. When required, MLVs were subjected to 7 freeze/thaw cycles in acetone bath (in dry ice)/water bath (70 °C), holding samples in each medium for 3 minutes, to free entrapped vesicles and reduce vesicle size, yielding LUVs. LUVs of uniform size distribution were obtained by passing vesicles through a 100 nm pore size membrane extruder (Avanti Polar Lipids) 11 times at 70 °C.

Fluorescence Anisotropy

For anisotropy experiments, preformed vesicles were incubated at room temperature for one hour, aliquoted to quartz cuvettes and diluted to 50 μM in PBS. DPH, a lipophilic fluorescent probe, was dissolved in acetone and added to vesicles and inserted into the bilayer. DPH concentration in vesicles was 0.2 mol %. The fluorescence anisotropy of the probe was measured on a Fluorolog®3 spectrofluorometer from Horiba

Scientific (Edison, NJ). Anisotropy values were collected over a range of temperatures, from 4 °C – 44 °C, at 4-degree increments, with time for thermal equilibration.

27

Förster Resonance Energy Transfer

Vesicles of the same composition as in anisotropy experiments were prepared except during vesicle synthesis, the fluorescently labeled lipids NBD-DPPE (saturated) and rho-DOPE (unsaturated) were introduced at the lipid mixing step, at concentrations of 0.5 mol % and 2 mol%, respectively. NBD-DPPE and rho-DOPE form a FRET pair, where NBD has excitation and emission peaks at 460 nm and 535 nm, respectively, and rhodamine has excitation and emission peaks at 560 nm and 583 nm, respectively.

Vesicles were made using the same steps except that vesicle suspensions were not subjected to freeze and thaw, as it was found previously that freeze and thaw causes separation of probes into separate vesicles.49 For each FRET experiment, four vesicle samples were prepared. Samples labeled F had both donor and acceptor lipids.

Background F samples had only acceptor probe. F0 samples had only donor probe, and background for F0 had no probe, only plain vesicles. Vesicles were incubated in the dark at room temperature before transfer to quartz cuvettes. A Fluorolog®3 spectrofluorometer was used to excite NBD at 460 nm and measure its emission at 535 nm. The ratio of fluorescence intensity, F/F0 (FRET efficiency), was calculated.

Confocal Microscopy

GUVs were synthesized via electroformation.50 Equimolar of SM, DOPC, and sterol was mixed in chloroform at 500 μM concentration and 0.2 mol % rho-DOPE was added for purposes of visualization.51 A 1.5 μL volume of lipids was streaked across a tin indium oxide conductive coverslip. Lipids were dried in a desiccator for 5 minutes. The lipid coverslip and a second clean coverslip were placed across a sealed, DI-water filled 28 chamber (300 μL), conductive sides facing in, and hooked up to electrodes. Lipids were heated to 70 °C and an electric field was applied via a generator set at 10 Hz and 700 mV.

The electric field was applied for four hours, over which time lipids swelled from one coverslip toward the other, forming giant unilamellar vesicles. The GUV chamber was then transferred to an LSM 510 upright confocal microscope from Zeiss (Thornwood,

NY) for imaging.

Results and Discussion

Vesicle models were synthesized with equimolar ratios of a saturated lipid, an unsaturated lipid, and a sterol. This lipid composition was chosen to control partitioning of saturated lipids and unsaturated lipids into spatially distinct lipid domains and to observe the impact on silica nanoparticle interactions. Control of domain formation was achieved by altering sterol composition in vesicles, using the previously described ability of certain sterols to either induce (cholesterol and ergosterol) or hinder (coprostanol and androstenol) domain formation in lipid bilayers. Structures of the selected sterols are shown in Figure 2-3. In the case of coprostanol, we suspect that the lack of a double bond in the ring structure of the sterol reduces rigidity, and thus reduces the sterols ability to pack tightly in the lipid bilayer. Androstenol lacks an R group, and we hypothesize this reduces Van der Waals interactions with surrounding phospholipid acyl chain groups, leading to a less stabilized bilayer. Porcine brain sphingomyelin (SM, average formula weight of 760.223) was chosen as the saturated lipid and DOPC, which has 18 carbons and one cis double bond in each of its acyl chains, was chosen as the unsaturated lipid.

SM is a major component in cell membranes, especially in nerve cells and red blood 29 cells, and is known to form strong associations with cholesterol. Phosphatidylcholines, of which DOPC is a member, are also major components of the cell membrane.

Lipid packing in vesicle models containing varied sterol was measured using fluorescence anisotropy. Anisotropy values for those sterols are shown in Figure 2-4.

Figure 2-3. Selected sterol structures, drawn using Chemspace. Cholesterol and ergosterol are expected to induce formation of LO domains whereas coprostanol and androstenol are not.

30

Figure 2-4. Anisotropy values for vesicles containing 1:1:1 SM/DOPC/Sterol. Each line corresponds to a vesicle containing a unique sterol (A). The lipophilic probe DPH was dissolved in acetone and incorporated into preformed vesicles at 0.2 mol% concentration and anisotropy values were recorded (n = 3) across a range of temperatures. High anisotropy corresponds to high lipid packing order. Anisotropy values at selected temperatures for side by side comparison of sterol effect on anisotropy values (B). Asterisks denote statistical difference. Only for coprostanol and androstenol vesicles was there a statistical difference in anisotropy at increased temperature, suggesting these vesicles are more susceptible to perturbations in lipid order based on temperature.

As was expected, anisotropy values were higher for vesicles containing either cholesterol or ergosterol, indicating overall higher lipid packing order in the bilayer.

Coprostanol and androstenol vesicles showed low anisotropy, thus low lipid packing order. In the case of coprostanol this is likely due to the lack of a double bond in the ring structure allowing the molecule to adopt a non-planar conformation, reducing the sterol’s ability to pack tightly with surrounding phospholipids. Androstenol, which results in the lowest anisotropy, additionally has OH group in the α configuration and lacks an R group, which reduces the hydrophobic interactions of the molecule with the acyl chains.

Anisotropy values at 25 °C (room temperature) and at 37 °C (body temperature) are shown for comparison (Figure 2-4B). Only coprostanol and androstenol vesicles showed 31 significant decrease in anisotropy at higher temperature compared to respective values at

25 °C.

While anisotropy was used to investigate membrane fluidity, localization of saturated and unsaturated lipids into nanodomains was probed via FRET. Briefly, FRET results indicate vesicles with cholesterol as sterol form phase segregated domains and that vesicles with coprostanol as sterol do not form domains. However, experimental FRET efficiency values proved challenging to reproduce, although certain general trends in values were observed. The fluorescence of the saturated donor lipid probe NBD-DPPE, which packs in the ordered phase, was measured both in the presence (F) of unsaturated acceptor probe rhodamine-DOPE, which packs in the disordered phase, and in its absence

(F0). F0 is expected to be greater than F, as when the donor is in proximity to the acceptor, energy is transferred from the donor to the acceptor. The closer the value of

F/F0 is to unity, the less efficient the energy process is. High F/F0 indicates low energy transfer and separation of lipid probes into segregated domains. Low F/F0 indicates high energy transfer and proximity of donor and acceptor probes. F/F0 values for cholesterol

(green) and coprostanol (red) vesicles are shown in Figure 2-5. 32

Figure 2-5. Temperature dependent FRET between the saturated lipid probe NBD-DPPE (donor) and the unsaturated lipid probe rho-DOPE (acceptor). NBD was inserted into SM/DOPC/sterol vesicles at 0.5 mol% while rho-DOPE was inserted at 2 mol%, n = 1 trials.

Note that at low temperatures, cholesterol vesicles exhibit high F/F0 values, indicating low FRET efficiency and segregation of lipid probes into separate domains. As temperature increases, F/F0 decreases, until domains are completely dissolved at approximately 36 °C. For coprostanol, F/F0 values are low and nearly temperature independent, suggesting domains are not formed even at low temperatures.

GUV imaging was used to examine the existence of large-scale domains in the structure of the model membranes. For GUV imaging, rhodamine-DOPE, a fluorescent lipid probe that preferentially packs into LD domains, was incorporated into vesicles.

Vesicles were made using the previously described electroformation method. A confocal microscope was used to take images. Images of cholesterol, ergosterol, and coprostanol vesicles are shown in Figure 2-6. Regions that appear red are portions of the bilayer 33 where rhodamine-DOPE has packed and indicate LD regions. In images taken of cholesterol and ergosterol vesicles, separate dark and illuminated regions exist, showing the segregation of liquid ordered domains and liquid disordered domains. For coprostanol vesicles, there was no separation of dark and illuminated regions, suggesting the entire bilayer exists as one uniform LD phase. Androstenol GUVs were unable to be formed using the electroformation method, lipids stuck to coverslip and were unable to swell out towards oppositely charged electrode.

Figure 2-6. GUV images of cholesterol containing vesicles (left), coprostanol containing vesicles (middle), and ergosterol containing vesicles (right). Dark regions indicate ordered domains where the unsaturated lipid probe rho-DOPE is excluded.

Conclusion

Lipid domain characterization of vesicle models consisting of 1:1:1

SM/DOPC/sterol revealed the ability of various sterols to induce or hinder lipid domain formation in the lipid bilayer. Fluorescence anisotropy revealed that vesicles containing either cholesterol or ergosterol were more ordered and less fluid than vesicles containing either coprostanol or androstenol. FRET experiments revealed that cholesterol vesicles 34 exhibited a temperature dependent ability to form domains on the basis that at low temperature, F/F0 values were high, suggesting FRET probes were separated by space, but gradually, as temperature increase, F/F0 values decreased and levelled out, suggesting that at higher temperatures domains were dissolved and FRET probe was brought into proximity. For coprostanol, energy transfer was efficient and independent of temperature, suggesting coprostanol vesicles lacked the ability to form domains in the entire range of temperatures tested. Results from FRET compliment anisotropy data. Lastly, GUV imaging allowed for the visualization of lipid domains in cholesterol and ergosterol vesicles, while coprostanol vesicles did not form domains. Having completed the characterization of vesicles with various degrees of domain formation, tt was decided to probe the impact of membrane lipid domains on nanoparticle-membrane interactions.

35

Chapter 3: Role of Lipid Domains and Sterol Structure in Silica Nanoparticle-

Membrane Interactions

Introduction

The interactions of silica nanoparticles with cells are of great interest to biomedical researchers. A greater understanding of nanoparticle-cell interactions will allow for the design of safer and more efficacious nanomaterials for use in such applications as disease diagnosis, drug delivery, and tissue engineering. Nanoparticle-cell interactions depend on several different factors, including the physicochemical properties of the nanoparticles themselves and the cell type.52 These factors determine the initial interaction of the nanoparticle with the cell membrane (whether particles adhere to, fuse with, or penetrate the cell membrane) and ultimately determine the response of the cell to nanoparticle exposure. The biopersistence of the nanoparticle in the cell, the possible exocytosis of the nanoparticle, and the interactions of the nanoparticle with intracellular membranes and organelles are also of great consequence.52 Nanoparticles have been shown to alter many cell functions upon uptake, including mitochondrial activity, cytoskeletal function, and gene expression.53 These consequences provide the motivation to further study nanoparticle-cell interactions.

As described previously, membrane models are extremely useful for investigating the mechanistic detail of the nanoparticle-membrane interaction. The interactions of nanoparticles with membrane models have been extensively studied; however, most studies overlook the impact of ordered lipid domains on the interaction. Amongst those few studies that investigate the role of lipid domains, results are often inconclusive. A 36 study by Melby et al. found that gold nanoparticles attached to supported lipid bilayers with LO domains (composition 60:20:20 DOPC/SM/chol) more readily than to those that lacked domains (pure DOPC).34 However, removing cholesterol and SM from the vesicle to elucidate the role of lipid domains also drastically alters lipid mole balance. Atukorale et al. studied the interactions of ligand coated gold nanoparticles with giant multilamellar vesicles of varied composition to observe the effect of membrane fluidity on nanoparticle-membrane interactions.35 They found that nanoparticles interacted strongly with vesicles composed of pure DOPC and vesicles with a 50/50 ratio of DOPC and the saturated lipid DPPC, however nanoparticles did not interact with pure DPPC vesicles.

Lastly, Sheavly et al. used molecular simulations to suggest that cationic nanoparticles more readily interact with single phase LD bilayers than single phase LO bilayers, but that the particles attach most strongly at the phase boundary of a two-phase lipid bilayer.36

Each of these studies utilize different nanoparticle models, different membrane models with differing lipid compositions, and different techniques to draw their conclusions.

Further investigation is required to consolidate nanoparticle-lipid domain interactions.

Here, we investigate the interactions of silica nanoparticles with the vesicle models characterized in chapter two. It has already been shown that by altering the sterol composition of the vesicle model, the lipid domain properties can be controlled.

Cholesterol and ergosterol vesicles exhibit ordered lipid domain formation while androstenol and coprostanol vesicles do not. Importantly, only the sterol identity has been altered between the membrane models, and the overall lipid mole balance (1:1:1

SM/DOPC/sterol) remains the same. By exposing these vesicles to silica nanoparticles, 37 we can observe the effect of lipid domains on the nanoparticle-membrane interaction. We hypothesize that vesicles exhibiting ordered lipid domains will show reduced disruption by silica nanoparticles, based on stronger lipid-lipid interactions.

Materials and Methods

Plain silica nanoparticles with green fluorescence (λex: 569 nm, λem: 585 nm) and nominal diameter of 30 nm were purchased from Mikromod Partikel Technologie GMBH

(Rostock, Germany). Hydrodynamic radius in PBS was measured using a Zetasizer from

Malvern Panalytical, Ltd. (Malvern, United Kingdom).

Commercial Reagents

In addition to the commercial reagents previously used, epicholesterol and 5- androsen-3β-ol were purchased from Steraloids, Inc. (Newport, RI). 5(6)- carboxyfluorescein (CF) and Triton X-100 were purchased from Sigma Aldrich (St.

Louis, MO).

Fluorescence Anisotropy

For the new vesicle models, fluorescence anisotropy was carried out in the same fashion described previously.

Leakage Assays

The leakage assay provides an effective means to measure vesicle disruption upon nanoparticle exposure. During synthesis, a self-quenching fluorescent dye is incorporated into the interior of the vesicle (Figure 3-1). Upon disruption of the lipid bilayer by nanoparticles, dye leaks from the interior, is diluted by the surrounding aqueous solution, and begins to fluoresce. The intensity of fluorescence is directly proportionally to the 38 volume of dye released from the vesicle interior, and thus indicates the level of membrane disruption.

Figure 3-1. The leakage assay detects nanoparticle induced disruption of the vesicle lipid bilayer. Self-quenching fluorescent dye is released upon disruption, becomes dilute, and fluoresces.

A per cent leakage of dye at any time can be quantified by treating vesicles with strong detergent following nanoparticle exposure and fluorescence measurement. Strong detergent completely disrupts the vesicle and the corresponding fluorescence represents the maximal fluorescence and complete leakage of dye. By taking the ratio of fluorescence at any time prior to treatment with detergent and the maximum fluorescence after treatment, per cent leakage is established, providing a measure of membrane disruption.

Lipids from stock solutions were mixed in chloroform at a 1:1:1 ratio of

SM/DOPC/sterol, with choice of sterol ranging from either cholesterol, ergosterol, coprostanol, or androstanol, as in previous experiments. Lipids were dried in a

SpeedVac™ vacuum concentrator and rehydrated. For the leakage assay, rehydration solution consisted of 30 mg of carboxyfluorescein (CF) dissolved in 790 μL of PBS and 39

210 μL of 1 M NaOH (to increase pH to allow CF to dissolve). The concentration of CF in the rehydration solution was approximately 80 mM, above the self-quenching concentration of the fluorescent dye. Lipids were vortexed to cause spontaneous assembly into MLVs, with encapsulated CF dye. MLVs were subjected to 7 cycles of freeze/thaw for size reduction and to form LUVs. Extrusion was carried out to obtain a uniform size distribution of vesicles. Lastly, LUVs were passed through PD-10 desalting columns with Sephadex™ G-25 medium from GE Healthcare (Chicago, IL) to remove un-encapsulated CF. As the vesicle suspension flows through the column, the large vesicles are excluded from pores in the Sephadex™ matrix and eluted first, while the unencapsulated lower molecular weight CF is entrapped in pores and separated from vesicles. Isolated vesicles were transferred to quartz cuvettes and diluted to 50 μM. Silica nanoparticles were added to the vesicle suspensions at 0.01 g/L. Control samples were not treated with nanoparticles. CF fluorescence was measured using the same

Fluorolog®3 spectrofluorometer as in previous experiments. CF was excited at 492 nm and emission was measured at 517 nm. Measurements were taken at regular intervals over an hour-long period, holding temperature constant. After the one-hour testing period, Triton X-100 was added to samples and fluorescence was measured a final time.

Per cent leakage was calculated using the formula:

% 퐿푒푎푘푎푔푒 = (퐹푇 − 퐹0)/(퐹100 − 퐹0) (3-1) where FT is fluorescence at any timepoint prior to treatment with detergent, F0 is fluorescence of the control sample at time 0, and F100 is fluorescence after complete disruption of vesicles following treatment with Triton X-100. 40

Results and Discussion

Plain silica nanoparticles were received from the manufacturer dispersed in water at 25 mg/mL. For particle characterization, nanoparticles were diluted in PBS at 0.1 mg/mL (to match experimental conditions) and particle size/charge was measured using a

Zetasizer. Particle diameter was 37.5 ± 1.8 nm and particle surface charge was -12.6 ±

1.8 mV, due to the presence of silanol groups on the nanoparticle surface.

Following exposure to nanoparticles, cholesterol, ergosterol, and, surprisingly, coprostanol vesicles showed a similar low maximal leakage after 1 hour for temperature at 25 °C. Cholesterol vesicles experienced 20.1 ± 6.0 per cent leakage, ergosterol vesicles

16.4 ± 1.1, and coprostanol 11.4 ± 2.5 per cent leakage. Androstenol vesicles showed high leakage at 75.3 ± 9.5% leakage (Figure 3-2). Cholesterol vesicle leakage increased significantly at 37 °C, and, interestingly, androstenol vesicles showed decreased disruption from nanoparticles at higher temperature. No significant change in leakage for ergosterol or coprostanol was observed at higher temperature. Results from the leakage assay do not support the hypothesis that liquid ordered lipid domains significantly impact the nanoparticle-membrane interaction. Cholesterol and ergosterol vesicles exhibit liquid ordered domains (as determined by anisotropy, FRET, and GUV imaging) while coprostanol vesicles do not form domains, yet there is no significant difference in disruption experienced by the different vesicle models. Unpredictably, coprostanol and androstenol vesicles, both lacking domains, had extremely opposite response to nanoparticle exposure: coprostanol vesicles experienced very little disruption while androstenol vesicles experienced the greatest disruption. 41

Figure 3-2. Per cent leakage of carboxyfluorescein dye from vesicle models following silica nanoparticle exposure. A high per cent leakage indicates greater disruption of the lipid bilayer by nanoparticles. Asterisks denote statistical difference as determined by ANOVA with post-hoc Tukey analysis (n=3).

To investigate the role of sterol structure in modulating the nanoparticle- membrane interaction, an additional two vesicle models were studied. Consistent with previous experiments, the vesicles were composed of 1:1:1 SM/DOPC/sterol. The new vesicle models contained either epicholesterol or 5-androsten-3β-ol as a sterol constituent. With these additions, the entire range of membrane models included six vesicle types: cholesterol, ergosterol, coprostanol, androstenol, epicholesterol, 5- androsten-3β-ol. The structures of these sterols are shown in Figure 3-3. 42

Figure 3-3. Selected sterols used as constituents in vesicle models, including two additional sterols: epicholesterol and 5-androsten-3β-ol (drawn using Chemspace).

Key sterol structural differences include the lack of a double bond between carbons 5-6 in coprostanol and androstenol, the α-orientation of the OH group in epicholesterol and androstenol (as opposed to β-orientation in the rest), and the lack of an

R group at carbon 17 in androstenol and 5-androsten-3β-ol. Ergosterol has an additional double bond in its ring structure and a double bond in the R group. It is possible that the orientation of the OH group affects sterol interactions with lipid headgroups, and the identity of the R group impacts sterol interactions with lipid tails. Double bonds may also impact sterol packing by altering sterol rigidity. To gain insight on how sterol impacts the membrane and interactions between the membrane and nanoparticles, the additional vesicle models were investigated using fluorescence anisotropy and the leakage ass

Results from fluorescence anisotropy with the additional two vesicle models are 43 shown in Figure 3-4. Epicholesterol vesicles exhibit high anisotropy, statistically similar to anisotropy exhibited by cholesterol, thus the vesicle models have similar lipid packing density. Epicholesterol varies from cholesterol only in the orientation of its OH group. 5- androsten-3β-ol vesicles have intermediate anisotropy, less than the strong domain forming cholesterol, ergosterol, and epicholesterol vesicles, but greater than the weakly domain forming androstenol vesicles. 5-androsten-3β-ol differs from cholesterol in that it lacks an R group extending from the ring structure.

Figure 3-4. Anisotropy values for ternary vesicle models, including those containing epicholesterol and 5-androsten-3β-ol.

Extension of the leakage assay to incorporate two additional vesicle models provided interesting insights into how sterol structure impacts nanoparticle induced disruption of the vesicle membrane (Figure 3-5). Most surprisingly, when analyzing results of leakage assays conducted at 25 °C and 37 °C, only two vesicle models 44 experienced a significant decrease in disruption at the greater temperature. Both containing sterols, epicholesterol and androstenol, that had the α configuration of the OH group, this despite that epicholesterol shows propensity to form liquid ordered domains and androstenol does not. It should also be noted that the two models experiencing the greatest disruption, 5-androsten-3-βol and androstenol, both lacked an R group.

Figure 3-5. Per cent leakage of carboxyfluorescein dye from vesicle models following silica nanoparticle exposure, including epicholesterol and 5-androsten-3β-ol vesicles. For each sterol, the bar on the left represents data taken at 25 °C, the bar on the right represents data taken at 37 °C.

When analyzing the results from anisotropy and leakage assays combined, it does not appear the lipid domain profile of the bilayer can be used to predict whether nanoparticles will disrupt the membrane or not. Sterol chemistry seems to have a greater impact, although just exactly what features of the sterol molecule are important for 45 modulating nanoparticle-membrane interactions are unclear, especially considering that some research groups have proposed an umbrella model for sterols in the membrane that implies sterols are tucked away beneath larger phospholipid headgroups and inaccessible to extracellular materials.17 It has been shown that epicholesterol, due to its axial OH position, inserts into the bilayer at a tilted angle when compared to cholesterol, and this disrupts the packing of surrounding lipids. Interestingly, at higher temperatures degrees of motion for the molecule are increased and the tilt of the molecule is decreased, leading to more perpendicular placement of epicholesterol with respect to the bilayer normal.54

This suggests one reason why only epicholesterol and androstenol, with α-OH configuration, experience reduced disruption at higher temperature, although it is unclear why this effect does not manifest in a more disordered membrane and lower anisotropy values in the case of epicholesterol vesicles. Because ergosterol, which has an additional double bond in the sterol ring structure, shows only marginal decrease in disruption compared to cholesterol, and because coprostanol, which lacks a double bond, shows even less disruption, the flexibility of the sterol ring seems to have little effect on nanoparticle-membrane interactions. The lack of an R group in androstenol and 5- androsten-3β-ol seems to have the most critical impact on nanoparticle-membrane interactions. It is perhaps the case that as with epicholesterol, these sterols pack into the bilayer at exaggerated angles as a result of having fewer conformational restrictions that would be imposed by having long acyl chains that interact with other phospholipid tails.

The energy barriers to adopt extreme angle positions in the bilayer would most likely be 46 lower, allowing these two sterols to pack with less order, and additionally putting greater stress on the surrounding membrane.

Conclusion

The leakage assays conducted on the vesicle models did not indicate a relationship between lipid domains and nanoparticle interactions. Because the membrane models varied only by choice of sterol constituent, it was then hypothesized that sterol structural elements must be modulating the nanoparticle-membrane interaction. The decision to expand the range of vesicle models to include additional sterols was made to further probe the impact of sterol structure on nanoparticle-membrane interactions.

After expanding experimentation to include two additional vesicle models, those containing epicholesterol or 5-androsten-3β-ol, some insights into how sterol structure impact nanoparticle-membrane interactions have been established. It appears that the orientation of the sterol OH group plays a significant role in modulating the lipid bilayer’s response to increased temperature. While most vesicle models experienced greater disruption upon nanoparticle exposure at higher temperatures, those containing sterol molecules with the α-OH orientation showed a reduction at higher temperature.

47

Chapter 4: Conclusions and Future Work

The growing trend in the 21st century to incorporate nanotechnology into an ever- increasing number of industrial and domestic applications has been fueled by advances in nanoparticle synthesis and characterization. Nanoparticles may now be found in the electronics, cosmetics, and food industries, to name a few. In addition, nanoparticles show promise for use in the biomedical field for such applications as drug delivery, tissue engineering, and medical imaging. The development of nanoparticles for these applications calls for greater understanding of how nanoparticles impact health and safety. Understanding how the cell membrane responds to these foreign substances is of great importance. The cell membrane is complex and dynamic, and many factors may affect whether nanoparticles adsorb to, fuse with, or penetrate the membrane. A little explored aspect of the cell membrane is the role that lipid domains play in modulating the nanoparticle-membrane interaction.

To elucidate how lipid domains alter nanoparticle-membrane interactions, the interactions of silica nanoparticles with ternary lipid vesicle models was studied. Sterol composition of the models was altered to control lipid domain formation. The ability to synthesize vesicles exhibiting varied domain formation by altering sterol composition was verified using fluorescence anisotropy, FRET, and confocal microscopy.

Fluorescence anisotropy revealed that cholesterol, epicholesterol, and ergosterol vesicles had high lipid order, while coprostanol and androstenol vesicles showed low lipid order.

Lipid order in 5-androsten-3β-ol vesicles was intermediate. FRET analysis revealed that cholesterol vesicles exhibited segregated liquid ordered and liquid disordered domains, 48 consistent with high lipid order revealed by anisotropy. Coprostanol vesicles showed no separation of lipid probes into distinct membrane regions. It was possible to visualize distinct lipid domains in cholesterol and ergosterol vesicles using confocal microscopy, while coprostanol vesicles appeared homogenous. Results of the leakage assay showed no correlation between lipid bilayer disruption and presence of lipid domains. While vesicle models exhibiting high lipid order (cholesterol, epicholesterol, ergosterol) showed similar response to nanoparticle exposure, vesicle models with low lipid order

(coprostanol, androstenol, 5-androsten-3β-ol) show a widely varying response.

Coprostanol vesicles showed the lowest disruption in all models, while androstenol showed the greatest.

Based on the testing, it appears there is no ability to predict membrane response to nanoparticles based on the lipid domain profile of the membrane. However, there appears to be a relationship between sterol structure and response to nanoparticle exposure.

Vesicles containing sterols that have α-OH group configuration display reduced nanoparticle disruption at high temperature. This may relate to the way in which the OH group interacts with the polar phospholipid headgroups of the surrounding lipids. When vesicles contain sterols lacking an R group, disruption by nanoparticles is high. This may be a result of fewer Van der Waals interactions with the surrounding phospholipid acyl chains. The impact of sterol double bonds on the nanoparticle-membrane interaction is uncertain. 49

Future Work

To better understand the interactions of nanoparticles with the cell membrane, research must be continued to elucidate the full mechanistic details governing these phenomena. The impact of lipid domains and sterol structure is only one small portion of full picture. Furthermore, the scope of the present study has been limited. Future exploration of a more expanded suite of sterols would allow for more and better conclusions to be drawn. Similarly, there are other membrane complexities aside from lipid domains that have rarely been studied to determine their impact on nanoparticle- membrane interactions. These complexities include asymmetry of lipid composition across in the inner and outer leaflets of the lipid bilayer. Recently, techniques have been developed to synthesize asymmetric vesicles by extracting outer leaflet lipids and replacing them with desired lipids to cyclodextrins.51 Additionally, many in vitro studies on nanoparticle-membrane interactions are done with bare nanoparticles, neglecting that when in the blood, nanoparticles are exposed to a wide variety of proteins, which form a corona around the particle, modulating its surface properties.55

Ultimately, through the combined efforts of the scientific research community, the complexities of the nanoparticle-membrane interaction will be worked out. As the collective knowledge grows, scientists and engineers will be better able to design engineered nanomaterials for biomedical, industrial, and domestic applications. The development of more relevant membrane models will be a crucial part of the journey.

50

References

(1) Schaming, D.; Remita, H. Nanotechnology: From the Ancient Time to Nowadays. Found. Chem. 2015, 17 (3), 187–205. https://doi.org/10.1007/s10698-015-9235-y. (2) Reibold, M.; Paufler, P.; Levin, A. A.; Kochmann, W.; Pätzke, N.; Meyer, D. C. Carbon Nanotubes in an Ancient Damascus Sabre. Nature 2006, 444 (7117), 286. https://doi.org/10.1038/444286a. (3) Tolochko, N. K. History of Nanotechnology. Encyclopedia of Life Support Systems; Nanoscience and Technologies; UNESCO; p 6. (4) Feynman, R. P. There’s Plenty of Room at the Bottom: An Invitation to Enter a New Field of Physics - Proceedings from the Annual American Physical Society Meeting at Caltech; Pasadena, California, 1959. (5) Bayda, S.; Adeel, M.; Tuccinardi, T.; Cordani, M.; Rizzolio, F. The History of Nanoscience and Nanotechnology: From Chemical–Physical Applications to Nanomedicine. Molecules 2019, 25 (1), 112. https://doi.org/10.3390/molecules25010112. (6) Roco, M. C. The Long View of Nanotechnology Development: The National Nanotechnology Initiative at 10 Years. J. Nanoparticle Res. 2011, 13 (2), 427–445. https://doi.org/10.1007/s11051-010-0192-z. (7) Subcommittee on Nanoscale Science, Engineering, and Technology - Committee on Technology of the National Science and Technology Council. The National Nanotechnology Initiative—Supplement to the President’s 2020 Budget. Executive Office of the President of the United States August 2019. (8) Khan, I.; Saeed, K.; Khan, I. Nanoparticles: Properties, Applications and Toxicities. Arab. J. Chem. 2019, 12 (7), 908–931. https://doi.org/10.1016/j.arabjc.2017.05.011. (9) Salata, O. Applications of Nanoparticles in Biology and Medicine. J. Nanobiotechnology 2004, 2 (1), 3. https://doi.org/10.1186/1477-3155-2-3. (10) Gonçalves, M. C. Sol-Gel Silica Nanoparticles in Medicine: A Natural Choice. Design, Synthesis and Products. Mol. Basel Switz. 2018, 23 (8), 2021. https://doi.org/10.3390/molecules23082021. (11) Alberts, B.; Johnson, A.; Lewis, J.; Raff, M.; Roberts, K.; Walter, P. The Lipid Bilayer. In Molecular Biology of the Cell. 4th edition; W. W. Norton & Co: New York, 2002. 51

(12) Lodish, H.; Berk, A.; Zipursky, S. L.; Matsudaira, P.; Baltimore, D.; Darnell, J. Membrane Proteins. In Molecular Cell Biology. 4th edition; W. H. Freeman and Company, 2000. (13) Harayama, T.; Riezman, H. Understanding the Diversity of Membrane Lipid Composition. Nat. Rev. Mol. Cell Biol. 2018, 19 (5), 281–296. https://doi.org/10.1038/nrm.2017.138. (14) Phua, S. C.; Chiba, S.; Suzuki, M.; Su, E.; Roberson, E. C.; Pusapati, G. V.; Setou, M.; Rohatgi, R.; Reiter, J. F.; Ikegami, K.; Inoue, T. Dynamic Remodeling of Membrane Composition Drives Cell Cycle through Primary Cilia Excision. Cell 2017, 168 (1–2), 264-279.e15. https://doi.org/10.1016/j.cell.2016.12.032. (15) Berg, J. M.; Tymoczko, J. L.; Gatto Jr., G. J.; Stryer, L. Biochemistry, 8th ed.; Macmillan, 2015. (16) Hankins, H. M.; Baldridge, R. D.; Xu, P.; Graham, T. R. Role of Flippases, Scramblases, and Transfer Proteins in Phosphatidylserine Subcellular Distribution. Traffic Cph. Den. 2015, 16 (1), 35–47. https://doi.org/10.1111/tra.12233. (17) Dai, J.; Alwarawrah, M.; Huang, J. Instability Of Cholesterol Clusters In Lipid Bilayers And The Cholesterol’s Umbrella Effect. J. Phys. Chem. B 2010, 114 (2), 840. https://doi.org/10.1021/jp909061h. (18) Mouritsen, O. G.; Bagatolli, L. A. Lipid Domains in Model Membranes: A Brief Historical Perspective. Essays Biochem. 2015, 57, 1–19. https://doi.org/10.1042/bse0570001. (19) Parasassi, T.; Loiero, M.; Raimondi, M.; Ravagnan, G.; Gratton, E. Absence of Lipid Gel-Phase Domains in Seven Mammalian Cell Lines and in Four Primary Cell Types. Biochim. Biophys. Acta BBA - Biomembr. 1993, 1153 (2), 143–154. https://doi.org/10.1016/0005-2736(93)90399-K. (20) Simons, K.; Toomre, D. Lipid Rafts and Signal Transduction. Nat. Rev. Mol. Cell Biol. 2000, 1 (1), 31–39. https://doi.org/10.1038/35036052. (21) London, E.; Brown, D. A. Insolubility of Lipids in Triton X-100: Physical Origin and Relationship to /Cholesterol Membrane Domains (Rafts). Biochim. Biophys. Acta BBA - Biomembr. 2000, 1508 (1–2), 182–195. https://doi.org/10.1016/S0304-4157(00)00007-1. (22) Delle Bovi, R. J.; Kim, J.; Suresh, P.; London, E.; Miller, W. T. Sterol Structure Dependence of Insulin Receptor and Insulin-like Growth Factor 1 Receptor Activation. Biochim. Biophys. Acta BBA - Biomembr. 2019, 1861 (4), 819–826. https://doi.org/10.1016/j.bbamem.2019.01.009. 52

(23) LaRocca, T. J.; Pathak, P.; Chiantia, S.; Toledo, A.; Silvius, J. R.; Benach, J. L.; London, E. Proving Lipid Rafts Exist: Membrane Domains in the Prokaryote Borrelia Burgdorferi Have the Same Properties as Eukaryotic Lipid Rafts. PLoS Pathog. 2013, 9 (5), e1003353. https://doi.org/10.1371/journal.ppat.1003353. (24) Liang, H.; Jin, C.; Tang, Y.; Wang, F.; Ma, C.; Yang, Y. Cytotoxicity of Silica Nanoparticles on HaCaT Cells. J. Appl. Toxicol. 2013, 34 (4), 367–372. https://doi.org/10.1002/jat.2953. (25) Armstrong, J. S.; Steinauer, K. K.; Hornung, B.; Irish, J. M.; Lecane, P.; Birrell, G. W.; Peehl, D. M.; Knox, S. J. Role of Glutathione Depletion and Reactive Oxygen Species Generation in Apoptotic Signaling in a Human B Lymphoma Cell Line. Cell Death Differ. 2002, 9 (3), 252–263. https://doi.org/10.1038/sj.cdd.4400959. (26) Siontorou, C. G.; Nikoleli, G.-P.; Nikolelis, D. P.; Karapetis, S. K. Artificial Lipid Membranes: Past, Present, and Future. Membranes 2017, 7 (3). https://doi.org/10.3390/membranes7030038. (27) Adib, A. A.; Nazemidashtarjandi, S.; Kelly, A.; Kruse, A.; Cimatu, K.; David, A. E.; Farnoud, A. M. Engineered Silica Nanoparticles Interact Differently with Lipid Monolayers Compared to Lipid Bilayers. Environ. Sci. Nano 2018, 5 (2), 289–303. https://doi.org/10.1039/C7EN00685C. (28) Baoukina, S.; Monticelli, L.; Risselada, H. J.; Marrink, S. J.; Tieleman, D. P. The Molecular Mechanism of Lipid Monolayer Collapse. Proc. Natl. Acad. Sci. 2008, 105 (31), 10803–10808. https://doi.org/10.1073/pnas.0711563105. (29) Bechinger, B. Supported Lipid Bilayers. In Encyclopedia of Biophysics; Roberts, G. C. K., Ed.; Springer Reference; Springer: Berlin, Heidelberg, 2013; pp 2522–2528. https://doi.org/10.1007/978-3-642-16712-6_566. (30) Iglic, A.; Kulkarni, C.; Rappolt, M. Advances in Planar Lipid Bilayers and Liposomes, 1st ed.; Elsevier, 2015; Vol. 21. (31) Stein, H.; Spindler, S.; Bonakdar, N.; Wang, C.; Sandoghdar, V. Production of Isolated Giant Unilamellar Vesicles under High Salt Concentrations. Front. Physiol. 2017, 8, 63. https://doi.org/10.3389/fphys.2017.00063. (32) Nazemidashtarjandi, S.; Vahedi, A.; Farnoud, A. M. Lipid Chemical Structure Modulates the Disruptive Effects of Nanomaterials on Membrane Models. Langmuir 2020, 36 (18), 4923–4932. https://doi.org/10.1021/acs.langmuir.0c00295. (33) Adib, A. A.; Nazemidashtarjandi, S.; Kelly, A.; Kruse, A.; Cimatu, K.; David, A. E.; Farnoud, A. M. Engineered Silica Nanoparticles Interact Differently with Lipid Monolayers Compared to Lipid Bilayers. Environ. Sci. Nano 2017, 5 (2), 289–303. https://doi.org/10.1039/C7EN00685C. 53

(34) Melby, E. S.; Mensch, A. C.; Lohse, S. E.; Hu, D.; Orr, G.; Murphy, C. J.; Hamers, R. J.; Pedersen, J. A. Formation of Supported Lipid Bilayers Containing Phase- Segregated Domains and Their Interaction with Gold Nanoparticles. Environ. Sci. Nano 2015, 3 (1), 45–55. https://doi.org/10.1039/C5EN00098J. (35) Atukorale, P. U.; Guven, Z. P.; Bekdemir, A.; Carney, R. P.; Van Lehn, R. C.; Yun, D. S.; Jacob Silva, P. H.; Demurtas, D.; Yang, Y.-S.; Alexander-Katz, A.; Stellacci, F.; Irvine, D. J. Structure–Property Relationships of Amphiphilic Nanoparticles That Penetrate or Fuse Lipid Membranes. Bioconjug. Chem. 2018, 29 (4), 1131–1140. https://doi.org/10.1021/acs.bioconjchem.7b00777. (36) Sheavly, J. K.; Pedersen, J. A.; Lehn, R. C. V. Curvature-Driven Adsorption of Cationic Nanoparticles to Phase Boundaries in Multicomponent Lipid Bilayers. Nanoscale 2019, 11 (6), 2767–2778. https://doi.org/10.1039/C8NR07763K. (37) Sezgin, E.; Levental, I.; Mayor, S.; Eggeling, C. The Mystery of Membrane Organization: Composition, Regulation and Roles of Lipid Rafts. Nat. Rev. Mol. Cell Biol. 2017, 18 (6), 361–374. https://doi.org/10.1038/nrm.2017.16. (38) Munro, S. Lipid Rafts: Elusive or Illusive? Cell 2003, 115 (4), 377–388. https://doi.org/10.1016/S0092-8674(03)00882-1. (39) Farnoud, A. M.; Toledo, A. M.; Konopka, J. B.; Del Poeta, M.; London, E. Raft- Like Membrane Domains in Pathogenic Microorganisms. Curr. Top. Membr. 2015, 75, 233–268. https://doi.org/10.1016/bs.ctm.2015.03.005. (40) Bramkamp, M.; Lopez, D. Exploring the Existence of Lipid Rafts in Bacteria. Microbiol. Mol. Biol. Rev. MMBR 2015, 79 (1), 81–100. https://doi.org/10.1128/MMBR.00036-14. (41) Ouweneel, A. B.; Thomas, M. J.; Sorci-Thomas, M. G. The Ins and Outs of Lipid Rafts: Functions in Intracellular Cholesterol Homeostasis, Microparticles, and Cell Membranes. J. Lipid Res. 2019, 61 (5), 676–686. https://doi.org/10.1194/jlr.TR119000383. (42) Jähnig, F. Structural Order of Lipids and Proteins in Membranes: Evaluation of Fluorescence Anisotropy Data. Proc. Natl. Acad. Sci. U. S. A. 1979, 76 (12), 6361–6365. (43) Lakowicz, J. R. Principles of Fluorescence Spectroscopy, 3rd ed.; Springer: New York, 2006. (44) Selvin, P. R. The Renaissance of Fluorescence Resonance Energy Transfer. Nat. Struct. Biol. 2000, 7 (9), 730–734. https://doi.org/10.1038/78948. 54

(45) Sahoo, H. Förster Resonance Energy Transfer – A Spectroscopic Nanoruler: Principle and Applications. J. Photochem. Photobiol. C Photochem. Rev. 2011, 12 (1), 20–30. https://doi.org/10.1016/j.jphotochemrev.2011.05.001. (46) Lipid Rafts: Methods and Protocols; Bieberich, E., Ed.; Methods in Molecular Biology; Springer US, 2020. https://doi.org/10.1007/978-1-0716-0814-2. (47) Witkowska, A.; Jablonski, L.; Jahn, R. A Convenient Protocol for Generating Giant Unilamellar Vesicles Containing SNARE Proteins Using Electroformation. Sci. Rep. 2018, 8 (1), 9422. https://doi.org/10.1038/s41598-018-27456-4. (48) Zhang, X.; St. Clair, J. R.; London, E.; Raleigh, D. P. Islet Amyloid Polypeptide Membrane Interactions: Effects of Membrane Composition. Biochemistry 2017, 56 (2), 376–390. https://doi.org/10.1021/acs.biochem.6b01016. (49) Pathak, P.; London, E. Measurement of Lipid Nanodomain (Raft) Formation and Size in Sphingomyelin/POPC/Cholesterol Vesicles Shows TX-100 and Transmembrane Helices Increase Domain Size by Coalescing Preexisting Nanodomains But Do Not Induce Domain Formation. Biophys. J. 2011, 101 (10), 2417–2425. https://doi.org/10.1016/j.bpj.2011.08.059. (50) Veatch, S. L.; Keller, S. L. Separation of Liquid Phases in Giant Vesicles of Ternary Mixtures of Phospholipids and Cholesterol. Biophys. J. 2003, 85 (5), 3074–3083. (51) Nazemidashtarjandi, S.; Farnoud, A. M. Membrane Outer Leaflet Is the Primary Regulator of Membrane Damage Induced by Silica Nanoparticles in Vesicles and Erythrocytes. Environ. Sci. Nano 2019, 6 (4), 1219–1232. https://doi.org/10.1039/C8EN01267A. (52) Rothen-Rutishauser, B.; Bourquin, J.; Petri-Fink, A.; Gehr, P.; Zellner, R. Nanoparticle-Cell Interactions: Overview of Uptake, Intracellular Fate and Induction of Cell Responses. In Biological Responses to Nanoscale Particles: Molecular and Cellular Aspects and Methodological Approaches; NanoScience and Technology; Springer, Cham, 2019; pp 153–170. https://doi.org/10.1007/978-3-030-12461-8_6. (53) Panariti, A.; Miserocchi, G.; Rivolta, I. The Effect of Nanoparticle Uptake on Cellular Behavior: Disrupting or Enabling Functions? Nanotechnol. Sci. Appl. 2012, 5, 87–100. https://doi.org/10.2147/NSA.S25515. (54) Dufourc, E. J. Sterols and Membrane Dynamics. J. Chem. Biol. 2008, 1 (1–4), 63–77. https://doi.org/10.1007/s12154-008-0010-6. (55) Chen, F.; Wang, G.; Griffin, J. I.; Brenneman, B.; Banda, N. K.; Holers, V. M.; Backos, D. S.; Wu, L.; Moghimi, S. M.; Simberg, D. Complement Proteins Bind to Nanoparticle Protein Corona and Undergo Dynamic Exchange in Vivo. Nat. Nanotechnol. 2016, 12 (4), 387–393. https://doi.org/10.1038/nnano.2016.269. ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! ! !

Thesis and Dissertation Services ! !