1 Toxicity of Food-Relevant Nanoparticles in Intestinal Epithelial

Toxicity of Food-Relevant Nanoparticles in Intestinal Epithelial Models

DISSERTATION

Presented in Partial Fulfillment of the Requirements for the Degree Doctor of Philosophy in the Graduate School of The Ohio State University

Christie McCracken

Biomedical Sciences Graduate Program

The Ohio State University

2015

Dissertation Committee:

Dr. W. James Waldman, Advisor

Dr. Estelle Cormet-Boyaka

Dr. Prabir K. Dutta

Dr. Narasimham L. Parinandi

Christie McCracken

2015

Abstract

Nanoparticles are increasingly being incorporated into common consumer products, including in foods and food packaging, for their unique properties at the nanoscale. Food-grade silica and titania are used as anti-caking and whitening agents, respectively, and these particle size distributions are composed of approximately one-third nanoparticles. Zinc oxide and silver nanoparticles can be used for their antimicrobial properties. However, little is known about the interactions of nanoparticles in the body upon ingestion. This study was performed to investigate the role of nanoparticle characteristics including surface chemistry, dissolution, and material type on toxicity to the intestinal epithelium.

Only mild acute toxicity of zinc oxide nanoparticles was observed after 24-hour treatment of intestinal epithelial C2BBe1 cells based on the results of toxicity assays measuring necrosis, apoptosis, membrane damage, and mitochondrial activity. Silica and titanium dioxide nanoparticles were not observed to be toxic although all nanoparticles were internalized by cells. In vitro digestion of nanoparticles in solutions representing the stomach and intestines prior to treatment of cells did not alter nanoparticle toxicity. Long- term repeated treatment of cells weekly for 24 hours with nanoparticles did not change nanoparticle cytotoxicity or the growth rate of the treated cell populations. Thus, silica,

ii titanium dioxide, and zinc oxide nanoparticles were found to induce little toxicity in intestinal epithelial cells.

Fluorescent silica nanoparticles were synthesized as a model for silica used in foods that could be tracked in vitro and in vivo. To maintain an exterior of pure silica, a silica shell was hydrolyzed around a core particle of quantum dots or a fluorescent dye electrostatically associated with a commercial silica particle. The quantum dots used were optimized from a previously reported microwave quantum dot synthesis to a quantum yield of 40%. Characterization of the silica particles showed that the surface properties resembled pure silica. These particles were able to be detected in vitro as well as in vivo after oral administration of nanoparticles to mice by gavage. After four daily administrations, nanoparticles were detected by fluorescence confocal microscopy in intestines as well as liver, kidney, spleen, lung, and brain. Thus, silica nanoparticles were able to traverse the intestinal epithelium. Further investigation is needed to determine nanoparticle accumulation and potential functional consequences throughout the body.

Silver nanoparticles were particularly toxic to proliferating (subconfluent) C2BBe1 cells plated at low density, inducing 15% necrosis and a 76% decrease in mitochondrial activity. Silver nanoparticle treatment induced oxidative stress in cells based on increased

GSH/GSSG ratios. In addition, silver nanoparticles induced G2/M phase cell cycle arrest and inhibited cell proliferation at doses forty times lower than those at which silica, titanium dioxide, and zinc oxide nanoparticles had inhibitory effects. Silver nanoparticles subjected to in vitro digestion before cell exposure required higher doses to induce toxicity, likely due to slower dissolution because of greater surface species adsorption. Silver

iii nanoparticles did not cause toxicity or oxidative stress in confluent (stationary) cells. Thus, upon ingestion, silver nanoparticles may be especially toxic to proliferating stem cells in intestinal crypts, particularly in disease states with a compromised epithelium.

Dedication

This document is dedicated to all consumers currently eating nanoparticles, particularly

those with gastrointestinal diseases. I share just a small piece of your pain.

Acknowledgments

There are many people who have significantly contributed to this work and I would be remiss if I did not take this opportunity to express my sincere appreciation. First and foremost, this dissertation would not exist without the mentorship of my advisor, Dr. Jim

Waldman. I would like to thank him for his invaluable patience with me and unfailing encouragement and advice. He is the kind of scientist I strive to be. I also want to thank my labmate and mentor, Debbie Knight, without whom my research and graduate career would not have been the same. I am greatly appreciative of the time she spent teaching me and then helping me with many lab procedures, serving as a sounding board, asking many questions, and keeping the lab running smoothly.

The entirety of this work was completed in close collaboration with Dr. Prabir

Dutta’s lab. Dr. Dutta’s mentorship has helped to shape me into the scientist I am today. I want to thank him for taking so much time to discuss data and experiments, teaching me how to approach writing a manuscript, and pushing me ever further. Many people from Dr.

Dutta’s group helped me throughout grad school. Dr. Andrew Zane was my partner in crime through most of this project. In addition to all of the nanoparticle synthesis, optimization, characterization, and digestion that he performed, which make up a good portion of my dissertation, I want to express my appreciation for the time he spent explaining chemical concepts to me, synthesizing and digesting particles for me vi

(sometimes at the last minute), and for continuing to be very responsive to my questions even after graduating. I am also grateful to Bo Wang for helping me find things in the lab, lending me centrifuge tubes, and taking TEM of my silver samples. I want to thank Dr.

Mike Severance for his help in characterizing particles, and all of the other Dutta group members for sharing their lab space with me from time to time.

Dr. Parinandi (“Pari”) was unfailingly generous with his lab space and reagents, as well as advice about data and suggestions of experiments that would further my research.

I am truly grateful for his investment in me. I would also like to thank Travis Gurney and

Dr. Sainath Kotha from Pari’s lab for advice and help with multiple assays as well as all past and present Parinandi lab members for graciously sharing their lab with me and especially for allowing me to use their plate reader.

I would sincerely like to thank my remaining committee member, Dr. Estelle

Cormet-Boyaka, for all of her experimental advice and thoughtful questions which added to the strength of my research. I greatly appreciate the time she spent investing in me as a scientist and her flexibility in scheduling committee meetings.

Many others played valuable roles in helping me to obtain and analyze data. I had the pleasure of getting to teach Mallory McMullen a little of the research I was doing in the lab one summer and I would like to thank her for her help in running experiments and for giving me the opportunity to briefly serve as a mentor. I would like to particularly thank

Ed Calomeni for his help in performing TEM on cells treated with nanoparticles and then explaining his images to me and graciously answering my questions. I would also like to thank Eric Jackson for his work on our cell counter and the Lehman and Byrd labs for

vii allowing me to use their cell counters when ours was out of commission. I would like to thank the Biomedical Sciences Graduate Program directors and staff over my time at Ohio

State, particularly Amy Lahmers, for their support and the opportunities they provided to further my education.

Finally, I need to thank my family and friends for their support and encouragement throughout grad school and particularly as I wrote my dissertation. I am especially appreciative of those in the Christian Graduate Student Alliance who best understand the joys and tribulations of graduate school. And most of all, I want to thank God for the incredible blessing of receiving this education from so many amazing people and for His constant steadfastness.

I also want to acknowledge funding from USDA /NIFA (2011-67021-30360) without which this research could not have taken place. Large portions of this research have been published in peer-reviewed scientific journals and reprinting of the material appears here with the permission of the editors of Chemical Research in Toxicology

(Chapter 3), The Journal of Physical Chemistry C (Chapter 4), and The International

Journal of Nanomedicine (Chapter 4).

viii

Vita

2006...... Peters Township High School

2010...... B.S. Biochemistry and Molecular Biology,

Penn State University

2010 to present ...... Graduate Research Associate, Biomedical

Sciences Graduate Program, The Ohio State

University

Publications

McCracken C., Zane A., Knight D.A., Hommel E., Dutta P.K., and Waldman W.J.

Oxidative stress-mediated inhibition of intestinal epithelial cell proliferation by

Ag nanoparticles. 2015. Toxicol. In Vitro (In Press)

Zane A., McCracken C., Knight D.A., Young T., Lutton A.D., Olesik J.W., Waldman

W.J., and Dutta P.K. Uptake of bright fluorophore core-silica shell nanoparticles

by biological systems. 2015. Int. J. Nanomed. 10:1547-1567.

Zane A., McCracken C., Knight D.A., Waldman W.J., and Dutta P.K. Spectroscopic

evaluation of the nucleation and growth for microwave-assisted CdSe/CdS/ZnS

quantum dot synthesis. 2014. J. Phys. Chem. C 118(38):22258-22267.

McCracken C., Zane A., Knight D.A., Dutta P.K., and Waldman W.J. Minimal intestinal

epithelial cell toxicity in response to short- and long-term food-relevant inorganic

nanoparticle exposure. 2013. Chem. Res. Toxicol. 26(10):1514-1525.

Fields of Study

Major Field: Biomedical Science

Area of Emphasis: Immunology

Table of Contents

Abstract ...... ii

Dedication ...... v

Acknowledgments...... vi

Vita ...... ix

List of Tables ...... xiv

List of Figures ...... xv

List of Abbreviations ...... xix

Chapter 1: Introduction ...... 1

1.1 Rationale...... 1

1.2 Hypothesis and Approach ...... 2

Chapter 2: Background and Literature Review ...... 5

2.1 Nanoparticles ...... 5

2.2 Nanoparticles in food ...... 7

2.3 Intestinal environment ...... 9

2.4 Nanoparticle characteristics important to toxicity ...... 34

2.5 Nanoparticle-induced oxidative stress ...... 44

2.6 Silica nanoparticles ...... 48

2.7 Titania nanoparticles ...... 55

2.8 Zinc oxide nanoparticles ...... 61

2.9 Silver nanoparticles ...... 67

2.10 Summary ...... 75

Chapter 3: Minimal intestinal epithelial cell toxicity in response to short- and long- term food-relevant inorganic nanoparticle exposure ...... 77

3.1 Introduction ...... 77

3.2 Experimental Procedures...... 79

3.3 Results ...... 91

3.4 Discussion ...... 115

Chapter 4: Spectroscopic evaluation of the nucleation and growth for microwave- assisted CdSe/CdS/ZnS quantum dot synthesis and uptake of bright fluorophore core- silica shell nanoparticles by biological systems...... 123

4.1 Introduction ...... 123

4.2 Experimental Procedures...... 128

4.3 Results ...... 136

4.4 Discussion ...... 160

xii

Chapter 5: Oxidative stress-mediated inhibition of intestinal epithelial cell proliferation by silver nanoparticles ...... 172

5.1 Introduction ...... 172

5.2 Experimental Procedures...... 175

5.3 Results ...... 188

5.4 Discussion ...... 216

Chapter 6: Conclusions and Future Directions ...... 228

6.1 Summary of Findings ...... 228

6.2 Importance of nanoparticle physicochemical characteristics ...... 229

6.3 Silver nanoparticle toxicity and disease ...... 232

6.4 Safety of nanoparticles for oral ingestion in foods ...... 234

6.5 In vivo studies ...... 237

6.6 Concluding Remarks ...... 238

References ...... 239

xiii

List of Tables

Table 2.1. Summary of literature showing an immune response to NP treatment in intestinal epithelial cells...... 23

Table 3.1. Hydrodynamic radii and zeta potential of nanoparticles...... 95

Table 5.1. Peaks corresponding to specific elements detected in XPS spectra of pristine and digested Ag NP...... 191

Table 5.2. Proliferating cell toxicity assay data...... 198

Table 5.3. GSH assay data...... 202

Table 5.4. Cell cycle data...... 204

Table 5.5. Confluent cell toxicity assay data...... 214

Table 6.1. Estimates of human nanoparticle exposure to intestinal epithelial cells to compare to doses used for in vitro studies...... 236

xiv

List of Figures

Figure 2.1. Scale of nanomaterials...... 6

Figure 2.2. Nanoparticle surface area...... 6

Figure 2.3. Intestinal epithelial structure...... 14

Figure 2.4. Junctions between intestinal epithelial cells...... 28

Figure 2.5. Nanoparticle characteristics important in determining behavior...... 35

Figure 2.6. Size-dependent endocytosis of nanoparticles...... 39

Figure 2.7. Oxidative stress responses corresponding to changes in redox couples...... 45

Figure 3.1. Representative TEM images of commercial nanoparticles...... 92

Figure 3.2. X-Ray diffraction patterns of commercial nanoparticles...... 94

Figure 3.3. Dissolution of and protein adsorption to ZnO nanoparticles...... 96

Figure 3.4. Difference infrared spectra of TiO2 and SiO2 in cell culture media...... 97

Figure 3.5. Difference infrared spectra of digested SiO2 and TiO2 nanoparticles...... 98

Figure 3.6. DRIFTS spectra of freeze-dried digestive enzyme solutions...... 99

Figure 3.7. Cell morphology after nanoparticle treatment...... 100

Figure 3.8. Tight junction structure of TiO2-treated cells...... 101

Figure 3.9. Adherens junction structure of TiO2-treated cells...... 102

Figure 3.10. Representative TEM images of nanoparticle-treated C2BBe1 cells...... 103

Figure 3.11. Results of cytotoxicity assays for C2BBe1 cells treated with nanoparticles in media for 24 hours...... 105

Figure 3.12. Cytotoxicity of C2BBe1 cells after treatment with SiO2 and TiO2 nanoparticles exposed to simulated digestive solutions...... 107

Figure 3.13. Interference of digestive solutions with LDH assay...... 108

Figure 3.14. ELISA assay to measure IL-8 secretion induced by nanoparticles...... 109

Figure 3.15. Cytotoxicity after repeated nanoparticle exposure...... 111

Figure 3.16. Nanoparticle toxicity after 84 repeated exposures...... 112

Figure 3.17. Growth curves after 11 nanoparticle exposure cycles...... 113

Figure 3.18. Growth curves after 29 nanoparticle exposure cycles...... 114

Figure 3.19. Summary of the study...... 121

Figure 4.1. QD synthesis and optimization strategy...... 137

Figure 4.2. Biological imaging of QDs...... 140

Figure 4.3. Representative confocal fluorescence images comparing 19% and 40% QY

QDs...... 141

Figure 4.4. Dye/Silica synthetic pathway...... 143

Figure 4.5. QD/Silica synthetic pathway (arginine-driven synthesis)...... 143

Figure 4.6. TEM images and fluorescent properties of dye/silica particles...... 144

Figure 4.7. Representative TEM images of QD/silica particles...... 146

Figure 4.8. Quenching of QD/silica particles in PBS...... 147

Figure 4.9. Zeta potential titration of silica cores, Rhodamine 6G/silica, Rhodamine

800/silica, and QD/silica...... 147

xvi

Figure 4.10. Optical spectra of fluorescent core/shell silica particles...... 148

Figure 4.11. Infrared and 29Si nuclear magnetic resonance spectroscopy of arginine- driven Rhodamine 6G and QD/silica...... 150

Figure 4.12. QD and Rhodamine 6G/silica particle DRIFTS spectra...... 151

Figure 4.13. QD/silica and Rhodamine 6G/silica nanoparticles in macrophages...... 152

Figure 4.14. QD/silica and Rhodamine 6G/silica in intestinal epithelial cells...... 153

Figure 4.15. Rhodamine 6G/silica in mouse tissues...... 154

Figure 4.16. Rhodamine 6G/silica particles in small intestine tissue...... 156

Figure 4.17. Rhodamine 6G/silica particles in cecum tissue...... 157

Figure 4.18. Rhodamine 6G/silica particles in colon tissue...... 158

Figure 4.19. QD/silica particle detection in mouse tissue...... 159

Figure 4.20. Synthetic pathway of CdSe/CdS/ZnS microwave-assisted synthesis...... 162

Figure 4.21. Potential pathways of QD/silica coating...... 167

Figure 5.1. Characterization of pristine and digested Ag NP...... 189

Figure 5.2. Zeta potential of pristine and digested Ag NP...... 190

Figure 5.3. XPS spectra of pristine and digested Ag (BE = binding energy)...... 191

Figure 5.4. XPS spectra for both pristine and digested Ag NP corresponding to individual elements...... 192

Figure 5.5. Ag NP dissolution in cell culture media...... 194

Figure 5.6. Internalization of Ag NP by intestinal epithelial cells in vitro...... 195

Figure 5.7. Toxicity of Ag NP to proliferating cells...... 197

Figure 5.8. Interference of Ag NP with the LDH assay...... 198

xvii

Figure 5.9. Time course of GSH/GSSG ratios after treatment of cells with Ag NP...... 200

Figure 5.10. GSH/GSSG after 24-hour Ag NP treatment...... 201

Figure 5.11. Cell cycle phase distribution (DNA content) of Ag NP-treated C2BBe1 cells.

...... 203

Figure 5.12. Inhibition of intestinal epithelial cell proliferation by Ag NP...... 205

Figure 5.13. Sensitivity to Ag NP-induced inhibition of proliferation depends on cell passage...... 206

Figure 5.14. Inhibition of intestinal epithelial cell proliferation by food-relevant NP. .. 207

Figure 5.15. Recovery of SiO2, TiO2, and ZnO-treated cell populations after NP removal.

...... 208

Figure 5.16. Growth of cells treated with 40, 200, and 2000 nm zeolite particles...... 208

Figure 5.17. N-acetylcysteine protection from Ag NP-mediated inhibition of cell proliferation and induction of oxidative stress...... 210

Figure 5.18. Cell growth and GSH/GSSG ratios after pretreatment of cells with NAC prior to Ag NP exposure...... 211

Figure 5.19. Impact of Trolox on Ag NP-mediated inhibition of cell proliferation...... 212

Figure 5.20. Acute toxicity in intestinal epithelial cells induced by Ag NP...... 213

Figure 5.21. Toxicity of 24- and 48-hour exposure of Ag NP to confluent cells...... 215

Figure 5.22. IL-8 secretion induced by Ag NP treatment of confluent cells...... 216

Figure 5.23. Mechanism of Ag NP toxicity...... 224

xviii

List of Abbreviations

Ag Silver

ALT Alanine aminotransferase

AMPs Antimicrobial peptides

AST Aspartate transaminase

ATP Adenosine triphosphate

BE Binding energy

BrdU Bromodeoxyuridine

BSA Bovine serum albumin

CP-MAS Cross-polarization magic angle spinning

Cys Cysteine

CySS Cystine

DAPI 4’,6’-diamidino-2-phenylindole

DMEM Dulbecco’s modified Eagle medium

DRIFTS Diffuse reflectance infrared Fourier transform spectroscopy

E-cadherin Epithelial cadherin

ELISA Enzyme-linked immunosorbent assay

xix

FAE Follicle-associated epithelium

FBS Fetal bovine serum

FFA Free fatty acid

FITC Fluorescein isothiocyanate

G0 phase Gap 0 phase

G1 phase Gap 1 phase

G2 phase Gap 2 phase

GI Gastrointestinal

GSH Glutathione

GSSG Glutathione disulfide (oxidized glutathione)

HBSS Hank’s balanced salt solution

HO-1 Heme oxygenase-1

HRTEM High resolution transmission electron microscopy

IBD Inflammatory bowel disease

ICP-MS Inductively coupled plasma mass spectrometry

IEP Isoelectric point

IFN-γ Interferon-γ

IgA Immunoglobulin A

IL Interleukin

IR Infrared

JAM Junctional adhesion molecule

LDH Lactate dehydrogenase

M phase Mitosis phase

MPA 3-mercaptopropionic acid

MUC2 Mucin 2

NAC N-acetylcysteine

NMR Nuclear magnetic resonance

NP Nanoparticles

PA Palmitic acid

PBS Phosphate-buffered saline

PLGA Polylactic polyglycolic acid

PVP Polyvinylpyrrolidine

QD Quantum dot

QY Quantum yield

ROS Reactive oxygen species

S phase Synthesis phase

SAS Synthetic amorphous silica

SERS Surface enhanced Raman spectroscopy

SiO2 Silicon dioxide, silica

SOD Superoxide dismutase

TCSPC Time-correlated single photon counting

TEER Transepithelial electrical resistance

TEM Transmission electron microscopy

TEOS Tetraethyl orthosilicate

xxi

TGF-β Transforming growth factor β

TH1 T helper type 1

TH2 T helper type 2

TiO2 Titanium dioxide

TMPyP meso-Tetra(N-methyl-4-pyridyl)porphine tetratosylate salt

TNF-α Tumor necrosis factor α

TrSS Oxidized disulfide thioredoxin

Trx Thioredoxin

TSLP Thymic stromal lymphopoietin

XPS X-ray photoelectron spectroscopy

XRD X-ray diffraction

ZnO Zinc oxide

ZO Zonula occludens

ZOT Zonula occludens toxin

xxii

Chapter 1: Introduction

1.1 Rationale

Nanotechnology is an exponentially growing field. The food industry has been increasingly interested in utilizing nanotechnology in both food and food packaging.

Although many of the potential nanoparticle (NP) applications have not yet been incorporated into foods, it has been shown that food-grade silica (SiO2) and titanium dioxide (TiO2) have broad particle size distributions in which approximately one-third of the particles are in the NP size range.1, 2 Silica is used in foods as an anticaking agent and

TiO2 as a whitening agent. It has been estimated that the daily intake for an adult is 0.2-0.7

2 mg TiO2/kg body weight/day. Silver (Ag) NP have applications as antimicrobial agents and are of particular interest to the food packaging industry. Ag has been incorporated into some food storage containers and used as antimicrobial coatings on utensils, cookware, and appliances,3 and there is the potential for Ag NP in food contact materials to migrate into food. From these applications, it is clear that humans are ingesting NP of SiO2 and

TiO2 and it is likely that there will be increasing ingestion of other NP including Ag.

The small size of NP vastly increases their surface area per unit mass and thus the atoms available on the particle surface to react. This can increase reactivity of a material and change its physical and chemical properties which may alter how it interacts with cells in the body. Due to their small size, NP have been shown to be internalized by cells better 1 than larger particles.4, 5 Thus, in the body, NP will likely interact more with cells and this may increase their toxic potential over larger particles of the same materials. Thus, even if these materials have been used in foods and determined to be safe at larger sizes, it is necessary to ensure that these materials are safe in the nanoparticulate form.

Research on food-relevant NP including SiO2, TiO2, ZnO, and Ag to date has indicated that certain NP are able to cause toxicity, particularly in inhalation models.

Studies conducted with these NP in intestinal models both in vitro and in vivo to represent oral NP exposure have shown that these NP can induce toxicity, particularly ZnO and Ag.

However, based on discrepancies in the literature, it is still incompletely understood how the physicochemical characteristics of these NP contribute to their cytotoxicity.

1.2 Hypothesis and Approach

In this dissertation work, we further investigated the toxicity of the food-relevant

NP SiO2, TiO2, ZnO, and Ag in intestinal models to test the hypothesis that NP characteristics including surface chemistry and particle type would determine the toxicity of these NP.

The experiments performed in Chapter 3 investigated the toxicity of SiO2, TiO2, and ZnO NP in vitro in the intestinal epithelial C2BBe1 cells. NP were subjected to in vitro digestion in enzymatic solutions representing the stomach and intestinal environments in order to determine the effects of altered surface chemistry that occurs during the digestive process in vivo. In this investigation, digested and nondigested NP were characterized by transmission electron microscopy (TEM), dynamic light scattering, zeta potential measurements, X-ray diffraction, and diffuse reflectance infrared Fourier transform

2 spectroscopy (DRIFTS) prior to use in experiments. Acute toxicity after treatment of cells with NP for 24 hours was determined using Sytox Red and Annexin V FITC staining and fluorescence detection by flow cytometry to determine cell necrosis and apoptosis, respectively. The LDH assay was also used to determine lactate dehydrogenase release due to cell membrane damage as well as the MTT assay to determine mitochondrial activity of cells. The same toxicity assays were used to evaluate toxicity after repeated NP exposure to cells in addition to experiments measuring cell growth.

Synthesis and application of food-relevant silica NP with fluorescent cores are detailed in Chapter 4. The previously reported microwave synthesis of quantum dots (QDs) was optimized using real-time monitoring of QD nucleation and growth, and by light illumination after QD formation. Both QD and silica NP bound to rhodamine 6G or rhodamine 800 were used as cores on which silica shells were formed by hydrolysis and condensation of TEOS. The resulting particles were characterized using TEM, zeta potential measurements, UV/Vis absorption and fluorescence spectra, DRIFTS analysis, and nuclear magnetic resonance spectra. These particles were visualized by confocal fluorescence microscopy both in vitro in intestinal epithelial cells and mouse alveolar macrophages and in vivo using tissue sections from SKH1-E mice orally administered the fluorescent silica NP.

Chapter 5 contains experiments used to determine the mechanism of Ag toxicity to both confluent (stationary phase) and proliferating C2BBe1 cells. Ag NP were synthesized by reduction of Ag ions with borohydride. The Ag NP were also subjected to in vitro digestion by sequential incubation in solutions representing stomach and intestines. Both

3 pristine and digested Ag NP were characterized by TEM, DRIFTS, Raman, and X-ray photoelectron spectroscopy prior to biological experiments. Acute toxicity after 24-hour exposure to Ag NP was determined by Sytox Red (necrosis), Annexin V (apoptosis), LDH

(cell membrane damage), and MTT (mitochondrial activity) assays. Ratios of reduced to oxidized glutathione (GSH/GSSG) were measured to indicate redox potential and oxidative stress in cells. Cell cycle analysis and growth curves were performed after continual Ag

NP exposure for up to 10 days in order to determine cell cycle progression and cell growth, respectively. Cells were treated with the antioxidants N-acetylcysteine and Trolox in order to further elucidate the mechanism of Ag NP toxicity.

Chapter 2: Background and Literature Review

2.1 Nanoparticles

Despite the recent explosion of nanoparticle (NP) research, synthesis, and use, NP are not a recent innovation. A NP is defined as any particle with at least one dimension less than 100 nm, and thus both organic and inorganic NP exist naturally. Naturally occurring organic NP include proteins and some viruses (Figure 2.1). It has also been discovered that people have used NP for their optical properties dating back to Roman times. Copper and gold NP were used by Romans to color glass.6 Beginning in the Middle Ages, glazes used to make lusterware contained copper and silver (Ag) NP for their iridescent properties.7, 8

Michael Faraday’s experiments on the interaction of light with gold NP dispersions are considered the beginning of modern colloid science and thus nanoscience.9

Today there is interest in the use of nanotechnology because of the unique properties of nanosized materials. For instance, at the nanoscale, gold is red rather than yellow. Nanosized gold can also serve as a catalyst.10 As illustrated in Figure 2.2, NP have a greatly increased surface area per unit mass than the bulk material, which means that there are many more exposed atoms available to interact.11 These atoms on the surface of a particle are less stable than interior atoms and thus are more reactive,10 so NP can have different reactivity than larger particles of the same material. The quantum confinement effect also contributes to the unique electronic and optical properties of NP. Confinement 5

Figure 2.1. Scale of nanomaterials. This scale places the nanosize range (1-100 nm) into the size of some well-known materials and shows that NP are in the same size range as proteins and viruses. At the bottom of the image is pictured some commonly-used manmade NP.12

Figure 2.2. Nanoparticle surface area. This illustration depicts the vastly increased surface area of NP over the same mass of larger particles of the same material.11

6 of electrons within a particle leads to size-dependent behavior. Changing the size of particles, and thus changing the band gaps, allows for great tunability for particles such as quantum dots, which fluoresce differently based on size.10

2.2 Nanoparticles in food

Nanotechnology is currently of interest to both the food and food packaging industries. NP applications are being considered to enhance flavor and texture, improve nutrient bioavailability, control flavor, color, or nutrient release in consumer-activated products, and reduce the fat needed to create the same taste, texture, and consistency.13 The technologies that make these applications possible include nano-emulsions, micelles, and nanoencapsulation. In food packaging, nanomaterials are of interest to improve the stability, flexibility, and gas barrier properties of packaging, to actively utilize antimicrobial or oxygen scavenging NP to keep food fresh longer, to add nanosensors which can detect and respond to the freshness of the package contents (i.e. color change), and to make packaging biodegradable. Some NP of interest for these applications include silver, zinc oxide (ZnO), and magnesium oxide (MgO) for their antimicrobial properties, titanium dioxide (TiO2) for its UV absorption properties, and nanoclay which can limit gas permeation.13

Although there is great interest by the food industry in utilizing nanotechnology, many of the aforementioned products are not yet on the market. The Project on Emerging

Nanotechnologies has attempted to inventory consumer products using NP.3 This inventory currently identifies 117 food and beverage products incorporating nanotechnology

7 including food, food storage products, supplements, and products used for cooking.3 One of the most common NP used in these products is Ag for its antimicrobial properties.

It has also been shown that the commonly-used food-grade silica and titanium dioxide contain NP. TiO2 is often used as a whitening agent. A 2012 study analyzed 89 consumer food products and found titanium in many food products, especially sweet foods such as candies, gums, icing, and powdered sugars.2 They also showed that although food- grade TiO2 particles, E171, have an average size of 110 nm, the size distribution is broad and 36% of these particles were less than 100 nm and thus can be classified as NP. They estimated a daily exposure to TiO2 in the US to be 0.2-0.7 mg TiO2/kg body weight/day for adults and 1-2 mg TiO2/kg body weight/day for children under the age of 10 due to greater consumption of sweets, and 36% of this TiO2 would be nanosized. A study that isolated TiO2 from chewing gum found that >93% of TiO2 particles in the gum were less than 200 nm and 18-44% were less than 100 nm. It was also shown that approximately

95% of the TiO2 in the gum is swallowed when it is chewed, so consumers are being

14 exposed to much of this TiO2. Food-grade silica, E551, is used as an anti-caking agent and was detected in many powdered food mixes.1 Their analysis indicated that between 4-

43% of the silica was less than 200 nm depending on the food product. This study estimated exposure to nanosilica to be 1.8 mg/kg body weight/day for the average adult.1

An analysis of 86 wheat breads and 49 wheat biscuits from 14 different countries

(the majority were from Italy) by Environmental Scanning Electron Microscopy to detect inorganic micro- and nano-scaled contaminants revealed that 40% of the samples contained environmental or industrial ceramic and metallic contaminants. The major elemental

8 components of the microparticulate and nanoparticulate contaminants were iron, lead, tungsten, titanium, aluminum, silicon, and silver.15 This study suggests that consumers may be exposed to NP in foods arising as environmental contaminants even when they are not being intentionally added to food products.

2.3 Intestinal environment

Digestion

Foreign substances are constantly ingested in food. The digestive tract creates a barrier that prevents most unwanted food components from penetrating further into the body. However, the main purpose of the digestive tract is to break food into its more basic components, specifically sugars, lipids, amino acids, vitamins, minerals, etc., which can then be absorbed and used by our bodies. Mastication of food in the mouth and the mixture of food with saliva begins enzymatic digestion. Lingual lipase digests lipids and α-amylase digests starches.16 Enzymatic digestion is then continued in the stomach, largely by pepsin, which digests proteins and is active in the acidic environment ranging from pH 1 to pH 5 depending on whether the stomach is empty or full. The acidity also denatures proteins and allows them to partially unfold, contributing to the protein digestion process, and additionally helps to sterilize the ingested contents. Peristalsis adds to digestion by mechanically mixing stomach contents. Gastric lipase continues the breakdown of fats.16

After 1-2 hours in the stomach, the resulting chyme is released into the duodenum where it mixes with pancreatic digestive enzymes, bicarbonate, and bile from the liver, which neutralize the acidic chyme.17 Periodic peristalsis pushes the chyme through the intestines as it digests further. Trypsin and chymotrypsin as well as other proteases

9 produced by the pancreas continue to digest proteins to form oligopeptides which are broken into amino acids by exopeptidases.16 Bile salts form micelles with hydrophobic cores where lipids and solubilized fats can be digested by pancreatic lipase and other enzymes. Pancreatic α-amylase continues carbohydrate digestion.16 Lactase, β- glucoamylase, isomaltase, trehalase, and sucrase found in the brush borders of intestinal epithelial cells further digest carbohydrates to monosaccharides.16 DNase and RNase from the pancreas break down DNA and RNA into mononucleotides. Once these nutrients are digested to primary units, they are absorbed through the intestine and enter the bloodstream for use by the body.

Most nutrient absorption takes place in the jejunum (middle region) of the small intestine by intestinal enterocytes. Specific transport proteins such as the Na+-glucose transporter facilitate transport of common nutrients across the plasma membrane.16 Lipids are absorbed through diffusion across the cellular membrane or temporary incorporation into the cell membrane. After absorption, nutrients are then transported out of the basolateral side of the enterocytes so that they can enter the bloodstream and be distributed throughout the body. An adenosine triphosphate (ATP)-driven Na+/K+ exchanger located on the basolateral membrane transports Na+ out of the cell to create an electrochemical gradient conducive to Na+ entry, which is necessary for sodium-dependent transport of many nutrients, including glucose, into cells.18

In vitro models of digestion

Due to the complexity of the digestive process, in vitro experiments can provide a simpler system to address questions of NP toxicity. In order to make in vitro models more

10 closely resemble in vivo ingestion of NP, some researchers have developed in vitro systems to mimic digestion. These systems usually consist of incubation of NP in solutions containing some combination of enzymes and ionic concentrations resembling the mouth, stomach, and intestines. Digestion of the NP is considered to be important because the low pH environment of the stomach and interaction with digestive enzymes and molecules may change the surface properties of the NP and thus change how they interact with cells.

Various researchers have developed in vitro digestion models and thus there is variation in what treatment has been considered “digestion” among the studies that have been performed to date. Despite this variation, studies have shown that digestion may affect nanoparticle dissolution and aggregation and may alter NP toxicity. A recent study that added silica NP to sample food matrices and subjected these NP to an in vitro digestion procedure showed that both the food matrix and the stage of digestion will likely play large roles in the bioavailability of silica NP. Their digestion consisted of saliva (pH 6.8), gastric juice (pH 1.3), duodenal juice (pH 8.1), and bile juice (pH 8.2) components, all of which were composed of many representative salts, sugars, enzymes, mucins, etc., that make up these digestive tract environments. They added the commercially-used food-grade E551 silica to model food matrices (hot coffee, instant soup, and pancake). Before digestion,

30% of E551 in coffee was nano-sized (5-200 nm) while 13% and 5% were nano-sized in soup and pancakes, respectively. Large agglomeration of silica was observed in the gastric digestion stage due to the low pH (the isoelectric point where silica is neutrally charged occurs at pH 2-3) and the high electrolyte concentration. However, after the full digestion,

80% of E551 in coffee was in the nano-size range (5-200 nm) while approximately 15% of

E551 added to soup and pancake was nano-sized.19 Despite the differences in available NP depending on the food matrix, this study did detect some E551 available as NP in the intestinal environment in all foods tested.

Several studies have analyzed NP toxicity after in vitro digestion and found it to be slightly different than that of NP that did not undergo digestion. One study simulated gastric and intestinal digestion of 14 nm silica and < 10 nm ZnO NP in solutions consisting of representative ionic and pH conditions but no digestive enzymes. They compared NP which were not subjected to this digestion to the digested NP in Caco-2 intestinal epithelial cells. Analysis of the NP suspensions before and after digestion revealed that less than 10% of both silica and ZnO NP dissolved and that soluble fraction changed very little after digestion. Both silica and ZnO NP were found to be taken up by cells and induced reactive oxygen species (ROS) formation in cells, particularly at doses of 80 µg/cm2, but simulated digestion of the silica NP largely inhibited ROS production. However, digested NP induced similar toxicity to cells based on the reduction of the tetrazolium salt WST-1 by cellular enzymes (at doses of 5-20 µg/cm2) and were able to induce similar induction of interleukin-

8 (IL-8) secretion at 20 µg/cm2 as the undigested NP.20 Thus, digestion of NP seems to be able to change NP toxicity mildly, but not inhibit its toxicity altogether. Another study subjected 7 nm Ag NP to an in vitro digestion in saliva for 5 minutes at pH 6.4, gastric juice for 2 hours at pH 2, and intestinal juice for 2 hours at pH 7.5. These digestive solutions contained ionic components as well as mucins, enzymes, salts, and other digestive components. They observed a slight reduction (requiring 10 µg/mL vs. 5 µg/mL) and 12-

24 hour delay in Ag NP cytotoxicity after the digestion as detected by impedance

12 measurements, but this difference was not observed using the CellTiter Blue assay.21 They suggest this is due to an increase in particle aggregation which hinders Ag ion release, once again showing that digestion is able to alter NP toxicity, but only slightly. Based on these studies, it is clear that both the food matrix and the digestive process could contribute to the toxicity of a given NP.

Cells of the intestinal epithelium

Because of the importance of nutrient absorption within the small intestine, intestinal architecture is specialized for maximal surface area. The epithelial layer is composed of villous projections and crypts of Lieberkühn between villi (Figure 2.3). The villi are vascularized for quick delivery of nutrients to the bloodstream.18 Near the base of intestinal crypts reside pluripotent intestinal epithelial stem cells. These cells are self- renewing and constantly proliferate to repopulate the intestinal epithelium, which turns over every 3-5 days.18 Undifferentiated crypt cells created from stem cell replication make up the majority of the crypt and have the ability to differentiate into all small intestinal epithelial cell types including enterocytes, goblet cells, and enteroendocrine cells. These undifferentiated cells have minimal absorption capabilities but play an important role in the intestinal immune response by producing an immunoglobulin A (IgA) receptor and transporting the IgA receptor from the basolateral to apical side of the cell after IgA binding. The complex is then released from the lumen and this IgA helps to control the immune response to pathogenic as well as nonpathogenic microbiota.18 As these cells differentiate, they migrate up the crypt-villus axis, eventually undergoing apoptosis and being shed from the tips of villi into the lumen.

Villus

Microvilli

Crypt

Figure 2.3. Intestinal epithelial structure. Intestinal epithelial stem cells (IESCs) in intestinal crypts constantly proliferate to renew the epithelium. Undifferentiated crypt cells differentiate and migrate up the crypt-villus axis until they undergo apoptosis and are shed from villus tips. The majority of the epithelium is composed of enterocytes which mainly function to absorb nutrients from food. Paneth cells in the small intestinal epithelial crypts secrete antimicrobial peptides (AMPs) to protect the epithelium from bacterial invasion. Goblet cells secrete mucins which form the mucus coating that helps protect the epithelium from bacterial contact. The small intestine has one relatively thin loosely adherent layer of mucus to promote absorption while the colon has a thicker mucus covering composed of an underlying firmly adherent layer covered by a loosely adherent layer which bacteria are better able to penetrate.22, 23 Enteroendocrine cells in the small intestine secrete hormones and other products that help to regulate digestion. M cells within the follicle-associated epithelium (FAE) connect the intestines to the underlying Peyer’s patch (small intestine) or lymphoid follicles (colon) and function to sample luminal contents and present antigens to the underlying immune cells. The colon is composed of similar enterocytes and goblet cells to those in the small intestine, but does not contain Paneth cells or enteroendocrine cells. Macrophages and dendritic cells reside in the lamina propria underlying the epithelium as well as in lymphoid follicles and can directly sample bacteria in the lumen as well as respond to bacteria transported across the epithelium.24 These macrophages and dendritic cells will present antigens to lymphoid cells which can then be activated to aid in inflammatory or tolerogenic responses. Plasma cells secrete IgA released into the lumen which helps to regulate the microbiota and contributes to intestinal homeostasis.25 14

The majority of the intestinal epithelium is composed of enterocytes. These cells are largely responsible for nutrient absorption and are covered in microvilli to further increase the absorptive surface of the intestine (Figure 2.3). The structure of these microvilli are maintained by bundles of actin filaments and hydrolyases including sucrase- isomaltase, lactase, and dipeptidyl-peptidase IV, which are found on their membranes to aid in digestion (as mentioned previously). Vesicles containing some of these membrane proteins are released from the microvilli into the lumen. These vesicles also contain intestinal alkaline phosphatase which can detoxify lipopolysaccharide and protects the epithelium from bacterial transcytosis.26

The secretory cells of the intestine contribute to the barrier function of the epithelium. These include goblet cells, Paneth cells, enteroendocrine cells, and M cells.

Paneth cells and enteroendocrine cells are found in the small intestine but not the colon while goblet cells and M cells are found in both epithelia. Enteroendocrine cells produce hormones, enzymes, and other products important for regulation of digestion.25 These products are stored in granules that are released from the basolateral side of the cells.18

Paneth cells reside at the base of intestinal crypts and secrete antimicrobial peptides

(AMPs) such as defensins, cathelicidins, and lysozyme.25 These AMPs protect the epithelium and particularly the stem cells (due to their proximity) from bacterial damage and penetration of bacteria across the epithelium. Enterocytes are also able to secrete some

AMPs.

Goblet cells are the second major cell type of the intestinal epithelium and contain mucus granules which are released into the lumen to create the epithelial mucus lining

(refer to Figure 2.3).18 The stomach and intestines are heavily coated with mucus to protect cells and also to serve as a barrier to keep food contents and bacteria from penetrating the epithelium. The mucus produced by goblet cells forms a layer that varies in thickness throughout the gastrointestinal (GI) tract. Within the intestine, mucin 2 (MUC2) is the major component of the mucus. MUC2 mucins are heavily glycosylated within cells, making them resistant to proteases and able to bind water. MUC2 monomers polymerize into net-like polymers which, upon secretion, form sheets that lead to stratified mucus layers.22 There is generally a firmly adherent layer of mucus just above the cells adjacent to a thicker, more loosely adherent layer that is constantly being turned over. This loose layer contains some bacteria, but generally bacteria do not penetrate the firmly adherent layer.22 However, there is a relatively thin layer of firmly adherent mucus in the small intestine in comparison to that found in the stomach and colon.23 This aids in nutrient absorption but also allows for greater potential penetration through the mucus. The mucus layer is renewed over a period of several hours by more mucin secretion from goblet cells of the epithelium.22 Small intestinal goblet cells have also been shown to transport soluble luminal antigens across the epithelium to present to dendritic cells, thus contributing to the immune response in the intestine.25

In the small intestine and colon, the epithelial layer also includes a certain fraction of M cells.27 Membranous or microfold (M) cells are relatively infrequent cells that connect the gut lumen to the underlying Peyer’s patch (small intestine) or lymphoid follicle (colon).

Their main function is to sample gut luminal contents and present antigens to the underlying immune cells, thus connecting the intestinal lumen to the immune system.25, 28

In addition to random sampling of luminal contents, M cells also express surface receptors for specific binding and transport of certain bacterial proteins.25, 29 Dendritic cells and macrophages can take up antigens transported across the M cells and can then present those antigens to lymphocytes (T cells and B cells) of the Peyer’s patch.28 The epithelium overlying lymphoid follicles of the intestines is referred to as follicle-associated epithelium

(FAE) and differs from the rest of the intestinal epithelium in that there is weak production of mucus, weak expression of digestive enzymes, and a large number of infiltrating immune cells.30 M cells make up approximately 10% of the FAE in humans.29 Factors expressed by B cells seem to stimulate M cell differentiation overlying lymphoid follicles.29

In vitro intestinal epithelial models

Several continuous cell lines are commonly used to study the intestinal epithelium, including Caco-2, HT-29, and T84 cells. If cultured under appropriate conditions, these cells form tight junctions, produce mucus, form microvilli on the apical side of the cells, and generally exhibit the characteristics of in vivo intestinal epithelial cells.31-33 Caco-2 cells were isolated from a patient with colon adenocarcinoma and observed to form a brush border of microvilli at confluency that expressed some small intestinal hydrolases including sucrase-isomaltase, alkalkine phosphatase, and aminopeptidase.34 However, this cell line is made up of morphologically heterogeneous cells, not all of which formed the brush border structure. Thus, Caco-2 cells were cloned for brush border-expressing cells and clone 1 formed the C2BBe1 cell line still used today.35 These cells have been extensively characterized and used to study brush border formation. When grown on filter

17 supports, confluent C2BBe1 cell monolayers become polarized with distinct apical and basolateral sides. The brush borders of confluent C2BBe1 cells are very similar to those in vivo and express many of the same microvillar proteins including villin, fimbrin, sucrase- isomaltase, myosin I, fodrin, and myosin II.35, 36 A recent study compared gene expression of xenobiotic-metabolizing enzymes, transporters, and nuclear receptors and transcription factors, important for metabolism of xenobiotics, between human intestinal biopsy samples and Caco-2, C2BBe1, HT-29, T84, and FHC cells. They found significant differences between the cell lines and biopsy samples, but T84 and HT-29 cells were the most similar to the intestinal biopsies with correlation coefficients to the ascending colon of 0.783 and

0.747, respectively. Caco-2 and C2BBe1 cell lines had correlation coefficients of 0.711 and 0.715 to the ascending colon, respectively. All cell lines were more closely correlated to colon than ileum samples.37 From this study, it is important to keep in mind that although these cell lines do have many similarities to the human intestine, they will not always behave the same as in vivo intestinal epithelium.

One major drawback of these in vitro models is that they do not contain the full diversity of epithelial cell types found in vivo. Although there is some mucus production by these cell lines, the mucus layer is often patchy and not as dense as you would find in vivo. Also, epithelial permeability may be underestimated in cell models without M cells even though they make up only a small proportion of the epithelium. To address some of these issues, more sophisticated models have been developed to better represent the full complement of epithelial cells. The HT-29 epithelial cell line was differentiated into Goblet cells that produce mucus by culturing cells with methotrexate.38 These cells are referred to

18 as HT-29 MTX cells. To model the protective mucus layer in vitro, some studies with NP have been performed in Caco-2/HT-29 MTX cocultures, particularly studies examining NP bioavailability since mucus is an important barrier preventing NP from reaching the epithelium.39

It has been shown that M cells can be differentiated by culturing Caco-2 cells with

Raji B cells.40 If these cells are physically separated using a transwell system, the epithelial layer can be maintained and provide a reasonable model of the intestinal epithelium where many of the epithelial cells will have differentiated into M cells.40 A similar model was also established with co-cultured Raji cells and C2BBe1 cells.41 This M cell model has the added benefit of incorporating immune cells which may be important in epithelial responses. Studies using these M cell co-culture models have shown increased transport of

NP across the epithelium than what is seen with Caco-2 monolayers due to the function of

M cells to sample luminal contents. For instance, a co-culture model of M cells was able to transport five times as many chitosan-DNA NP across the epithelial layer than Caco-2 cells alone.42

These models can be combined to provide an even more realistic in vitro intestinal epithelial model as was done in a study that used a tri-culture of Caco-2 and HT29-MTX cells cultured above Raji B cells to create monolayers containing M cells and Goblet cells.43

This study was investigating the effects of polystyrene NP on iron absorption and found that treatment of cells with 50 or 200 nm NP at doses of 2×1013 50 nm particles/mL or

1.25×1012 200 nm particles/mL (estimated as equivalent to potential pharmaceutical doses) increased iron uptake and transport across the epithelial layer in this tri-culture model. This

19 was accompanied by decreased transepithelial electrical resistance (TEER) values suggesting disruption of the epithelial barrier which is likely allowing the increased iron transport. This increase in iron transport was confirmed in vivo in broiler chickens fed polystyrene NP for 2 weeks, indicating that this in vitro model may be representative for what will occur in vivo.43 Thus, these models provide a way to more closely represent relevant features of the intestinal epithelium in an in vitro system.

Mucosal immune system of the gut

Immune cells such as lymphocytes, dendritic cells, and macrophages are found in the lamina propria beneath the intestinal epithelium in addition to those that reside within lymphoid follicles. These cells can respond to antigens transported across the epithelium.

There is also evidence that dendritic cells of the lamina propria extend processes between cells of the epithelium to sample antigens in the lumen while preserving the tight junctions between the cells.24 Luminal IgA plays a large regulatory and defensive role in the intestinal immune system. Activated plasma cells are induced to class switch and produce

IgA through signaling from the cytokine milieu and particularly transforming growth factor

β (TGF-β). T helper cells can also aid B cells in IgA production. The IgA secreted by these plasma cells is transported into the lumen by epithelial cells as previously discussed. The so-called “immune exclusion” role of IgA prevents pathogens and other antigens from gaining access to the surface of the epithelium. Nonspecific IgA can bind to glycans on the bacterial surface which allows for cross-linking and agglutination of bacteria. IgA can also cause microbes to get stuck in the mucus because of interactions between the secretory component of the IgA and the mucus. Binding of IgA to pathogens or antigens can prevent

20 their interaction with epithelial cells by steric hindrance or by direct IgA binding to cellular receptor-binding domains. It has also been shown that IgA can sensitize certain bacteria to the oxidative burst response.44, 45 M cells bind IgA and internalize IgA-immune complexes, and thus IgA helps to present antigens to the mucosal immune system. In addition to its defense against pathogens, microbe-specific IgA is the main adaptive response used to control the microbiota. IgA-coated bacteria have been found to downregulate proinflammatory cytokines which suggests that IgA-antigen complexes may be important in establishing and maintaining tolerance. Regulatory T cells are also necessary to prevent excessive immune responses and promote tolerance of the microbiota.44, 45 IgA secretion, along with the outlined barrier functions of the various epithelial cells, including the mucus layer and AMP production, together promote intestinal homeostasis with the microbiota to keep this population in check but not mount a more extensive immune response to these microbes.45

Since these barriers do not prevent all bacteria from reaching the intestinal epithelial surface, epithelial cells also express pattern-recognition receptors that allow them to detect any microbes able to bind to their surface. Signaling through these receptors contributes to intestinal homeostasis as well as repair of epithelial damage.25 When these cells recognize pathogenic bacteria or in pro-inflammatory disease states, intestinal epithelial cells promote an inflammatory response by secreting the chemokine IL-8 and the pathogen-elicited epithelial chemoattractant to recruit neutrophils.46 Intestinal epithelial cells secrete various factors which regulate immune cells in the intestine. Production of thymic stromal lymphopoietin (TSLP), TGF-β, and retinoic acid promote development of

21 tolerogenic dendritic cells and macrophages which can migrate to secondary lymphoid tissues and present antigens to adaptive immune cells or remain in contact with the epithelial cells and clear any bacteria that make it across the epithelium.25 TSLP and IL-25 production by intestinal epithelial cells helps to induce a T helper type 2 cell (TH2) response in helminth infection or allergic reaction. Thus, intestinal epithelial cells play a large role in detecting and responding to bacteria, both commensal and pathogenic, and other antigens in the intestines.

Immunogenicity of ingested nanoparticles

Upon introduction of foreign material such as NP into the body, one question that arises is whether that material will be immunogenic. This has been explored for NP used in foods which will come into contact with the intestinal immune system. Because one of the major chemokines released by intestinal epithelial cells is IL-8, many in vitro studies have evaluated IL-8 secretion as an indication of an immune response elicited by NP treatment. Multiple studies have shown that NP do induce IL-8 secretion and some of these papers are summarized in Table 2.1. A 3-fold induction in IL-8 secretion was observed in undifferentiated Caco-2 cells treated with 15 nm but not 55 nm silica NP at 32 µg/mL in serum free media.47 In another study, both 14 nm silica NP and 10 nm ZnO NP induced

IL-8 secretion in Caco-2 cells at a dose of 20 µg/cm2. IL-8 expression and secretion was similar using NP that had undergone in vitro digestion and those that had not undergone digestion. Silica NP increased IL-8 mRNA expression much more strongly in undifferentiated cells than differentiated cells and this trend held for IL-8 protein secretion as well. For ZnO NP, there was only slight induction of IL-8 mRNA expression in either

Nanoparticles Dose Cells IL-8 Ref. 15 nm silica 32 µg/mL Undifferentiated 3-fold induction in IL-8 47 Caco-2 secretion 55 nm silica 32 µg/mL Undifferentiated No induction in IL-8 secretion 47 Caco-2 14 nm silica, 20 µg/cm2 Caco-2 IL-8 secretion, greater 48 digested and increase in IL-8 mRNA and undigested IL-8 secretion in undifferentiated cells 19 nm silica 27 µg/mL Caco-2 Little IL-8 release 52 21 nm TiO2 10 and Caco-2 Slight increase in IL-8 49 100 SW480 production µg/mL Slight increase in IL-8 generation < 25 nm TiO2 1 and 2.5 Caco-2 Increased IL-8 production 50 µg/cm2 15 nm TiO2 27 µg/mL Caco-2 Little IL-8 release 52 20 nm ZnO 10 and Caco-2 IL-8 production 49 100 µg/ml SW480 No IL-8 production (likely due to toxicity) 50-70 nm ZnO 1 and 2.5 Caco-2 Increased IL-8 production 50 µg/cm2 50-70 nm ZnO 3 and 5 LoVo ~ Double the amount of IL-8 53 µg/cm2 release in untreated cells 20-30 nm Ag 10 and Caco-2 Significant IL-8 production 49 100 SW480 No IL-8 production µg/mL PVP-capped < 39 µg/cm2 Caco-2/THP- Induced IL-8 release in 51 20 nm Ag 1/MUTZ-3 coculture but not Caco-2 coculture monoculture. Less IL-8 induction was observed after pretreating cells with IL-1β to induce an inflammatory state prior to Ag NP treatment 25 nm Ag 27 µg/mL Caco-2 Significant IL-8 release 52 spheres and 80- 90 nm Ag rods Table 2.1. Summary of literature showing an immune response to NP treatment in intestinal epithelial cells.

23 differentiated or undifferentiated cells, but IL-8 secretion was strongly induced in differentiated cells. The fact that there was no induction of IL-8 secretion in undifferentiated cells is likely because ZnO NP treatment at this dose induces significant

48 cytotoxicity to cells. Another study showed that TiO2 (21 nm), Ag (20-30 nm), and ZnO

NP (20 nm) all induced IL-8 secretion in Caco-2 cells at doses of 10 and 100 µg/mL after

48 hour treatment. Ag NP induced the most IL-8 production, followed by ZnO NP and

49 finally TiO2 NP which induced only slightly increased IL-8 production. In SW480 cells

(another colon carcinoma-derived cell line), only TiO2 NP slightly increased IL-8 generation while almost no IL-8 was detected in ZnO-treated cells, likely due to toxicity of the ZnO to cells.49 Treatment of Caco-2 cells with 50-70 nm ZnO NP increased IL-8 production after 6 and 24 hours at doses of 1 and 2.5 µg/cm2. Similarly, IL-8 production increased after 24-hour treatment of cells with < 25 nm TiO2 NP at doses of 1 and 2.5

µg/cm2. However, there was no release of IL-6 or tumor necrosis factor-α (TNF-α)

50 observed after treatment with ZnO or TiO2 NP. Polyvinylpyrrolidine (PVP)-capped Ag

NP (< 20 nm) induced IL-8 release beginning at doses of 39 µg/cm2 in a co-culture model containing Caco-2 cells grown on top of human macrophages (THP-1) and dendritic cells

(MUTZ-3) embedded in collagen, but did not induce IL-8 release in monocultures of Caco-

2 cells until a dose of 312.5 µg/cm2 Ag NP.51 Co-cultures treated with IL-1β for two days to induce an inflammatory state also increased IL-8 production after treatment with Ag NP, but to a lesser extent than in the non-inflamed co-cultures.51 A study interested in NP use in paints found that 19 nm SiO2 and 15 nm TiO2 NP induced only slight IL-8 release in

Caco-2 cells, but Ag NP (25 nm spheres and 80-90 nm rods) induced a more significant

IL-8 release at 27 µg/mL after 48-hour treatment.52 Treatment of LoVo cells (colon carcinoma-derived epithelium) with 50-70 nm ZnO NP at 3 and 5 µg/cm2 for 24 or 48 hours induced an increased IL-8 release of approximately double the amount released in untreated cells.53 These studies suggest that ingested NP may be able to induce some immune response although this likely depends on the NP properties.

Several studies outside of the intestines have linked induction of inflammation by

NP to oxidative stress. A study in J774 murine macrophages showed that IL-1α mRNA expression (which leads to TNF-α release) after treatment of cells with PM10 was dependent on both calcium signaling and ROS production.54 Work with silica NP has suggested that ROS induced by silica NP treatment can trigger proinflammatory responses.55 Treatment of RAW264.7 macrophages with 12 nm silica NP showed that silica induced ROS generation and depletion of glutathione (GSH) beginning at 5 ppm silica NP and these macrophages also generated NO which plays a role in signaling related to inflammation. Administering silica NP intraperitoneally to mice at 50 mg/kg caused an increase in NO as well as inflammatory cytokines in blood, suggesting that this inflammatory reaction could be related to ROS generation.55 From these and similar studies, it is not clear whether oxidative stress induced by NP is solely responsible for induction of inflammatory responses or if NP can also directly induce inflammation, but these responses do seem to be related and likely feed into one another.

Several of the previously reported studies that observed IL-8 induction in intestinal models also observed oxidative stress in these cells, suggesting that the immunogenicity may be connected to oxidative stress.47, 48, 50, 53 However, other studies did not observe

25 oxidative stress in cells,49, 52 suggesting that an inflammatory response is not necessarily dependent on oxidative stress. Although studies investigating immunogenicity in vivo are more limited, there is some evidence that NP administered orally may cause inflammatory

® responses. TiO2 microparticles (Kronos 1171, 260 nm) and NP (66 nm) administered to mice by gavage at a dose of 100 mg/kg body weight/day for 10 days all increased inflammatory cytokines in the ileum including interferon-γ (IFN-γ), TNF-α, IL-4, IL-23, and TGF-β, and increased CD4+ T cells in duodenum, jejunum, and ileum of treated mice.56

However, other studies have suggested that some NP may not stimulate an immune response or may even induce a tolerogenic response. Administration of 10, 75, and 110 nm

Ag NP to rats at doses of 9, 18, and 36 mg/kg body weight/day for 13 weeks resulted in decreased expression of mucins, toll like receptors, and the T cell regulation genes FOXP3,

GPR43, IL-10, and TGF-β in the ileum, suggesting that Ag NP may be leading to a greater tolerance of bacteria by the intestines.57 Based on these studies, further investigation is required to determine the ability of specific NP to induce an immune response when ingested.

Regulation of intercellular junctions

Most nutrient absorption is carried out via transcellular transport through intestinal epithelial cells which is mediated by specific transporters or channels. However, luminal contents can also traverse the epithelial layer by the paracellular transport pathway between cells. Diffusion via the paracellular pathway may help maintain necessary gradients across the intestinal epithelium that promote transcellular transport.58 Under normal conditions, only water, ions, and certain solutes are able to pass between cells, but paracellular

26 permeability can be increased in response to inflammation or in disease states. Increased transport of luminal contents including allergens or pathogens across the epithelium can lead to increased immune activation. Increased paracellular permeability has been shown to be involved in disease initiation and development.59

Desmosomes and adherens junctions between epithelial cells help to adhere cells together, but tight junctions regulate the paracellular permeability (Figure 2.4). In addition to regulating permeability, these junctions help polarize epithelial cells.18 Tight junctions and adherens junctions form the apical junctional complex surrounding cells just below the base of the microvilli. Desmosomes are located basally to this apical junctional complex and are made of desmoglein and desmocollin which are connected to keratin filaments intracellularly by desmoplakin, plakofilin and plakoglobin proteins.58, 60 Adherens junctions are formed by the transmembrane epithelial cadherin (E-cadherin) proteins which interact with p120 catenin and β-catenin within the cell. β-catenin binds to α-catenin 1 which can regulate actin assembly.58 Tight junctions encircle the apical side of cells to prevent unwanted luminal contents from traversing the epithelial layer.59 The four transmembrane protein families occludin, claudin, junctional adhesion molecule (JAM), and tricellulin form the anastomosing strands of the tight junctions between cells which create these barriers to macromolecules.61 The claudin family is comprised of at least 24 members. Some of these proteins interact to create barriers to macromolecules while others form channel pores with radii of 4 angstroms that are selectively permeable to cations, anions, or water. Thus, claudin expression can differentially regulate the permeability of tight junctions. The intracellular domains of these transmembrane proteins interact with

Figure 2.4. Junctions between intestinal epithelial cells. This illustrates the major components of desmosomes, adherens junctions, and tight junctions that form between intestinal epithelial cells. Tight junctions regulate the paracellular permeability by impeding the passage of molecules between cells. They are composed of occludin, claudin, junctional adhesion molecule (JAM), and tricellulin proteins which can create barriers to macromolecules as well as channels which can selectively allow passage of cations, anions, or water. The intracellular domains of these proteins interact with zonula occludens (ZO) proteins which anchor the tight junction complex to cellular actin. Adherens junctions are just basal to the tight junctions and help to adhere cells together. The adherens junctions are formed by E-cadherin proteins that interact with p120 catenin and β-catenin within the cell to anchor this complex to the actin cytoskeleton. Desmosomes are located most basally and serve to adhere cells together along with the adherens junctions. Desmosomes are composed of desmoglein and desmocollin which bind to desmoplakin, plakofilin, and plakoglobin proteins to connect the desmosome to the intracellular intermediate filament network.60

zonula occuldens (ZO) proteins that form a scaffold which anchors the tight junction complex to the actin cytoskeleton in the cell.59 Contraction of the perijunctoinal actomyosin ring is a major regulator of tight junction permeability.62 The ZO proteins also interact with 28 actin-associated proteins such as cingulin and can recruit signaling proteins to the tight junctions.59

Tight junction permeability can be modulated by cytokines, growth factors, pathogens, and food constituents that come into contact with epithelial cells. Pro- inflammatory cytokines such as IFN-γ, TNF-α, and IL-1β increase intestinal permeability by changing expression of the tight junction-associated proteins. Conversely, cytokines such as IL-10, IL-17, and TGF-β decrease permeability by modulating expression of tight junction-associated proteins or neutralizing pro-inflammatory cytokines. Pathogens can alter tight junction permeability by direct interaction with epithelial cells or secretion of toxins such as Zonula occludens toxin (ZOT) by Vibrio cholerae. Food constituents such as the gliadin protein found in wheat and other grains can transiently increase tight junction permeability while other food factors including probiotics can restore tight junction permeability.59

Although many extrinsic factors are known to regulate intestinal permeability, the molecule zonulin is the only known human protein that reversibly regulates tight junction permeability.63 Zonulin was discovered as a human analog to the Vibrio cholerae ZOT toxin. In response to treatment with pathogenic bacteria, secreted zonulin disrupted tight junction permeability, resulting in ZO-1 redistribution away from the cell border.64 Zonulin likely regulates tight junction permeability to allow for the transport of macromolecules and immune cells into or out of the intestinal lumen.65

Transport of nanoparticles across the epithelium

In order for ingested NP to penetrate into the body beyond the GI tract, NP will have to be transported across the intestinal epithelium by one of several methods of transport. NPs can be endocytosed by cells and transported transcellularly. A study that treated Caco-2 cells with < 40 nm TiO2 NP found NP both within and below cells of the monolayer without detecting changes in intercellular junction structure. Based on the intact cellular junctions, lack of cell death to create gaps in the epithelium, and the presence of

NP within cells, the authors concluded that TiO2 NP transport was most likely transcellular.66 Many other studies have reported internalization of NP by Caco-2 cells, suggesting that transcellular transport is possible across the intestinal epithelium.47, 52, 67-70

Ingested NP can also undergo paracellular transport between cells when there is a gap in the epithelium or increased permeability of the tight junctions. This is likely to play a role in NP transport in diseased intestines which may have greater epithelial permeability and, depending on disease severity, more gaps in the epithelium. Finally, NP can be phagocytosed by M cells. Because of their function to sample luminal contents, an in vitro co-culture model of M cells was able to transport five times as many chitosan-DNA NP across the epithelial layer as Caco-2 cells alone.71 This suggests that M cells may be an important means of NP transport across the epithelium. However, since there are vastly more enterocytes than M cells, enterocytes will still play a large role in NP transport. It has been observed for many years that Peyer’s patches contained pigmented macrophages and this pigment was determined to consist of granular aluminum, silicon, and titanium, which was presumably taken up from the diet, and was generally in the size range of a few

30 hundred nanometers in diameter.72 This further suggests that NP transport by M cells will be important and this accumulation of inorganic food additives in macrophages and potentially other cells of the intestinal epithelium will likely only amplify as NP are more frequently used and these particles are more readily transported across the epithelium.

Intestinal diseases

Many diseases involve intestinal dysfunction and dysregulation, but inflammatory bowel diseases (IBD; Crohn’s disease or ulcerative colitis) are some of the more prevalent intestinal-specific diseases. Although much of the etiology of IBD is still unknown, it has been shown that both genetic and environmental factors play a role in disease initiation and progression. Crohn’s disease has been associated with polymorphisms in NOD2

(nucleotide-binding oligomerization domain) and autophagy-related genes ATG16L1 and

IRGM which may cause IBD patients to be ineffective at autophagic response to pathogens.73 In ulcerative colitis, genes involved in the IL-23 pathway have been implicated in disease progression.73 Interaction of the host with pathogens plays a role in the inflammatory response of IBD, but it is unclear if this is due to the persistence of certain potentially pathogenic bacteria, increased permeability of the mucosal barrier allowing pathogens more direct access to the body, or an imbalance in pathogenic vs. protective bacteria composing the microbiota. The host immune system is also impaired at multiple levels including disruption of the mucosal barrier and dysregulation of the innate and adaptive immune responses to pathogens. Together all of this promotes an inappropriate inflammatory response (Crohn’s disease is characterized by a T helper type 1 (TH1)

31 response while a TH2 response is characteristic of ulcerative colitis) that severely affects a patient’s quality of life.73

Upregulation of zonulin leading to an increase in intestinal permeability has been shown to play a role not only in IBD but also many other diseases including celiac disease, type I diabetes, multiple sclerosis, and various cancers. It has been hypothesized that autoimmune diseases may require the loss of effective muscosal barriers, especially in the intestine, in order to allow for exposure to the triggering environmental antigens. This theory is best exemplified by celiac disease where the environmental trigger, gliadin (a gluten protein found in grains such as wheat), induces upregulation of zonulin expression, thereby allowing gliadin to traverse the epithelium and induce the adaptive immune response leading to the autoimmune-induced intestinal injury.74 Gliadin induces zonulin release and thus transient increase in intestinal permeability by binding to the chemokine receptor CXCR3. However, CXCR3 is overexpressed in patients with celiac disease, which leads to a greater disruption in tight junctions and allows exposure to the specific gliadin peptide which initiates the autoimmune response.75

While the environmental factor leading to disease is not known for most autoimmune diseases, zonulin may play a role in the initiation of diseases such as type I diabetes, IBD, and multiple sclerosis, all of which have been associated with increased zonulin expression.74 Multiple other diseases have also been associated with increased intestinal permeability. Intestinal disease development may induce defects in the intestinal barrier, but increased intestinal permeability can lead to exposure of the body to pathogens, allergens, and toxic materials, initiating or further promoting disease. Increased intestinal

32 permeability has been observed in patients with diseases including inflammatory bowel diseases (IBD), irritable bowel syndrome, celiac disease, type I diabetes, alcoholic liver disease, and food allergies.75-81 While cytokines produced by patients with inflammatory bowel disease are likely to maintain hyperpermeability of the intestinal epithelium, there is evidence suggesting that increased intestinal permeability prior to disease onset is necessary for disease and may be involved in disease initiation.82 Increased intestinal permeability has been shown in first-degree relatives of patients with celiac disease who do not display the celiac-specific villous atrophy83 as well as asymptomatic first-degree relatives of Crohn’s patients, 84 suggesting in both cases that hyperpermeability may precede disease onset. The exposure of the mucosal immune system to luminal antigens as a result of this hyperpermeability induces an elevated inflammatory response that will lead to cytokine production and increased barrier dysfunction in this self-amplifying cycle. It is clear that regulation of intestinal permeability plays an important role in normal intestinal function.

Nanoparticle ingestion in diseased intestines

Increased intestinal permeability found in diseased intestines may have significant implications for the ability of ingested NP to traverse the epithelium and gain access to the circulation. NP may also have different effects when the intestine is already inflamed.

Several in vitro models have been developed to investigate NP toxicity in an inflamed environment by treating cells with an inflammatory cytokine prior to treatment with NP.51,

85 One study using a co-culture model with THP-1 macrophages and MUTZ-3 dendritic cells embedded in a collagen scaffold under Caco-2 cells found that co-cultures treated

33 with IL-1β to induce an inflamed state were more sensitive to < 20 nm PVP-capped Ag NP at a dose of 312.5 µg/cm2 based on LDH release and at 78.125 µg/cm2 based on TEER measurements.51 However, IL-8 release was slightly decreased in the inflamed co-culture

51 model compared to the non-inflamed co-culture upon Ag NP treatment. TiO2 NP (7-10 nm) did not induce toxicity in this model.51 These models may be helpful in determining differences in NP toxicity in the presence of inflammation.

However, these questions also need to be addressed in vivo. In addition to greater transport across an inflamed intestine with increased permeability, ingested NP may also exacerbate inflammation and disease. Clinical studies have been performed to assess whether Crohn’s disease patient symptoms were alleviated by a low microparticle diet

(which eliminated any sources of TiO2 and silica among other natural contaminants or additives). An initial study with 18 participants showed efficacy of a low microparticle diet in decreasing Crohn’s disease activity index,86 but a larger study with 70 participants did not replicate these findings.87 In the larger study, patients in the control group were supplemented with 5 mg/day TiO2 to establish some baseline intake. This is well below the intake estimated by Weir et al.2 (0.2-0.7 mg/kg body weight) and thus with increasing NP use in foods, it will be necessary to ensure that normal dietary NP intake is not harmful, particularly to populations with diseased intestines where it is still unclear whether NP may play a role in disease triggering or progression in some individuals.

2.4 Nanoparticle characteristics important to toxicity

In health and safety studies with NP to date, it has been recognized that characteristics including size, shape, material, surface charge, solubility, and surface

34 chemistry are all important in determining the toxicity of NP (Figure 2.5). Differences in these properties may also explain discrepancies observed between studies with the same type of NP. The importance of these characteristics highlights the need for thorough NP characterization in order to fully understand the toxic potential of various types of NP. The

NanoRelease Food Additive project recently published an article outlining some of the NP properties important in the GI tract including particle size, shape, and surface properties.88

Not only are these properties important to consider for their biological effects, but they may also be modified by the conditions of the GI tract.88 Thus, it is important to not only investigate the role of NP characteristics in vitro, but also to confirm the importance of these characteristics using in vivo studies where the NP will be interacting with the entire

GI tract and the body as a whole.

Figure 2.5. Nanoparticle characteristics important in determining behavior. This scheme represents some of the many characteristics of NP which can be important for NP behavior and toxicity. Variations in any combination of these characteristics including NP size, shape, agglomeration, and surface functionality can alter NP behavior and thus all of these characteristics are important for NP toxicity. This also illustrates the vast potential variation between NP of the same material and the need for thorough characterization and identification of the NP being used in a particular study.89

Solubility

One important NP characteristic that may determine toxicity is solubility. In vivo, the solubility of NP will help to determine their biopersistence. Soluble or degradable NP may not remain in cells long enough to accumulate over chronic exposure. It has been suggested that solubility of silica will play a large role in determining its toxicity due to its limited persistence.90 The solubility of amorphous silica in water is approximately 115 ppm at 25°C91 which is greater than other forms of silica including quartz.90 In vitro, NP may dissolve in the cell culture media. Soluble NP are likely to begin dissolving in the storage solution. One study showed that ~14% of citrate-stabilized Ag NP in water will dissolve at

25°C after 3000 hours while 70% dissolves at 37°C.92 Thus, it is important to be aware of the possible dissolved species present before in vitro or in vivo NP exposure. In vivo, NP will quickly come into contact with digestive enzymes and fluids. The low pH environment of the stomach may accelerate dissolution of NP. One study with various Ag NP showed that exposure to simulated stomach fluid for 15 minutes caused aggregation and partial dissolution of Ag NP along with precipitation of AgCl on the surface of the particles.93

In addition to determining biopersistence and bioavailability of NP, solubility also plays a role in NP toxicity. Studies have shown that intracellular as well as extracellular

ZnO NP dissolution can mediate toxicity and the role of these pathways may vary by cell type.94 Thus, dissolution occurring both within and outside of cells may be important to provide toxic ions. It is generally thought that dissolution of ZnO NP is necessary for toxicity. Ag NP toxicity is also thought to be mediated through Ag ions and similar toxicity has been observed using Ag NP and Ag ions at comparable doses based on dissolved Ag

36 ions.95 Dissolved ions, particularly Ag ions, may form complexes with other species within the culture media or in cells, but these ions are still largely available to interact with cells.

NP are likely internalized by endocytosis and transported through the endocytic pathway to lysosomes. In lysosomes, the pH is lowered to degrade cellular debris as well as extracellular materials and this acidic environment may increase the rate of NP dissolution within cells. It has been shown that silica NP solubility in artificial lysosomal fluid (at pH 5.5) was 5.0%, ZnO NP solubility was 100%, and TiO2 and Ag NP solubility were both 0%.96 The lack of Ag NP solubility in lysosomal fluid may be due to formation of precipitates that remain associated with the Ag NP surface,93 and this has been proposed in other studies as the reason they were unable to detect free Ag ion release in artificial lysosomal fluid.97, 98 Studies in alveolar macrophages have shown that silica NP can lead to disruption of lysosomal membranes. This is likely a result of the inability of silica NP to be degraded in the lysosomes rather than a consequence of lysosomal NP dissolution. This increased lysosomal permeability may be involved in the induction of apoptosis.99, 100 A study in THP-1 macrophages showed that 10.7 nm ZnO treatment led to lysosomal destabilization which led to cell death.101 This was proposed to be an effect of dissolution of ZnO within the lysosomes, creating Zn2+ which can damage the lysosomal membrane and be released into the cell where they can damage organelles and potentially lead to cell death.101 Thus, toxicity of soluble NP may depend on whether the ions produced upon dissolution are toxic.

It is known that Ag NP dissolution intracellularly is important and may provide a continual source of Ag ions102 and there is evidence that Ag NP dissolution also occurs

37 within lysosomes. Ag NP have been localized to lysosomes after cell internalization in spermatogonial stem cells103 and BEAS-2B lung epithelial cells.98 In earthworms, citrate- coated Ag NP were shown to decrease lysosomal stability.104 One study showed that lysosomal degradation of GSH-Ag NP released Ag ions when incubated in a lyososmal- mimic mixture for 12 hours and lysosomal degradation of Ag NP was necessary in human fibroblasts to induce oxidative stress.105 Collectively, it is clear that NP solubility both extracellularly and intracellularly will play a role in the toxicity of the NP. Ions likely mediate toxicity in the cases of Zn2+ and Ag+ and NP dissolution can prevent the persistence and accumulation of NP in vivo.

Size

NP size can determine the speed of diffusion of NP through the mucus layer and the proportion of NP internalized by intestinal epithelial cells. Latex NP (14 nm) administered into the distal colon of Lewis rats were observed to diffuse through the colonic mucus layer in 2 minutes to the epithelial cell surface whereas 415 nm particles took 30 minutes to diffuse to the epithelial surface and 1.09 µm particles were distributed throughout the mucus layer and did not accumulate at the epithelial surface after 30 minutes.106 Biodegradable polylactic polyglycolic acid (PLGA) copolymer particles containing bovine serum albumin (BSA; 100 nm, 500 nm, 1 µm, 10 µm) were used to study particle uptake in a rat in situ intestinal loop model. They observed 15-250-fold higher uptake of 100 nm particles over the larger microparticles depending on the tissue type.4

Thus, NP seem to be more likely to be internalized by cells than larger particles. Optimal particle diameters for the most efficient endocytosis of NP have been estimated

38 theoretically to be ~28-30 nm for spherical particles,107 54-60 nm for cylindrical particles,107 and 50-60 nm for spherical particles in another study.108 NP much smaller than this may have to form aggregates in order to be taken up by cells, as illustrated in Figure

2.6. These theoretical estimations of NP uptake efficiency were confirmed by an in vitro study in human cervical cancer cells which found that 50 nm spherical gold particles were internalized more efficiently than 14, 30, 74, or 100 nm particles.109 However, a study in human colon adenocarcinoma cells found that 100 nm polystyrene particles were taken up better than 50, 200, 500, or 1000 nm particles.110 This difference may indicate that NP material type or other characteristics such as coating may alter the optimal size for cellular uptake.

Figure 2.6. Size-dependent endocytosis of nanoparticles. A) Larger NP (estimated to be approximately 30-60 nm)107, 108 can trigger receptor- mediated endocytosis of a single particle. B) Smaller NP must cluster together in order to trigger receptor-mediated endocytosis and be internalized by the cell. This may lead to size-dependent differences in NP uptake efficiency.111

In vivo, when gold NP of 1.4, 5, 18, 80, and 200 nm were administered by intraesophageal instillation, significantly more 1.4 nm NP entered the circulation and accumulated in blood, kidney, and urine reticulendothelial system after 24 hours, suggesting that smaller particles may be able to be distributed most readily. However, the

18 nm particles accumulated the most in the heart and brain,112 suggesting that accumulation may depend on other factors that may regulate exocytosis. In agreement with this conclusion, the study with gold NP showed that although 50 nm particles were internalized most efficiently, the smallest NP were exocytosed most efficiently such that

14 nm particles were exocytosed more than 30, 50, 74, and 100 nm particles.109 This difference in efficiency of endocytosis and exocytosis of NP may be important for NP accumulation and mediation of toxicity.

Shape

NP shape may also affect NP behavior in biological systems. Similarly to NP size,

NP shape can play a role in the efficiency of NP uptake. Internalization efficiency based on NP shape may also be dependent on cell type. A study in HeLa cells showed that transferrin-coated gold spherical NP were found to be internalized more (14 and 50 nm sizes) than 20×30, 14×50, or 7×42 nm transferrin-coated gold nanorods, but the nanorods were exocytosed more efficiently than the spheres.113 Similarly, 14×40 and 14×74 nm gold nanorods were taken up less efficiently by HeLa cells than 14 and 74 nm gold spherical

NP.109 The cell internalization of nanorods with a lower aspect ratio was higher in both of these studies than for nanorods with higher aspect ratios. This was confirmed in another study showing that shorter rods (with lower aspect ratios) were internalized more by HeLa

40 cells than longer rods.114 However, once NP were internalized, cytotoxicity was independent of the aspect ratio or length of nanorods.114 Contrary to these studies, internalization of 15×50 nm cetyltrimethylammonium bromide-coated gold nanorods was found to be more efficient than that of 15 or 50 nm gold spheres in monocytes and macrophages isolated from human blood.115 It is possible that these differences are due to the professional phagocytic character of monocytes and macrophages which may internalize NP differently than HeLa cells. Regardless, NP shape may play a role in the bioavailability of NP to cells and the body.

Surface charge

The surface charge of NP may also play a role in their interactions with cells and other biological components. Some studies have shown differences between positively- and negatively-charged NP. Administration of negative and positive 2.8 nm gold particles by intraesophageal instillation to rats revealed greater entry of the negatively-charged particles into the circulation and a trend of greater accumulation of negatively-charged particles in the examined tissues, but this was only significant in the liver and urine.112 A change in NP surface charge which destabilizes particles and leads to particle aggregation was shown to decrease NP transport across an intestinal epithelial cell model in vitro, as would be expected for larger particle aggregates.71 Neutral iron oxide NP (dextran-coated) were poorly internalized by cells. Either negatively-charged (heparin or DMSA-coated) or positively-charged (aminodextran-coated) iron oxide NP were better internalized by cells, but the positively charged particles accumulated more in cells.116 A study using negatively- charged carboxymethyl chitosan-grafted NPs and positively-charged chitosan

41 hydrochloride-grafted NPs (150 nm, rhodamine B-labeled) found that the positively- charged NP were internalized better by murine macrophages, but less negatively-charged

(-15 mV vs. -40 mV) and more positively-charged (35 mV vs. 15 mV) NP tended to be internalized more in L02 (human hepatic) cells and SMMC-7721 (human hepatocarcinoma) cells, indicating differences between cell types.117 This preferential interaction of non-phagocytic cells with positively-charged particles may be due to particle uptake using faster clathrin-dependent endocytosis while phagocytic preference for negatively-charged particles may be due to their affinity for negatively charged bacteria.118

Another study showed differences in the interaction of positive vs. negative quantum dots with cell surface scavenger receptors on murine macrophages in serum-free media and negatively-charged quantum dots were internalized more slowly by cells, suggesting different methods of internalization.119

Surface charge of particles may dictate how they interact with other biological fluids such as the mucus lining the intestinal tract. It has been shown that net neutral or near neutral but highly charged polyelectrolyte NP composed of polyacrylic acid and chitosan (mimicking the densely charged but net neutral surface of many viruses) can migrate better through mucus than similar particles with different ratios of polyacrylic acid and chitosan giving them a net positive or negative charge.120 Negatively-charged particles can bind to mucin strands through hydrophobic adhesive interactions and positively- charged particles can bind strongly to the negatively-charged glycosylated mucins.121 Thus, charge may be an important factor in determining the fraction of NP able to reach the intestinal epithelial surface. In addition to this, surface charge can affect the association of

NP with molecules and proteins in biological fluids which can alter the interaction of NP with cells.

Nanoparticle surface coatings/corona

The surface properties of NP will play a large role in determining NP behavior. In a study analyzing transport of chitosan-DNA NP across an intestinal epithelial model, functionalization of NP with ligands dramatically affected trancytosis. For example, transferrin modification enhances transport three- to five-fold.71 This suggests that the molecules and functional groups on the outer surface of the NP can determine particle behavior and NP coatings will be important in determining their toxicity. Indeed, NP functionalization is being explored to create particles optimized for specific applications such as drug delivery.122

It is known that proteins and other biomolecules will adsorb to the surface of NP in biological systems.123 The protein corona is often composed of two layers, a tightly bound hard corona covered by a more transient, loosely-associated soft corona layer.123 Some studies have observed that NP will induce cell damage only in the absence of a corona.

Silica NP were found to be internalized more by HeLa and A549 cells treated with serum- free media to prevent formation of a corona on the NP.124 The silica NP in serum-free media were also shown to have a stronger adhesion to the cell membrane and caused greater cell damage and toxicity in A549 cells. After one-hour exposure of silica NP in serum-free medium to cells, coronas composed of cytosolic, cytoskeletal, and cell membrane proteins had formed on NP, which may be the cause of the cell damage and toxicity observed.124

Agglomeration of NP in the presence of serum may also play a role in decreased toxicity

43 of these particles to cells, as was observed with silica NP in mouse fibroblast cells.125

Protein adsorption to NP and the composition of the corona can alter how NP are recognized and internalized by cells126 as well as NP interactions with intracellular components. Thus, the protein corona and what is on the surface of NP is likely to play an important role in the response to NP. The presence of serum in cell culture media used for in vitro studies may be an important factor in the subsequent NP toxicity.

2.5 Nanoparticle-induced oxidative stress

Many NP have been shown to mediate cytotoxicity through induction of oxidative stress, and it is the most commonly proposed mechanism of NP toxicity.67, 127-130 Normal cellular metabolism results in production of ROS and low levels of oxidative stress.

However, when oxidative stress increases beyond what the cellular antioxidants are equipped to handle, this can cause serious problems, including cell death. Low levels of oxidative stress, or ROS production, can cause a mitogenic response and increase cell proliferation. Higher levels of ROS will lead to a temporary growth arrest that will protect cells from excess energy use and DNA damage. Cells may also transiently adapt to protect themselves from oxidants by altering gene expression and protein translation to increase resistance to oxidative stress, which has been shown after exposure to mild levels of hydrogen peroxide.131 At higher levels of ROS, cells are unable to adapt and enter a permanent growth arrest or senescence. Further increasing ROS will lead to induction of apoptosis and finally necrotic cell death will occur at the highest ROS levels (Figure

2.7).132-135 This wide continuum of cellular responses to oxidative stress indicates that the

Figure 2.7. Oxidative stress responses corresponding to changes in redox couples. It has been determined experimentally that changes in the potential of glutathione, cysteine, and thioredoxin redox couples both intracellularly and extracellularly accompany various stages of oxidative stress including cell proliferation, differentiation and arrest, apoptosis, and necrosis.136

dose of NP used and the proliferation state of the cells may determine the cellular response observed.

Growth arrest is often associated with oxidative stress. Cell cycle arrest can allow cells time to repair DNA damage and protect from further DNA damage. During growth arrest, cells are often observed to accumulate in specific phases of the cell cycle. This accumulation may correspond to certain cell cycle checkpoints which ensure that cells are ready to continue the process of mitosis by halting cell division if problems are encountered in the cell cycle. The cell cycle is composed of four phases. In gap 1 (G1) phase, which 45 cells enter following division, cells grow in size and prepare for DNA replication. In the synthesis (S) phase of the cell, cells replicate DNA. The gap 2 (G2) phase allows time for cells to continue to grow and prepare for division. The mitosis (M) phase of the cell cycle is where sister chromatids separate and the cell divides into two new cells. Cell populations no longer dividing due to quiescence or senescence are in a resting phase termed gap 0

(G0).137

It has been shown that the redox couples of glutathione (GSH), cysteine (Cys), and thioredoxin (Trx) are the major cellular redox systems in cells. Collectively, their reducing potential and capacity determine the intracellular redox environment.135, 136 The

GSH/GSSG ratio can approximate the intracellular redox environment because it constitutes the largest component of the total redox pool. Further understanding of the roles these redox couples can play in controlling cellular metabolism and growth has led to the expansion of the understanding of oxidative stress as disruption of redox signaling in cells rather than merely an imbalance between prooxidants and antioxidants. Within cells, glutathione exists in its reduced form and formation of the oxidized disulfide GSSG form is associated with oxidative stress. In addition to intracellular redox systems, both Cys and

GSH are found in the intestinal lumen (GSH released from cells will be converted to Cys) and serve as a thiol pool that may reduce disulfide bonds to aid in protein digestion, help regulate mucus fluidity, and participate in luminal detoxication.138 This also means that cells can increase GSH through de novo synthesis, regeneration from GSSG, and import of luminal GSH.136 As shown in Figure 2.7, changes in the intracellular GSH/GSSG,

Cys/CySS, and Trx/TrSS redox potentials are known to be associated with oxidative stress- induced cell proliferation, differentiation, apoptosis, and necrosis.136

There are various means through which NP may induce oxidative stress in cells.

Some transition metal NP, including iron, copper, and silica, are able to participate in

Fenton-type reactions and directly create ROS.139, 140 Prooxidant reactive groups on the

•- surface of NP including ZnO and SiO2 NP can react with molecular oxygen to form O2 and further reactions with bound free radicals can generate additional ROS. Free radicals

• • • •- bound to the particle surface such as SiO or SiO2 can generate OH and O2 .

Environmental matter such as ozone or NO2 adsorbed to the NP surface can also produce oxidative damage. Dissolved ions from reactive NP such as silica will also add to ROS generation.139, 140

The other major component of oxidative stress generation is the interaction of NP with cells. The internalization of NP by immune cells can activate those immune cells so that they induce a ROS/reactive nitrogen species response.139, 140 Once internalized by cells,

NP or ions dissolved from the NP will interact with intracellular enzymes and proteins, disrupting their cellular functions. This disruption leads to cellular damage that can contribute to ROS generation. For instance, Ag NP have been shown to inhibit the glutathione (GSH)-synthesizing enzymes glutamate cysteine ligase and glutathione synthetase, leading to increased ROS in human Chang liver cells.141 Ag NP are also thought to be able to indirectly induce oxidative stress through interaction with thiol-containing enzymes in the respiratory chain or with superoxide radical-scavenging enzymes.142 NP have been shown to be capable of physically damaging mitochondria.143 The increase in

47 oxidative stress in cells beyond what can be managed by cellular antioxidants can then contribute to inflammation through upregulation of cytokines or other inflammatory mediators and can lead to DNA damage and eventually cell death.

2.6 Silica nanoparticles

E551, the common food-grade silica additive, is a form of synthetic amorphous silica (SAS). SAS exists in four forms including pyrogenic silica, precipitated silica, silica gel, and colloidal silica.144 Although E551 has not been used in the majority of in vitro models, most studies have used amorphous silica NP and thus these NP should reasonably represent the NP portion of E551. Because different forms of amorphous silica are used between studies, it is important to recognize that differences in the NP characteristics may alter NP toxicity in the same cell type. As previously discussed, NP size, charge, surface chemistry, etc., will all play significant roles in determining NP toxicity and thus results may vary between experiments utilizing different silica NP.

Toxicity to the intestinal epithelium

It has been known for some time that silica inhalation can cause silicosis in the lungs. However, several recent studies have evaluated silica NP toxicity in both in vitro and in vivo intestinal epithelial models. A study of 25-30 nm colloidal silica NP in six cell lines (both mouse and human lung, colon, and macrophage cell lines) showed that macrophages were most sensitive to silica NP-induced toxicity while the colon cells were the least sensitive.145 This suggests that silica NP toxicity likely differs between cell and tissue types and silica NP may be less toxic in the intestines than in the lungs.

Silica NP have been shown to be able to mediate toxicity through oxidative stress generation resulting in DNA damage and induction of apoptosis.146 This mechanism of toxicity has been observed in multiple intestinal epithelial cell studies to date. A study treating Caco-2 cells with 15 nm silica NP showed decreased cell viability and an increase in caspase 3 activation in cells, indicating initiation of apoptosis.47 The silica NP were genotoxic to cells and increased micronucleus frequency. There was also a slight increase observed in ROS production after 15 nm silica NP treatment.47 Thus, they concluded that oxidative stress is likely involved in induction of cell death and DNA damage by silica NP.

A study in Caco-2 and human gastric epithelial cells (GES-1) observed that 10-50 nm silica

NP induced LDH release, a decrease in cell viability based on the Cell Counting Kit-8

(CCK-8) assay, a partial inhibition of cell proliferation, a slight S phase cell cycle arrest in

GES-1 cells and S and G2/M phase arrest in Caco-2 cells, and a slight increase in ROS generation. This study did not observe induction of apoptosis or necrosis in cells, but the observed ROS generation may be contributing to the silica NP-induced decrease in viability and cell cycle arrest. Since these effects were observed at high silica NP doses

(often 200 µg/mL), they conclude that silica NP are safe below 100 µg/mL doses.147 A study in differentiated and undifferentiated Caco-2 cells showed that 14 nm silica NP were taken up by cells and induced ROS formation in cells. Silica NP induced toxicity in undifferentiated cells based on the reduction of the tetrazolium salt WST-1 by cellular enzymes after 4 and 24 hour treatments, but no toxicity was apparent in differentiated cells.48 From this study it is clear that cell differentiation state can play an important role in NP cytotoxicity. A study with various NP in serum-starved Caco-2 cells showed that 14

49 nm silica NP were toxic to cells at doses of 20 and 80 µg/cm2 based on both LDH release after 4 and 24 hours and a decrease in metabolic activity (WST-1 assay). The silica NP also induced DNA damage and glutathione depletion in cells, indicating oxidative stress.148 A study measuring HT-29 cell growth by impedance observed similar growth as control cells after treatment with 25 nm rhodamine B-embedded silica particles, but slight inhibition of cell proliferation after treatment with 100 nm meso-Tetra(N-methyl-4-pyridyl)porphine tetratosylate salt (TMPyP)-embedded silica particles. They observed an increase in cell death, especially after treatment with 100 nm silica particles (up to 40%). There was a slight increase in double-strand DNA breaks as determined by phosphorylated histone

H2Ax foci, particularly with 100 nm silica.149 Although they do not evaluate ROS production or initiation of oxidative stress, their observations are also consistent with a

ROS-mediated mechanism of toxicity which may lead to DNA damage and cell death.

It is also thought that there are other components of silica toxicity beyond ROS.146

Another possible mechanism of toxicity is silica NP-induced increase in lysosomal permeability and destabilization of cellular lysosomes. The inability of the lysosome to degrade silica NP causes a loss of lysosomal membrane integrity and can lead to ROS- independent initiation of apoptosis in cells.146 Although there has not been direct evidence for lysosomal destabilization contributing to silica NP toxicity in Caco-2 cells, there is some evidence that oxidative stress may not be the only mechanism of silica toxicity. A study performed in HT-29 cells with 12 and 40 nm silica particles reported an increase in cell growth (significant but relatively small) after treatment with 12 nm silica NP for 24,

48, or 72 hours as measured by the sulforhodamine B assay (measures sulforhodamine B-

50 bound protein to determine cell number). Increased LDH release and decreased mitochondrial activity was observed and effects were more pronounced in proliferating cells than confluent cells. ROS production was not observed and little DNA damage detected after silica treatment, indicating that toxicity was likely not mediated through oxidative stress. Treatment with 12 nm silica NP increased total cellular glutathione and γ- glutamylcysteine-ligase needed to synthesize glutathione at a dose of 31.3 µg/cm2 and above. This increase in GSH seems to be due to an upregulation in the ERK1/2

(extracellular signal-regulated kinases) pathway rather than due to oxidative stress. The

MAPK (mitogen-activated protein kinases)/ERK1/2 pathway can help regulate the cell cycle and thus may also be responsible for the increased cell proliferation.150

Other studies have observed no toxicity of silica NP in intestinal epithelial cells. A study using Caco-2 and colon carcinoma RKO cells showed minimal toxicity of 10 nm silica NP based on formazan assays up to 100 µg/cm2 doses. Four-hour exposure of cells to silica NP also showed minimal changes in gene expression as determined by a whole genome microarray.151 Another study performed in undifferentiated Caco-2 cells in serum- containing media (20%) treated with 32 and 83 nm fluorescent silica NP did not observe silica NP toxicity. These NP were internalized by cells and agglomerated in cells near the cell nucleus, but no cytotoxicity (WST-1 assay) or genotoxicity (comet assay) was observed in cells up to a dose of 200 µg/mL.69 These differences may stem from use of serum in the media which has been shown to ameliorate silica NP toxicity152 or differences in the particles themselves. Based on the current research, it appears that silica NP do

51 induce some toxicity in the intestinal epithelium, but the toxicity usually does not result in complete cell death.

In vivo toxicity

While in vitro studies are an important step in approximating NP toxicity after ingestion, in vivo studies are more representative of the entire environment to which NP will be exposed. Several studies have orally administered silica NP to rodents to estimate toxicity. From these studies, it has been determined that there is minimal silica absorption through the intestines. Repeated oral administration (20 mg SAS/kg body weight/day) for up to 5 days in rats revealed very little accumulation of silicon in any organs as detected by inductively coupled plasma-dynamic reaction cell-mass spectrometry (ICP-DRC-MS) although there seemed to be slightly more accumulation of silicon in liver and spleen of females over males.153 From this they concluded that there is very little absorption of silica from the GI tract. A long-term study administered 20 and 100 nm negatively-charged colloidal silica NP to Sprague-Dawley rats by gavage at doses of 500, 1000, and 2000 mg/kg body weight/day for 90 days. They observed no clinical signs of toxicity over the treatment period. Necropsy results did show differences in organ weights of lung and liver in 20 nm NP treatment groups and kidney, lung, submandibular gland, and ovary in the

100 nm NP treatment groups. However, these changes did not seem to be dose-dependent or correspond to histopathological observations, so it was concluded that they were not related to silica ingestion.154

There is evidence in the literature for some oral absorption of silica NP. One study orally administered 70, 300, and 1000 nm silica particles to mice for 28 days at 2.5 mg/day.

They detected no indication of NP toxicity or damage after 28-day silica administration to mice and concluded that silica administration induced no toxicity in these animals.

However, absorption studies using an everted gut sac method did demonstrate silica NP absorption across the epithelium. The everted intestinal segment was incubated in a solution of silica for 45 minutes. They used 70 nm particles that had been functionalized with both carboxyl and amine groups to analyze the effects of NP surface charge and found greater absorption of the carboxyl-functionalized silica, although the amine-labeled particles absorbed better than the unlabeled 70 nm silica particles. Surprisingly, there was no significant difference in the absorption of 70 nm particles over 300 or 1000 nm particles, but there was some absorption of all particles over background levels.155 These experiments suggest that there is likely some absorption of silica NP across the epithelium. This observation was confirmed by a study utilizing silica NP with incorporated rhodamine B isothiocyanate to allow in vivo imaging of the NP distribution in mice which showed significant localization of fluorescent signal to kidneys as well as some to lungs and liver of treated mice starting as early as two hours after administration.156 Thus, there are silica

NP that are being absorbed from the GI tract and distributed to tissues although they may not significantly accumulate in tissues.

Some studies have also observed enough absorption of silica NP to cause toxicity.

A study that administered 30 nm and 30 µm silica particles (silica obtained from rice husk) to mice in their diet for 10 weeks (total silica intake was 140 g silica/kg body weight) observed apparent liver toxicity in the nano-fed group based on increased alanine aminotransferase (ALT) and evidence of fatty liver in H&E staining.157 Another study

53 administering two silica NP materials, SAS (7 nm particles) and NM-202 (10-25 nm particles), orally to mice for 28 days at doses of 100, 500, or 1000 mg/kg body weight/day

NM-202, or 100, 1000, or 2500 mg/kg body weight/day SAS observed minimal silica accumulation in tissues although the silica levels in the liver, kidney, and spleen did reach significance at the lowest NM-202 dose and middle NM-202 dose in liver only. After an

84-day exposure with the highest doses, they observed an accumulation of SAS but not

NM-202 in the spleen. There was increased liver fibrosis determined by histology after 84- day exposure which reached significance only in the NM-202-treated mice as well as a moderate increase in expression of fibrosis-related genes.158 These studies suggest that at certain doses and lengths of exposure, at least some silica NP can induce pathological changes upon oral ingestion.

Greater effects at low doses

Some studies with silica have reported that lower doses of NP allow greater absorption and lead to greater toxicity. The van der Zande et al.158 study observed a trend that more silica NP seemed to be absorbed through the GI tract at lower doses and found that in simulated in vitro intestinal conditions, the silica had stronger gel-like properties at higher concentrations which may decrease the bioaccessibility of the silica and thus the absorption into the body. The high pH and salt concentrations of the intestine will add to the gelating behavior of silica. Thus, this may be an important factor for studies performed with high doses of silica NP, but will be less of an issue at human exposure levels of silica.158 In another study, nanoparticulate SAS was administered orally to rats to observe genotoxicity and a slight increase in micronucleated cells in colon crypts in rats treated

54 with only the lowest dose, 5 mg/kg body weight/day of NM-200 and NM-201, was observed.159 Based on the van der Zande et al. paper, this effect which was observed at the lowest dose could be due to gel-like behavior of the silica and less absorption at higher doses. However, the suspension of silica used to administer NP to rats in this study (6 mg/mL) is a lower concentration than what was tested by van der Zande et al. (9 mg/mL), so it seems unlikely that there would be much gelation of silica at the administered doses.158, 159 In vitro, the study using 100 nm TMPyP-embedded silica particles observed greater toxicity at lower doses (10 µg/mL) than the highest doses (150 µg/mL) in all of their assays. They suggest that this could be a protection response by cells that is initiated at a certain level of toxicity.149 However, this could also be due to gelation of the silica NP at the higher doses such that less of them were available to be internalized by cells and cause toxicity.

From these studies both in vitro and in vivo, it is apparent that silica NP may induce toxicity in cells. However, the low absorption of silica NP in vivo greatly decreases the potential for NP toxicity. Thus, silica NP are generally assumed to be biocompatible and nontoxic for medical and food additive uses.

2.7 Titania nanoparticles

Titanium dioxide is used as a white pigment in many applications. TiO2 is the naturally occurring oxide of titanium and can be purified from ilmenite ore (FeTiO3). TiO2 exists in multiple crystal structures including rutile, anatase, and brookite. Anatase and

160 rutile TiO2 phases are the most commonly synthesized and used. Anatase TiO2 has been reported to be more reactive than rutile TiO2, and studies have observed differences in

161, 162 toxicity between rutile and anatase TiO2. The majority of TiO2 NP consist of a mixture of rutile and anatase phases,163 but it may be an important factor to determine the proportions of rutile and anatase when assessing NP toxicity. Food-grade TiO2, E171, can be obtained as either rutile or anatase and has a broad size distribution with average particle size of 110 nm.2

Toxicity in the intestinal epithelium

TiO2 NP have been shown to cause cytotoxicity via oxidative stress-dependent pathways leading to DNA damage, cell cycle arrest or delay, and mitochondrial dysfunction, particularly in pulmonary and inhalation models.163 However, experiments in intestinal epithelial cells are generally in agreement that TiO2 NP are nontoxic. A study in

Caco-2 cells treated with TiO2 NP (<40 nm) showed a decrease in epithelial monolayer integrity by decreased TEER measurements and a loss of localization of γ-catenin to cell adherens junctions beginning at 6 days after continuous TiO2 NP treatment and continuing to 10 days at a dose of 1000 µg/mL. No decrease in TEER was observed after acute exposure and no induction of cell death was observed after acute or chronic exposure.66

They also observed cell internalization of TiO2 and apparent transepithelial transport of

TiO2 across the epithelium based on confocal microscopy of the cell monolayer at different depths. Studies of transport across the epithelium in a transwell system revealed that a similar total mass of TiO2 was transported through the epithelial layer regardless of dose.

Cells responded to TiO2 treatment by increasing intracellular-free calcium which can regulate calcium-dependent enzymes, but this response was downregulated over 14 days.66

In another study, investigation of P25 TiO2 treatment on the intestinal epithelial cell lines

Caco-2 and SW480 revealed slight toxicity based on the MTT assay in SW480 cells with

TiO2 suspended in buffered synthetic freshwater (to mimic NP contamination in drinking water) but not in cell culture media or in Caco-2 cells.49 The differences in toxicity observed between the cell culture media and synthetic freshwater could be due to the absence of or differences in composition of the protein corona on cells in the freshwater versus the culture medium. However, little toxicity was observed in either cell type under any conditions. No induction of ROS production was observed by TiO2 in either cell type, but there was a slight increase in IL-8 production after TiO2 treatment in both cell lines at doses of 10 and 100 µg/mL.49

A study in Caco-2 cells investigated differences between cellular treatment with bulk, P25 (25% rutile and 75% anatase), nano-anatase, and nano-rutile forms of TiO2.

There was no change in LDH release after treatment with any of the TiO2 particles. They observed cellular accumulation of TiO2 that seemed to be due to active cellular uptake.

Also, anatase was absorbed faster than rutile forms of TiO2, suggesting that the form of

TiO2 will make a difference in effect on cells. They also observed some effects of TiO2 on electrolytes in cells. All forms of TiO2 caused a depletion in potassium and elevation of

164 magnesium, and P25 or anatase TiO2 forms caused an elevation of calcium. Another study comparing anatase (4, 7, or 215 nm particle sizes) to anatase/rutile TiO2 NP mixtures

(22 or 25 nm) in Caco-2 cells found that the anatase/rutile but not anatase only samples induced LDH release, decreased metabolic activity (to ~70%) at 80 µg/cm2, and induced

DNA damage at 20 µg/cm2. No decrease in total cellular glutathione was observed, suggesting that this toxicity may not be mediated by oxidative stress.162 Although the

57 toxicity induced by the anatase/rutile TiO2 NP was mild, this study suggests that rutile TiO2

NP may be necessary for toxicity in these cells. However, this is opposed to research performed in other cells including fibroblasts and lung epithelial cells which showed that

161 anatase TiO2 was 100 times more toxic than rutile TiO2. Thus, some other difference between these NP formulations may be contributing to the differences observed in cytotoxic potential.

Treatment of Caco-2 cells with 20-60 nm TiO2 NP did not induce cytotoxicity up

2 to a dose of 20 µg/cm even though TiO2 NP did increase cellular ROS. There was a decrease in cell proliferation to ~80% of the control as measured by the colony forming assay at 20 µg/cm2, but overall toxicity was considered to be negligible.50 However, the increased ROS production indicates that these NP may be capable of inducing toxicity at higher doses. Another study used a 3D intestinal model where human macrophages (THP-

1) and dendritic cells (MUTZ-3) were embedded in collagen on a transwell insert and

Caco-2 cells were then grown on top of this layer. TiO2 NP (7-10 nm) did not induce toxicity or inflammation in the co-culture model in inflammatory or noninflammatory

51 conditions, confirming the safety of TiO2 NP in a more complex in vitro model incorporating immune cells. Nano-TiO2 isolated from chewing gum and exposed to Caco-

2 and GES-1 (gastric epithelium) cells elicited minimal toxicity based on LDH release and cell viability determined by WST-8 assay up to a dose of 200 µg/mL although pure P25

165 TiO2 decreased cell viability by approximately 20% in GES-1 cells. There was a slight increase in ROS production after treatment with the nano-TiO2 from gum, but they concluded that the NP were relatively safe. Thus, although TiO2 NP are mostly nontoxic

58 in these studies, there is some indication that mild toxicity induced by TiO2 NP may be mediated by oxidative stress. This is of particular significance using TiO2 isolated from chewing gum since consumers are being exposed to this form of TiO2.

Several studies have shown that TiO2 NP can disrupt normal microvilli structure in intestinal epithelial cells. Even if TiO2 NP do not induce acute toxicity, disruption of important structures such as microvilli may significantly affect normal cellular functions.

A study which treated C2BBe1 cells with food-grade E171 TiO2 and TiO2 particles isolated from the candy coating of chewing gum found that both forms of TiO2 could disrupt the brush border (microvilli) in these cells at a dose of 0.1 µg/cm2 and this effect is not simply due to sedimentation of aggregated particles.70 This disruption resulted in limp rather than erect microvilli and fewer total microvilli. They also observed cellular internalization of

TiO2 24 hours after treatment with this dose which could require remodeling of the actin

70 cytoskeleton and further disruption of the brush border. Further, the low dose of TiO2 used is more representative of doses epithelial cells will be exposed to in vivo. In another study, scanning electron microscopy of Caco-2 cells treated with <40 nm TiO2 NP revealed disruption of cell microvilli after TiO2 treatment leading to decreased numbers of microvilli and changes in shape. Microvilli no longer stood erect and seemed to be absorbed into cells. Unlike toxicity (decrease in TEER was observed at 1000 µg/mL), this effect was observed at doses as low as 10 µg/mL, which may again represent more physiologically relevant doses.66 These studies necessitate in vivo studies to further investigate the effects of TiO2 NP exposure on microvilli.

TiO2 NP absorption after oral exposure

Despite little in vitro toxicity of TiO2 NP in intestinal epithelial cells, the in vivo toxicity depends greatly on NP absorption through the GI tract into the body. A study estimating the ability of NP to cross a Caco-2 monolayer using a transwell system and 18 nm TiO2 NP found that an amount near the detection limit of 0.4% of the applied concentration of NP (100 µg/mL) were able to cross the Caco-2 monolayer.166 When this study orally administered 100 mg/kg body weight TiO2 NP to mice, they observed no

166 significant increases in TiO2 in tissues 24 hours following administration, which is in agreement with the low transport of TiO2 NP across the epithelial layer observed in vitro.

However, they did observe significant TiO2 uptake in at least one isolated Peyer’s patch by

TEM analysis. Another study orally administered 21 nm TiO2 (80% anatase, 20% rutile)

NP to rats daily for 13 weeks at 260.4, 520.8, and 1041.5 mg/kg body weight/day. There was a slight increase in the titanium content of blood in treated animals which was only significant in male rats. There was no increase in TiO2 content of liver, kidney, spleen, or brain, as measured by ICP-MS. They also did not detect increased titanium in the urine but there was considerable titanium in feces, suggesting that little TiO2 was absorbed through the GI tract.167

Other studies have detected TiO2 NP in tissues after oral administration in vivo, suggesting greater oral absorption. Mice administered a single oral dose of 5 g/kg body weight of 25, 80, or 155 nm TiO2 particles displayed no acute toxicity after 2 weeks, but liver injury was induced based on increased ALT/aspartate transaminase (AST) and LDH serum levels, hydropic degeneration around the central vein, and spotty necrosis in

60 hepatocytes detected by histology. Also, increases in blood urea nitrogen and pathological changes in the kidney (swelling in renal glomerulus, renal tubule filled with proteinic

168 liquids) indicated renal injury. A study examining toxicity of oral exposure to TiO2 and lead also showed liver and kidney damage 7 days after a 5 g/kg body weight administration of TiO2 NP as well as slight increases in ROS production and decreases in glutathione peroxidase and superoxide dismutase (SOD) levels in various tissues.169 This indicates that if TiO2 traverses the intestinal epithelium and is transported throughout the body, there may be more global toxicity than that expected in the intestinal epithelium alone.

Therefore, further studies are needed to more completely determine the potential for toxicity upon TiO2 ingestion.

2.8 Zinc oxide nanoparticles

ZnO NP are also being used in many different applications and are of interest to the food industry. ZnO NP are used as chemical and biosensors, light emitting diodes, photo- detectors, and in sunscreens to absorb UV.170 ZnO also has antimicrobial properties that make it useful in biological applications. ZnO particles have a wurtzite crystal structure and NP are often synthesized using the sol-gel method or hydrothermal method which allows for control of NP shape and size.170

Toxicity to intestinal epithelial cells

Increased use of ZnO NP in various industries has led to increased human exposure to ZnO NP. Various research models have suggested that ZnO NP toxicity is due to dissolution of the NP either outside cells or within cells leading to increased availability of zinc ions able to interact with enzymes and various cell components. It has been

61 demonstrated in various models that ZnO NP mediate toxicity through oxidative stress, and lysosomal destabilization and mitochondrial dysfunction contribute to the cytotoxic response.94 This mechanism seems to apply to intestinal epithelial cells as well. In one study, 90 nm ZnO NP decreased Caco-2 cell viability and inhibited cell proliferation based on live/dead cell staining, and decreased cell activity to a level less than 30% of control cells after 24-hour treatment. ZnO NP induced ROS production up to 2.5 times that of control cells, increased cellular glutathione, and decreased SOD levels, suggesting a mechanism of oxidative stress.171 In another study, MTT and LDH assays revealed dose- dependent toxicity of 26, 62, and 90 nm ZnO NP to Caco-2 cells (in serum-free media).

All ZnO NP increased ROS production in cells and decreased intracellular glutathione levels. ZnO NP treatment also increased cell death. Propidium iodide staining to assess

DNA content of cells revealed a slight S and G2/M phase cell cycle arrest at 50 µg/mL doses of all ZnO NP (although most obvious with 26 nm ZnO NP) which may allow time for DNA damage repair.172 This again seems to suggests that ZnO NP toxicity is mediated through oxidative stress. A study analyzing the effects of ZnO NP on Caco-2 cells observed a dose-dependent increase in LDH release and decrease in metabolic activity (WST-1 assay) at 5, 20, and 80 µg/cm2 doses after 24-hour treatment.148 ZnO treatment also increased DNA damage and oxidative DNA damage and decreased total cellular glutathione content, again consistent with an oxidative stress-dependent mechanism of toxicity. Based on differences they observe between toxicity detected by the WST-1 assay and the LDH assay, they suggest that it is important to use multiple toxicity assays in determining NP cytotoxicity.148

In another study, 50-70 nm ZnO NP toxicity in Caco-2 cells was detected via decreased viability (Neutral Red uptake assay) and decreased cell proliferation (colony formation assay).50 Since the Neutral Red assay measures ability of cells to maintain

Neutral Red in lysosomes, requiring ATP to maintain the pH gradient, this suggests that

ZnO may interfere with lysosomal integrity, as has been shown previously. ZnO NP treatment was also found to increase ROS production after 6 hours and increase IL-8 production after 6 and 24 hours, suggesting that ZnO cytotoxicity is being induced through oxidative stress.50 A study comparing differentiated and undifferentiated Caco-2 cells showed less than 10% dissolved Zn in the ZnO suspension, but that population may be important for toxicity. After an in vitro digestion in pH and salt-based gastric and intestinal solutions, they did not see an increase in the soluble fraction of ZnO NP.48 ROS production was found to be induced by ZnO NP but in vitro NP digestion prevented ROS induction.

The WST-1 assay revealed significant toxicity of ZnO NP after both 4 and 24-hour treatments in undifferentiated cells, but toxicity was induced only after 24 hours and at higher concentrations in differentiated cells.48 This suggests a ROS-dependent mechanism of toxicity, but also shows the importance of the differentiation state of the cells as well as the surface chemistry of the NP which was changed by in vitro digestion. ZnO incorporated into food products will be ingested as a component of the food matrix. Treatment of cells with ZnO NP and fatty acids, a common food component, was examined in Caco-2 cells.173

ZnO NP decreased cell proliferation based on incorporation of the thymidine analog bromodeoxyuridine (BrdU) into DNA and cell viability based on MTT and WST-1 assays in a threshold-like pattern at a dose of 32 µg/mL. Addition of palmitic acid (PA) or a free

63 fatty acid (FFA) mixture increased the potency of ZnO in the MTT and WST-1 assays but not the BrdU assay. Also at a dose of 32 µg/mL ZnO, most cells were observed to have detached from the cell culture plate. ZnO NP and PA together induced mitochondrial ROS production while ZnO NP and FFA decreased intracellular ROS production but had no effect on mitochondrial ROS. ZnO NP or FFA were able to destabilize lysosomes based on a decrease in LysoTracker fluorescence but this effect was slightly more pronounced with ZnO and PA or FFA. This study suggests that the interactions of ZnO NP with other food components will add to the complexity of the NP-induced toxicity and necessitates studies with NP in specific food matrices.

Despite all the studies showing ZnO NP toxicity mediated through oxidative stress, there are also some studies in intestinal epithelial cells which did not observe oxidative stress. The study investigating toxicity of NP in drinking water observed toxicity of ZnO

NP to Caco-2 cells at a dose of 100 µg/mL and to SW480 cells at 10 and 100 µg/mL in cell culture medium and toxicity at lower doses with ZnO in synthetic freshwater (10 µg/mL in

Caco-2 cells and 1 µg/mL in SW480 cells). ZnO NP treatment did not induce production of ROS in Caco-2 or SW480 cells, suggesting that toxicity may not be due to oxidative stress, but ZnO NP treatment did increase IL-8 production in Caco-2 cells.49 Treatment of

Caco-2 and RKO cells with ZnO NP revealed toxicity based on a formazan viability assay and the lethal concentration 50 (LC50) for ZnO NP was determined to be 27 µg/cm2 in

RKO cells and 28 µg/cm2 in Caco-2 cells.151 Pretreating with TNF-α to mimic inflammation did not significantly alter ZnO effects on cell viability. ZnO was localized within cells by TEM, but it was difficult to find and seemed to be associated with

64 autophagic responses. This is consistent with lysosomal dissolution of ZnO. Gene expression studies revealed changes in genes responsible for metal metabolism, cellular stress responses, chaperonin proteins, and protein folding, but no indication of a proinflammatory or oxidative stress response.151 Thus, although oxidative stress seems to often be responsible for ZnO NP-induced toxicity, this is likely in conjunction with other mechanisms of toxicity. Regardless, it is clear that ZnO NP induce toxicity in intestinal epithelial cells and thus further studies need to be completed in vivo to investigate toxicity in response to ZnO NP ingestion.

ZnO NP toxicity in vivo

Since in vitro studies have consistently observed significant ZnO NP toxicity, investigating the toxicity of ZnO NP in vivo will help to determine whether ZnO NP ingestion is safe using models more representative of human ingestion. The studies of oral administration in rodents that have been performed so far suggest that the proportion of

ZnO NP absorbed from the intestines may be enough to cause toxic responses in various regions of the body. However, as with most NP that are administered orally, the absorption from the intestines is likely small. A study orally administering hexagonal 40 nm ZnO NP to rats daily for 13 weeks at doses of 134.2, 268.4, and 536.8 mg/kg body weight/day observed decreased body weight in male rats but not female rats at the highest administration dose after 13 weeks. They also observed a dose-dependent increase in zinc concentrations in serum. ICP-MS analysis showed increased Zn in liver and kidneys at the highest dose, but not in spleen and brain. They observed a dose-dependent increase in Zn in urine as well as feces, with significantly more Zn in feces than in any tissue or bodily

65 fluid, suggesting that the majority of ZnO is not being absorbed in the GI tract.167 However, there are indications that ZnO NP do accumulate upon repeated oral administration.

Another study which administered 20 nm ZnO NP to Sprague-Dawley rats at doses of 125,

250, and 500 mg/kg daily by oral gavage for 90 days found increased concentrations of zinc in the plasma in a dose-dependent manner after prolonged NP exposure.174 Although they did not look at tissue accumulation of ZnO NP, this suggests that there is accumulation of orally administered ZnO NP.

Other studies have observed toxicity after ZnO administration. A study administering one bolus of ZnO nano- and micro-sized particles to Sprague-Dawley rats by oral gavage at 5, 50, 300, 1000, and 2000 mg/kg body weight analyzed animals 14 days after administration and found that ZnO NP increased ALT and AST values, indicating liver damage.175 This effect was inversely related to dose, meaning that the greatest toxicity was observed at the lowest ZnO dose. They explain this as a possible effect of greater NP aggregation at higher doses and thus decreased absorption and toxicity. All treated animals showed greater incidence of microlesions in the liver and pancreas and the animals administered nano-ZnO also had lesions in the stomach and heart.175 In another study,

Wistar rats were administered 20 nm ZnO NP by gavage daily at a dose of 100, 200, or 400 mg/kg body weight/day for 14 days. They found that oral administration of ZnO NP increased serum inflammatory markers TNF-α and IL-6 as well as LDH levels in a dose- dependent manner. Serum immunoglobulin G (IgG) was increased at all doses. They found reduced GSH in lung homogenates and also observed lung damage by histology and a dose- dependent increase in lung congestion.176 This study suggests that ZnO is being absorbed

66 after oral administration enough to cause significant lung damage and although this study did not include other tissues, damage will need to be evaluated in other tissues as well.

A study administering 100 nm negatively- (citrate-coated) or positively- (L-serine- coated) charged ZnO NP to Sprague-Dawley rats by daily oral gavage at doses of 31.25 mg/kg, 125 mg/kg, or 500 mg/kg for 90 days observed decreases in blood biochemical parameters including mean cell volume, mean cell hemoglobin, and mean cell hemoglobin concentration as well as total protein and albumin levels which may be due to ZnO-induced anemia. Histopathological changes were observed in the stomach, eye, pancreas, and prostate gland, again showing the potential for ZnO toxicity in vivo.177 Oral administration of 50 nm ZnO NP to mice at a dose of 2.5 g/kg body weight showed increases in zinc concentration beginning 30 minutes after gavage and increased levels of zinc in liver, spleen, and kidney. Increased levels of AST, ALT, and LDH in the serum and histopathological lesions indicative of hepatic swelling and vacuolization in the liver of mice treated with ZnO suggest that ZnO NP are causing liver damage.178 Thus, it is clear that ZnO NP may cause toxicity in vivo and more studies need to be done to determine the toxic potential at doses relevant for human ingestion.

2.9 Silver nanoparticles

Ag NP have recently been used in a multitude of applications spanning a wide number of industries. There are many different ways in which Ag NP can be synthesized.

When Ag NP are synthesized by chemical reduction, protective polymers are often used to stabilize the NP and keep the NP from aggregating.179 This means that different preparations of Ag NP will have different surface properties based on the capping agent

67 used, which may change the behavior of the Ag NP. Thus, characterization of Ag NP samples is important when they are used for research purposes. Ag NP are used for their optical and catalytic properties in many different industries. Another major use of Ag NP, especially in biological applications, is for their antimicrobial properties. This has allowed for use of Ag NP in wound healing and dressings and for anti-inflammatory properties. The food and food packaging industries are also interested in Ag NP for their antimicrobial activity.

Toxicity of Ag NP in intestinal epithelial cells

Although Ag NP have a long history of use in medicine, there is also recent evidence showing that Ag NP can be toxic to mammalian cells at the right doses. With the increasing use of Ag NP in many industries, it is necessary to investigate the effects of environmental exposure to Ag NP. Although there is some disagreement in the literature, it is thought that Ag NP dissolution is required for toxicity which is then mediated by Ag ions.180 The discrepancies observed may stem from the difference between dissolution of

Ag NP outside of cells versus inside cells. In the latter case, Ag NP would still be necessary to cause toxicity in cells.

In many cell models, Ag NP have been shown to mediate toxicity through an oxidative stress-dependent mechanism although Ag NP likely also mediate oxidative stress-independent intracellular effects. In human lung fibroblasts and glioblastoma cells,

Ag NP increased ROS production, induced mitochondrial damage and DNA damage, and caused G2/M phase cell cycle arrest.127 Similar toxicity has been observed in intestinal epithelial cells although fewer studies have been performed in these cells. Peptide-coated

(L-cysteine L-lysine L-lysine) 20 nm Ag NP decreased cell viability, as determined by the

CellTiter-Blue assay, and increased LDH release in Caco-2 cells.129 Impedance measurements showed an initial increase in proliferation over the first 6 hours of Ag NP treatment and then a decrease to a cell index of close to 0. Proliferating cells were much more sensitive to Ag NP than differentiated cells based on viability and cell number assays.

There was also a dose-dependent increase in ROS production, but no detected micronucleus formation.129 Thus, the observed toxicity may be due to induction of oxidative stress. A study that subjected Ag NP to an in vitro digestion in saliva for 5 minutes at pH 6.4, gastric juice for 2 hours at pH 2, and intestinal juice for 2 hours at pH

7.5 observed a slight reduction in Ag NP cytotoxicity after the digestion as detected by impedance measurements. They suggest this is due to an increase in particle aggregation which is hindering Ag ion release. This paper once again shows greater sensitivity of proliferating cells to Ag NP over differentiated cells.21 This also highlights the importance of studying Ag NP toxicity in models representative of in vivo conditions such as digestion.

In a more mechanistic study of the cytotoxicity of Ag NP on Caco-2 cells, Ag NPs were shown to be internalized by cells and a dose-dependent decrease in cell viability was observed by MTT and trypan blue exclusion assays. No increase in ROS production was detected by the 2’,7’-dichlorofluorescein or Mitotracker assays, but a decrease in total cellular glutathione levels and depolarization of the mitochondrial membrane potential indicate oxidative stress. Ag NP treatment also induced activation of the stress-responsive gene Nrf2 and the expression of heme oxygenase-1 (HO-1) which is cytoprotective and controlled by Nrf2.67 Pretreatment of cells with the antioxidant N-acetylcysteine was able

69 to attenuate Nrf2 and HO-1 mRNA expression, demonstrating that they are controlled through an oxidative stress-dependent pathway. They observed in cell culture media that

Ag ions made up 2.18% of the solution at 0 hours which increased to 3.47% after 24 hours, and they suggested that Ag ions mediate much of the Ag NP toxicity through Ag ion release either outside of or within cells after internalization of Ag NP.67

Not all studies have demonstrated oxidative stress in response to Ag NP treatment.

A study interested in NP use in paints observed dose-dependent toxicity of nanosilver to

Caco-2 cells resulting in decreased viability and activity of cells and increased cell necrosis. However, no induction of ROS production was observed after Ag NP treatment for 4 hours.52 This may be due to limited sensitivity of the ROS release assay to detect oxidative stress or may suggest that the observed toxicity is due to an oxidative stress- independent mechanism. Several studies have also shown limited Ag NP toxicity which may be due to differences in the Ag NP formulations or doses used. A study interested in

NP in drinking water observed a trend of decreased viability of Caco-2 cells after 24-hour treatment with 100 µg/mL Ag NP in cell culture media and buffered synthetic freshwater but it was not significantly different than controls. That trend was not apparent in SW480 cells treated with Ag NP in cell culture media, but there was significant toxicity of Ag NP in buffered synthetic freshwater.49 Ag NP treatment did not induce ROS production in either cell type, but did induce significant IL-8 production in Caco-2 cells only. This may again indicate an oxidative stress-independent mechanism of toxicity.

Another study showed Ag NP (35 nm) uptake by Caco-2 cells and transport of these

NP across the epithelial layer using confocal microscopy (detecting Ag particles by their

70 reflectance). However, there did not appear to be toxicity of Ag NP at 31.25 µg/cm2 (50

µg/mL).181 A study in the M cell coculture model observed minimal decrease in viability of cells at doses up to 50 µg/mL using 20, 34, 61, and 113 nm Ag NP. They then treated cells with 5 and 25 µg/mL doses of the different Ag NP sizes for 4 hours and analyzed whole-genome mRNA expression. From this they observed changes in genes involved in the oxidative stress response, endoplasmic stress response, and apoptosis. When treating cells with Ag ions only, they saw a similar gene expression response and thus concluded that Ag NP toxicity is likely mediated through ion release.95 Another study in Caco-2 cells observed a Ag NP-mediated decrease in cell activity as measured by the CCK-8 assay (to about 60-70% of control with multiple doses of Ag NP), but no accompanying induction of cell death. They also found no ROS production or change in SOD levels, but they did observe a slight increase in total GSH and thus antioxidant capacity in cells. Thus, the induction of oxidative stress in cells by these Ag NP is minor.171 Another study in Caco-2 cells detected a decrease in cell viability with the Alamar Blue reduction assay upon treatment with Ag NP. They also measured decreased dsDNA content indicating decreased cell proliferation and increased DNA damage induced by Ag NP treatment. No ROS production was detected, but they did observe a decrease in mitochondrial membrane potential indicating mitochondrial injury.68 They conclude that the toxicity is a result of mitochondrial membrane damage leading to mitochondrial dysfunction in cells rather than oxidative stress-dependent.

A study in the M cell co-culture model where Raji B cells are cultured under the

Caco-2 monolayer investigated toxicity of Ag NP administered with several phenolic

71 compounds that may be found in food matrices. Ag NPs alone were found to be cytotoxic to this cell model and induce oxidative stress. Decreased TEER measurements after Ag NP treatment and increased transport of Lucifer Yellow across the epithelial layer indicate disruption of the epithelial barrier. Tight junction disruption was indicated by changes in occluden and ZO-1 localization. Co-administration of the phenolic compounds quercetin and kaempferol were partially protective against Ag NP toxicity but resveratrol was not protective.130 This once again suggests that NP toxicity will be more complicated in the presence of the food matrix. A study using a 3D coculture transwell model with Caco-2 cells cultured on top of THP-1 macrophages and MUTZ-3 dendritic cells found dose- dependent PVP-capped <20 nm Ag NP toxicity based on LDH release and decreased TEER values as well as IL-8 production induced by Ag NPs.51 The Caco-2 monolayer was more sensitive to Ag NP than the co-culture. However, co-cultures treated with IL-1β to represent inflamed conditions were more sensitive to Ag NP than non-inflamed cells.51

This study suggests that the cell microenvironment will be very important for Ag NP toxicity and highlights the need for in vivo studies to more completely investigate Ag NP toxicity.

In vivo studies

Multiple in vivo studies have orally administered Ag NP to rodents or other animal models in order to further investigate Ag NP toxicity. It has been shown that some Ag NP are absorbed through the intestines and are disseminated throughout the body.182-186 It was determined from a case study in a woman suffering from argyria that 18% of an orally administered dose of Ag was retained187 and this has been used to estimate Ag NP

72 absorption in humans.188 Some studies have also shown Ag NP-induced intestinal epithelial damage after oral administration. In one study, mice were administered 3-20 nm Ag NP at doses of 5, 10, 15, and 20 mg/kg body weight for 21 days and Ag NP were shown to decrease body weight at all doses. The Ag NP also damaged intestinal epithelial cell microvilli and intestinal glands and increased numbers of inflammatory cells were observed in the lamina propria underlying the epithelium.189 They suggest that the damage to microvilli may have impaired nutrient absorption, leading to the decreased body weight.189 Another study administering 60 nm Ag NP to Sprague-Dawley rats at doses of

30, 300, and 1000 mg/kg for 28 days found that there was a dose-dependent accumulation of Ag NP in several regions throughout the intestines including the lamina propria of the small and large intestines. Ag NP treatment also caused greater numbers of goblet cells to release their mucus and there was a change in mucus composition. Greater cell shedding from the tips of the villi was also observed.190 Thus, damage to the intestines may be of particular concern after Ag NP ingestion.

Other studies have found damage to organs including liver after oral administration of Ag NP. Oral administration of 56 nm Ag NP to F344 rats at doses of 30, 125, and 500 mg/kg for 13 weeks induced a mild decrease in body weight of male rats over controls after

4 weeks and dose-dependent accumulation of silver in testes, liver, kidneys, brain, lungs, and blood (all tissues examined). Dose-dependent increases in alkaline phosphatase and cholesterol levels for male and female rats at the highest dose and a greater incidence of bile-duct hyperplasia were observed after Ag NP treatment.191 Thus, it appears that Ag NP may induce liver toxicity. Another study orally administering 22, 42, and 71 nm Ag NP to

73 mice at 1 mg/kg for 14 days found distribution of Ag NP to the brain, lung, liver, kidney, and testis. They also found increased levels of TGF-β in serum and increased B cell distribution after Ag NP treatment, indicating an inflammatory response. Mild liver and kidney toxicity as determined by increases in ALP and AST in the serum and mild cell infiltration in the kidney cortex were observed after 28-day oral administration at 1 mg/kg

Ag NP. This was accompanied by increases in IL-1, IL-6, IL-4, IL-10, IL-12, and TGF-β levels.186 Thus, Ag NP oral administration may lead to liver and kidney toxicity as well as inflammation.

Another concern is whether Ag NP will be transferred across the placenta and could be toxic to the growing fetus. This was examined in a study of pregnant rats. Females were orally administered 7.9 nm citrate-coated Ag NP at 250 mg/kg beginning 14 days before mating through the mating and gestation period until 4 days after birth. Silver concentration was determined in pups at 4 days after birth and found to significantly accumulate in liver, kidney, lung, and brain tissue.192 Thus, further experiments are necessary to determine whether this passage of Ag NP to offspring will cause toxicity.

Since Ag NP have antimicrobial activity, ingestion of Ag NP may also change the composition of the intestinal microflora. In order to investigate this question, Ag NP were added to drinking water at doses of 5, 15, and 25 mg/kg given to quail as a model animal for poultry over 12 days. They found that there were not major changes in quail microflora, but the highest dose of Ag NP did increase the population of lactic acid bacteria in the cecum.193 Further experiments need to be done to confirm whether Ag NP can alter gut microflora, especially over longer exposures.

From this base of literature, it is clear that Ag NP ingestion may cause intestinal damage or liver and kidney damage in vivo. Further studies are needed to determine the potential toxicity at relevant doses for human exposure.

Anti-inflammatory properties of Ag NP

While many studies in cell models have shown that Ag NP are toxic, this is contrary to the studies showing anti-inflammatory properties of Ag NP in wound healing and for other medical applications. Ag NP produced by Nucryst Pharmaceuticals (NPI 32101) have been shown to be anti-inflammatory in a cream for allergic contact dermatitis tested in mouse and guinea pig models.194, 195 These NP were also tested in a rat model of ulcerative colitis and found to have anti-inflammatory effects (at a dose of 40 mg/kg orally) and reduce IBD symptoms more than a 100 mg/kg dose of the commonly used anti- inflammatory drug sulfasalazine.196 The vastly different responses to Ag NP may stem from different responses based on cell type. This has been reported in vitro in a study comparing macrophages and fibroblasts.197 The dose of Ag NP treatment may also play a role. Low levels of oxidative stress induced by Ag NP may stimulate cell proliferation and differentiation which may be important for wound healing. However, oxidative stress is often associated with inflammation and further research is needed to understand the mechanisms by which Ag NP can be both pro-inflammatory and anti-inflammatory.

2.10 Summary

From the reviewed literature, it is clear that NP are ingested in the human diet.

Although NP largely seem to mediate toxicity through oxidative stress-dependent mechanisms in vitro, it is clear that the complexities of the digestive system and the food

75 matrix in which NP are incorporated will have significant effects on NP toxicity in vivo.

Studies with well-characterized NP are needed to assess the role of physicochemical characteristics of NP in determining NP toxicity. Further work is also needed to allow for means to effectively determine NP accumulation and toxicity in vivo. This dissertation work attempts to address these remaining questions.

Chapter 3: Minimal intestinal epithelial cell toxicity in response to short- and

long-term food-relevant inorganic nanoparticle exposure1

3.1 Introduction

The food industry has identified many potential applications for nanotechnology.

Examples include nanoencapsulation to increase bioavailability of nutrients and provide controlled release, flavor enhancers, sensors using nanotechnology in food packaging, and nanoparticles (NP) with antimicrobial properties.199-201 Many of these applications are still being researched and may eventually find their way to the food products on the market.

The Project on Emerging Nanotechnologies has compiled and periodically updates an inventory of consumer products containing nanoscale components, including products in the “Food and Beverage” category.3 Despite the increased interest in using NP in food, the

FDA currently has no specific regulations related to NP. A 2012 draft guidance suggested that food manufacturers investigate the safety of foods incorporating NP, but as of now, many of the inorganic NP are generally regarded as safe for use in foods.202

Silica (SiO2) is used as an anti-caking agent and to clarify liquids, while titania

(TiO2) is commonly used as a whitening agent. The average size of food-grade TiO2 (E171)

1 Previously published in modified form.198 Text and figures appear with permission from the editors of Chemical Research in Toxicology. 77 and SiO2 (E551) particles is a few hundred nanometers (nm), but both powders have broad size distributions. In an analysis of 89 consumer food products for TiO2 content, 36% of

2 E171 particles were found to have one dimension less than 100 nm. Based on the TiO2

2 detected in the tested food products, Weir et al. estimated dietary exposure to TiO2 for the

US population to be 0.2-0.7 mg TiO2/kg body weight/day for consumers over the age of

10 and 1-2 mg TiO2/kg body weight/day for children under age 10 due to greater consumption of candies and sweets that tend to contain TiO2. Food-grade SiO2 was found to contain up to 33% silica with a size between 10-200 nm,1 some of which would be considered NP according to the conventional definition of a NP as less than 100 nm. A study assessing the silica content of various foods estimated a “worst-case” intake of 1.8 mg nanosilica/kg body weight/day for the average adult (9.4 mg total silica/kg body weight/day).1 These studies reveal that consumers are currently being exposed to nanosized fractions of both food-grade TiO2 and SiO2. However, individual exposure will vary greatly with diet.

As orally ingested NP traverse the digestive tract through the mouth, stomach, and intestines, they will be exposed to a number of digestive enzymes. The epithelial cells lining the small intestine absorb most nutrients obtained from food by endocytosis or diffusion and transport them across the epithelium where they can enter the bloodstream.

As NP move through the small intestine, they will come into contact with intestinal epithelial cells, which may be able to internalize NP as they take up nutrients.

Previous in vitro and in vivo research has shown that NP do cross the intestinal epithelium, apparently by transcytosis, i.e. transport through the enterocytes.4, 66, 203 Several

78 studies have begun to investigate the toxicity of inorganic NP to intestinal epithelial cells both in vitro66 and in vivo.168, 169, 175, 204 Based on the toxicity studies that have been conducted in different systems, it is apparent that toxicity depends upon many different properties of NP. Particle size influences toxicity, with NP showing greater toxicity than their larger counterparts.5 Smaller particles have also been shown to be more readily internalized by intestinal epithelial cells.4, 5, 203 Recently, NP characteristics other than size, including composition, solubility, crystal structure, surface charge, surface modifications, and shape have been recognized to substantially influence NP toxicity.119, 205-207

Here we investigated the internalization of commercially available nanoparticulate

SiO2, TiO2, and ZnO, and examined their toxicity on C2BBe1 intestinal epithelial cells using several assays. To simulate the in vivo environment before interaction with the intestinal epithelium, NP were incubated with representative gastric and small intestinal digestive enzyme solutions before cell exposure. Long-term exposure during which cells were repeatedly treated with NP was also conducted. Data generated by these studies, the first to include long-term exposure in an in vitro intestinal epithelial cell model, suggest that SiO2 and TiO2 exposed to culture media do not cause significant toxicity in vitro in their interactions with intestinal epithelial cells, whereas ZnO exhibits mild toxicity.

However, TiO2 treated with simulated digestion media do exhibit mild toxicity.

3.2 Experimental Procedures

All NP characterization (dynamic light scattering and zeta potential measurements,

TEM analysis, infrared spectroscopy, X-ray diffraction, and atomic absorption spectroscopy) and simulated gastrointestinal digestion of NP was performed by Dr.

Andrew Zane in Dr. Prabir Dutta’s laboratory in The Ohio State University Department of

Chemistry and Biochemistry.

Nanoparticles

Zinc oxide, titania, and silica NP were purchased from Sigma-Aldrich (St. Louis,

MO). The specifications of the ZnO particles were size ≤ 100 nm, with a specific surface

2 area of 15-25 m /g. For TiO2, the particle size was specified as 21 nm, with surface area of

35-65 m2/g, and purity of ≥ 99.5%, with a trade name of Aeroxide P25. Silica particle specification included size of 12 nm, with surface area of 175-225 m2/g, and purity of

99.8%.

Nanoparticle characterization

Particle size and surface charge

A Zetasizer Nano ZS (Malvern, Westborough, MA) was used to determine the size and zeta potential of the particles. The Nano ZS uses a 633 nm laser as its light source. For size measurements, a 173° backscatter angle was used for collecting scattered light. The instrument was set to automatically determine the number of runs, the run duration, and the optimal focal point for each sample. The analysis model was set to general purpose, with default size limits of 0.4-10,000 nm in diameter. Three replicate measurements were taken for all samples and averaged. All particle sizes are given as radii. For zeta potential measurements, a forward angle of 12° was used for collecting light. The default

Smoluchowski model in the software program was used. Each measurement included 20 runs and monomodal analysis provided by the vendor was used to analyze. Three replicate measurements were taken for all samples and averaged. Samples were titrated versus pH

80 using an attached MPT-2 Autotitrator. The titrator was supplied with 0.1 M HCl and 1.0

M HCl. Three replicate measurements were taken at each pH with a two minute pause between all measurements.

Transmission electron microscopy (TEM) images of commercial nanoparticles

Images were taken using a Tecnai F20 Transmission Electron Microscope. Ten mg/mL solutions of each particle in ethanol were prepared and sonicated for 30 minutes.

The solutions were dropped onto lacey carbon copper TEM grids (Ted Pella, Inc., Redding,

CA) and allowed to dry for several hours.

Infrared spectroscopy of treated particles

Diffuse reflectance infrared Fourier transform spectroscopy (DRIFTS) was performed on particles which had undergone simulated digestion treatment. After the final digestion step, the particles were washed twice with water. The digested particles were isolated by centrifugation and frozen with liquid nitrogen, then placed in a Millrock Bench

Top Manifold Freeze Dryer (Millrock, Inc. Kingston, NY) to preserve any potential protein coating. The digestion protocol was performed on another set of particles using water in place of digestive enzymes to use as reference. DRIFTS analysis was performed with a

Spectrum 400 FTIR Imaging System (Perkin Elmer, Waltham, MA). Scans were performed in percent reflectance mode, an air background (mirror in the diffuse reflectance cell) was collected, and then the water and digested samples (50 mg) were sequentially added to the cell (undiluted), and the spectra recorded. Kubelka-Munk analysis was done on these two samples using the air spectrum as background with a program developed in- house. The Kubelka-Munk spectrum obtained from the water-exposed sample was

81 subtracted from the digested sample spectrum to analyze the coating on the particles. The

Kubelka-Munk spectrum of each water-exposed particle was subtracted from the spectrum of each digested particle using Origin software, leaving only peaks related to the digestion

-1 enzymes and media (normalized relative to the 1260 cm band observed in both TiO2 and

- ZnO, and for SiO2, the subtraction was done to maximize/symmetrize the band at 1300 cm

1 ). An example of the subtraction process for TiO2 is shown in Figure 3.5B.

X-ray diffraction

X-Ray diffraction (XRD) analysis of the commercial particles was performed using a Rigaku Geigerflex diffractometer. Particles were packed onto glass slides with no further preparation. The diffraction angle was scanned from 0 to 100o (2θ) with a 0.2 step size and

10 second dwell time.

Atomic absorption spectroscopy

Atomic absorption spectroscopy was performed to test for the presence of dissolved

Zn ions after exposure of ZnO particles to the acidic pepsin enzyme solution. A Buck

Scientific Accusys 211 spectrophotometer (East Norwalk, CT) was used with a Zn hollow cathode lamp (Heraeus). Standard solutions were diluted from 1000 ppm standard (Buck

Scientific). The digestion protocol was performed on ZnO particles and stopped after exposure to pepsin solution. Following this, the solution was centrifuged at 209,000×g for

30 minutes. The supernatant was removed and analyzed for Zn ion concentration.

Simulated gastrointestinal nanoparticle digestion

Pepsin, pancreatin, and bile salts were used to simulate the gastric and small intestinal digestive environments in vitro. The concentrations used were based on in vitro

82 digestion methods used in previously published studies.208-211 The stomach enzyme pepsin

(146 U/mL, Sigma-Aldrich) was dissolved in water (pH 2, adjusted with 1N HCl). The small intestinal enzyme mixture pancreatin (2mg/mL, Sigma-Aldrich) was dissolved in water (adjusted to pH 7 with 1M NaHCO3 and HCl). A solution of intestinal bile extract

(porcine, Sigma-Aldrich) was made at a concentration of 0.024mg/mL in water (adjusted to pH 7 with 1M NaHCO3 and HCl).

To simulate the digestive process, NP (50 mg/L) were incubated first in the pepsin solution for one hour (37oC). The NP were pelleted by centrifugation (209,000×g for 30 minutes) and resuspended in the pancreatin solution. After a one-hour digestion (37oC), the particles were pelleted and resuspended in bile extract (one hour, 37°C). NP were centrifuged, resuspended in phosphate-buffered saline (PBS), and used in biological experiments.

Cell culture

C2BBe1 cells were obtained from the American Type Culture Collection

(Manassas, VA) at passage 47 (as specified by the vendor). For short-term exposure assays, cells were used between passages 55-70, consistent with previous studies.37, 212, 213

Cells were cultured in Dulbecco’s Modified Eagle Medium (DMEM; Life

Technologies, Grand Island, NY) supplemented with 10% fetal bovine serum (FBS; Serum

Source International, Inc., Charlotte, NC), 1 mM sodium pyruvate, 2 mM L-glutamine,

0.3% penicillin/streptomycin, 0.3 ug/mL Amphotericin B (fungizone), and 10 µg/mL transferrin (all from Life Technologies). Cells were incubated in 10% CO2/90% room air at 37°C. Cells were passaged every 5-7 days and plated on flasks or plates pre-coated with

83 collagen I (0.05 mg/mL, rat tail, Life Technologies). Cells were incubated for at least 24 hours after plating prior to treatment with NP.

Treatment of cells with nanoparticles

The NP were weighed and placed in a glass vial prior to steam-sterilization. Sterile

NP were suspended in PBS to make a 1 mg/mL solution. Immediately prior to treating cells, NP solutions were sonicated using a VC130 Sonics Vibra-Cell sonicator (Sonic

Materials, Inc. Norwalk, CT) pulsing for one second on, one second off for approximately

15 seconds in order to break up NP agglomerates before exposing them to cells. NP were added to cells at a dose of 10 µg/cm2 (100 µL of the 1 mg/mL suspension per well of a 6- well plate in 2 mL total volume per well, or 20 µL per well in a 24-well plate in 0.5 mL total volume per well). For the assays described below, cells were seeded in 24-well tissue culture plates at a density of 2×104-1×105 cells/well and grown to 75-100% confluency. To promote contact of NP with cells, plates were centrifuged at 300×g for 15 minutes immediately following NP treatment. Cells were incubated with NP for 24 hours at 37°C at 10% CO2/90% ambient air, and then washed twice with PBS. Toxicity assays were performed immediately after washing. For long-term NP exposure studies, culture media was added and cells were incubated at 37°C at 10% CO2/90% air.

Long-term nanoparticle exposure studies

For long-term exposure studies, one exposure cycle consisted of plating cells in a

6-well tissue culture plate (Corning-Costar, Tewksbury, MA) pre-coated with collagen.

After two days to allow the cells to firmly attach, cells were treated with 10 g/cm2 NP and centrifuged to promote NP-cell contact. After a 24-hour incubation at 37°C, cells were

84 washed with PBS and media was replaced. Cells were cultured 4-6 days to reach confluency. Cells were then passed 1:10 into new 6-well plates and the NP treatment was repeated after each passage. This was continued for 84 NP exposure cycles. To perform toxicity assays, a portion of these long-term cultured cells were plated into 24-well tissue culture plates (Corning-Costar). Cells were treated with NP (10 g/cm2) for 24 hours prior to each assay, as in single exposure studies. Long-term exposure studies were only conducted with commercially-obtained NP (i.e., no enzymatic digestion).

Cell morphology

Digital images of cell morphology after treatment with NP were taken using a

Nikon Coolpix 4500 digital camera (Nikon, Chiyoda, Tokyo) attached to an Olympus CK2 microscope (Olympus, Tokyo, Japan) by a Leitz adapter photo tube (Leica, Wetzlar,

Germany).

Confocal microscopy was performed to assess the formation of intercellular junctions. For confocal experiments, cells were plated on coverslips or 8-chamber slides and grown to confluence. After 24-hour treatment of cells with NP, cells were fixed in 4% paraformaldehyde for 10-45 minutes. Washes in PBS were performed between staining steps and all staining was performed at room temperature. Cells were permeabilized in a

0.2% Triton X-100 (Sigma-Aldrich) solution in PBS for 15 minutes. Cells were incubated in a 1% bovine serum albumin (BSA; Sigma-Aldrich) blocking solution for 1 hour. To look at adherens junction structure, cells were incubated with a 1:250 dilution of Alexa Fluor

647 mouse anti-E-cadherin (BD Biosciences) in 1% BSA in PBS for 90 minutes. Cells were stained with a 0.25 µg/mL solution of 4’,6’-diamidino-2-phenylindole (DAPI; Life

Technologies) in 1% BSA in PBS for 10 minutes. To look at tight junction structure, cells were incubated with a 1:200 dilution of mouse anti-ZO-1 (Life Technologies) for 12 hours as the primary antibody and subsequently incubated with a 1:100 dilution of an anti-mouse

AlexaFlour 488-conjugated secondary antibody for 1 hour. Coverslips were mounted using

ProLong Gold Antifade Reagent (Life Technologies). Mounting media was allowed to cure overnight at room temperature before analysis using a confocal fluorescence microscope.

Assay of cellular growth

To assess the effects of long-term NP exposure on treated cells, we assessed their proliferation over a 10 day period. The cells were seeded at a density such that they reach confluence after 10 days. Cells were plated in triplicate into 6-well tissue culture plates at a density of 1×105 or 5×104 cells/well. The time at which cells were plated was considered time zero. Cells were counted daily using a Z2 Beckman Coulter Counter (Indianapolis,

IN) and mean counts of triplicate wells were plotted versus time.

Toxicity assays

Sytox Red staining

To evaluate necrotic cell death (cellular membrane damage) in NP-treated cells, a flow cytometric analysis was performed using Sytox Red Dead Cell Stain (Life

Technologies). Cells were treated with NP and briefly centrifuged. After a 24-hour incubation at 37°C, cells were washed twice with PBS and detached from the culture plate with trypsin. Each cell sample was suspended in 1 mL calcium- and magnesium-free Hanks

Balanced Salt Solution (Life Technologies) and stained with 1 µL of the Sytox Red Dead

Cell Stain. Samples were incubated for at least 15 minutes at room temperature before

86 analyzing for fluorescence using a FACScalibur flow cytometer (BD Biosciences) at an excitation wavelength of 635 nm. All experimental conditions were performed in triplicate.

As a positive control for cell death, a subset of wells were treated with 10 or 20 mM hydrogen peroxide (H2O2) in DMEM for one hour. Culture medium was replaced and cells were incubated for 23 hours prior to staining and flow cytometric analysis.

Annexin V-FITC staining

To examine whether cells treated with NP underwent apoptosis, cells were stained with fluorescein isothiocyanate (FITC)-conjugated Annexin V (BD Biosciences, San Jose,

CA). Cells were plated, treated with NP, and harvested with trypsin as described above.

Cells were resuspended in 100 µL of Annexin binding buffer (Life Technologies). Five µL of FITC Annexin V was added to each tube and the samples were incubated for 15 minutes in the dark at room temperature before adding an additional 400 µL of Annexin binding buffer. Fluorescence of bound FITC Annexin V was then detected by flow cytometry, excitation wavelength of 488 nm. All experimental conditions were performed in triplicate.

As a positive control for apoptosis, a subset of wells were treated with 10 or 20 mM H2O2 in DMEM for one hour. Culture medium was replaced and cells were incubated for 23 hours prior to flow cytometric staining and analysis.

LDH assay

To further assess cell death by detecting cellular membrane damage, release of the cytosolic enzyme lactate dehydrogenase (LDH) into the culture medium was examined by colorimetric assay. Cells were plated and treated with NP as described above. After a 24- hour incubation, 50 µL of culture media was collected from each well and placed in a flat-

87 bottom 96-well plate (BD Falcon). LDH activity was assessed using a commercially available kit (LDH Assay Kit, Sigma-Aldrich) according to manufacturer’s instructions.

Briefly, LDH Assay Substrate Solution, LDH Assay Cofactor Preparation, and LDH Assay

Lysis Solution were added to samples and incubated at room temperature for 15-30 minutes. The absorbance was read on a microplate reader at 490 nm and 690 nm. Sample absorbance values were corrected by subtracting the background reading at 690 nm from the 490 nm reading. As a positive control for LDH release, cells were treated with a 1% solution of Triton X-100 (Sigma-Aldrich) in DMEM for 5 minutes to lyse all cells before collecting the supernatants. Sample absorbance values were normalized to the Triton positive control (considered as 0% cell viability) and untreated negative control

(considered as 100% cell viability). All experimental conditions were performed in quadruplicate.

MTT assay

To measure the metabolic activity of NP-treated cells, we used a commercially available MTT assay kit (Cayman Chemical, Ann Arbor, MI). In this assay, the tetrazolium dye, MTT, is reduced by mitochondrial NAD(P)H-dependent oxidoreductase enzymes to an insoluble purple crystal, formazan. Cells were plated and treated with NP as described above and incubated for 24 hours at 37°C prior to the assay. As a positive control for complete cell death, cells were treated with a 1% solution of Triton X-100 in DMEM for five minutes immediately before the assay was performed. After the incubation, all but 150

µL of the cell media was removed from each culture well before the addition of 15 µL of the MTT reagent. The cells were incubated for 3-4 hours at 37°C. The culture media/MTT

88 reagent was removed and formazan crystals were dissolved with the provided MTT solvent. The solution was transferred to a flat-bottom 96-well plate and absorbance values were read using a microplate reader at 570 nm and 690 nm. Absorbance values were corrected by subtracting the background absorbance reading at 690 nm from the absorbance at 570 nm. To analyze and compare the data, absorbance values were normalized to the untreated control cells such that the untreated control represented 100% mitochondrial activity and an absorbance of 0 represented the absence of mitochondrial activity (i.e., 0%). All experimental conditions were performed in quadruplicate.

Enzyme-linked immunosorbent assay (ELISA)

Secretion of IL-8 by NP-treated cells was measured using a human IL-8

SABiosciences Single Analyte ELISArray Kit (QIAGEN, Valencia, CA). The assay was performed using supernatants from cells that were then used for Sytox Red and FITC

Annexin V flow cytometry-based toxicity experiments. Thus, cells were plated and treated with NP as described previously. After a 24-hour incubation with NP, supernatants were collected, any dead cells removed by centrifugation, and samples were frozen at -80°C until the assay could be performed. The assay was performed according to the manufacturer’s instructions. Briefly, a serial dilution of standards was prepared using the antigen standards provided. The measurements of these standards were used to create a standard curve in order to calculate the concentration of IL-8 in the cell supernatants. Fifty

µL of each standard or supernatant sample was added to wells of the provided ELISArray plate in duplicate and incubated for two hours. All incubations were performed at room temperature. Wells were washed three times with Wash Buffer before adding 100 µL of

Detection Antibody to each well and incubating for one hour. Wells were again washed three times with Wash Buffer, 100 µL of Avidin-HRP was added to each well, and the plate was incubated for 30 minutes in the dark. Wells were washed four times with Wash

Buffer, 100 µL of Development Solution was added to each well, and the plate was incubated for 15 minutes in the dark. Finally, 100 µL of Stop Solution was added to each well and the absorbance was read at 450 nm and 570 nm. The absorbance readings at 570 nm were subtracted from the absorbance at 450 nm to correct for background. The standard curve was plotted as absorbance vs. concentration and the best fit linear curve was calculated for the linear portion of the curve. The equation for this line was used to calculate the concentration of IL-8 in the cell supernatants.

TEM of nanoparticle-treated cells

To verify that treated cells internalized NP, TEM was performed. Cells were grown to near confluence in a 6-well tissue culture plate. Cells were then treated with NP and briefly centrifuged. After a 24-hour incubation at 37°C, the treated cells were washed twice with PBS before detachment with trypsin. Detached cells were washed with PBS and resuspended in 3% glutaraldehyde in PBS. Incubation steps were carried out at room temperature on a Lab Line orbital shaker (Barnstead/Thermolyne, Melrose Park, IL) operating at 700 rpm. The cell suspension was centrifuged at 1000×g for 5 minutes between each processing step. Fixed cells were washed twice with sodium cacodylate buffer (pH

7.4, 10 minutes each), then post-fixed in 1% osmium tetroxide in sym-collidine buffer (pH

7.6) for 1 hour at room temperature. Following two washes with s-collidine buffer (10 minutes each) the cells were en-bloc stained with a saturated aqueous uranyl acetate

90 solution (pH 3.3) for 1 hour. Cells were dehydrated in a graded ethanol series up to absolute

(10 minutes each). Acetone was used as the transitional solvent for two 10-minute washes.

Cells were infiltrated overnight with a 1:1 mixture of acetone and Spurr’s epoxy resin

(Electron Microscopy Sciences, Fort Washington, PA). Finally, the cells were centrifuged and the pellet was placed into BEEM embedding capsules containing 100% Spurr’s resin.

Polymerization of epoxy blocks was carried out at 70°C overnight. Polymerized blocks were sectioned with a Leica Ultracut UCT ultramicrotome (Leica Microsystem GmbH,

Wein, Austria). Ultrathin (80 nm) sections were collected on 200 mesh copper grids

(Electron Microscopy Sciences) and post-stained with lead citrate (3 minutes). Electron micrographs were generated with a JEOL JEM-1400 TEM (JEOL Ltd. Tokyo, Japan) equipped with a Veleta digital camera (Olympus Soft Imaging Solutions GmbH, Műnster,

Germany).

3.3 Results

Nanoparticle characteristics

The toxicity experiments were conducted with readily available commercial TiO2,

SiO2, and ZnO NP. Particle specifications provided by the vendor appear above (3.2

Experimental Procedures, Nanoparticles) and results of characterization studies conducted in our laboratory are as follows. Figure 3.1A-C show the TEM images of the three particles; the primary particles of silica and titania are almost spherical with radii of ~10-

15 nm for SiO2 and ~20-25 nm for TiO2. For ZnO, there is a polydispersity in size and morphology with both spherical (~10-20 nm) and rod-like (5-10×50-200 nm) particles.

There is clearly some aggregation of these primary particles, and we discuss below the size

Support

Figure 3.1. Representative TEM images of commercial nanoparticles.

A) SiO2, B) TiO2, and C) ZnO. Long strands seen in the images are from the carbon support films, as indicated by arrows.

based on dynamic light scattering, which is used in this study rather than the primary size.

Figure 3.2 shows the X-ray powder diffraction patterns, indicating that the SiO2 is

92 amorphous, TiO2 is a mixture of anatase and rutile (marked on the figure, as is expected for P25), and ZnO is primarily wurtzite.

To simulate the digestion process that the particles would undergo in the gastrointestinal tract, particles were sequentially treated with pepsin at pH 2 to mimic the stomach, and pancreatin and bile salts, both at pH 7, to mimic the small intestine. Particle characterization by light scattering and zeta potential measurements at each of these stages was measured and is summarized in Table 3.1. In the culture medium DMEM, all NP were negatively charged, with zeta potentials of -11 millivolt (mV) for TiO2 and SiO2, and -19 mV for ZnO. The radii varied with ~96 nm for silica, ~112 nm for ZnO, and ~115 nm for titania. Titania particles remained stable in pepsin solution at approximately 119 nm radii, but with a positive zeta potential of 14 mV. At the acidic pH (~2) of the pepsin solution, the surface of TiO2 is protonated, consistent with the isoelectric point (IEP) of TiO2 being pH ~6.214 Silica particles had surface charge close to zero in the pepsin solution and aggregated to ~3000 nm particle radii, consistent with their IEP reported as pH 2,215 and with our pH titration studies. ZnO particles dissolved during incubation in pepsin at pH 2.

This is consistent with reports in the literature216, 217 and was confirmed by measuring the dissolved Zn in the supernatant by atomic absorption spectroscopy, and the dissolved zinc correlated with the amount of NP added to the medium. Figure 3.3A is a plot of the size and zeta potential as a function of pH for ZnO in water and in the presence of media. In water, the ZnO particles are not tracked by light scattering below pH 6.5, indicating dissolution, whereas in the media there is a protective effect on the ZnO, and particle size only begins to decrease below pH 4. Dissolution studies were not done on the SiO2 and

A 2000

1000

Intensity

0 50 100 2

C 10000

5000

Intensity

0 50 100 2 Figure 3.2. X-Ray diffraction patterns of commercial nanoparticles.

A) SiO2, B) TiO2 (anatase and rutile peaks are marked by “A” and “R,” respectively), and C) ZnO.

SiO2 TiO2 ZnO

Size Zeta Size Zeta Size Zeta (r, nm) Potential (r, nm) Potential (r, nm) Potential (mV) (mV) (mV)

DMEM/FBS 96 ± 4 -11 ± 0 115 ± 2 -11 ± 1 112 ± 2 -19.3 ± 1

Pepsin 3039 ± +1 ± 2 119 ± 2 +14 ± 2 - - (pH 2) 321

Pancreatin 109 ± 1 -38 ± 2 110 ± 0 -39 ± 2 - - (pH 7)

Bile Salts 109 ± 4 -29 ± 2 104 ± 2 -35 ± 1 - - (pH 7)

Table 3.1. Hydrodynamic radii and zeta potential of nanoparticles. Summary of dynamic light scattering data showing hydrodynamic radius and zeta potential of commercial NP in cell culture medium (DMEM/FBS) and during each step of a simulated digestion process.

TiO2 particles, since these particles could be isolated after the various treatments and were used for the in vitro studies. Figure 3.3B shows the difference infrared spectrum between

ZnO that was exposed to media and that which was exposed to water (both samples lyophilized prior to infrared measurements as detailed in Experimental Procedures). Weak bands at 1661, 1534, 1444, 1395 cm-1 are characteristic of adsorbed protein, with the

Amide 1 band at 1661 cm-1 suggesting the presence of β-sheet structures.218 Similar bands were also noted for SiO2 and TiO2 exposed to media (Figure 3.4), indicating that protein adsorption is occurring on all particles exposed to media.

Upon treatment with pancreatin and bile salts, the zeta potentials of the TiO2 and

SiO2 reversed and became strongly negative, in the range of -30 to -40 mV. TiO2 particles retained their radii of ~ 104 nm, whereas SiO2 deagglomerated and had radii of ~ 109 nm.

DRIFTS Spectra of DMEM Exposed ZnO B 1661 1534 1444 1395

Arbitrary KubelkaArbitrary MunkUnits

1800 1700 1600 1500 1400 1300 1200 Wavenumber (cm-1)

Figure 3.3. Dissolution of and protein adsorption to ZnO nanoparticles. A) Particle size and zeta potential of ZnO during pH titration in media (diamond markers) and water (squares; refer to symbol legend on top left of figure). The water samples are circled for clarity. B) Difference infrared spectrum in diffuse reflectance mode between ZnO particles exposed to DMEM/FBS media and water (y-axis in Kubelka-Munk arbitrary units). The detailed subtraction procedure can be found in the Experimental Procedures.

Figure 3.4. Difference infrared spectra of TiO2 and SiO2 in cell culture media.

Difference infrared spectra in diffuse reflectance mode between TiO2 and SiO2 particles exposed to DMEM/FBS media and water (y-axis in Kubelka-Munk arbitrary units). Labeled peaks are 1661, 1534, 1444, and 1395 cm-1.

Figure 3.5A shows the difference infrared spectra between the SiO2 and TiO2 that were exposed to pepsin, pancreatin and bile salts and those just exposed to water, with an example of the subtraction process shown in Figure 3.5B. The bands observed at 1682,

1527 and 1415 cm-1 are different from those observed in the media and stronger in intensity by about two orders of magnitude, indicating that the pepsin treatment alters the surface and promotes adsorption of material from the pancreatin/bile salt mix. Just based on the spectra, it is difficult to assign the exact types of species adsorbed on these particles.

A DRIFTS Spectra of Digested Particles 1682 1527 1415

TiO 2

Arbitrary KubelkaArbitrary MunkUnits SiO 2

1800 1700 1600 1500 1400 1300 1200 Wavenumber (cm-1)

TiO Water Exposed 2

TiO Digested 2

Arbitrary KubelkaArbitrary MunkUnits TiO Subtracted 2

1800 1700 1600 1500 1400 1300 1200 Wavenumber (cm-1)

Figure 3.5. Difference infrared spectra of digested SiO2 and TiO2 nanoparticles.

A) Difference infrared spectrum in diffuse reflectance mode between TiO2 and SiO2 NP treated with simulated digestion media and water (y-axis in Kubelka-Munk arbitrary units). B) An example of DRIFTS subtraction process: spectra of TiO2 exposed to water and freeze-dried, TiO2 treated with digestive solutions and freeze-dried, and the subtraction of the water-exposed spectra from the digested sample, leaving peaks related to the digestive enzymes. The procedure used for the subtraction is detailed in the Experimental Procedures.

The infrared spectra for the pepsin, pancreatin, and bile salts alone are shown for comparison (Figure 3.6). Infrared spectra reported on bile salts show presence of a band at

1680 cm-1,219 and IR spectrum of bile salts used in this study show bands at 1625 and 1670 cm-1 (Figure 3.6). These bands can be assigned to carboxylate groups of the bile salts, indicating that such species are adsorbed on the NP surface after treatment with the digestion medium.

Figure 3.6. DRIFTS spectra of freeze-dried digestive enzyme solutions.

Three major peaks from the digested TiO2 and SiO2 NP spectra are marked for comparison.

Nanoparticle interaction with C2BBe1 cells

The intestinal epithelial cell line C2BBe1 was established as a sub-clone from

Caco-2 cells, originally isolated from a human colon cancer,220 to be a more homogeneous population of brush border-expressing cells.35 These cells have been shown to form polarized monolayers of cells with microvilli maintained by cytoskeletal proteins, as is observed for intestinal epithelium in vivo.36, 221 For these studies, we used actively proliferating (but nearly confluent) cells rather than differentiated cells since proliferating cells are generally more sensitive to toxic agents including nanoparticles as has previously been observed.20 This level of sensitivity likely more accurately simulates the sensitivity of proliferating intestinal stem cells in vivo. As shown in Figure 3.7, morphology of

C2BBe1 cells following exposure to TiO2 (Figure 3.7B) is indistinguishable from that of

2 untreated cells (Figure 3.7A). Confocal microscopy of cells treated with 10 µg/cm TiO2 or SiO2 showed no apparent changes in tight junction structure after treatment as

A B

Figure 3.7. Cell morphology after nanoparticle treatment. Phase contrast micrographs of A) untreated control cells were compared to B) cells treated with TiO2. Images were taken after TiO2 was washed off cells.

100 determined by ZO-1 staining (Figure 3.8). No changes in adherens junction structure was observed after E-cadherin staining of TiO2-treated cells (Figure 3.9).

A B

Figure 3.8. Tight junction structure of TiO2-treated cells. C2BBe1 cells were plated on coverslips, allowed to grow for two days, and treated with 2 2 A) no NP, B) 10 µg/cm TiO2, and C) 10 µg/cm SiO2 for 24 hours before fixing, staining with an anti-ZO-1 antibody (green), and performing confocal microscopy.

101

A B

Figure 3.9. Adherens junction structure of TiO2-treated cells. Cells were plated on 8-chamber slides, allowed to grow to confluence, and treated with 2 A) no NP or B) 5 µg/cm TiO2 for 48 hours before fixing cells in 4% paraformaldehyde. Cells were stained with an anti-E-cadherin antibody (red, adherens junctions) and DAPI (blue, nuclei) and analyzed by confocal fluorescence microscopy.

In order to observe whether these NP are internalized by C2BBe1 cells, TEM was performed on cells that were treated with NP for 24 hours. Cells were treated with 10

2 µg/cm of SiO2, TiO2, or ZnO and centrifuged. After 24 hours, cells were washed and processed for TEM. Each type of NP was able to be visualized within C2BBe1 cells, as shown in Figure 3.10, although there seemed to be differences in the amount of NP uptake as a function of the particle type. For all of the treated cells, NP were only found in a fraction of the cells. TiO2 seemed to be taken up most readily by cells. All of the NP inside cells seemed to be clustered together, probably in vesicles. No particles were observed within the nucleus, only in the cell cytoplasm.

102

A B

Figure 3.10. Representative TEM images of nanoparticle-treated C2BBe1 cells. Cells were treated with NP for 24 hours before being harvested, fixed in glutaraldehyde, and further processed for TEM. A) SiO2-treated cell, B) TiO2-treated cell, C) ZnO-treated cell.

Toxicity

Cells were treated with NP at a dose of 10 µg/cm2, centrifuged to promote NP-cell interaction, and incubated for 24 hours. Toxicity assays were then performed on the treated

103 cells. Four different toxicity assays were performed. These short-term treatment experiments were conducted over a range of 5-12 cell passages (varying by assay).

Sytox Red is a nuclear stain that will bind DNA. However, it will only enter cells with damaged membranes and is thus a marker for cell necrosis or late apoptosis. Cells were stained with Sytox Red and analyzed by flow cytometry to measure the percent of stained cells (Figure 3.11A). Results are shown as percent viability based on normalization with unstained cells. The NP-treated cells showed comparable viabilities to the untreated control cells, approximately 100% viability, whereas H2O2 (used as a positive control) reduced viability of cells to 25%. Thus, SiO2, TiO2, and ZnO do not seem to cause necrotic cell death in C2BBe1 cells.

Flow cytometric analysis was also performed on cells stained with Annexin V to label apoptotic cells. Annexin V will bind to phosphatidylserine that is exposed on the outer surface of cells, which occurs in early apoptosis when phosphatidylserine is flipped to the outer layer of the cell membrane. C2BBe1 cells were stained with Annexin V and analyzed by flow cytometry (Figure 3.11B). NP-treated cells revealed approximately 100% cell viability, which was similar to the untreated control cells. Inducing apoptosis with H2O2 as a positive control reduced cell viability to 10%. These results indicate that SiO2, TiO2, and

ZnO do not induce apoptotic cell death in C2BBe1 cells.

The LDH assay indirectly measures cell death based on the amount of lactate dehydrogenase released by cells with damaged membranes. LDH release was measured in supernatants from C2BBe1 cells treated with NP for 24 hours and data were converted to percent viability of cells based on LDH release from positive control cells (treated with

104

A 120 B 120 100 100 80 80 60 60

40 40 % viability% % viability% 20 20 0 0 no NP H2O2 SiO2 TiO2 ZnO no NP H2O2 SiO2 TiO2 ZnO C D * 140 140 ** 120 120 100 100 80 80 60 60

40 activity 40 % viability%

20 mitochondrial % 20 0 0 no NP Triton SiO2 TiO2 ZnO no NP Triton SiO2 TiO2 ZnO Figure 3.11. Results of cytotoxicity assays for C2BBe1 cells treated with nanoparticles in media for 24 hours. Hydrogen peroxide was used as a positive control to cause cell death in flow cytometric assays (A and B) while Triton X-100 served as a positive control for cell death in the colorimetric assays (C and D). A) Cells were stained with Sytox Red and analyzed by flow cytometry to monitor necrosis. B) Cells were stained with Annexin V and analyzed by flow cytometry to monitor apoptosis. C) LDH release into the supernatant was measured to detect membrane leakage. D) Reduction of MTT to formazan by mitochondrial enzymes was measured as an indicator of mitochondrial activity. Significance as compared to untreated controls in the LDH and MTT assays was measured by Student’s t-test (* indicates p<0.01; ** indicates p<0.0001). Data are a compilation of five experiments, each containing three replicates.

Triton X-100) and untreated control cells (Figure 3.11C). Both TiO2 and SiO2 showed comparable viability to the untreated control (near 100% viability) which supports the results of the flow cytometric assays. However, cells treated with ZnO displayed an approximately 20% (p < 0.01) decrease in viability from the control cells, indicating that

ZnO is causing toxicity.

105

We also used the MTT assay to evaluate mitochondrial activity of these cells. The

MTT assay measures the ability of cells to reduce a tetrazolium dye to insoluble formazan, which can be measured colorimetrically. SiO2 and TiO2 particles induced no change in the mitochondrial activity, but ZnO treatment reduced mitochondrial activity to approximately

70% of untreated controls (p < 0.0001; Figure 3.11D).

Toxicity of nanoparticles treated with digestive enzymes

C2BBe1 cells were also treated with SiO2 and TiO2 NP that had undergone simulated in vitro digestion. Since the ZnO dissolved in pepsin at pH 2, cells could not be treated with digested ZnO. Cells were treated with the digested NP, centrifuged, incubated for 24 hours, and toxicity assays were performed as described above. Flow cytometric analysis after Sytox Red staining revealed no decrease in cell viability below that of the untreated control cells, suggesting no necrotic cell death induced by the digested NP

(Figure 3.12A). Annexin V staining also revealed comparable cell viability between cells treated with digested NP and untreated control cells, suggesting that the digested NP do not induce cell apoptosis (Figure 3.12B). The digestion solutions appeared to interfere with the LDH assay (increase in apparent cell viability was observed after treatment of cells directly with the digestive solutions; Figure 3.13). Thus, LDH release could not be used as an indicator of digested NP toxicity. In the MTT assay, only the digested TiO2 induced a slight decrease (10%) in mitochondrial activity (Figure 3.12C, p < 0.02), indicating that treatment with the digestion media caused minor NP toxicity.

106

A 120 100 80 60 40

% viability% 20 0 no NP H2O2 SiO2 TiO2

B 120 100 80 60 40

% viability% 20 0 no NP H2O2 SiO2 TiO2 * C 120 100 80 60

activity 40 20 % mitochondrial mitochondrial % 0 no NP Triton SiO2 TiO2

Figure 3.12. Cytotoxicity of C2BBe1 cells after treatment with SiO2 and TiO2 nanoparticles exposed to simulated digestive solutions. Hydrogen peroxide was used as a positive control for cell death in flow cytometric assays (A and B) while Triton X-100 served as a positive control for cell death in the MTT assay (C). A) Cells were stained with Sytox Red and analyzed by flow cytometry to measure necrosis. B) Cells were stained with Annexin V and analyzed by flow cytometry to measure apoptosis. Flow cytometry assay data are a compilation of three experiments, each containing three replicates. C) Reduction of MTT to insoluble formazan was measured to indicate mitochondrial activity of these cells. Significance as compared to untreated controls in the MTT assay was measured by Student’s t-test (* indicates p<0.02). MTT data are a compilation of four experiments, each containing four replicates.

107

140 ** ** 120 * ** * * * ** 100

80 5 µM

60 25 µM % viability% 40

0 untreated Triton pepsin pancreatin bile salts all

Figure 3.13. Interference of digestive solutions with LDH assay. Cells were treated with 5 or 25 µM of each of the digestive solutions or all of them combined for 24 hours before performing the LDH assay. Data were normalized to untreated control cells as 100% viability and represent means of four replicate wells ± one standard deviation. Significance as compared to untreated control cells was determined using Student’s t-test (* represents p-value < 0.05, ** represents p-value < 0.0005).

Immunological effects of nanoparticle exposure

There are many foreign materials that are not acutely toxic to cells in the body, but that may elicit an immune response. Intestinal epithelial cells secrete the chemokine IL-8 in order to recruit immune cells such as neutrophils to control a perceived threat.222 In order to determine whether these food-relevant NP elicit an immune response in intestinal epithelial cells, IL-8 secretion by cells was measured by ELISA. This assay was performed with the supernatants of the cells used for Sytox Red and FITC Annexin V experiments.

The ELISA was performed with both as-purchased and digested NP. As shown in Figure

3.14, IL-8 was only detected in the supernatants of cells treated with digested TiO2 and

ZnO. The detected concentrations of IL-8 were below 20 pg/ml, which is relatively little

108

100 90 80 70 60 50 40 30 20

Concentration Concentration (pg/ml) 10 0 untreated SiO digested TiO digested ZnO 2 2 SiO2 TiO2

Figure 3.14. ELISA assay to measure IL-8 secretion induced by nanoparticles. The concentration of the chemokine IL-8 was measured in supernatants of cells treated with NP for 24 hours. Supernatants were taken from the same cells used to measure acute toxicity by Sytox Red and Annexin V FITC staining and flow cytometric analysis. Measurement for each treatment was performed in duplicate and is presented as means ± 1 standard deviation.

IL-8 production by cells. However, this assay suggests that digested TiO2 and ZnO NP are inducing slight IL-8 secretion.

Long-term C2BBe1 exposure to nanoparticles

Consumption of NP in foods can realistically be expected to result in repeated exposure of the intestinal epithelium to these particles over a protracted period. To examine the effects of long-term NP exposure on intestinal epithelial cells, C2BBe1 cells were treated with NP for 24 hours after each cell passage. Cells were plated and allowed to adhere and acclimate for 2 days before being treated with NP. After 24-hour treatment, free

NP were removed and cells were grown to confluency. Cells were then passed into new plates and the particle exposure process was repeated. To examine the effects of repeated

NP exposure on cells, some cells were also plated in 24-well plates for toxicity assays after 109 certain cell passages. These cells were treated with NP for 24 hours before performing the toxicity assays described previously.

Toxicity of NP after long-term exposure (26 NP exposures) was determined using the same toxicity assays used for the acute exposure studies (Figure 3.15). Staining with

Sytox Red and flow cytometric analysis revealed no changes in cell viability from the untreated control. Flow cytometric analysis of cells stained with Annexin V likewise showed no decrease in viability of the cells. This suggests that the NP do not induce necrosis or apoptosis in C2BBe1 cells even after repeated exposure. No decrease in viability was observed for cells treated with TiO2 or SiO2 by LDH assay. Measurement of

LDH release from cells revealed a 30% decrease in viability in cells treated with ZnO (p <

0.01). A similar pattern was observed for ZnO treatment after all exposure cycles evaluated. This demonstrates that ZnO again may be inducing modest toxicity in cells exposed to ZnO long-term. No change in mitochondrial activity was observed after treatment with TiO2 or SiO2, however, the MTT assay revealed a 55% decrease (p <

0.0001) in mitochondrial activity in cells treated with ZnO.

Long-term exposure was continued up to 84 total NP exposures. As shown in

Figure 3.16, there was a similar pattern of toxicity after 84 NP exposures as was observed after 26 NP exposures (Figure 3.15). However, in this experiment the reduction in cell viability and mitochondrial activity in the LDH and MTT assays by ZnO was of greater magnitude. Also, there was a slight but significant reduction in cell viability by TiO2 in the

LDH assay.

110

A A

A 120 B 120 100 100 80 80 60 60

40 40 % viability% % viability% 20 20 0 0 no NP H2O2 SiO2 TiO2 ZnO no NP H O SiO TiO ZnO 2 2 2 2 C D ** 120 * 120 100 100 80 80 60 60

40 activity 40

% viability% 20 20 0 mitochondrial % 0 no NP Triton SiO2 TiO2 ZnO no NP Triton SiO2 TiO2 ZnO

Figure 3.15. Cytotoxicity after repeated nanoparticle exposure. Toxicity assays were performed on C2BBe1 cells that had been repeatedly exposed to NP, approximately weekly. The assays were performed 24 hours after the most recent NP treatment. Representative data shown were obtained after 26 NP exposures. A) Cells were stained with Sytox Red and analyzed by flow cytometry to measure necrosis. B) Cells were stained with Annexin V and analyzed by flow cytometry to measure apoptosis. Flow cytometry assay data (A and B) consist of 3 replicates for each treatment. C) LDH release by cells due to membrane damage was measured. D) Mitochondrial activity was assayed by reduction of MTT to formazan. LDH and MTT data consist of 4 replicates for each treatment. Significant differences in viability or mitochondrial activity as compared to untreated control were measured by Student’s t-test (* indicates p<0.01; ** indicates p<0.0001) in LDH and MTT assays.

Because long-term exposure to NP may induce more subtle effects than cell death,

growth curves were performed on long-term-exposed cells to measure changes in cell

proliferation. Growth curves were conducted with cells repeatedly exposed to NP

subsequent to the most recent NP treatment. Cell proliferation was measured by counting

cells daily for 10 days, and growth curves were performed after 7, 11, 16, and 29 NP

111

A 120 B 120 100 100 80 80 60 60

40 40 % viability % % viability% 20 20 0 0 no NP H2O2 SiO2 TiO2 ZnO no NP H2O2 SiO2 TiO2 ZnO

C * D 120 * 140 ** 100 120 80 100 80 60 60 40 activity 40 % viability% 20 20 0 mitochondrial % 0 no NP Triton SiO2 TiO2 ZnO no NP Triton SiO2 TiO2 ZnO Figure 3.16. Nanoparticle toxicity after 84 repeated exposures. Toxicity assays were performed on C2BBe1 cells that had been repeatedly exposed to NP, approximately weekly, for 84 total exposures. The assays were performed 24 hours after the most recent NP treatment. A) Cells were stained with Sytox Red and analyzed by flow cytometry to measure necrosis. B) Cells were stained with FITC Annexin V and analyzed by flow cytometry to measure apoptosis. Flow cytometry assay experiment data (A and B) consist of 3 replicates for each treatment. C) LDH release by cells due to membrane damage was measured by colorimetric assay. D) Mitochondrial activity was assayed by reduction of MTT to formazan. LDH and MTT data consist of 4 replicates for each treatment. Significant differences in viability or mitochondrial activity as compared to the untreated control were measured by Student’s t-test (* indicates p<0.001; ** indicates p<0.0005) in LDH and MTT assays.

exposures. Figure 3.17 shows the growth curves obtained after 11 cycles. The lack of differences among these growth curves suggests that long-term NP exposure did not affect cell proliferation. Two replicate growth curve experiments performed after 29 NP exposures are shown in Figure 3.18. Although individual curves indicated slight differences between differentially treated cell populations, no reproducible trends were

112 observed between the curves performed at any time points, and thus no further analysis was performed.

2.50E+06

2.00E+06

1.50E+06 Untreated

1.00E+06 SiO2SiO2

Cell count Cell TiO2TiO2 5.00E+05 ZnO 0.00E+00 0 50 100 150 200 250 300 Time (hours)

Figure 3.17. Growth curves after 11 nanoparticle exposure cycles. Growth was measured using cells that had undergone 11 cycles of repeated exposure to NP of SiO2, TiO2, or ZnO, along with untreated controls. Cells were plated at a starting concentration of 5×104 cells in 6-well tissue culture plates at time zero such that three replicate wells for each treatment could be counted daily for 10 days. These data are representative of growth curves performed after 7, 11, 16, and 29 NP exposures.

113

2.50E+06

2.00E+06

1.50E+06 Untreated SiOSiO2 1.00E+06 2 Cell count Cell TiOTiO22

5.00E+05 ZnO

0.00E+00 0 50 100 150 200 250 300 Time (hours)

1.60E+06 1.40E+06 1.20E+06 1.00E+06 Untreated 8.00E+05 SiOSilica2 6.00E+05 Cell count Cell TiOTitania2 4.00E+05 ZnO 2.00E+05 0.00E+00 0 50 100 150 200 250 300 Time (hours)

Figure 3.18. Growth curves after 29 nanoparticle exposure cycles. Growth of cells repeatedly exposed to NP was measured. These cells had undergone 29 cycles (replicate curves are shown) of repeated exposure to NP of SiO2, TiO2, or ZnO, along with untreated controls. Cells were plated at a starting concentration of 5×104 cells in 6-well tissue culture plates at time zero such that three replicate wells for each treatment could be counted daily for 10 days.

114

3.4 Discussion

Particle Characteristics

From the X-ray diffraction patterns shown in Figure 3.2, silica is amorphous, ZnO

223 224 has the wurtzite-type structure, and TiO2 is a mixture of anatase and rutile. Table 3.1 shows that the size and charge of the particles undergo significant changes in different environments. In serum-containing media, the particles appear aggregated as compared to the primary particle size (Figure 3.1), indicating that the as-obtained particles were aggregated, consistent with previous reports.49, 50 Titania becomes positively charged and the silica particles are almost neutral in the “stomach” fluid (pepsin, pH 2.0). In the case of silica, the neutrality of the particles results in formation of large aggregates (~ 3 µm radii).

The charges revert back to negative in the neutral intestinal fluid and in the case of silica, the particles deagglomerate.

ZnO dissolved completely in the pepsin solution (pH 2.0). There are conflicting reports in the literature, with most studies suggesting that treatment at pH ~2 (simulating gastric conditions) will increase the solubility of ZnO, but at least one study suggests otherwise.20 Light scattering studies as a function of pH shown in Figure 3.3A indicate that

ZnO is indeed stabilized by the serum, and dissolution occurs at pH < 4, as compared to

ZnO in water, where rapid dissolution occurs at pH < 6.5. The infrared spectrum (Figure

3.3B) shows that upon suspension in serum, characteristic protein bands218 are observed on the ZnO. It has been proposed that the presence of serum proteins will inhibit dissolution of ZnO,50 however, it is unlikely that serum can stabilize against dissolution in the highly

115 acidic pepsin solution. Protein coronas are also observed on the TiO2 and SiO2 particles

(Figure 3.4).

Particle Uptake by Cells

Our choice of a sub-clone (C2BBe1) of the Caco-2 cell line is based on their ability to mimic human intestinal cells. The Caco-2 cell line has been extensively characterized, and is recommended for use as an in vitro model of the GI tract by an international toxicity screening workgroup.225 Figure 3.7 demonstrates that these cells form a confluent layer with intercellular tight junctions, the morphology of which is unaffected by NP exposure.

This was further confirmed by staining with junctional proteins ZO-1 and E-cadherin which appear undisturbed after treatment with TiO2 (Figures 3.8 and 3.9). The TEM data in Figure 3.8 indicate that particles are being internalized. Previous studies have also noted that silica with primary particle sizes of 32 and 83 nm were taken up by Caco-2 cells, formed larger agglomerates (200-300 nm), and even showed up in the nucleus after 72- hour exposure with the smaller particles.69 Silica NP uptake by Caco-2 cells has been reported based on elemental analysis of the cells, although surface binding versus

20 internalization is not readily distinguished by this method. Internalization of TiO2 NP in

Caco-2 cells has also been reported.66

Toxicity: Undigested Particles

Engineered NP have been reported to interfere with optical assays due to light absorption/scattering effects.226-228 In particular, for the MTT and LDH assays, it was suggested that loading levels be kept below 50 µg/cm2. With ZnO at a loading level of ~10

µg/cm2, slight effects were even observed with the LDH assay, but not with the MTT

116

226 227 assay. TiO2 were found to adsorb on LDH above loading levels of 100 µg/mL. SiO2 has been reported to interfere with the MTT assay in HeLa cells by promoting exocytosis of the formazan crystals, even at loading levels of 10 µg/mL.228 Because of these literature reports, the experiments in the present study were conducted at a dose of 10 µg/cm2.

Silica/titania dispersed in cell media did not exhibit any toxicity by any of the assays, including the long-term study with repeated exposures (~29 cycles) (Figures 3.11,

3.15). There is conflicting data in the literature on toxicity of silica. Lack of toxicity of

SiO2 in Caco-2 cells was observed despite particle internalization by cells and even nuclear localization of particles.69 Mitochondrial activity (WST assay) in Caco-2 cells treated with

SiO2 NP showed no significant changes as compared to controls, even at dosage levels of

69 200 µg/mL. In another study, SiO2 NP were found to reduce cell viability in undifferentiated Caco-2 cells, even at loadings of 5 µg/cm2.20

With NP of TiO2, there appears to be agreement of a lack of toxicity towards Caco-

49 2 cells. Commercial TiO2 used in sunscreens, made up of a rutile core surrounded by

Al(OH)3 shell and fibrous-type morphology with lengths of 50 ±10 nm and widths of 7 ±

2 nm were not internalized, and no toxic effects were noted in response to loadings of up

229 2 to 100 µg/mL. TiO2 at concentrations of 1-20 µg/cm had no effect on Caco-2 cell

50 viability in the presence or absence of serum. TiO2 at loadings of 0.1-100 mg/L did not

49 influence Caco-2 cell viability, as measured by MTT assay. TiO2 at loading levels of 1-

10 µg/mL did not cause cell death after either acute or chronic exposure in Caco-2 cells.66

Anatase/rutile mixtures, similar to the NP examined in the present study, did not induce toxicity (by LDH assay) to Caco-2 cells at loading levels of 20 µg/cm2, but toxicity was

117 observed at 80 µg/cm2. The WST-1 assay indicated reduction in metabolic activity only at loadings of 80 µg/cm2.162 Using the MTT assay, different mammalian cell types have been

230 shown to respond differently to TiO2 NP, with only some showing a toxic response.

Zinc Oxide

In response to ZnO exposure, we found evidence of low-level cell toxicity as determined by the LDH assay (Figure 3.11C), but this was not corroborated by assays of necrosis or apoptosis which demonstrated no toxicity (Figure 3.11A,B). Mitochondrial activity was also diminished by ZnO treatment (Figure 3.11D). Both these effects of ZnO were also observed in response to long-term treatment (Figure 3.15C,D). In addition, low- level secretion of IL-8 by cells was detected after ZnO treatment (Figure 3.14). There is agreement in the literature that ZnO induces some cytotoxicity in Caco-2 cells.49 ZnO have been reported to be cytotoxic to both undifferentiated and differentiated Caco-2 cells at doses above 5 µg/cm2 and to induce increased IL-8 secretion in differentiated cells but not undifferentiated cells.20

Considering that TEM demonstrated that ZnO is internalized by cells, one hypothesis is that if the internalized ZnO NP are taken up in endosomes, the particles can dissolve and provide a burst of Zn2+ within the cell. The kinetic profile of pH change in

Caco-2 endosomes upon internalization of particles (endosomal maturation en route to the lysosome) indicates a drop in pH from 7.4 to ~5.231 The intracellular dissolution hypothesis could explain the decreased mitochondrial activity observed in ZnO-treated cells. Zinc ions have been shown to be pro-apoptotic under certain circumstances232 and the solubilized

118

ZnO released from the endosomes could be causing toxicity by inhibiting energy metabolism.232-234

However, as Figures 3.17 and 3.18 show, none of the three particles had any influence on the rate of proliferation of the cells after repeated exposure.

Toxicity: Digested Particles

The MTT experiments suggest a slight decrease in metabolic activity induced by titania NP only after exposure to the simulated digestion solutions (Figure 3.12C). Based on the ELISA, treatment with TiO2 but not SiO2 NP exposed to the simulated digestion solutions induced slight but detectable IL-8 secretion in cells (Figure 3.14). IR data shown in Figure 3.5 indicates that in both silica and titania, there is strong adsorption of material from the simulated digestive media. Bile salts/proteins appear to be on the surface of these particles, with the acidic pepsin treatment promoting the amount of adsorbed material

(since particles exposed to media have much less material). Micelles of lecithin and the bile salt sodium deoxycholate at concentrations > 0.2 mM have also been reported to decrease cell viability of Caco-2 cells.235 Thus, the NP could be transporting toxic material from the digestion media into the cells. It is unclear at this point why silica exposed to the simulated digestion media is not showing the same type of toxicity as the titania. It is important to note that the toxicity manifested by titania, as measured by the MTT assay, is mild and possibly can be related to the different levels of adsorbed bile salts being transported by the two particles. The IL-8 secretion detected by ELISA is minimal and may also be attributable to differences in NP adsorption of proteins from the digestion solutions,

119 particularly if this enhances NP uptake by cells, as has been reported for polystyrene NP after an in vitro digestion treatment.236

Long-term nanoparticle exposure

Cells that had undergone up to approximately 35 NP treatments displayed very similar toxicity to the cells exposed once to NP in the acute exposure experiments (data shown after 26 NP exposures in Figure 3.15, compared to Figure 3.11). This repeated exposure to NP also did not affect cell growth (Figures 3.17 and 3.18). This suggests that repeated exposure does not sensitize or desensitize cells to NP. However, toxicity seemed to be slightly amplified in cells that had undergone 84 exposure cycles, including a mild but significant decrease in viability of TiO2-treated cells in the LDH assay (Figure 3.16).

This difference is likely due to less confluent cells used for this particular NP treatment or further transformations of the C2BBe1 cells over the course of repeated cell passages rather than a result of the repeated NP treatment, especially since the increased toxicity is minimal.

Implications of this Study

The implications of this in vitro study of NP of SiO2, TiO2 and ZnO in media towards C2BBe1 cells are as follows. The main findings of the study are summarized in

Figure 3.19. The cells internalized all three particles. ZnO were the only NP that induced mild toxicity by the LDH and MTT assay. Repeated exposure of cells to SiO2, TiO2 or ZnO did not alter their growth patterns or render them any more susceptible to toxicity. With the SiO2 and TiO2 treated with the simulated digestion medium, there was no indication of necrosis or apoptosis, but diminished mitochondrial activity was noted with titania,

120 possibly due to transport of bile salts/proteins into the cell. The importance of the corona formed on NP is being recognized as important in toxicity,123 since not only is the response of the living cell to the particle influenced by the composition of the corona layer, but it is also a mechanism to get material into the cells.

150

100

Culture media % % viability 0 none SiO2 TiO2 ZnO

Nanoparticles

Digestive enzymes Minimal toxicity Figure 3.19. Summary of the study. Cells were treated with NP suspended in cell culture media, some of which had undergone simulated gastrointestinal digestion where NP were incubated with digestive enzyme solutions. This treatment affected the composition of the protein corona on the surface of the NP. All tested NP were internalized by cells, but minimal toxicity was observed. ZnO NP exhibited mild cytotoxicity based on several toxicity assays.

Even though we did not find SiO2 and TiO2 to be toxic, it is clear that they are internalized by the epithelial cells as NP and may subsequently enter the circulation and migrate to other parts of the body. The surface charge of SiO2 and TiO2 particles becomes neutral/positive within the “stomach” solution and can therefore interact with the negative mucus proteins and influence their transport. With ZnO, toxicity is not relevant in its nanoparticulate form, unless stabilized against dissolution in the stomach. This is consistent with the observation that when ZnO was administered to rats by oral gavage,

121

Zn2+ were found in tissues, but not ZnO particles.223 Thus, at least for silica and titania, it becomes essential to monitor transport of these NP in vivo to determine whether they do indeed enter the circulation and if so, to map their route(s) of distribution and potential sites of accumulation. Several studies have examined the fate of ingested NP. TiO2 injected into rat ileum did not disrupt the structure of the epithelial layer, though the particles were taken up by the cells and eventually reached the liver.204 In vivo studies with mice fed with

157 SiO2 exhibited liver toxicity. Dynamics of the motion of NP into systemic circulation, localization in tissues, and clearance from the body still needs to be mapped out.

122

Chapter 4: Spectroscopic evaluation of the nucleation and growth for microwave-

assisted CdSe/CdS/ZnS quantum dot synthesis and uptake of bright fluorophore

core-silica shell nanoparticles by biological systems2

4.1 Introduction

Nanoparticles (NP) are being increasingly used in a wide variety of industrial, environmental and consumer applications. This has led to considerable interest in understanding how NP interact with biological systems, as well as their effects and fate in the environment. NP interactions with biological/environmental systems are dependent on their material properties, including size4, 5, 203, 239 and surface charge.119, 205, 206

Quantum dots (QDs) are nanosize semiconductors that fluoresce due to quantum confinement effects. In a quantum dot, the particle size is smaller than the Bohr exciton radius, and this leads to an increase in the band gap energy.240 Using this strategy with various semiconducting materials, nanometer-sized quantum dots have been made which fluoresce across the entire visible spectrum241-243 and in the infrared.244 QDs are useful in a variety of fields because they are less prone to photo-bleaching and chemical degradation, have discrete emission wavelengths based on size, and have a wide absorption profile

2 Previously published in modified form.237, 238 Text and figures appear with permission from the editors of The Journal of Physical Chemistry C and the International Journal of Nanomedicine. 123 which makes them very suitable for biological imaging.245 Materials scientists are exploring QDs for solar cells, sensors, and improved lighting.246-248

Control of nucleation and growth of QDs is essential to optimize size and optical properties. Conventional organic-based hot-injection synthesis involves rapid nucleation by introducing precursor solutions to a hot (>300°C) solvent. Controlled growth then occurs more slowly at a lower temperature and the final size of the particles is determined by the temperature during the growth phase and the organic capping agent which sterically inhibits growth.249 Such methods of quantum dot production are well-developed and produce high-quality quantum dots with quantum yields (QY) between 50-80%.250 The QY typically decreases with surface defects that create trap states resulting in nonradiative decay and a broad red-shifted trap-state emission.251 To enhance the fluorescence and stability of the core particle, protective shells with higher band gap are often grown around them.250, 251

Organically prepared quantum dots do not disperse in water and require ligand exchange, which can significantly alter the optical properties of the particles.252 Aqueous- based methods of quantum dot synthesis have been developed which yield water-soluble quantum dots suitable for use in biological research. In these methods, cadmium chalcogenides (Se, Te) are nucleated by reacting cadmium salts (CdCl2) with NaH(Se, Te) in the presence of a passivating ligand.253, 254 These reaction methods have the benefit of being safer, less expensive, and yielding water-soluble particles without the need for ligand exchange, however, their quantum yield is typically lower compared to organic methods.

124

Microwave irradiation has been explored to decrease the time of aqueous quantum dot synthesis, and synthesis of CdSe(S), ZnSe(S), and CdSe/CdS/ZnS have been reported.255-258 Microwave heating of the precursors that are typically used in conventional convective heating has also been reported to produce QDs at shorter reaction times.259, 260

While QDs can have specialized biological applications, silicon dioxide (silica) is a commonly used material in food, pharmaceutical, chemical and consumer goods. The

Project on Emerging Nanotechnologies Consumer Products Inventory attempts to catalog these products, which as of the time of writing lists 41 products using or claiming to use nanosilica.3 Seven of the listed products are cosmetics and dietary supplements. In the food industry, silica finds use as an anti-caking agent and clarifying agent.19 While silica particles used in foods (commonly E551) typically have an average size of a few hundred nanometers, they have a broad size distribution. Food-grade silica is reported to contain up to 33% silica with a size less than 200 nm, leading to an estimated 1.8 mg/kg of body weight/day intake of nanosize silica particles.1, 261 Other uses of silica are in coatings, paints, and cleaning supplies. Another NP commonly found in many foods is titania.2

Despite the increased interest in using NP in food, the FDA currently has no specific regulations related to NP. A 2012 draft guidance suggested that food manufacturers investigate the safety of foods incorporating NP, but at present, based upon studies of their larger counterparts, many of the inorganic NP are generally regarded as safe for use in foods.202 There is increasing evidence that ingested NP can traverse the intestinal epithelium, enter the portal and systemic circulation, and accumulate in tissues and

125 organs.4, 203, 205 In particular, silica has been found in liver tissues and a transport model has been proposed.144, 158

Thus, tracking NP as they interact with the biological milieu and in the environment is of interest. Fluorescence spectroscopic methods can be used to study NP fate in real time, but many NP used in the commercial world, such as titania, zinc oxide, and silica, are intrinsically nonfluorescent. These NP can be functionalized with a fluorescent component, and used as trackable model NP simulating the real particles. However, because NP interactions with the host system depend on surface properties, it is important that the model particle’s surface characteristics are similar to those of the commercial particles.

This necessitates that the fluorescent probes remain within these model particles and not leach into the solution or migrate/bind to the particle surface. In such cases, the interactions of the model NP with the biological systems may not be representative of the commercial samples.

Silica NP synthesis is a well-developed field, typically involving an ammonia- catalyzed hydrolysis of an alkyl silicate followed by condensation. The sol-gel-based

Stöber synthesis is the most common,262, 263 and modifications of this process where the hydrolysis takes place in a reverse microemulsion water-in-oil droplet is also widely practiced.264 These methods have been used to incorporate fluorescent probes in silica NP.5,

265-270 Recent publications show evidence of silica NP dissolution, including from the inside, leaving hollow spheres.271 This poses an obvious problem for dye encapsulation: the dye will leach out of the particles. The instability is size-dependent and more pronounced with smaller NP, especially in biological media.272 A more recent development

126 in nanosilica synthesis is the use of weakly basic amino acids such as lysine and arginine in place of ammonia.273 Using this strategy, dense, stable coatings of dye-incorporated silica have been synthesized which minimize dye leakage.272

There are a variety of fluorophores that can be encapsulated within the silica coating for biological studies. The variables desired are good fluorescence stability, lack of photobleaching, stability in biological media, bright fluorescence/high quantum yield, and a range of emission wavelengths. Fluorescence emission in the near-IR region is beneficial for analysis of biological tissues.274 Here we focus on Rhodamine 6G, Rhodamine 800, and

CdSe/CdS/ZnS quantum dots. Rhodamine 6G is a commonly available organic dye, with very strong emission (90% quantum yield, QY) at 558 nm.275 Rhodamine 800 is a dye with a 25% QY, and emission maximum at 700 nm.276 The near-IR fluorescence emission allows for non-invasive whole body in vivo imaging.

We have reported a fast, “one-pot” synthesis that produces stable, aqueous

CdSe/CdS/ZnS core/shell quantum dots.257 While they are relatively easy to produce and show promising stability during biological experiments, these particles have a low quantum yield of 13%. In this research, we examined strategies for improving the quality of the microwave-produced QDs for biological applications including encapsulation within silica coating. By coupling the microwave reactor to a fluorescence spectrometer via fiber optics, the growth of the QD can be monitored during synthesis, and this real-time data has provided insight into the growth mechanism. We also examined the effect of temperature during both the initial nucleation phase and the microwave-assisted crystal growth, and found a way to form QDs with varying emission wavelengths. When combined with UV-

127 visible light exposure during nucleation and after microwave-based QD growth,

CdSe/CdS/ZnSe QDs with quantum yields of 40% were obtained.

Using three fluorophores (Rhodamine 6G, Rhodamine 800 and CdSe/CdS/ZnS

QDs), we defined the conditions for synthesis of stable, bright fluorescent core/silica shell

NP. The surface characteristics, morphology, and optical properties of these composite NP are established. Internalization of the optimally prepared fluorophore/silica NP by a murine alveolar macrophage cell line and a human intestinal epithelial cell line, C2BBe1, relevant to the digestive system, is studied. Finally, the fate of orally administered fluorophore/silica

NP in mice, especially the movement of particles from the gastrointestinal tract to tissues, is examined. With the QD/silica, fluorescence microscopy can confirm the location within tissues, and if combined with elemental analysis of the Cd/Si ratio by inductively coupled plasma mass spectrometry (ICP-MS), we can unequivocally determine that the silica NP rather than its dissolved counterpart has been transported to the tissues from the digestive system.

4.2 Experimental Procedures

All QD and silica NP synthesis and characterization was performed by Dr. Andrew

Zane in Dr. Prabir Dutta’s laboratory in The Ohio State University Department of

Chemistry and Biochemistry. Measurement of Si and Cd in liver tissue of silica/QD-treated mice by ICP-MS performed by Dr. Anthony Lutton in Dr. John Olesik’s laboratory in The

Ohio State University School of Earth Sciences.

128

Summary of typical microwave synthesis

To nucleate CdSe cores, 0.25 mL of a 20 mM NaHSe solution was added to 19 mL of a Cd/MPA solution (1.05 mM Cd and 5.26 mM MPA) and stirred for 1 hour. To create

2+ the cap, 0.75 mL of a 26.67 mM Zn(NH3)4 solution was added for a final QD formulation of 1 mM Cd, 1 mM Zn, 0.25 mM Se, and 5 mM MPA. The solution was heated in a

Discover SP (CEM Corp.) microwave system for appropriate times and temperatures. UV-

Visible light irradiation was performed for 2 hours after initial nucleation, for 6 hours after microwave crystal/shell formation, or both.

Fluorophore entrapment by arginine-driven reaction

For dye/silica particle synthesis, commercial silica NP (30 mg, Sigma Aldrich, ~20 nm) were added to a 30 mL, 10 ppm solution of dye and sonicated for 30 minutes. For

QD/silica particle synthesis, quantum dots were diluted to 1 mg particles/mL (30 mL total).

To the dye/silica or QD solutions, 15 mg L-arginine was added and dissolved prior to addition of tetraethyl orthosilica (TEOS, 600 µL). This solution was stirred at 100 rpm for

48 hours in a 70°C oil bath. Reaction concentrations were 6.82×10-4 M Rhodamine 6G,

6.59×10-4 M Rhodamine 800, 478 nM QDs, 2.81×10-3 M arginine, and 8.78×10-5 M TEOS in water. The resulting particles were washed by centrifugation at 50,000×g until fluorescence was not detected in the discarded supernatant.

Particle characterization

Optical properties

UV/Visible absorption and fluorescence spectra were measured using a Shimadzu

UV-2501PC spectrophotometer (Shimadzu Corporation, Kyoto, Japan) and a Horiba

129

Jobin-Yvon Fluorolog 3, respectively, for particles diluted 10:1 with water. Excitation wavelengths were 375 nm for QDs and QD/silica, 480 nm for Rhodamine 6G/silica, and

635 nm for Rhodamine 800/silica.

Quantum yields were measured using a Quanta-Phi integrating sphere attachment

(Horiba Jobin-Yvon) for the Fluorolog 3. Spectra were measured from 460 nm to 750 nm and the areas under the excitation and emission curves in each spectrum were integrated, allowing for calculation of the quantum yield.

Fluorescence lifetime was measured with the Fluorolog 3 time-correlated single photon counting (TCSPC).

In situ fluorescence

In situ fluorescence measurements were taken during the microwave irradiation step of the QD synthesis by coupling the fluorometer to the microwave reactor using a fiber optic cable attachment. Batch runs were set up to take approximately one spectrum every minute using an excitation wavelength of 480 nm, and the emission wavelength was scanned from 500 to 750 nm.

X-ray diffraction

As-prepared QD solutions were washed twice with water, dried overnight in a vacuum oven, loaded into a 0.5 mm capillary, and XRD spectra were recorded using a

Bruker D8-advance system.

Infrared spectroscopy

Diffuse reflectance infrared Fourier transform spectroscopy measurements were taken using a Perkin Elmer Spectrum 400 spectrometer (PerkinElmer Inc., Waltham, MA).

130

Samples were freeze-dried for 2 days (Millrock Benchtop Manifold Freeze Drier, Millrock

Technology, Inc., Kingston, NY, USA) followed by vacuum drying for at least 12 hours.

Nuclear magnetic resonance spectroscopy

High-resolution nuclear magnetic resonance (NMR) spectra were collected at room temperature using a Bruker DSX spectrometer (Bruker, Billerica, MA) operating at 300.13 and 59.62 MHz for 1H and 29Si, respectively. The 29Si cross-polarization magic angle spinning (CP-MAS) NMR spectra were obtained using a Bruker two-channel probe in a 4 mm rotor spun at 5 kHz.

Zeta potential and isoelectric point measurement

Zeta potentials were measured (Zetasizer Nano ZS, Malvern Instruments Ltd,

Malvern, UK) using particles suspended at 50 ppm in water. Samples were titrated against

1 M HCl using the attached autotitrator (Malvern MPT-1), and isoelectric points (IEPs) were determined by the Malvern software.

Electron microscopy

High resolution transmission electron microscopy (HRTEM) images were obtained using a Tecnai-F20 system (FEI™, Hillsboro, OR). Particles were washed twice, resuspended in a dilute solution in ethanol, and deposited onto a lacey-carbon-coated copper grid.

Biological imaging

Cell culture

Human intestinal epithelial cells (C2BBe1) and murine macrophage cells (MH-S) were obtained from the American Type Culture Collection (Manassas, VA). C2BBe1 cells

131 were cultured in Dulbecco’s Modified Eagle Medium (DMEM; Life Technologies, Grand

Island, NY) supplemented with 10% fetal bovine serum (FBS; Serum Source International,

Inc., Charlotte, NC), 1 mM sodium pyruvate, 2 mM L-glutamine, 0.3% penicillin/streptomycin, 0.3 µg/mL Amphotericin B (fungizone), and 10 µg/mL transferrin

(all from Life Technologies). MH-S cells were cultured in RPMI Medium 1,640 (Life

Technologies) supplemented with 10% fetal bovine serum (Serum Source International,

Inc.), 0.3% penicillin/streptomycin, and 0.3 μg/mL Amphotericin B (fungizone; both from

Life Technologies). Cells were incubated in 5% CO2/95% ambient air at 37°C. Cells were passaged every 5-7 days and plated on flasks or plates pre-coated with collagen I (0.05 mg/mL, rat tail, Life Technologies).

Cells were plated at a density of 70,000 cells/well in 24-well tissue culture plates

(Corning-Costar, Tewksbury, MA) for comparison of QD fluorescence intensity by flow cytometry. Cells were plated in 8-chamber slides (Thermo Scientific) at a density of

90,000-100,000 cells/chamber for analysis by confocal microscopy. Cells were incubated for at least 24 hours after plating before treatment with QDs, Rhodamine 6G/silica, and

QD/silica. Directly prior to treatment on cells, QD, Rhodamine 6G/silica, and QD/silica solutions were sonicated using a Sonics Vibra-Cell sonicator (Sonic Materials, Inc.,

Norwalk, CT) pulsing for one second on, one second off for approximately 15 seconds in order to minimize QD agglomeration. Cells were treated with the appropriate dose of particles (generally 100 µg/cm2) for Rhodamine 6G/silica and QD/silica. Experiments with

40% QY QDs were generally performed at a dose of 157 nM which was achieved by adding

107 µL of the QDs (732 nM) per well in a 24-well plate with a 0.5 mL total volume per

132 well with cell culture media and 64 µL of QDs per chamber in an 8-chamber slide with a total volume of 0.3 mL per chamber. After treating cells with the appropriate dose of QDs, cells were centrifuged to promote contact of QDs with cells. The 24-well plates were centrifuged at 300×g for 15 minutes. Due to greater fragility, 8-chamber slides were centrifuged at 75×g for 15 minutes. Cells were incubated with QDs for 24 hours before performing experiments.

In vivo studies

All animal experiments were approved by The Ohio State University Institutional

Animal Care and Use Committee (IACUC, protocol 2012A00000020). Female SKH1-E mice of 8-12 weeks of age were purchased from Charles River Laboratories (Raleigh, NC).

When mice were received, they were provided food and water ad libitum. For the first experiment (where mice were administered Rhodamine 6G/silica and Rhodamine

800/silica), mice were fed Teklad Global Rodent Diet 2016 (Harlan Laboratories,

Indianapolis, IN) for 10 days prior to NP administration in order to minimize autofluorescence in the mice from their diet. In a later experiment (with Rhodamine

6G/silica and QD/silica), mice were fed the Teklad AIN-93M Purified Diet (Harlan

Laboratories) for 10 days prior to NP administration to further reduce the autofluorescence due to their diet.

Rhodamine 6G/silica and QD/silica were administered by oral gavage to mice once daily for four total administrations. Food was taken from mice several hours prior to each administration in order to empty the stomach. In the first experiment, Rhodamine 6G/silica and Rhodamine 800/silica were administered in 100 µl total volume at doses of 1 mg each.

133

In the second experiment, Rhodamine 6G/silica and QD/silica were administered in doses of 0.1 mg and 0.67 mg, respectively (again, 100 µl total volume). Rhodamine 6G/silica,

Rhodamine 800/silica, and QD/silica were suspended in water and sonicated using a Sonics

Vibra-Cell sonicator (Sonic Materials, Inc.) pulsing for one second on, one second off for approximately 15 seconds in order to minimize particle agglomeration immediately prior to administration. Mice were euthanized three hours after the final silica particle administration by carbon dioxide asphyxiation. Organs were excised and frozen in Tissue-

Tek O.C.T. Compound embedding medium (Sakura Finetek, Torrance, CA). Organ blocks in OCT were sectioned onto microscope slides using a Microm HM505E cryostat (Microm

International GbmH, Walldorf, Germany) at a thickness of 0.4 µm for fixation and staining for confocal microscopy.

A control mouse was kept under the same conditions as the mice administered

Rhodamine 6G/silica and Rhodamine 800/silica or Rhodamine 6G/silica and QD/silica for the duration of the experiment, but was not administered particles. This mouse was euthanized first to prevent contamination of harvested organs with fluorescent NP. The organs of the control mouse were sectioned, stained, and examined by confocal fluorescence microscopy in the same manner as the organs from the silica NP-treated mice.

Minimal green fluorescence was observed in control tissues, but any signal observed in control tissue was assumed to be autofluorescence and used to determine the microscope settings needed to be able to differentiate real signal and for comparison with tissue of treated animals. Tissues including kidney, spleen, and especially liver displayed greater

134 background autofluorescence due to blood cells, but the gastrointestinal tract tissues, lung, and brain tissue showed very little background fluorescence.

Flow cytometry

After 24-hour treatment of cells with QDs, cells were washed twice with phosphate- buffered saline (PBS) and detached from the culture plate with trypsin. Cells were suspended in PBS and analyzed for fluorescence using a FACScalibur flow cytometer (BD

Biosciences, San Jose, CA) at an excitation wavelength of 488 nm. All experimental treatments were performed in triplicate. Mean fluorescence intensity values were used to compare the fluorescence of QDs made by different syntheses in cells.

Confocal microscopy

After 24-hour treatment of cells with QDs, the chambers were removed from 8- chamber slides for staining. All staining was performed at room temperature. Slides were washed twice with PBS and fixed in 4% paraformaldehyde for 45 minutes. Cells were washed again in PBS before permeabilization in a 0.2% Triton X-100 (Sigma-Aldrich) solution in PBS for 15 minutes. Cells were washed in PBS and incubated in a 1% bovine serum albumin (BSA; Sigma-Aldrich) blocking solution for 1 hour. Cells were incubated with a 1:150 dilution of Alexa Fluor 647 mouse anti-E-cadherin (BD Biosciences) or Alexa

Fluor 647 mouse anti-GFAP (brain tissue; Cell Signaling Technology, Danvers, MA) in

1% BSA in PBS for 90 minutes. Cells were washed with PBS and stained with a 0.25

µg/mL solution of 4’,6’-diamidino-2-phenylindole (DAPI; Life Technologies) in 1% BSA in PBS for 10 minutes. Cells were washed with PBS before coverslips were mounted using

ProLong Gold Antifade Reagent (Life Technologies). Mounting media was allowed to cure

135 overnight at room temperature before analysis using a Zeiss LSM 700 confocal fluorescence microscope (Jena, Germany).

Si, Cd measurements in liver by inductively coupled plasma-mass spectrometry

Approximately 0.3 g of each of the wet liver samples was digested in 7 mL of concentrated nitric acid (GFS Chemicals, Veritas Low Trace Metals, Columbus, OH, USA) and 3 mL of concentrated ultrapure hydrogen peroxide (GFS Chemicals, Veritas Low

Trace Metals, Columbus, OH, USA) at 190°C for 10 minutes in a closed vessel using an

Ethos 320 microwave digestion system (Milestone, Bergamo, Italy) and then diluted to a total volume of 100 mL with deionized water. The digested samples were measured using an Element 2 Inductively Coupled Plasma-Sector Field Mass Spectrometer

(ThermoFinnigan, Bremen, Germany). Si was measured at a resolving power (R= m/Δm) of 300 while Cd was measured at a resolving power of 4000. Solution calibration standards were prepared by dilution from commercial standard solutions (CPI International, Inc, Palo

Alto, CA, USA). Ten ppb In was added to each standard and sample solution and used as an internal standard. A dilute suspension of QD/silica NP was also measured in order to compare the Cd/Si concentration ratio to those measured in the digested liver samples.

4.3 Results

QD synthetic strategies

Figure 4.1 shows the synthetic pathways for QD optimization discussed below. For

QD optimization and characterization data, refer to the published paper.237 The overall composition Cd4:Se1:Zn4:MPA20 was maintained for all experiments. Reactions were conducted using microwave temperatures between 130 and 155°C as those outside this

136

Figure 4.1. QD synthesis and optimization strategy.

range produced weakly fluorescent products. Optimal reaction times at each temperature were determined based on quantum dot fluorescence in situ and quantum yields of the recovered samples. UV-visible light illumination was carried out during nucleation as well as after microwave treatment.

QD nucleation

Room temperature nucleation

In situ fluorescence monitoring of samples nucleated at room temperature detected an initial emission at 502 nm which decreased in intensity after the first 20 minutes while an emission at 540 nm increased in intensity and red-shifted to 574 nm with peak intensity at 60 minutes. An isosbestic point at 550 nm between 40 and 60 min indicates that two species are interconverting.

The optimal microwave time for temperatures between 130 and 155°C was determined using in situ fluorescence to monitor the times around which the fluorescence reached its maximum. Similar quantum yields (QY, 17-19%) were obtained at all temperatures in this range using optimized microwave times. Lower temperatures required longer times to reach the optimum QY.

137

Plots of the rate of change in fluorescence intensity with time provide insight into the stages of particle growth.237 In particular, the data at 135°C shows four distinct regions with differing slopes throughout particle growth while these stages merge into one another at 155°C.

Nucleation at various temperatures

Varying nucleation temperature between 0-100°C systematically altered the emission maximum, with the 0°C sample emitting at 546 nm and the 100°C sample at 574 nm. All samples except that nucleated at 0°C exhibited QY of 17-18%. The in situ fluorescence experiments over the range of temperatures 130-155°C were repeated with the sample nucleated at 100°C, and the data appear similar to the spectra recorded with the room temperature nucleation.

Effects of light illumination

UV-visible irradiation of QDs during room temperature nucleation for two hours decreased the QY considerably (19 to 3%), while the QY remained unchanged during

100°C nucleation (18 to 19%). Light illumination after the microwave treatment (150°C for 80 min) improved the QY for both room temperature and 100°C nucleated samples from 19 to 28% and 18 to 31%, respectively. With illumination both during nucleation and after microwave treatment, the room temperature nucleation QY almost recovered (15%), but QY of the 100°C nucleated sample consistently reached 40-41%.

Characterization of final QDs

Characterization data of the QD sample with the 40% QY can be found in Zane et al. 2014.237 Absorption maximum of the QD sample with the 40% QY was observed at 542

138 nm, with peak emission at 574 nm. High resolution TEM showed lattice fringes indicative of a single crystal and was used to calculate an average QD diameter of 4.6 ± 0.3 nm.

Lifetime (TCSPC) data fell in a range typically reported for QDs in literature (majority contribution at 16 ns lifetime).258, 277, 278 The XRD pattern showed four broad peaks at 26.6,

43.9, 45.8, and 52.2 degrees (2θ), and based on the observed peaks, we assign the QD to have a wurtzite structure with all three chalcogenides of this form.241, 279

Biological imaging of QDs

We have reported previously imaging of microwave-based QDs with QY < 20 % in macrophages.257 With the present QDs having twice the QY, we examined the improvement possible for biological imaging. These experiments were performed in

C2BBe1 cells,35 which are an intestinal epithelial cell line originally cloned from the human colon cancer cell line Caco-2. These cells are widely used as an in vitro model for normal intestinal epithelium and provide a model for uptake of food-relevant NP.198 Cells were treated with QDs for 24 hours and fluorescence of the cells was measured by flow cytometry in order to detect the amount of QDs internalized by cells. Comparing mean fluorescence intensities (MFI) of treated cells revealed that fluorescence could be reliably detected only in cells treated with the QDs with 40% QY at concentrations of 8-39 nM, whereas the QDs with 19% QY were unable to be detected at concentrations of 11-57 nM by flow cytometry (Figure 4.2A). This method will detect fluorescence in cells with both internalized and membrane-bound QDs. To further investigate internalization of QDs, fluorescence confocal microscopy was carried out after 24-hour treatment of cells with

QDs. The QDs with 40% QY were detected within cells at concentrations as low as 0.8 nM

139

(Figure 4.2B). A comparison of the 19% QY (220 nM) and 40% QY (160 nM) QDs shows that the QDs of 40% QY are visibly brighter and were easier to detect by confocal microscopy (Figure 4.3). QDs localized in the cytoplasm of cells and may be contained in vesicles which are visible as depressions in the differential interference contrast (DIC) image in Figure 4.3B.

Figure 4.2. Biological imaging of QDs. A) Flow cytometry showing mean fluorescence intensity of cells treated for 24 hours with QDs nucleated at room temperature (19% QY, old) and optimized QDs nucleated at 100oC (40% QY, new). B) Confocal microscopy of cells treated with 16 nM (left) and 0.8 nM (right) QDs (40% QY) for 24 hours. Arrows point to QDs. Blue staining is DAPI (nuclei), red staining is E-cadherin (cell junctions), and green is QD signal. 140

Figure 4.3. Representative confocal fluorescence images comparing 19% and 40% QY QDs. Cells were treated with A) 19% QY QDs at 220 nM or B) 40% QY QDs at 160 nM for 24 hours before staining and performing confocal microscopy. Arrows point to QDs. The block arrow points to a depression visible in the differential interference contrast (DIC) image that may be a vesicle containing QDs. Blue staining is DAPI (nuclei), red staining is E-cadherin (cell junctions), and green is QD fluorescence. The gray textured background is the DIC image.

141

Fluorophore entrapment by base-driven hydrolysis

Although it is useful to be able to track QDs in biological systems, it would also be useful to pair the fluorescence of QDs with other biologically relevant types of NP. For food-relevant studies, fluorescently-labeled NP that could model the NP used in foods are of interest. Making silica-encapsulated fluorescent NP would allow these particles to be tracked by their fluorescence but behave like food-grade silica in interactions with cells.

This was done with both QDs and dyes encapsulated within a silica coating.

Fluorescent silica particles were synthesized using hydrolysis of TEOS by arginine in the presence of organic dyes or QDs, as depicted in Figures 4.4 and 4.5. Due to the positive charge of the two dyes, they electrostatically bind to the surface of commercially obtained nanosilica cores, and the entire composite is then encapsulated within a silica framework formed by the arginine (base)-catalyzed hydrolysis and condensation of TEOS

(Figure 4.4). The commercial silica particles were used as the core because their morphology is more representative of the particles present in real samples. When using the

QDs prepared with 40% QY as the core, arginine was used to catalyze direct growth of a silica shell on the QD (Figure 4.5).

Experiments with arginine as base

Rhodamine 800

Fluorescence intensity as a function of dye and TEOS concentration was used to optimize Rhodamine 800/silica synthesis with arginine as base.238 Figure 4.6A shows the

TEM of the starting silica particle (diameter 20 ± 5 nm), and Figure 4.6B shows the

Rhodamine 800/silica particle (diameter of 32 ± 15 nm). There was no loss in fluorescence

142 intensity upon incubation of Rhodamine 800/silica in PBS (pH 7.4) for one week (Figure

4.6C).

Figure 4.4. Dye/Silica synthetic pathway. Positively charged dye molecules electrostatically bind to commercial silica cores, and are exposed to TEOS and arginine (base) to form silica shells.

TEOS QD QD Arginine SiO2

Figure 4.5. QD/Silica synthetic pathway (arginine-driven synthesis). QDs are reacted with TEOS and arginine without the addition of an emulsion. Arrows indicate individual quantum dots in TEM image. The TEM inset shows a single CdSe/CdS/ZnS QD particle.

143

Figure 4.6. TEM images and fluorescent properties of dye/silica particles. A) TEM images of initial core silica NP; typical diameter is 20 ± 5 nm. B-E) Arginine- driven silica coating of core silica NP shown in A). B) TEM images of Rhodamine 800/silica particles, 32±15 nm in diameter (n=54). C) Fluorescence spectra of particles after 1 week in PBS (black) and supernatant after PBS incubation (red), indicating no leakage of dye. D) Transmission electron microscopy images of Rhodamine 6G/silica particles, 28±11 nm (n=55). E) Fluorescence spectra of particles after 1 week in PBS (black) and supernatant after PBS incubation (red), indicating no leakage of dye. 144

Rhodamine 6G

Rhodamine 6G/silica particle synthesis conditions were optimized based on quantum yields. Figure 4.6D shows the TEM of the Rhodamine 6G/silica particle, with an average size of 28 ± 11 nm. These Rhodamine 6G/silica samples were stable against dye leakage in PBS (pH 7.4) for one week (Figure 4.6E).

Quantum dots

As the concentration of arginine used in the QD/silica synthesis increased, the particles became more aggregated, with more QDs per silica particle (Figure 4.7). The optimized QD/silica (2.81×10-3 M arginine) based on QY (25%) displayed little particle aggregation and exhibited stable fluorescence throughout the 7 days of suspension in PBS

(pH 7.4), as shown in Figure 4.8.

Characterization

Zeta potential

Similar zeta potentials for dye and QD/silica particles in comparison to commercial silica across a range of pH values (approximately -30 mV at a pH of 5, approaching 0 mV near pH 1 to 2, Figure 4.9) suggests that the synthesized dye/silica and QD/silica particles are comparable in surface functionality with no positively charged dye molecules on the silica surface (dye/silica particles).

Optical properties

Upon silica coating, slight changes in the UV/Vis absorption spectra and shifts in the fluorescence spectra of Rhodamine 6G and Rhodamine 800 were observed (Figure

145

4.10A-D), but the absorption and emission spectra of QD/silica remain unchanged (Figure

4.10E,F).

Figure 4.7. Representative TEM images of QD/silica particles. TEM of QD/silica particles synthesized for 48 hours with a) 9.38×10-4 M, b) 2.81×10-3 M, c) 5.63×10-3 M, and d) 1.13×10-2 M arginine. Primary particle sizes measured to be a) 22 ± 3 nm, b) 29 ± 6 nm, c) 29 ± 5 nm, and d) 31 ± 8 nm.

146

Figure 4.8. Quenching of QD/silica particles in PBS. Fluorescence quenching of silica-coated QDs prepared by arginine silica synthesis after 0 and 7 days in PBS. No quenching is observed. λex = 375 nm

Figure 4.9. Zeta potential titration of silica cores, Rhodamine 6G/silica, Rhodamine 800/silica, and QD/silica. Similar values for all three indicate that dye particles are completely encapsulated within a silica shell. IEP values were 1.91 for commercial silica, 1.34 for Rhodamine 6G/silica, and 1.70 for Rhodamine 800/silica. QD/silica did not cross the IEP.

147

Figure 4.10. Optical spectra of fluorescent core/shell silica particles. UV/Vis absorption spectra of A) Rhodamine 6G, C) Rhodamine 800, and E) QDs before and after silica coating. Fluorescence spectra of B) Rhodamine 6G, λex = 480, D) Rhodamine 800, λex = 635, and F) QDs, λex = 375, before and after silica coating.

148

Infrared

Diffuse reflectance infrared Fourier transform spectroscopy (DRIFTS) spectra of

Rhodamine 6G and QD/silica particles revealed in-plane stretching Si-O vibration attributed to Si-O-H groups at 963 cm−1, indicating silanol groups on these samples (Figure

4.11A). The entire spectra are shown in Figure 4.12, and peaks at 800, 963, 1,104, and

1,206 cm−1 closely match literature values.280

NMR

The 29Si CP-MAS NMR spectra of the arginine-derived Rhodamine 6G and

QD/silica particles show Q2 and Q3 silicon coordination peaks, indicating the presence of silanols (Figure 4.11B).281

Biological studies

In vitro studies

The internalization of QD/silica and Rhodamine 6G/silica by the mouse macrophage cell line MH-S was readily observed by confocal fluorescence microscopy.

After 24-hour treatment, both dye/silica and QD/silica were observed in almost all macrophage cells surrounding the nucleus but not within the nucleus (Figure 4.13).

Internalization of QD/silica and Rhodamine 6G/silica was also observed in C2BBe1 cells as a model of intestinal epithelium, cells that are likely to come into contact with ingested silica particles. After both QD/silica and Rhodamine 6G/silica treatments, silica particles were internalized by C2BBe1 cells and are visible in the cytoplasm of cells surrounding the nucleus (Figure 4.14).

149

Figure 4.11. Infrared and 29Si nuclear magnetic resonance spectroscopy of arginine-driven Rhodamine 6G and QD/silica. A) DRIFTS spectra of QD and Rhodamine 6G silica coating driven by arginine, focusing of ~800 cm-1 framework peak and ~960 cm-1 silanol peak. Samples were lyophilized and vacuum-dried at 110°C for at least 12 hours and then diluted to approximately 1% by weight/KBr. Sample runs were set to continue for 20 minutes from 600 to 4500 cm-1 with 2 cm-1 resolution. B) 29Si NMR spectra of silica-coated QD and Rhodamine 6G driven by arginine. Q4, Q3, and Q2 peaks are labeled. Samples were run with the same preparation as the DRIFTS experiment. 29Si chemical shifts were referenced to tetramethylsilane (δ 29Si= 0 ppm). The 29Si NMR spectra of the samples were obtained using a Bruker two- channel probe in a 4 mm rotor spun at 5 kHz. All spectra were carried out with a proton- to-silicon cross polarization time of 5 ms, relaxation delay of 5 sec, and ~100 kHz proton decoupling.

150

Figure 4.12. QD and Rhodamine 6G/silica particle DRIFTS spectra. DRIFTS spectra of silica-coated a) QD and b) Rhodamine 6G driven by arginine, showing full spectrum. Samples were lyophilized and vacuum dried at 110°C for at least 12 hours and then diluted to approximately 1% KBr by weight. Sample runs set to continue for 20 minutes from 600 to 4500 cm-1, with 2 cm-1 resolution.

In vivo studies

NP (Rhodamine 6G/silica and Rhodamine 800/silica, 1 mg each) were orally administered to mice. Figure 4.15A shows a fluorescence image of the stomach of a mouse

151

Figure 4.13. QD/silica and Rhodamine 6G/silica nanoparticles in macrophages. MH-S cells were grown to confluence in an 8-chamber slide and treated for 24 hours with a 100 µg/cm2 dose of A) QD/silica and B) Rhodamine 6G/silica particles. Cells were fixed in 4% paraformaldehyde, stained with DAPI (blue; cell nuclei), and analyzed by confocal fluorescence microscopy. Both particles are shown in green.

administered silica four times over the four hours prior to euthanasia. The stomach appears to be full of Rhodamine 6G/silica (Rhodamine 800 could not be imaged with the 152

Figure 4.14. QD/silica and Rhodamine 6G/silica in intestinal epithelial cells. The intestinal epithelial cells C2BBe1 were grown to confluence in an 8-chamber slide and treated for 24 hours with a 100 µg/cm2 dose of A) QD/silica and B) Rhodamine 6G/silica particles. Cells were fixed in 4% paraformaldehyde, stained with DAPI (blue; cell nuclei) and E-cadherin (red; junctions between cells), and analyzed by confocal fluorescence microscopy. Both particles are shown in green.

153

Figure 4.15. Rhodamine 6G/silica in mouse tissues. Mice were administered optimized Rhodamine 6G/silica particles orally every 24 hours for four days and organs were harvested three hours after the final administration. Confocal microscopy of frozen tissue sections are shown: a) stomach, b) kidney, c) lung, d) brain, and e) spleen. Sections were stained with DAPI (cell nuclei; blue) and E-cadherin (junctions between epithelial cells; red). Rhodamine 6G/silica particles are green.

154 microscope). These particles appeared to be contained in the stomach contents but generally not internalized by the cells lining the stomach. Rhodamine 6G/silica was also readily observed in the small intestine, cecum, and colon of a mouse treated daily for four days (1 mg/day) and euthanized three hours after the final treatment (Figures 4.16, 4.17,

4.18). The dye/silica particles visible in these sections are often near the edge of the lumen of the GI tract, indicating that they may be trapped in the mucus layer that protects the epithelial cell lining. It is not surprising to find orally-administered particles throughout the GI tract as they pass through the digestive system. To determine if particles traversed the intestinal epithelium and entered the circulation to gain access to other tissues/organs, confocal fluorescence images of sectioned kidney, lung, brain, and spleen were taken, as shown in Figure 4.15B,C,D,E, respectively. Rhodamine 6G/silica signal was found in all of these tissues. Liver sections were also examined and no Rhodamine 6G/silica particles were observed in this tissue. However, this may be due to high background fluorescence.

A control experiment with a mouse that was not administered any of the particles showed only a background fluorescence (described in Experimental Section).

In another experiment, a comparatively dilute mixture of Rhodamine 6G/silica (0.1 mg) and QD/silica (0.67 mg) were administered once a day for four total administrations.

Confocal fluorescence microscopy demonstrated the presence of particles in the gastrointestinal tract as well as in the liver. Figure 4.19A shows the fluorescence image of a liver tissue section, indicating the presence of silica particles. Liver tissues were also digested and analyzed by ICP-MS. Figure 4.19B shows the Cd/Si ratio of the liver as compared to the as-synthesized samples of the silica/QD. The absolute concentrations for

155

Figure 4.16. Rhodamine 6G/silica particles in small intestine tissue. Mice were administered optimized Rhodamine 6G/silica particles orally every 24 hours for four days and organs were harvested three hours after the final administration. Confocal microscopy of frozen small intestine tissue section is shown. Section was stained with DAPI (cell nuclei; blue) and E-cadherin (junctions between epithelial cells; red). Rhodamine 6G/silica particles are green.

each sample were QD/silica Cd-1.9 ppm, Si-150 ppm in a diluted silica/QD NP suspension and Cd-6.2 ppb, Si-480 ppb in the solution digests of liver tissue. The fact that the Cd/Si ratio is similar for the as-synthesized QD (79) and the liver digest (77) indicates unequivocally that the liver section examined by ICP-MS contains the QD/silica NP rather than a dissolved silicate.

156

Figure 4.17. Rhodamine 6G/silica particles in cecum tissue. Mice were administered optimized Rhodamine 6G/silica particles orally every 24 hours for four days and organs were harvested three hours after the final administration. Confocal microscopy of frozen cecum tissue section is shown. Section was stained with DAPI (cell nuclei; blue) and E-cadherin (junctions between epithelial cells; red). Rhodamine 6G/silica particles are green.

157

Figure 4.18. Rhodamine 6G/silica particles in colon tissue. Mice were administered optimized Rhodamine 6G/silica particles orally every 24 hours for four days and organs were harvested three hours after the final administration. Confocal microscopy of frozen colon tissue section is shown. Section was stained with DAPI (cell nuclei; blue) and E-cadherin (junctions between epithelial cells; red). Rhodamine 6G/silica particles are green.

158

Figure 4.19. QD/silica particle detection in mouse tissue. A) QD/silica particles in liver tissue section. Mice were administered QD/silica particles orally every 24 hours for four days and organs were harvested three hours after the final administration. Confocal microscopy of frozen liver tissue section is shown. Sections were stained with DAPI (cell nuclei; blue) and E-cadherin (junctions between epithelial cells; red). QD/silica particles are green. B) ICP/MS measurements of Cd and Si concentration from a solution of QD/silica particles (black) and from a liver tissue sample (red). Si was measured at a resolving power (R= m/Δm) of 300 while Cd was measured at a resolving power of 4000.

159

4.4 Discussion

The advantages of microwave heating for synthesis of QD have been enumerated.242, 256, 259 Besides uniform dielectric heating of the reaction volume, influence on the collisions between reactants and entropic effects due to the rotation of the dipoles are considered important in explaining kinetics and reaction yields. We have reported earlier that with the composition Cd4:Se1:Zn4:MPA20, microwave synthesis led to QD of nominal structure with a CdSe core and an alloyed shell involving CdS and ZnS, and represented the QD as CdSe/CdS/ZnS. The QY of these QDs was < 15%.257 In the present study, we examine the same composition, with the goal to further understand and optimize this synthesis. Fluorescence spectroscopy during the microwaving process is also being used for the first time; previous efforts have reported on the use of infrared spectroscopy.282

The higher QY QDs synthesized were then used as the core for encapsulation by silica to be able to track NP representative of commercial silica by fluorescence in biological systems.

QD growth mechanism

We propose a reaction mechanism for the microwave synthesis of the QDs. A population of CdSe seed NP forms immediately upon mixing cadmium and selenium ions.

These particles are protected from aggregation and further reactions by the thiol-linked protection of 3-MPA. The CdS cap forms after 3-MPA decomposition and the resulting release of sulfur. The growth process continues with the Zn2+ depositing as an outer shell of ZnS on the CdSe/CdS particles, resulting in a red-shift of the emission. The isosbestic point observed is assigned to the disappearance of the CdSe/CdS particle and formation of

160

CdSe/CdS/ZnS QD. This would indicate that the ZnS deposition occurs later than CdS

2+ because of the higher energy necessary to decompose the Zn(NH3)4 complex.

The four regions of the rate of change in fluorescence intensity plot for the room temperature nucleated sample heated in the microwave at 135oC can be correlated with the observations above as follows. Region A between 10-30 min depicts an upward slope, and is related to the growth of the CdSe nuclei. Region B has a downward slope (though there is still positive growth in fluorescence intensity) and is proposed to arise from the deposition of the CdS shell. Region C is the incorporation of the ZnS shell. Region D continues with the incorporation of the ZnS shell, but as the quality of the QD improves, the self-quenching of the QDs becomes more pronounced, and leads to a decreasing rate of fluorescence intensity. If the microwaving temperature is increased to 155°C, these regions merge into one another. Similar observations were also made with the sample nucleated at 100°C.

The reaction temperature during the microwave process does not impact the optical properties of the QDs, as long as the temperatures are within the range of 130 to 155°C and enough time is given to reach the optimized state. The unchanged λmax with microwave temperature indicates that the nuclei formed during the nucleation are responsible for the final optical properties of the QD. The microwave temperature is influencing the rates at which the various processes occur, including growth of nuclei by incorporation of the limiting nutrients and deposition of CdS followed by ZnS. With continued microwave treatment beyond the optimum time at a fixed temperature or at temperatures exceeding

155°C, the QY drops, which can be related to deposition of thicker ZnS layers241, 283 and

161 extensive MPA decomposition with loss of the surface passivation of QD.242 Figure 4.20 provides a mechanistic description of the growth.

Figure 4.20. Synthetic pathway of CdSe/CdS/ZnS microwave-assisted synthesis.

Controlling λmax of emission

The primary effect of the temperature at which CdSe cores are formed prior to microwave irradiation is to alter the emission maximum, with the peak emission wavelength red-shifting from 546 to 574 nm. At the lower temperature of nucleation, smaller-sized nuclei are formed, since the Cd2+ is coordinated with the MPA. The influence of ligand detachment on growth of ZnSe QDs has been carefully studied.284 As nucleation temperatures increase, more of the Cd2+ is released, forming larger nuclei, assisted by ligand detachment from the surface of the nucleated state. Once these nuclei are subjected to microwave, they grow into CdSe cores, with the lower temperature nucleated state producing smaller CdSe cores and more of them as compared to the higher temperature nucleation, since the nutrient pool is the same in all cases. So, for a fixed composition, manipulating the nucleation temperature controls the core particle size.

162

Improving quantum yield

There have been many reports on improvement in QY by light illumination after QD synthesis.285-289 The improvement arises from dissolution of surface defects and their passivation. We show that light illumination during nucleation is also a method to manipulate nuclei structure. Indeed, light illumination during nucleation (2 hours) and after microwave treatment (6 hours) had a profound effect on the QY, especially for the nucleation carried out at 100°C. The QY is improved considerably (18 → 40%), with unchanged λmax at 574 nm. For the nucleation at 22°C, illumination during nucleation followed by microwaving led to significant decrease in QY (19% → 3%), which could be improved by illumination after microwave treatment (QY: 3% → 15%). We explain these results as arising from the fact that the nuclei formed at 22°C are less stable when exposed to irradiation. Nanocrystals of CdTe with a large defect structure have been reported to dissolve during photochemical etching.290 However, for the nuclei prepared at 100 °C, the nuclei are larger and more stable such that light illumination restructures the surface, vacancies can be satisfied with the 3-MPA, and there will be minimal change in size. This restructured surface is further improved with post-microwave illumination. To date, the highest quality microwave synthesis of aqueous CdSe/ZnS particles were 38% QY, and thus the method reported here is an improvement.

Biological imaging of QDs

The biological imaging studies shown in Figures 4.2 and 4.3 clearly demonstrate that with the higher QY (40%), more dilute concentrations of QDs can be detected by both flow cytometry and confocal microscopy. From a biological perspective, the uptake of QDs

163 by the intestinal epithelial C2BBe1 cells is of interest, since NP used in food can enter the circulation and be further distributed once transported through these epithelial cells, and thus QDs can be used as a model for evaluating the rate of uptake of NP by these cells.

QDs can also be coated with food-relevant materials such as silica to serve as fluorescent models of commercial NP as is described below.

Fluorescent labeling of silica nanoparticles

Production of fluorescent dye-modified silica is an extensively researched field. A large number of dyes have been conjugated to silica NP, primarily by covalent linkage, and used in biological research for imaging and targeting of specific cells by appropriate surface modification of the silica.270 There are also reports of bioanalysis and biodetection using fluorescently labeled silica particles.270 Different synthesis protocols are reported, but most studies do not report quantum yield values for bound dye, which would allow for direct comparison of the syntheses. Besides the brightness, stability of the particles and their ability to retain dyes is crucial for certain applications. Ammonia-based Stöber or the microemulsion method for silica NP usually results in porous silica shells.271, 291 Recent studies have reported that use of basic amino acids such as lysine and arginine results in more dense shells.272, 273 Our interest is to simulate commercial silica particles, and therefore the dye needs to be encapsulated within the silica.

Stability of fluorophore/silica particles

Using arginine as base produced particles with stable fluorescence properties

(Figures 4.6 and 4.8), with no leaching of dye or quenching of the QD, even after one week in PBS. The stability of arginine-synthesized silica particles with entrapped dyes has been

164 noted before, and attributed to slower reaction kinetics, allowing for complete hydrolysis of TEOS in silica monomers and condensation to form larger particles.272 We report that the arginine method also leads to silica-entrapped QD with stable fluorescence.

Optical characteristics of dye/silica particles

Figure 4.4 shows the structures of the Rhodamine 6G and Rhodamine 800 dyes; both have a positive charge, and associate with the negatively charged silica seed surface by electrostatic interactions. The highest fluorescence intensity for the final dye/silica particles was obtained with concentrations of 1.32×10-3 M Rhodamine 800 and 1.36×10-3

M Rhodamine 6G. Above this concentration, the silica seeds aggregated because dye adsorption leads to a decrease in surface charge, reducing the repulsion between the silica particles. There was also an ideal TEOS concentration of 8.78×10-5 M for optimum fluorescence. We speculate that at higher concentrations, the anionic hydrolyzed silicate species can bind the positive dye molecules, removing them from the seed core prior to condensation on the core.

Mechanism of QD incorporation into the QD/silica system

The optimum QD/silica was obtained using 2.81×10-3 M arginine and has a QY of

25%. With CdTe QDs, an increase of initial QY from 8-10 to 15% upon silanization has been noted,268 but the stability of the particles was not examined. In another study focused on conferring stability to the CdSe QD, multiple shell layers were required, and a complex

7-layer shell structure had a QY of 47.8%, while in particles with simpler shell structures, the shell allowed penetration of reactants that quenched the fluorescence of the QDs.269

165

The TEM images in Figure 4.7 show that as the arginine concentration gradually increases from 9.38×10-4 to 1.13×10-2 M, the number of QDs in the silica particle increases, and at the highest concentrations of arginine, the silica coating actually occurs on chains of QDs. Figure 4.21 explains the presence of multiple QDs in a single silica particle. The initial Si-OH groups on the QDs (formed by replacing the MPA) can condense with similar groups on neighboring QDs, forming dimers and higher order chains. Thus, the extent of connectivity for a specific concentration of QDs will be dependent on the concentration of silicate species in the medium. The silicate concentration increases with arginine concentration. The larger number of silicate groups on the QD favors the connectivity between the QDs (Figure 4.21B). There have been several previous observations that have noted that increase in the hydrolyzing base concentration introduces QD aggregation.265,

267, 292, 293 With the lowest concentration of arginine (9.38×10-4 M), there was almost one

QD/per silica particle, but the coating was not thick enough for stable fluorescence.

Application of the fluorescent silica particles in biological studies

Particles that leach dye or in which the QD fluorescence is not stable are not suitable for imaging studies. Other relevant characteristics include NP size and surface charge which determine NP behavior in biological systems.4, 5, 119, 203, 205, 206, 239 Thus, if dye molecules are associated with the particle surface, they can change the surface characteristics. We have previously shown that negatively charged QDs quickly associated with cell surface scavenger receptors while positively charged QDs did not.119 Zeta potential versus pH titrations shown in Figure 4.9 confirm that the surface properties of the dye/silica as well as QD/silica NP are comparable to the commercial silica particles. Thus,

166

Figure 4.21. Potential pathways of QD/silica coating. A) Low concentration of arginine, silanol groups do not cross-link between QDs. B) High concentration of arginine, silanol groups cross-link, leading to QDs bound together within silica shell.

both the arginine-derived dye/silica and QD/silica particles developed in this study are suitable for simulating commercial silica particles as they traverse through biological systems, both in vitro and in vivo.

The dye/silica as well as the QD/silica particles are taken up by both macrophages and the intestinal epithelial cell line C2BBe1 (Figures 4.13, 4.14). The epithelial cell line in particular is important since it is relevant to NP ingestion.198 We have reported that

167 titania, silica, and zinc oxide are internalized by C2BBe1 cells, indicating that there is the potential for particles to translocate from the gastrointestinal tract into the circulation and gain access to other tissues and organs.198 We have also noted, using a simulated digestion, that silica particles aggregate in stomach conditions (pepsin solution, pH 2), but then deagglomerate in intestinal conditions (neutral solutions of pancreatin and bile salts).198

The present dye/silica and QD/silica particles with similar surface characteristics as commercial silica particles afforded a way to track these particles in vivo by fluorescence microscopy. Upon administration of Rhodamine 6G dye/silica particles to mice by gavage, we noted the presence of particles in multiple tissues both within and outside of the digestive system. In the gastrointestinal tract, we observed particles in the stomach, small intestine, cecum, and colon (Figures 4.15A, 4.16, 4.17, 4.18). We also observed particles in the kidney, lung, brain, and spleen, indicating that these particles do enter the systemic circulation and subsequently localize within several organs (Figure 4.15B,C,D,E). In comparison, tissues from a control mouse exhibited background fluorescence only. A previous study has noted that oral exposure to commercial silica NP in mice resulted in liver toxicity.144, 158 Uptake of thiol-coated negatively-charged silica particles of sizes ranging from 95-1050 nm by Peyer’s patches in mice have been examined.5 Though such particles can be targeted towards specific cells because of their surface functionality, their uptake is not representative of silica NP present in foods.

The QD/silica NP provided us with another novel opportunity that addresses a long- term issue in tracking NP in living systems.1, 261 Presence of ingested silica from foods in tissues is usually determined by elemental analysis. However, the elemental analysis does

168 not address whether detected silica came from a NP. This issue relates to how nanosilica particles present in the gastrointestinal tract are observed, whether as dissolved silicate species, or in their pristine nanoform, which will have major implications in their toxicity.1

We administered low levels of Rhodamine 6G/silica and QD/silica by gavage, and detected fluorescent particles in the liver by fluorescence confocal microscopy (Figure 4.19A).

However, there was no way to distinguish whether the dye/silica or the QD/silica (or both) were being observed by microscopy. Sections of the liver were also examined using ICP-

MS. As Figure 4.19B shows, the Cd/Si ratio from the liver sample is very similar to that of the free QD/silica particles, suggesting that the observed silica particle contains QD and not the dye. Thus, by exploiting the unique Cd/Si ratio of the QD, ICP-MS of the QD/silica- containing tissues did confirm that the silica indeed came from the particles that were introduced by gavage and not from the background. There are two possible reasons why we detect the QD, rather than the dye. The first reason could be the size: the QD/silica is

22 nm, and the dye/silica is 28 nm. Size-dependent internalization of NP in tissues in rats administered NP by gavage has been known for decades. A higher percentage of 50 nm polystyrene spheres as compared to 100 nm spheres was found in liver, spleen, blood and bone marrow.4 Intestinal tissue uptake of ~100 nm polymer particles was 15-250 fold higher than micron-sized particles.203 However, considering the closeness in size, this may not be the main reason behind the preferential uptake of the QD. A more likely reason is that the QD/silica is present at a concentration about seven times that of the dye/silica.

Two studies have noted the presence of CdTe QD/silica and CdHgTe QD/silica particles in various organs following direct injection into the blood stream via the tail

169 vein.268, 294 The present study is more relevant to ingested NP in food and passage of NP through the gastrointestinal tract.

Conclusions

We are reporting several synthetic strategies for QD synthesis all starting with the composition Cd4:Se1:Zn4:MPA20. In situ fluorescence spectroscopy during microwave synthesis of the QD provided insight into the growth process, including growth of the CdSe nuclei and deposition of the CdS and ZnS shell layers. The temperature during the microwave treatment (130 – 155°C) led to QDs with the same QY and λmax. Changing the nucleation temperature (0 – 100°C) prior to microwave growth produced QD of varying size with increasing λmax (range 546 – 574 nm). UV-visible light illumination in the nucleation step and post-microwave treatment led to optimal QDs with QY of 40 - 41%.

These QDs were examined as biological imaging agents. Uptake of QDs into intestinal epithelial cells could be readily observed at 0.8 to 8 nM QD concentrations by confocal microscopy and flow cytometry, respectively.

We also show that an arginine-based silica-shell synthesis is appropriate for encapsulation of quantum dots and Rhodamine-family dyes. Particles showed no dye leaching or fluorescence quenching within one week of exposure to biological conditions.

These particles were strongly fluorescent with quantum yields of 20 and 25% for

Rhodamine 6G and QD/silica particles, respectively. Zeta potential titrations indicate that the surface of these fluorophore/silica NP are similar to commercial particles, making them suitable for monitoring the fate of these particles in biological systems. The fluorophore/silica particles were readily imaged in macrophages and epithelial cells.

170

Particles were detected via fluorescence in several non-gastrointestinal organs of mice after oral exposure, indicating that some population of the silica NP translocates across the digestive system following ingestion. For the QD/silica particles, the ICP-MS and analysis of Cd/Si ratio can unequivocally identify the presence of NP in tissues. We have shown here that stable fluorescent silica NP with similar surface characteristics as commercial silica particles can enter circulation upon ingestion, indicating a need for further study of their bioaccumulation and potential toxicity.

171

Chapter 5: Oxidative stress-mediated inhibition of intestinal epithelial cell

proliferation by silver nanoparticles3

5.1 Introduction

Nanotechnology is being increasingly exploited by a variety of industries for the unique properties materials display at the nanoscale. Silver nanoparticles (Ag NP) are used for their increased catalytic activity. Optical properties of Ag NP are also being exploited in chemical and biological sensors.179 Another property of Ag NP is their strong antimicrobial activity179 which has found application in water purification, wound dressings, prevention and treatment of infection, and other medical uses.295 The Project on

Emerging Nanotechnologies, which attempts to inventory consumer products containing nanotechnologies, currently lists 438 products containing Ag NP.3 These include textiles such as sheets, towels, and socks, curling irons, air purifiers, a stuffed animal, and toothbrushes, all utilizing the antimicrobial properties of Ag NP. The food and food packaging industries are also interested in utilizing Ag for its antibacterial properties and

41 of the products listed by the Project on Emerging Nanotechnologies are in the Food and

Beverage category.3 These include Ag supplements, cookware, utensils, and appliances

3 Accepted in modified form by Toxicology In Vitro, July 2015. Text and figures appear with permission from the editors of Toxicology In Vitro. 172 coated with Ag NP, and Ag NP incorporated into food storage containers to prevent food spoilage. In addition to direct consumption of Ag NP used in foods, some fraction of Ag

NP incorporated into food packaging and food contact materials may be transferred to food, further increasing Ag NP consumption.

It has been shown that Ag NP mediate their antimicrobial effects through oxidative stress, which damages bacterial membranes. Electron spin resonance has been used to detect radicals on the surface of 13.5 nm Ag NP.296 While this microbial toxicity has great potential utility, the same mechanism may contribute to cytotoxicity to mammalian cells.

In medical applications, care can be taken to use doses of Ag NP that will be toxic to microbes but not human cells. However, the intake of Ag NP is less controllable when these NP are being ingested in food and food contact materials.

Ingested Ag NP will be transported through the digestive tract, exposed to digestive enzymes, and excreted from the body if not absorbed in the intestines, as occurs with most non-nutrient components of our food. Intestinal epithelial cells within the intestines are responsible for transport of nutrients from the intestinal lumen to the bloodstream to be used by the rest of the body. NP are likely to come into the most contact with epithelial cells after ingestion. In order to gain access to the circulation to be more widely distributed throughout the body, Ag NP will have to first be transported across the intestinal epithelium. In vivo studies with PLGA, polystyrene, and thiol-organosilica NP between 50-

100 nm have shown that NP are internalized by and transported across the intestinal epithelium into the circulation more readily than larger particles between 100-10,000 nm.4,

5, 203 Studies in which 10-60 nm Ag NP were orally administered to rodents have also shown

173 incorporation of Ag NP into intestinal epithelial cells and transport of NP across the epithelium.182, 183, 190 Thus, interaction of Ag NP with intestinal epithelial cells will be important for the systemic effects of Ag NP ingestion, and intestinal epithelial cells serve as a relevant in vitro model to study impact of NP ingestion.

As the use of Ag NP has increased, there has been concern about environmental accumulation of Ag NP and the potential health impacts of Ag NP in humans. This has led to studies in various systems exploring Ag NP cytotoxicity and the mechanism of response to Ag NP in several cell models. These studies have described an oxidative stress-mediated mechanism of cytotoxicity in cells which leads to DNA damage, mitochondrial damage, cell cycle arrest, and apoptosis.127, 128, 297 Studies in intestinal epithelial cell models have reported Ag NP toxicity including decreased cell viability, formation of reactive oxygen species (ROS), decreased mitochondrial function, increased IL-8 generation by cells, and

DNA damage.49, 67, 68, 129 However, there has been less thorough investigation of the mechanism of Ag NP toxicity in intestinal epithelial models and no investigation of Ag NP effects on cell proliferation. It is known that oxidative stress plays a role in inflammation and disease progression in the intestines, so an increase in oxidative stress by Ag NP in the intestines may be particularly problematic for those with intestinal pathologies.136

In this study, we utilize the intestinal epithelial C2BBe1 model to investigate the mechanism of ~ 23 nm Ag NP cytotoxicity on proliferating and confluent cells. Low dose

24-hour Ag NP treatment at 0.25 µg/cm2 induced ~ 15% necrosis and ~ 80% loss of metabolic activity in proliferating cells (defined as < 10,000 cells/cm2). Ag NP treatment of proliferating cells for 24 hours decreased the ratio of reduced to oxidized glutathione,

174 indicating induction of oxidative stress. Ag NP treatment at 0.25 µg/cm2 also completely inhibited cell proliferation and induced G2/M phase cell cycle arrest. Ag NP that underwent simulated in vitro digestion before cell exposure required a slightly higher dose (0.5

µg/cm2) to induce the same inhibition of cell proliferation. Inhibition of cell proliferation

2 at low doses (0.25 µg/cm ) was unique to Ag and not observed after SiO2, TiO2, or ZnO

NP treatment, although ZnO inhibited cell proliferation at higher doses (10 µg/cm2). Based upon these observations, we hypothesize that Ag NP increase cellular oxidative stress, leading to cell cycle arrest, and thus inhibit cellular proliferation. Ag NP were not directly toxic to confluent cells (defined as 100,000 cells/cm2) up to doses of 10 µg/cm2.

5.2 Experimental Procedures

All Ag NP synthesis and characterization (dynamic light scattering and zeta potential measurements, TEM analysis, infrared spectroscopy, Raman spectroscopy, X-ray diffraction, and atomic absorption spectroscopy) and simulated gastrointestinal digestion of NP was performed by Dr. Andrew Zane in Dr. Prabir Dutta’s laboratory in The Ohio

State University Department of Chemistry and Biochemistry.

Nanoparticles

Zinc oxide (ZnO), titania (TiO2), and silica (SiO2) NP were purchased from Sigma-

Aldrich (St. Louis, MO). The specifications of the ZnO particles were size ≤ 100 nm, with

2 a specific surface area of 15-25 m /g. For TiO2, the particle size was specified as 21 nm, with surface area of 35-65 m2/g, purity of ≥ 99.5%, and a trade name of Aeroxide P25.

Silica particle specification included a size of 12 nm, surface area of 175-225 m2/g, and purity of 99.8%.

175

Silver Nanoparticle synthesis

Ag NP (~20 nm) were synthesized as follows. A 200 mL solution of 0.25 mM

AgNO3 and 0.25 mM trisodium citrate was prepared in purified water and stirred for 30 minutes. Six mL of a 10 mM NaBH4 solution was slowly added while stirring. The solution was stirred for 30 minutes, then left at room temperature for 24 hours. Particles were washed twice by centrifugation at 209,000×g for 30 minutes, removal of supernatant, and redispersion in purified water.

Simulated gastrointestinal digestion of Ag

Pepsin, pancreatin, and bile salts were used to simulate the gastric and small intestinal digestive environments in vitro.198 The concentrations used were based on in vitro digestion methods used in previously published studies.208-211 Briefly, pepsin (stomach,

146 U/ml, Sigma-Aldrich), pancreatin (intestinal enzyme mixture, 2 mg/ml, Sigma-

Aldrich), and bile extract (porcine, 0.024 mg/ml, Sigma-Aldrich) were dissolved in water and adjusted to a pH of 2 (pepsin) or 7 (pancreatin and bile extract).

Ag NP (50 mg/L) were incubated sequentially in the pepsin, pancreatin, and bile extract solutions for one hour each at 37oC. Subsequent to each incubation, the NP were pelleted by centrifugation (209,000×g for 30 minutes) and resuspended in the next solution.

After incubation in the bile extract, NP were centrifuged, resuspended in phosphate- buffered saline (PBS), and used for biological studies. Ag NP treated in this manner are hereafter referred to as “digested NP” while untreated NP are referred to as “pristine.”

176

Nanoparticle Characterization

Particle Size and Surface Charge

A Zetasizer Nano ZS (Malvern, Westborough, MA) was used to measure the zeta potential of the pristine and digested Ag NP. The Nano ZS uses a 633 nm laser as its light source. For zeta potential measurements, a forward angle of 12° was used for collecting light. The default Smoluchowski model in the software program was used. Each measurement included 20 runs, and monomodal analysis provided by the vendor was used for analysis. Samples were titrated versus pH using an attached MPT-2 Autotitrator. The titrator was supplied with 0.1 M HCl and 1.0 M HCl. Three replicate measurements were taken at each pH with a two minute pause between all measurements.

TEM Images of Pristine Silver Nanoparticles

Images were taken using a Tecnai F20 Transmission Electron Microscope. A 10 mg/L solution of pristine Ag NP in ethanol was prepared and sonicated for 30 minutes. The solution was dropped onto lacey carbon copper TEM grids (Ted Pella, Inc., Redding, CA) and allowed to dry for several hours.

Infrared Spectroscopy

Diffuse reflectance infrared Fourier transform spectroscopy (DRIFTS) was performed on pristine Ag NP and digested Ag NP. For the digested Ag, after the final digestion step, the particles were washed twice with water. The digested particles were isolated by centrifugation and frozen with liquid nitrogen, then placed in a Millrock Bench

Top Manifold Freeze Dryer (Millrock, Inc. Kingston, NY) to preserve any potential coating. The pristine Ag NP were washed twice, redispersed in water, frozen with liquid

177 nitrogen, and placed in the freeze dryer. DRIFTS analysis was performed with a Spectrum

400 FTIR Imaging System (Perkin Elmer, Waltham, MA). Scans were performed in percent reflectance mode, a KBr background was collected, and then the samples were added to the cell (approximately 10% wt. in KBr) and the spectrum recorded. Kubelka

Munk analysis was performed with instrument software.

Raman Spectroscopy

Raman spectroscopy was recorded with a Renishaw InVia Raman Microscope using a 633 nm laser source and a CCD detector. For the digested Ag, after the final digestion step, the particles were washed twice with water. The digested particles were isolated by centrifugation and frozen with liquid nitrogen, then placed in a Millrock Bench

Top Manifold Freeze Dryer (Millrock, Inc., Kingston, NY) to preserve any potential coating. The pristine Ag NP were washed twice, redispersed in water, frozen with liquid nitrogen, and placed in the freeze dryer.

Atomic Absorption Spectroscopy

Atomic absorption spectroscopy was performed to test for the presence of dissolved

Ag ions after incubation of both pristine Ag and digested Ag NP in DMEM cell culture media for 0, 1, and 2 days. Supernatants used in the elemental analysis were isolated from the samples by centrifugation at 209,000×g for 45 mins. A Buck Scientific Accusys 211 spectrophotometer (East Norwalk, CT) was used with a Ag hollow cathode lamp

(Heraeus). Standard solutions were diluted from a 1000 ppm standard prepared by dissolving AgNO3.

178

X-ray Photoelectron Spectroscopy

X-ray photoelectron spectroscopy (XPS) surface analysis of pristine and digested

Ag NP was carried out using a Kratos Axis Ultra system (Kratos Analytical, Manchester,

UK) equipped with monochromated Al radiation. All binding energies for different elements were calibrated with respect to the C 1s line at 285 eV.

Cell culture

C2BBe1 cells were obtained from the American Type Culture Collection

(Manassas, VA) at passage 47 (as specified by the vendor). The majority of these experiments were performed with cells between passages 75-94. Cells were cultured in

Dulbecco’s Modified Eagle Medium (DMEM; Life Technologies, Grand Island, NY) supplemented with 10% fetal bovine serum (FBS; Serum Source International, Inc.,

Charlotte, NC), 1 mM sodium pyruvate, 2 mM L-glutamine, 0.3% penicillin/streptomycin,

0.3 ug/mL Amphotericin B (fungizone), and 10 µg/mL transferrin (all from Life

Technologies). Cells were incubated in 5% CO2/95% room air at 37°C. Cells were passaged every 5-7 days and plated on flasks or plates from Corning-Costar (Tewksbury,

MA) pre-coated with collagen I (0.05 mg/mL, rat tail, Life Technologies). Cells were incubated at least 24 hours prior to treatment with NP.

Treatment of cells with nanoparticles

Non-silver NP were weighed and sterilized by steam sterilization, then suspended in PBS to make a 1 mg/mL solution to use to treat cells. Ag NP were collected by ultracentrifugation and resuspended in water, generally to a concentration of 0.1 or 1 mg/mL. Immediately prior to treating cells, NP solutions were sonicated using a VC130

179

Sonics Vibra-Cell sonicator (Sonic Materials, Inc. Norwalk, CT) pulsing for one second on, one second off for approximately 15 seconds in order to break up NP agglomerates before treatment of cells. NP were added to cells at the appropriate dose (between 0.05 and

10 µg/cm2). To ensure consistency, doses were based on the area of the cell culture well

(available from Corning for each plate sized used) rather than volume of cell culture media added to cells. To promote contact of NP with cells, plates were centrifuged at 300×g for

15 minutes.

TEM of nanoparticle-treated cells

7.4, 10 minutes each), then post-fixed in 1% osmium tetroxide in sym-collidine buffer (pH

7.6) for 1 hour at room temperature. Following two washes with s-collidine buffer (10 minutes each) the pristine Ag NP-treated cells were en-bloc stained with a saturated aqueous uranyl acetate solution (pH 3.3) for 1 hour. Digested Ag NP-treated cells were not stained with the uranyl acetate to decrease contrast. All cells were dehydrated in a graded ethanol series up to absolute (10 minutes each). Acetone was used as the transitional

180 solvent for two 10-minute washes. Cells were infiltrated overnight with a 1:1 mixture of acetone and Spurr’s epoxy resin (Electron Microscopy Sciences, Fort Washington, PA).

Finally, the cells were centrifuged and the pellet was placed into BEEM embedding capsules containing 100% Spurr’s resin. Polymerization of epoxy blocks was carried out at 70°C overnight. Polymerized blocks were sectioned with a Leica Ultracut UCT ultramicrotome (Leica Microsystem GmbH, Wein, Austria). Ultrathin (80 nm) sections were collected on 200 mesh copper grids (Electron Microscopy Sciences) and post-stained with lead citrate (3 minutes). Electron micrographs were generated on a JEOL JEM-1400

TEM (JEOL Ltd. Tokyo, Japan) equipped with a Veleta digital camera (Olympus Soft

Imaging Solutions GmbH, Münster, Germany).

Toxicity assays

For the 24-hour acute toxicity assays described below (Sytox Red, Annexin V,

LDH, and MTT), cells were seeded in 24-well tissue culture plates at a density of 2×104-

1×105 cells/well and grown to 90-100% confluency. For the Sytox Red and Annexin V assays performed with proliferating cells, cells were plated in 6-well tissue culture plates at a density of 5×104-1×105 cells/well the day before treatment with Ag NP. The LDH and

MTT assays in proliferating cells were performed on cells seeded in 24-well tissue culture plates at a density of 1×104 cells/well the day before treatment with Ag NP. To normalize the cell densities we are defining as proliferating versus confluent/stationary cells, the cell density was calculated per cm2 based on the surface area of one well of a culture plate

(available from Corning).

181

Sytox Red staining

To evaluate NP-induced necrotic cell death (cellular membrane damage), NP- treated cells were stained with Sytox Red Dead Cell Stain (Life Technologies) and analyzed by flow cytometry. Cells were treated with NP for 24 hours, then detached from the culture plate with trypsin, suspended in 1 mL calcium- and magnesium-free Hank’s

Balanced Salt Solution (HBSS, Life Technologies), and stained with 1 µL of the Sytox Red

Dead Cell Stain. Samples were incubated for at least 15 minutes at room temperature before analyzing for fluorescence using a FACScalibur flow cytometer (BD Biosciences) at an excitation wavelength of 635 nm. All experimental conditions were performed in triplicate.

As a positive control for cell death, a subset of wells were treated with 20 mM hydrogen peroxide (H2O2) in DMEM at the same time as NP treatment, 24 hours prior to harvesting cells for flow cytometric analysis.

Annexin V-FITC staining

To examine whether cells treated with NP underwent apoptosis, cells were stained with FITC-conjugated Annexin V (BD Biosciences, San Jose, CA). Cells were plated, treated with NP, and harvested with trypsin as described above. Cells were resuspended in

100 µL of Annexin binding buffer (Life Technologies). Five µL of FITC Annexin V was added to each tube and the samples were incubated for 15 minutes in the dark at room temperature before adding an additional 400 µL of Annexin binding buffer. Fluorescence of bound FITC Annexin V was then detected by flow cytometry using an excitation wavelength of 488 nm. All experimental conditions were performed in triplicate. As a positive control for apoptosis, a subset of wells were treated with 20 mM H2O2 in DMEM

182 at the same time as NP treatment, 24 hours prior to harvest of cells for flow cytometric analysis.

LDH assay

To further assess cell death by detection of cellular membrane damage, release of the cytosolic enzyme lactate dehydrogenase (LDH) into the culture medium was examined by colorimetric assay. Cells were plated and treated with NP as described above. After a

24-hour incubation, 50 µL of culture media was collected from each well and placed in a flat-bottom 96-well plate (BD Falcon). LDH activity was assessed using a commercially available kit (LDH Assay Kit, Sigma-Aldrich) according to manufacturer’s instructions.

Briefly, LDH Assay Substrate Solution, LDH Assay Cofactor Preparation, and LDH Assay

Lysis Solution were added to samples and incubated at room temperature for 15-20 minutes. The absorbance was read on a microplate reader at 490 nm and 690 nm. Sample absorbance values were corrected by subtracting the background reading at 690 nm from the 490 nm reading. As a positive control for LDH release, cells were treated with a 1% solution of Triton X-100 (Sigma-Aldrich) in DMEM for 5 minutes to lyse all cells before collecting the supernatants. Sample absorbance values were normalized to the Triton positive control (considered as 0% cell viability) and untreated negative control

(considered as 100% cell viability). All experimental conditions were performed in quadruplicate.

MTT assay

To measure the metabolic activity of NP-treated cells, we used a commercially available MTT assay kit (Cayman Chemical, Ann Arbor, MI). In this assay, the tetrazolium

183 dye, MTT, is reduced by mitochondrial NAD(P)H-dependent oxidoreductase enzymes to an insoluble purple crystal, formazan. Cells were plated and treated with NP as described above and incubated for 24 hours at 37°C prior to the assay. As a positive control for complete cell death, cells were treated with a 1% solution of Triton X-100 in DMEM for five minutes immediately before the assay was performed. After the incubation, cell culture media was removed and cells were washed once with PBS to prevent interaction of NP with the MTT reagent. Fresh cell culture media (200 µL) was added to each culture well before the addition of 10 µL of the MTT reagent. The cells were incubated for 3-4 hours at 37°C. The culture media/MTT reagent was removed and formazan crystals were dissolved with the provided MTT solvent. The solution was transferred to a flat-bottom 96- well plate and absorbance values were read using a microplate reader at 570 nm and 690 nm. Absorbance values were corrected by subtracting the background absorbance reading at 690 nm from the absorbance at 570 nm. To analyze and compare the data, absorbance values were normalized to the untreated control cells such that the untreated control represented 100% mitochondrial activity and an absorbance of 0 represented the absence of mitochondrial activity (i.e., 0%). All experimental conditions were performed in quadruplicate.

Assay of cellular proliferation

To assess the effects of NP exposure on proliferating cells, we compared cell proliferation of untreated cells with those treated with NP continuously over a 10 day period. The cells were seeded at a density of 5×104 cells/well so that they would reach confluence after approximately 10 days. Cells were plated in triplicate in 6-well tissue

184 culture plates such that there were triplicate wells of each treatment to harvest and count daily for 10 days (total of 30 counts). The time at which cells were plated was considered time zero. One day after plating cells, cell culture media was changed and cells were treated with appropriate NP doses. NP were allowed to remain on cells for 48 hours, at which point cell culture media was changed and cells were re-treated with NP. Cells were re-treated with NP every 48 hours. Cells were counted daily using a Z2 Beckman Coulter Counter

(Indianapolis, IN) and mean counts of triplicate wells were plotted versus time. Growth curves were normalized to the growth of untreated control cells such that the maximum count measured for the untreated control cells was considered 100% growth.

The experiments performed using antioxidants were done similarly. Antioxidants were dissolved in cell culture media and added to the cells at the same time as the first NP treatment and at every subsequent cell culture media change and NP retreatment. N- acetylcysteine (NAC, Sigma-Aldrich) was used at 10 mM and Trolox ((±)-6-Hydroxy-

2,5,7,8-tetramethylchromane-2-carboxylic acid, Sigma-Aldrich) was used at 0.5 mM.

Optimal doses were determined based on preliminary experiments. After dissolving NAC in cell culture media, pH was adjusted to 7.4 using 1 N NaOH. After adding antioxidants and adjusting pH, cell culture media was filter sterilized before being added to cells.

Cell cycle analysis

In order to evaluate the effects of NP on cell cycle, proliferating cells were treated with NP and analyzed for DNA content by flow cytometry. Cells were plated in 6-well cell culture plates at a density between 5×104 and 1×105 cells/well. After 24-48 hour incubation, cells were treated with the appropriate doses of Ag and digested Ag. As was done for the

185 growth curves, NP were left on the cells for 48 hours, at which time cell culture media was changed and they were treated again with NP. Cells were grown for 5 days after plating such that the untreated cells were approaching confluence but still proliferating. Cells were harvested by trypsinization, washed with PBS, and suspended in 1 mL PBS. To fix cells, 4 mL absolute ethanol (Thermo Fisher Scientific, Waltham, MA) was slowly added to each sample while vortexing in order to minimize aggregation of cells and incubated at -20°C at least overnight and up to two weeks. Fixed cells were washed twice with PBS, suspended in 1 mL staining solution consisting of 0.1% Triton X-100 (Sigma-Aldrich), 100 µg/mL

RNase A (Life Technologies), and 40 µg/mL propidium iodide (Sigma-Aldrich) in PBS.

Cells were incubated in the dark at room temperature for 30 minutes and analyzed by flow cytometry at an excitation wavelength of 488 nm. All experimental conditions were performed in triplicate.

GSH assay

To quantitate cellular oxidative stress induced by Ag NP, total and oxidized

(disulfide form, GSSG) glutathione were measured using the GSH/GSSG-Glo™ Assay kit

(Promega, Madison, WI). In this assay, glutathione is required for the reaction converting

Luciferin-NT to luciferin (catalyzed by glutathione S-transferase, resulting in GSH-NT).

The luciferin is then converted to light in a firefly luciferase reaction and light formed is thus dependent on the amount of glutathione. Total glutathione is measured by lysing cells, reducing any GSSG to reduced glutathione (GSH), and allowing the GSH to react with

Luciferin-NT. GSSG is measured by blocking GSH with N-Ethylmaleimide and then reducing GSSG to GSH and allowing it to react with Luciferin-NT to eventually produce

186 light. Both total GSH and GSSG were measured in order to determine the GSH/GSSG ratio. For the assay, cells were plated in 6-well cell culture plates at a density of 5×104 or

1×105 cells/well. Cells were allowed to adhere to the plate for one day prior to treatment with Ag NP. Cells were treated with the appropriate dose of pristine and digested Ag NP, centrifuged to promote contact of NP with cells, and incubated for 24 hours. Cells were then harvested by trypsinization, resuspended in cell culture media, washed once in HBSS, and resuspended in HBSS at a cell density of 5×105 cells/ml. The assay was then performed according to the provided protocol, halving the volume of all reagents for each reaction.

Briefly, 10 µl of cell suspension was added to each well along with 2.5 µL HBSS. Total

Glutathione Lysis Reagent (12.5 µL) was prepared and added to wells to measure total glutathione while Oxidized Glutathione Lysis Reagent (12.5 µL) was added to a second set of wells to measure oxidized glutathione. The plate was shaken for 5 minutes before adding

25 µL Luciferin Generation Reagent to all wells. The plate was then incubated for 30 minutes at room temperature. After the incubation, Luciferin Detection Reagent (50 µL) was added to wells, the plate was incubated for 15 minutes at room temperature, and the luminescence was measured using a Packard TopCount (PerkinElmer, Waltham, MA). A standard curve was prepared at the same time with known concentrations of glutathione from 0-16 µM. The linear portion of the standard curve was used to determine the concentrations of GSH and GSSG measured. The Pierce BCA protein assay (Thermo

Scientific) was used to report concentration based on protein. The assay was performed according to the provided protocol. Briefly, 100 µL of cell suspension was added to a test tube followed by 2 mL of the prepared working reagent (50 parts BCA Reagent A and 1

187 part BCA Reagent B). Tubes were incubated at 37°C for 30 minutes, cooled to RT, and absorbance was measured at 562 nm. A standard curve was prepared with known concentrations of BCA and used to calculate total protein in each treatment. GSH/GSSG ratios were calculated as follows (there are two moles of GSH per one mole of GSSG):

GSH/GSSG = ([total glutathione] – [GSSG]*2) / [GSSG]. Each GSH assay was performed using three biological replicates of each treatment.

5.3 Results

Silver nanoparticle synthesis

Citrate-stabilized Ag NP were produced by reduction of Ag ions with borohydride using a previously reported synthesis.298 In order to simulate the in vivo gastrointestinal environment, Ag NPs were incubated sequentially with digestive enzyme solutions of pepsin (pH 2, stomach), pancreatin (pH 7, intestines), and bile extract (pH 7, intestines) for one hour each as has been previously described.198 After incubations, particles were centrifuged and re-suspended in PBS, and experiments were performed with both pristine and “digested” Ag.

Characterization

Transmission electron microscopy (TEM) of the synthesized Ag NP revealed largely spherical particles with an average diameter of 23 ± 8 nm (Figure 5.1A). Figure 5.2 compares the zeta potential over a range of pH values, with the pristine Ag NP having a more negative charge under all conditions. The pristine Ag NP had an average particle diameter of 90.9 ± 2.8 nm prior to digestion and 346 ± 105 nm after digestion, as measured by dynamic light scattering.

188

A) B) Pristine Ag 1165 Digested Ag 1078 862 951 1673 1369

KBM

1574 1376 811 1015

1800 1600 1400 1200 1000 800 Wavenumber (cm-1)

C) 80000 166 Pristine Silver 70000 1040 Digested Silver 60000 1270 1315 50000 1486 40000 1558 1600 2913 30000 766

Raman Intensity 20000 377 124 1589 10000 238 2926 0 0 1000 2000 3000 4000 Wavenumber (cm-1)

Figure 5.1. Characterization of pristine and digested Ag NP. A) TEM of pristine Ag NP. The average diameter of Ag particles is 23 ± 8 nm. B) DRIFTS spectra of freeze-dried pristine and digested Ag NP. C) Raman spectra of freeze-dried pristine and digested Ag NP.

Infrared spectra for pristine and digested Ag NP reveal differences in the surface of the NP (Figure 5.1B), with significantly higher intensity of the infrared bands on the digested Ag NP, indicating more adsorption. The peaks at 1376 and 1574 cm-1 in the pristine sample are the carboxylate vibrations from the adsorbed citrate.299 For the digested 189

10 Digested Silver Pristine Silver 0

-10

-20

-30

-40

Zeta Potential (mV)

-50

1 2 3 4 5 6 7 8 pH

Figure 5.2. Zeta potential of pristine and digested Ag NP. Zeta potentials were measured as Ag NP solutions were titrated from neutral to acidic pH with addition of HCl.

Ag NP spectrum, the peak at 1673 cm-1 is likely due to protein adsorption, specifically β- sheet structures,218 and the other peaks in the digested sample at 862, 951, 1078, and 1165 cm-1 correspond to bile salts adsorbed to the NP surface.219 Raman spectra differ between the pristine and digested Ag as well (Figure 5.1C), with very strong peaks in the pristine sample. These peaks are characteristic of surface enhanced Raman spectroscopy (SERS)- enhanced citrate and citrate photodecomposition products on the Ag NP surface (especially the strong bands at 1040, 1315, 1486, 1556, and 1600 cm-1),300 and are not observed on the digested Ag NP. Thus, during digestion other molecules have adsorbed on the Ag surface, and the SERS enhancement is lost. Figure 5.3 shows a survey scan of the X-ray photoelectron spectra (XPS) for pristine and digested Ag NP and specific peaks are identified in Table 5.1. Figure 5.4 shows a comparison of the various elements on the NP

190 surface, with only carbon and oxygen peaks on the pristine Ag NP, as expected from citrate ions, while there are much larger sodium, chlorine, phosphorus, and nitrogen peaks for the digested Ag NP (S is not observed on either sample).

Digested Silver Pristine Silver

1400 1200 1000 800 600 400 200 0 BE (eV) Figure 5.3. XPS spectra of pristine and digested Ag (BE = binding energy).

Identified elemental peak Pristine Ag (eV) Digested Ag (eV) P - 132.0 Cl - 196.9, 198.5, 268.0 C 283.3, 287.1, 291.1, 294.0 283.3, 291.1, 294.0 Ag 367.3, 373.3, 377.0 367.3, 373.3, 377.0 N - 398.3 O 530.5 530.5, 534.1 Na - 495.0, 1069.9 Table 5.1. Peaks corresponding to specific elements detected in XPS spectra of pristine and digested Ag NP.

191

A) Ag 3d B) C 1s Digested Silver Digested Silver Pristine Ag Pristine Silver

385 380 375 370 365 360 295 290 285 280 275 BE (eV) BE (eV) C) D) Na 1s O 1s Digested Silver Digested Silver Pristine Silver Pristine Silver

545 540 535 530 525 520 1080 1075 1070 1065 1060 BE (eV) BE (eV)

E) Cl 2p F) P 2p Digested Silver Pristine Silver Digested Silver Pristine Silver

208 206 204 202 200 198 196 194 192 190 140 138 136 134 132 130 128 126 124 BE (eV) BE (eV) Figure 5.4. XPS spectra for both pristine and digested Ag NP corresponding to individual elements. XPS spectra for A) silver, B) carbon, C) oxygen, D) sodium, E) chlorine, F) phosphorous, G) nitrogen, and H) sulfur. Continued

192

Figure 5.4: Continued G) N 1s H) S 2p Digested Silver Digested Silver Pristine Silver Pristine Silver

405 400 395 390 172 170 168 166 164 162 160 158 BE (eV) BE (eV)

The Ag ion concentration in the supernatant after ultracentrifugation (209,000×g) was determined for pristine and digested Ag NP after 0, 24, and 48-hour incubation in cell culture media. In all cases, the concentration of Ag ions in solution was very small, on the order of 0.2-0.6 ppm, as compared to the total concentration of 1250 ppm if all of the Ag

NP were to be dissolved (Figure 5.5).

Cellular internalization of NP

Internalization of NPs by C2BBe1 intestinal epithelial cells (a sub-clone of Caco-2 cells), was analyzed by transmission electron microscopy. As the TEM in Figure 5.6 shows, the cells internalize both pristine and digested Ag NP. The digested Ag NP were in larger agglomerates within cells. The internalized particles were not found in cell nuclei and were not contained within clear vesicles, but seemed to be clustered in the cell cytoplasm.

The amount of Ag NP uptake by cells treated with 10 µg/cm2 Ag NP was analyzed with ICP-MS. The untreated cells contained <0.015 ± 0.002 ppm Ag/ 106 cells while the pristine and digested Ag NP-treated cells contained 1.97 ± 0.34 ppm and 1.43 ± 0.16 ppm

193

Ag per 106 cells, respectively (means of three replicate samples). Because of the cell preparation, this measurement includes any Ag firmly attached to the cell surface (which did not wash off cells) but not internalized by the cells.

Digested Ag NPs Incubated in Media A) 0.6

0.5

0.4

0.3

0.2

0.1

Ag Concentration (ppm)

0.0 0 24 48 Time in Media (hrs)

B) As Prepared Ag NPs Incubated in Media

0.35

0.30

0.25

0.20

0.15

0.10

Ag Concentration (ppm) 0.05

0.00 0 24 48 Time in Media (hrs) Figure 5.5. Ag NP dissolution in cell culture media. A) 0.5 µg/cm2 digested Ag and B) 0.25 µg/cm2 pristine Ag NP were incubated in cell culture media and Ag ions in solution were detected by elemental analysis after incubation for 0, 24, and 48 hours.

194

Figure 5.6. Internalization of Ag NP by intestinal epithelial cells in vitro. Cells were treated with A) pristine and B) digested Ag NP for 24 hours (10 µg/cm2), then suspended, pelleted, and fixed for transmission electron microscopy.

Toxicity in proliferating cells

To examine toxicity of Ag NPs in cells, Sytox Red staining of DNA in cells and analysis by flow cytometry was used to detect necrotic cell death. Annexin V FITC was

195 used to label apoptotic cells for detection by flow cytometry. Colorimetric detection of lactate dehydrogenase (LDH) released from cells was used as a measure of membrane damage. Finally, the MTT assay was used as a measure of mitochondrial activity in cells.

Cells plated at low density (< 10,000 cells/cm2) were treated with 0.25 µg/cm2 pristine or

0.5 µg/cm2 digested Ag NP for 24 hours before performing each of these assays. No apoptosis and minimal increase in LDH release was detected (Figure 5.7B,C). The LDH assay was not performed with digested Ag as the digestive enzymes were previously observed to interfere with the assay.198 However, the Sytox Red staining detected 15.4% and 19.8% induction of necrotic cell death for pristine and digested Ag NP, respectively, and the MTT assay detected 76.2% and 85.7% reduction in cell metabolic activity for pristine and digested Ag NP, respectively (Figure 5.7A,D). Table 5.2 provides details of the proliferating cell toxicity assay data.

In order to test for Ag interference with the LDH assay, LDH detection was compared for untreated cell lysate and cell lysate exposed to Ag NP and a dramatically decreased amount of LDH was detected after incubation of cell lysate with Ag NP (Figure

5.8). This is likely due to Ag NP binding to LDH which interfered with its detection in the asasy. Thus, although the LDH assay data with Ag NP was included in this work, less LDH may have been detected after Ag NP treatment of cells than was truly there and thus Ag

NP may be more toxic than detected by this assay.

196

* A) 120 * 100 80 60 40

% viability% 20 0 no NP H2O2 Ag Digested Ag B) 120 100 80 60 40

% viability% 20 0 no NP H2O2 Ag Digested Ag C) 120 * 100 80 60 40

% viability% 20 0 no NP Triton Ag

D) * 120 * 100 80 60

activity 40 20 % mitochondrial mitochondrial % 0 no NP Triton Ag Digested Ag

Figure 5.7. Toxicity of Ag NP to proliferating cells. Cells were plated at low density and allowed to adhere overnight before treatment with 2 2 0.25 µg/cm pristine Ag, 0.5 µg/cm digested Ag, and 20 mM H2O2 or 1% Triton X-100 as a positive control. A) Sytox Red staining (necrosis) and B) FITC Annexin V staining (apoptosis) were analyzed by flow cytometry. Data represent the means of three replicates normalized to untreated control cells ± 1 standard deviation. C) LDH assay (cell membrane damage). D) MTT assay (mitochondrial activity). LDH and MTT data represent the means of four replicates normalized to untreated control cells ± 1 standard deviation. The LDH assay was not used with digested Ag NP because the digestive enzymes were previously found to interfere with the assay.198 Significance was calculated using Student’s t-test (* indicates p < 5×10-6). Data are compilations of two replicate experiments.

197

Assay Pristine Ag NP Digested Ag NP Sytox 84.6 ± 3.1 80.2 ± 2.5 (% viability) Annexin V 98.0 ± 5.2 97.3 ± 6.2 (% viability) LDH 91.0 ± 3.8 - (% viability) MTT 23.8 ± 4.9 14.3 ± 4.3 (% mitochondrial activity) Table 5.2. Proliferating cell toxicity assay data. Normalized percent viability and percent mitochondrial activity is reported for proliferating cells treated with 0.25 µg/cm2 pristine Ag and 0.5 µg/cm2 digested Ag for all toxicity assays. Data were normalized to untreated cells as 100% viability and 100% mitochondrial activity for all assays. The LDH assay was not performed with digested Ag NP because of digestive enzyme interference. Data are normalized means from 3 replicate wells for Sytox and Annexin V assays and 4 replicate wells for LDH and MTT assays. The LDH assay was not performed with digested Ag NP because of digestive enzyme interference. Data represent a compilation of 2 replicate experiments for Sytox, Annexin V, LDH, and MTT assays.

140

120

100

LDH (percent of control) of (percent LDH 20

0 control Ag

Figure 5.8. Interference of Ag NP with the LDH assay. Cells were grown to confluence, lysed, and equal volumes of cell lysates were used as untreated controls and exposed to Ag NP. The assay was then completed as normal. Data are the mean of four replicate wells ± one standard deviation. Data were normalized to the untreated control as 100% LDH.

198

Oxidative stress

In order to investigate the role of oxidative stress induction by Ag NP, total glutathione and oxidized glutathione levels were measured in cells and used to calculate the ratio of reduced to oxidized glutathione (GSH/GSSG). Cells were plated at low density

(< 10,000 cells/cm2), allowed to adhere to plates overnight, and treated with Ag NP (doses of 0.25 ug/cm2 for pristine and 0.5 ug/cm2 for digested Ag). The optimal Ag NP exposure time for GSH/GSSG measurement was determined to be 24 hours based on time course experiments comparing GSH/GSSG ratios after 30 minute, 6 hour, and 24 hour exposures

(Figure 5.9A) as well as 4 day exposure (Figure 5.9B). GSH/GSSG ratios were found to be significantly reduced only after 24-hour exposure (Figure 5.9A). After 24-hour treatment with Ag NP, an increase in GSH within cells (30%) was observed (Figure 5.10A), but the increase in GSSG was disproportionately greater (176% and 216% for pristine and digested Ag NP, respectively, Figure 5.10B), leading to a significant and repeatable decrease in the GSH/GSSG ratio after treatment of cells with pristine or digested Ag NP

(Figure 5.10C). Table 5.3 provides the measured concentrations of GSH and GSSG. This suggests that Ag NP treatment does induce oxidative stress in intestinal epithelial cells.

Cell cycle analysis

Cell cycle analysis was performed using propidium iodide staining and flow cytometry analysis on proliferating cells. Cells were plated at low density (< 10,000 cells/cm2). After adhering overnight, cells were treated with Ag NP and then treated again two days later when cell culture media was changed. Experiments were performed 5 days after plating cells. Cell cycle analysis revealed an increase in G2/M phase cells (from 23.3

199

A) 60 50 40 30

20 GSH/GSSG 10 * 0 untreated 30 minutes 6 hours 24 hours

B) 20

GSH/GSSG 5

0 untreated 4 days

Figure 5.9. Time course of GSH/GSSG ratios after treatment of cells with Ag NP. C2BBe1 cells were plated at low density and allowed to adhere to cell culture plates overnight. The following day cells were treated with 0.25 µg/cm2 Ag NP. GSH/GSSG ratios were calculated based on detection of total and oxidized glutathione levels by luminescence after incubations with Ag NP of A) 30 minutes, 6 hours, 24 hours, or B) 4 days (Ag NP were replaced along with cell culture media after a two-day incubation). Data shown in A) and B) represent separate experiments. Data were collected in triplicate and are portrayed as mean ± one standard deviation. Student’s t-test was used to measure significance in comparison to untreated control (* represents p-value<0.0005).

to 49.3%) paired with a decrease in G0/G1 phase cells (from 53.8 to 31.1%) at a dose of

0.25 µg/cm2 pristine Ag (Figure 5.11A). At a dose of 0.5 µg/cm2 digested Ag NP, an increase in G2/M phase cells from 23.3 to 47.4% with a decrease in G0/G1 phase cells from 53.8 to 28.5% was observed (Figure 5.11B). Table 5.4 provides the percentages of cells in each cell cycle phase at all doses. The fluorescence histograms (Figures 5.11C-E) 200

A) 80 * 70 * 60 50 40 30 20

10 GSH (µmol/g protein) (µmol/g GSH 0 no NP Pristine Ag Digested Ag B) 9 8 ** 7 ** 6 5 4 3 2

1 GSSG (µmol/g protein) (µmol/g GSSG 0 no NP Pristine Ag Digested Ag

C) 25

10 ** ** GSH/GSSG 5

0 no NP Pristine Ag Digested Ag

Figure 5.10. GSH/GSSG after 24-hour Ag NP treatment. Levels of A) reduced glutathione (GSH), B) oxidized glutathione (GSSG), and C) GSH/GSSG ratio in cells treated with Ag NPs. Cells were plated at low density and allowed to adhere overnight before treating with Ag NPs for 24 hours. The cells were then harvested and GSH and GSSG levels were measured by luminescence. Data are representative of 5 experiments for pristine Ag and 3 experiments for digested Ag. Data represent mean concentrations for 3 replicate wells ± 1 standard deviation. Significance in comparison to untreated control was determined using Student’s t-test (* represents p-value<0.01, ** represents p-value<0.0005).

201

Treatment GSH (µmol/g protein) GSSG (µmol/g protein) GSH/GSSG Untreated 47.2 ± 3.0 2.37 ± 0.20 17.9 ± 1.6 Pristine Ag NP 60.5 ± 3.3 6.55 ± 0.55 7.2 ± 0.4 Digested Ag NP 62.7 ± 4.6 7.48 ± 0.30 6.4 ± 0.9 Table 5.3. GSH assay data. GSH and GSSG levels are reported as µmol/g protein along with the GSH/GSSG ratios for proliferating cells treated with no NP, 0.25 µg/cm2 pristine Ag NP, and 0.5 µg/cm2 digested Ag NP. Data are means of 3 replicate wells ± 1 standard deviation.

clearly show the increase in G2/M phase cells and decrease in G0/G1 phase cells in response to treatment with 0.25 µg/cm2 pristine Ag (Figure 5.11D) or 0.5 µg/cm2 digested

Ag (Figure 5.11E) over untreated cells (Figure 5.11C). Interestingly, the histograms shown in Figures 5.11D and 5.11E also demonstrate that Ag NP-treated cells (both digested and pristine) included small, yet distinct sub-populations in which DNA appears to have aberrantly proceeded through a second cycle of replication, as indicated by peaks at approximately 7-8n. Mechanisms underlying this dysfunction remain to be resolved.

Alternatively, these peaks may represent adherent cell doublets.

Inhibition of cell proliferation

Growth curves were performed to determine the impact of Ag NP exposure upon cell proliferation. Cells were plated at low density (< 10,000 cells/cm2) and allowed to adhere to the plate overnight. After 24 hours, cells were treated with Ag NP and subsequently retreated with Ag NP every 48 hours when cell culture media was changed.

A subset of cells were counted each day. Untreated cells proliferated and reached a plateau

9 – 10 days following plating, as expected (Figure 5.12A). However, a dose of 0.25 µg/cm2 pristine Ag induced complete inhibition of cell proliferation over the course of the 10 days.

202

A) 100% B) 100% 80% 80% * 60% ** 60% * G2/M phase 40% 40%

20% S phase 20%

% % gatedcells % % gatedcells 0% G0/G1 phase 0% no NP 0.05 0.1 0.25 no NP 0.05 0.1 0.25 0.5 Treatment (µg/cm2) Treatment (µg/cm2) C)

G0/G1 S G2/M

Figure 5.11. Cell cycle phase distribution (DNA content) of Ag NP-treated C2BBe1 cells. Proliferating intestinal epithelial cells were treated with increasing doses of pristine and digested Ag NP for 3 days. Cells were stained with propidium iodide, and DNA content was measured by fluorescence flow cytometry. Cell cycle phase was determined by DNA content: G0/G1 phase - 2n DNA, G2/M phase - 4n DNA, and S phase - 2n increasing to 4n. Percentages of cells in each phase were determined for A) pristine Ag NP and B) digested Ag NP. Data are representative of 6 replicate experiments. Significance compared to G2/M phase untreated cells was measured by Student’s t-test (* indicates p<0.0005; ** indicates p<0.00005). Representative flow cytometry histograms are shown for cells treated with C) no NP, D) 0.25 µg/cm2 pristine Ag NP, and E) 0.5 µg/cm2 digested Ag NP.

203

Pristine Ag: Treatment G0/G1 phases (% cells) S phase (% cells) G2/M phases (% cells) No NP 53.8 22.9 23.3 0.05 µg/cm2 51.4 23.6 24.9 0.1 µg/cm2 54.6 18.3 27.0 0.25 µg/cm2 31.1 19.5 49.3 Digested Ag: Treatment G0/G1 phases (% cells) S phase (% cells) G2/M phases (% cells) No NP 53.8 22.9 23.3 0.05 µg/cm2 53.6 22.9 23.6 0.1 µg/cm2 52.6 21.1 26.3 0.25 µg/cm2 41.5 19.5 39.0 0.5 µg/cm2 28.5 24.1 47.4 Table 5.4. Cell cycle data. Percentages of cells treated with pristine and digested Ag NP in G0/G1, S, and G2/M phases of the cell cycle are listed as determined by flow cytometry. Data are means of 3 replicate wells.

Cells treated with digested Ag NP showed a similar inhibition of cell proliferation, but at a dose of  0.5 µg/cm2 Ag (Figure 5.12B).

Cell proliferation after treatment with NP was shown to be dependent on cell passage number. Growth curves where cells were counted every two days showed that cells at lower passage number (passage 55) were more sensitive to 1 µg/cm2 pristine Ag NP treatment than cells at higher passage number (passage 73, Figure 5.13). To minimize variation in our data due to passage number, experiments with proliferating cells were performed between passages 75-94. As this experiment shows, some variation in the dose of Ag NP required to inhibit cell proliferation was observed in earlier experiments and is most likely due to variation in the Ag NP size distribution or the proportion of dissolved ions in solution. Later experiments consistently showed complete inhibition of cell proliferation at a dose of 0.25 µg/cm2 and above. 204

A) 120 100 80 No NP 60 0.1 ug 40 0.25 ug 20 0.5 ug 0 Normalized cell count cell Normalized 1 ug 0 100 200 300 Time (hours)

B) 120 100 80 No NP 60 0.1 ug 40 0.25 ug 20 0.5 ug

Normalized cell count cell Normalized 0 1 ug 0 100 200 300 Time (hours)

Figure 5.12. Inhibition of intestinal epithelial cell proliferation by Ag NP. Cells were plated at low density, allowed to adhere for 24 hours, then treated with various doses of A) pristine Ag NP and B) digested Ag NP. At 48-hour intervals NP-supplemented culture medium was replaced with fresh NP-supplemented medium. Samples of cells were counted at 24-hour intervals. Data points represent mean normalized cell counts from 3 replicate wells ± 1 standard deviation (bars). Data were normalized to the maximum cell count for the untreated control cells as 100% cell growth. Data are representative of results of 3 replicate experiments with pristine Ag and 2 replicate experiments with digested Ag.

Growth curves were also performed with SiO2, TiO2, and ZnO, other food-relevant

NP. Figure 5.14 shows that SiO2 and TiO2 induced partial inhibition of cell proliferation at the highest dose of 10 µg/cm2 (Figure 5.14A, B). ZnO NP induced dose-dependent growth inhibition with complete inhibition of cell proliferation at a dose of 10 µg/cm2 (Figure

5.14C). Thus, Ag NP were able to completely inhibit cell proliferation at a dose 40 times lower than ZnO, and far lower yet as compared to SiO2 and TiO2. Removal of NP from 205

A) Passage 55 120

100

60 No NP 40 Ag

Normalized cell count cell Normalized 20

0 0 50 100 150 200 250 300 Time (hours)

B) Passage 73 120

100

60 No NP 40 Ag

Normalized cell count cell Normalized 20

0 0 50 100 150 200 250 300 Time (hours)

Figure 5.13. Sensitivity to Ag NP-induced inhibition of proliferation depends on cell passage. Cells were plated at low density, allowed to adhere overnight, and treated with 1 µg/cm2 Ag NP which were replaced each time cell culture media was changed. Cells were counted every other day up to 11 days. Data points represent normalized mean counts of three replicate wells ± one standard deviation. Data were normalized to the maximum cell count for the untreated control as 100% cell growth.

2 cells treated continuously with 10 µg/cm SiO2, TiO2, and ZnO NP over 10 days (NP were washed off at day 11 of the growth curve) allowed recovery of the cell populations. SiO2

206

A) 140 120 100 No NP 80 1 ug silica 60 2.5 ug silica 40 20 5 ug silica Normalized cell count cell Normalized 0 10 ug silica 0 100 200 300 Time (hours)

B) 120 100 80 No NP 60 1 ug TiO2 40 2.5 ug TiO2 20 5 ug TiO2

Normalized cell count cell Normalized 0 10 ug TiO2 0 100 200 300 Time (hours)

C) 120 100 80 No NP 60 1 ug ZnO 40 2.5 ug ZnO 20 5 ug ZnO

Normalized cell count cell Normalized 0 10 ug ZnO 0 100 200 300 Time (hours)

Figure 5.14. Inhibition of intestinal epithelial cell proliferation by food-relevant NP. Cells were plated at low density, allowed to adhere for 24 hours, then treated with various doses of A) SiO2, B) TiO2, or C) ZnO NP. At 48-hour intervals NP-supplemented culture medium was replaced with fresh NP-supplemented medium. Samples of cells were counted at 24-hour intervals. Data points represent mean normalized cell counts from 3 replicate wells ± 1 standard deviation (bars). Data were normalized to the maximum cell count for the untreated control cells as 100% cell growth. Data are representative of results of 3 replicate experiments.

207

140 120 100 80 No NP 60 Silica 40 Titania 20 Zinc oxide 0

Normalized cell growth cell Normalized 0 100 200 300 400 500 Time (hours)

Figure 5.15. Recovery of SiO2, TiO2, and ZnO-treated cell populations after NP removal. Growth curves were set up by plating cells at low density, allowing cells to adhere for 24 hours, and treating cells continuously with 10 µg/cm2 of the appropriate NP. Cells were counted daily through day 9. After 10 days of continuous NP treatment (day 11 of the growth curve), NP were washed off cells and cells were allowed to recover. Cell populations were counted again after 4 and 7 days of recovery (days 15 and 18 of the growth curve). Data points represent mean normalized cell counts from 3 replicate wells ± 1 standard deviation (bars). Data were normalized to the maximum cell count for the untreated control cells as 100% cell growth. Data are representative of two replicate experiments.

120 100 80 No NP 60 40 nm 40 200 nm 20 2000 nm 0 Normalized cell growth cell Normalized 0 100 200 300 Time (hours)

Figure 5.16. Growth of cells treated with 40, 200, and 2000 nm zeolite particles. As previously described, cells were plated at low density, allowed to adhere to plates for 24 hours, then treated continuously with 10 µg/cm2 of 40, 200, or 2000 nm zeolite particles for a total of 10 days. Cell growth was counted daily. Data points represent mean normalized cell counts from 3 replicate wells ± 1 standard deviation (bars). Data were normalized to the maximum cell count for the untreated control cells as 100% cell growth. Data are representative of two replicate experiments.

208 and TiO2-treated populations had plateaued after 7 days of recovery and ZnO-treated cells were growing exponentially (Figure 5.15). Growth curves with 40, 200, and 2000 nm zeolite particles also revealed partial inhibition of cell proliferation with all of these zeolite particles and no size-dependent differences were observed (Figure 5.16).

Influence of N-acetylcysteine (NAC) treatment

Growth curves were performed as previously described, but with the addition of 10 mM NAC into the cell culture media for the entire time the cells were treated with Ag NP.

Cells were treated with a 0.25 µg/cm2 dose of pristine Ag NP or a 0.5 µg/cm2 dose of digested Ag NP, previously shown to induce complete inhibition of cell proliferation. Thus, the curve with no added antioxidant displays complete inhibition of cell proliferation

(Figure 5.17A). When cells were treated with NAC in addition to Ag NP, there was significant protection from Ag NP-induced inhibition of proliferation and cells treated with

Ag NP appeared to be proliferating almost as rapidly as the untreated cells (Figure 5.17B).

GSH assays performed on cells treated for 24 hours with Ag NP and NAC showed very little change in GSH/GSSG ratio after Ag NP treatment, suggesting that NAC prevented the induction of oxidative stress (Figure 5.17C).

Cells were also pretreated with 10 mM NAC for 1 hour followed by removal of

NAC-containing media from cells before treatment of cells with Ag NP. Growth curves demonstrated that NAC pre-treatment had no impact upon proliferation inhibition of Ag

NP-treated cells within 5 days of Ag NP treatment (Figure 5.18A). The GSH/GSSG ratio of the pretreated cells was slightly higher than that of the cells not treated with NAC (Figure

5.18B). Thus, NAC pretreatment may be slightly protective from induction of oxidative

209

A) 120 100 80 60 No NP 40 Ag 20 digested Ag 0 Normalized cell count cell Normalized 0 100 200 300 Time (hours) B) 120 100 80 60 No NP 40 Ag 20 digested Ag

0 Normalized cell count cell Normalized 0 100 200 300 Time (hours) C) 25 20 15 10

GSH/GSSG 5 0 no NP Pristine NAC NAC + Ag Ag

Figure 5.17. N-acetylcysteine protection from Ag NP-mediated inhibition of cell proliferation and induction of oxidative stress. Intestinal epithelial cells were plated at low density, allowed to adhere for 24 hours, then treated with A) no antioxidant or B) 10 mM NAC concurrent with the addition of 0.25 µg/cm2 pristine Ag NP or 0.5 µg/cm2 digested Ag NP. Medium replenishment (including antioxidants and Ag NP) and cell counts were conducted as described. Data represents normalized mean cell counts from 3 replicate wells ± 1 standard deviation (bars). Data were normalized to the maximum cell count for the untreated control cells as 100% cell growth. Data are representative of 2 replicate experiments. C) GSH/GSSG ratios of proliferating cells treated with 0.25 µg/cm2 Ag NP and 10 mM NAC for 24 hours. Data represent mean ratios for 3 replicate wells ± 1 standard deviation. Data are representative of 3 replicate experiments.

210 stress, but it does not play a large role. Trolox is another common antioxidant, and Figure

5.19 shows that Trolox (0.5 mM) also has no effect on Ag NP-induced inhibition of cell proliferation.

A) 120

100

80 No NP 60 NAC No NP 40 Ag NAC + Ag

Normalized cell count cell Normalized 20

0 0 50 100 150 Time (hours)

B) 14 12 10 8 6

GSH/GSSG 4 2 0 untreated Ag NAC pre NAC pre + Ag

Figure 5.18. Cell growth and GSH/GSSG ratios after pretreatment of cells with NAC prior to Ag NP exposure. Cells were plated at low density and pre-treated with NAC for 1 hour before washing cells and treating with 0.25 µg/cm2 Ag NP. A) Growth of cells pre-treated with NAC and treated continuously with Ag NP over five days. Data represent mean normalized cell counts of 3 replicate wells ± 1 standard deviation. Counts were normalized to the maximum growth observed for the untreated control cells as 100% growth. B) GSH/GSSG ratio measurements of cells pre-treated with NAC before treatment of cells with Ag NP for 24 hours. Data are representative of 3 experiments. Data represent mean concentrations for 3 replicate wells ± 1 standard deviation.

211

120

100

60 No NP Ag 40 digested Ag

Normalized cell count cell Normalized 20

0 0 50 100 150 200 250 300 Time (hours)

Figure 5.19. Impact of Trolox on Ag NP-mediated inhibition of cell proliferation. Intestinal epithelial cells were plated at low density, allowed to adhere for 24 hours, and then treated with 0.5 mM Trolox concurrent with the addition of 0.25 µg/cm2 pristine Ag NP or 0.5 µg/cm2 digested Ag NP. Medium replenishment (including Trolox and Ag NP) and cell counts were conducted as described. Data represent normalized mean cell counts from 3 replicate wells ± 1 standard deviation (bars). Data were normalized to the maximum cell count for the untreated control cells as 100% cell growth. Data are representative of results of 2 replicate experiments.

Cytotoxicity of silver nanoparticles in confluent cells

To examine toxicity of Ag NP in confluent cells (cell density > 100,000 cells/cm2), the cells were treated with 10 µg/cm2 pristine or digested Ag NP for 24 hours. No decrease in cell viability was detected after Ag NP treatment (Figure 5.20A-C). There was a slight decrease in mitochondrial activity after treatment with pristine and digested Ag NP but it was not significant (Figure 5.20D). The LDH assay was not performed with digested Ag as the digestive enzymes were previously observed to interfere with the assay.198

Measurement of GSH/GSSG after 24-hour treatment of confluent cells with 0.25 µg/cm2

212

A) 120 100 80 60 40

% viability% 20 0 no NP H2O2 Ag Digested Ag B) 120 100 80 60 40

% viability% 20 0 no NP H2O2 Ag Digested Ag C) 120 100 80 60 40

% viability% 20 0 no NP Triton Ag D) 140 120 100 80 60

activity 40 20 % mitochondrial % mitochondrial 0 no NP Triton Ag Digested Ag E) 50 40 30 20

10 GSH/GSSG 0 No NP Ag

Figure 5.20. Acute toxicity in intestinal epithelial cells induced by Ag NP. Cells were treated with pristine or digested Ag NP for 24 hours (10 µg/cm2), then analyzed by A) Sytox Red staining (necrosis), B) Annexin V binding (apoptosis), C) LDH assay (cell membrane injury), and D) MTT assay (mitochondrial activity). Data represent means of three (Sytox Red, Annexin V) or four (LDH, MTT) replicate wells normalized to untreated control cells ± 1 standard deviation. Data are compilations of two replicate experiments for Sytox Red, Annexin V, and LDH assays, and three replicate experiments for the MTT assay. E) Confluent cells were treated for 24 hours with pristine Ag NP before measuring the GSH/GSSG ratio. Data represent mean concentrations for 3 replicate wells. Data are representative of 3 experiments. 213

Ag NP revealed no change from the untreated cells (Figure 5.20E). Table 5.5 provides details of the confluent cell toxicity assay data.

Treatment of cells with 10 µg/cm2 pristine Ag NP for 48 hours induced a 12% increase in necrotic cell death, 20% increase in apoptotic cell death, and 41% decrease in mitochondrial activity of cells whereas only a 12% decrease in mitochondrial activity and no apoptosis or necrosis was induced after 24-hour treatment of cells with Ag NP (Figure

5.21). Slight IL-8 production (18.4 pg/mL) was measured by ELISA after 24-hour treatment of confluent cells with 5 µg/cm2 Ag NP whereas no IL-8 was detected in untreated cells (Figure 5.22).

Assay Pristine Ag NP Digested Ag NP Sytox 100.9 ± 1.2 101.8 ± 0.5 (% viability) Annexin V 100.7 ± 1.3 97.5 ± 1.6 (% viability) LDH 109.7 ± 7.9 - (% viability) MTT 90.6 ± 41.3 86.1 ± 30.7 (% mitochondrial activity) Table 5.5. Confluent cell toxicity assay data. Normalized percent viability and percent mitochondrial activity ± standard deviation is reported for confluent cells treated with 10 µg/cm2 pristine and digested Ag NP for all toxicity assays. Data are normalized means from 3 replicate wells for Sytox and Annexin V assays and 4 replicate wells for LDH and MTT assays. The LDH assay was not performed with digested Ag NP because of digestive enzyme interference. Data represent a compilation of 2 replicate experiments for Sytox, Annexin V, and LDH assays and 3 experiments for the MTT assay.

214

A) 120 ** 100 80 60 40

% viability% 20 0 No NP 24 hours 48 hours

B) 120 ** 100 80 60 40

% viability% 20 0 No NP 24 hours 48 hours

C) 140 120 100 80 60 40 % viability% 20 0 No NP 24 hours 48 hours

D) ** 120 * 100 80 60

40 activity 20

% % mitochondrial 0 No NP 24 hours 48 hours

Figure 5.21. Toxicity of 24- and 48-hour exposure of Ag NP to confluent cells. Cells were treated with 10 µg/cm2 pristine Ag NP for 24 or 48 hours before harvest and analysis by A) Sytox Red staining (necrosis), B) Annexin V binding (apoptosis), C) LDH assay (cell membrane injury), and D) MTT assay (mitochondrial activity). Data represent means of three (Sytox Red, Annexin V) or four (LDH, MTT) replicate wells normalized to untreated control cells ± 1 standard deviation. Significance in comparison to untreated cells was determined by Student’s t-test (* represents p-value < 0.05, ** represents p-value < 0.001).

215

100

Concentration Concentration (pg/ml) 0 untreated Ag

Figure 5.22. IL-8 secretion induced by Ag NP treatment of confluent cells. Confluent cells were treated for 24 hours with 5 µg/cm2 Ag NP and IL-8 production was measured by ELISA. Data are presented as means of 2 replicate wells ± 1 standard deviation.

5.4 Discussion

Cellular internalization of Ag NP

The C2BBe1 cell line was cloned from the epithelial Caco-2 cell line for more homogeneous brush border (microvilli) formation. C2BBe1 cells have been shown to express the same brush border proteins found in vivo and thus form a good model of normal intestinal epithelium.35 We observed by TEM that Ag NP were internalized by confluent

C2BBe1 cells after 24-hour exposure to 10 µg/cm2 (50 µg/mL) Ag NP and that these internalized NP localized to the cytoplasm of cells (Figure 5.6), consistent with earlier studies.67 Internalization of Ag NP by Caco-2 cells has previously been shown by TEM with 15 µg/mL of Ag NP < 100 nm67 and 20 µg/mL of 20 nm Ag NP.68 ICP-MS also indicates comparable uptake of Ag for both the pristine and digested samples.

216

Effects of Ag NP on proliferating cells

Exposure of proliferating C2BBe1 cells (< 10,000 cells/cm2) to pristine Ag NP at a dose of 0.25 µg/cm2 (1.25 µg/mL) for 24 hours led to necrotic cell death in 15% of cells and a 76% decrease in metabolic activity (Figure 5.7A,D). Comparable effects were noted with a higher loading of digested Ag NP (0.5 µg/cm2 or 2.5 µg/mL). This induction of necrotic cell death is consistent with several previous studies of Ag NP treatment in Caco-

2 cells.52, 129 The decrease in metabolic activity is also in agreement with several studies of

Caco-2 cells treated with Ag NP.52, 67, 68, 129, 171 We report interference of Ag NP with the

LDH assay (Figure 5.8) which has also previously been observed.227, 301 Treatment of proliferating cells with Ag NP for 24 hours decreased the GSH/GSSG ratio to less than half the level in untreated cells, indicating oxidative stress (Figure 5.10). Other studies in

Caco-2 cells have shown induction of oxidative stress by detection of ROS species129, 130 and upregulation of genes involved in oxidative stress.95 However, some studies have not detected ROS formation in Caco-2 cells upon Ag NP exposure.49, 52, 68 This may be a result of lower sensitivity of ROS detection over measurement of the GSH/GSSG ratio. A study of Caco-2 cells treated with Ag NP < 100 nm did not demonstrate an increase in ROS production but did detect oxidative stress by a decrease in total cellular GSH.67

Alternatively, these studies may not have detected increased ROS due to the different type of Ag NP used (20 nm Ag stored in 2 mM citrate,68 25 nm spheres and 80-90 Ag NP rods,52), treating more confluent stationary phase cells which may be less sensitive to Ag

NP treatment, or using less serum (2% FBS) in media during Ag NP treatment, thereby changing the surface species on the Ag NP.49

217

Ag NP treatment of proliferating cells at 0.25 µg/cm2 (1.25 µg/mL) for 96 hours induced G2/M phase cell cycle arrest (increase in G2/M phase cells from 23 to 49%)

(Figure 5.11). Comparable effects were noted with a higher loading of digested Ag NP (0.5

µg/cm2 or 2.5 µg/mL). To our knowledge, this is the first paper reporting Ag NP-induced cell cycle arrest in intestinal epithelial cells although G2/M phase cell cycle arrest has been reported previously in other cell types including lung epithelial cells, fibroblasts, and glioblastoma cells.127, 302-304 There is a DNA damage checkpoint at the end of the G2 phase of the cell cycle and thus G2/M phase arrest may be allowing cells with DNA damage time to repair their DNA. This model of DNA damage response has previously been proposed as the reason for cell cycle arrest after Ag NP treatment.127, 304

Cell proliferation as measured by growth curves was completely inhibited with 0.25

µg/cm2 (1.25 µg/mL) treatment of cells with Ag NP (Figure 5.12), whereas ZnO NP

2 2 required a dose of 10 µg/cm (40 µg/mL), and 10 µg/cm SiO2 or TiO2 NP only partially inhibited cell proliferation (Figure 5.14). Recovery and growth of SiO2, TiO2, and ZnO- treated cell populations was observed upon removal of NP (Figure 5.15), indicating that this inhibition of proliferation is reversible and not permanent. Inhibition of cell proliferation by ZnO NP is consistent with previous studies in Caco-250, 173 and human lung

305 cells. Inhibition of cell proliferation by high doses (200 µg/mL) of SiO2 has been observed previously in human gastric epithelial cells (GES-1) and Caco-2 cells.147

However, although TiO2 has been shown to inhibit proliferation in non-intestinal cell lines,306, 307 it has generally been observed to be non-toxic in intestinal cells.50, 308 Thus, Ag

NP is the most proficient at inhibition of cell proliferation.

218

Zeolite particles of various sizes (40, 200, and 2000 nm) were also observed to partially inhibit cell proliferation at a dose of 10 µg/cm2 (Figure 5.16). The absence of a size-dependent response to zeolite treatment suggests that this effect may be independent of cell internalization. The partial inhibition of cell proliferation may instead be some kind of mechanical effect of this quantity of material on the cells or association of the particles with the collagen coating the cell culture plate so cells cannot associate as readily with the collagen, resulting in slower growth. The similar pattern of inhibition observed after treatment with SiO2 and TiO2 NP may be due to a similar mechanism and not be a specific response to these NP. However, fine clinoptilolite zeolite with an average size of 2.9 µm has been shown to inhibit cancer cell proliferation by mechanisms including changes in intracellular signaling.309 Thus, the zeolite used here may be able to induce cytotoxicity by interfering with intracellular metabolism similarly to the food-relevant NP even at sizes above the nanoscale.

Since we observed complete inhibition of cell proliferation with 0.25 µg/cm2 (1.25

µg/mL) treatment of cells with Ag NP (Figure 5.12A) but not did not observe all cells arrested in the G2/M phase, this would suggest that some cells are being arrested in other phases of the cell cycle. Since the majority of the cell death was observed to occur within the first several days of the growth curve experiments, it is more likely that cells are being arrested at various stages of the cell cycle. Although they used other methods to detect cell proliferation, several other studies have also detected a decrease in cell proliferation after

Ag NP treatment in Caco-2 cells68 as well as other cell models.310, 311 It has been suggested that Ag NP effects on cell cycle may be separate from oxidative stress.128, 312 However,

219 many researchers have associated induction of cell cycle arrest and inhibition of proliferation by Ag NP with oxidative stress.127, 297, 304, 313 Because Ag NP are likely inducing oxidative stress by interactions with thiol-containing enzymes that may lead to sequestration of intracellular ROS scavengers142 or disruption of the electron transport chain,127, 142 it is possible that this interference is directly inhibiting cell proliferation. We conclude that inhibition of cell proliferation and cell cycle arrest is related to induction of oxidative stress,310 and that Ag NP are most proficient in inducing this effect.

Mechanism of Ag NP toxicity exposed to proliferating cells

There has been some debate as to whether the effects of Ag NP are mediated through the NP themselves or through Ag ions arising from dissolution of the Ag NP. One study determined that 70% of citrate-stabilized NP will dissolve after 3000 hours at 37°C, but only 14% will dissolve at 25°C.92 Many studies have assessed toxicity of both Ag NP and Ag ions and have generated mixed results. A study in a Caco-2/M cell model demonstrated that treatment with 20, 34, 61, and 113 nm Ag NP and 1.5 µg/mL Ag ions

(from AgNO3) all caused similar changes in gene expression and concluded that the toxicity of Ag NP is mediated through Ag ions.95 Another study using 90 nm Ag NP in

Caco-2 cells showed that 2.18% (3.57 µM) of the Ag of a 15 µg/mL suspension of Ag NP was in the form of Ag ions when initially suspended in cell culture media but ions increased to 3.47% (4.82 µM) after 24 hours.67 They compare these amounts of Ag to a study that showed that 5 µM AgNO3 could induce oxidative stress and cytotoxicity in human skin fibroblasts.314

220

However, another study showed that treatment of Caco-2 cells with supernatants from 20 nm Ag NP incubated for 24 hours in cell culture medium caused only slight decrease in cell number at a concentration of 100 µg/mL.129 An in vitro study using a simulated digestion model showed that after intestinal digestion, AgNO3 would form Ag

NP.315 Several in vivo studies found Ag NP in tissue even of rats treated with Ag ions, which suggests that Ag ions can also form Ag NP in vivo.183, 184

The apparent discordance of these observations could stem from the definition of

“dissolved” Ag ions. We note above that if Ag NP dispersed in cell culture media are subjected to high speed ultracentrifugation, the level of Ag ions in solution is 0.2-0.6 ppm

(Figure 5.5), though there is over 1250 ppm of Ag being introduced as Ag NP. Previous studies have reported much higher levels of Ag ions in solution, including a study that defined any silver in supernatant filtered through 25 nm filters as ionic.316 In agreement with previous literature, we also propose that the Ag NP are dissolving, but in the complex media used for cell culture, it is likely that these dissolved Ag ions are forming colloidal particles, and are not available as free Ag ions.

For Ag NP incubated with synthetic digestive enzyme solutions representing the environments of the stomach, and separately, the intestines, a higher dose was required (0.5

µg/cm2 or 2.5 µg/mL) to cause the same inhibition of proliferation as the pristine Ag NP

(0.25 µg/cm2 or 1.25 µg/mL; Figure 5.12). A previous study found that digestion of Ag NP delayed toxicity by 12 hours and required a higher dose, as measured by impedance monitoring.21 It is well known that NP can adsorb proteins to their surface,317 and the infrared/Raman spectra, XPS, and zeta potential (Figures 5.1, 5.2, 5.3, and 5.4) indicate

221 that there is adsorption of molecules from the digestion media onto the Ag NP. We propose that this “corona” decreases the dissolution of the digested Ag NP, resulting in fewer colloidal particles.

The experiments with N-acetylcysteine (NAC) also support the Ag NP dissolution model as the vehicle of toxicity. In our studies, exposure of cells to 10 mM NAC along with Ag NP was completely protective against Ag NP-induced inhibition of cell proliferation (Figure 5.17). In C. elegans, NAC was able to rescue growth inhibition induced by Ag NP and AgNO3 treatment and it was also concluded that chelation of Ag ions by NAC was mainly responsible for the protective effects of NAC.316 NAC, via its thiol group, can bind to Ag ions (formation constant of Ag complexes with cysteine, a thiol

318-320 closely related to NAC, has been reported to be log Kf = 11.5 or log β = 11.9 ). This complexation reaction of thiols with Ag+ takes place immediately to form molecular

321 complexes, however, over time (hours), a gel-like colloidal (Ag-SR)n material is formed with Ag bound very strongly.321, 322 Thus, we propose that the 24-48 hour incubation of Ag

NP with NAC leads to formation of these colloidal gels, which will not release Ag+ readily even if taken up by the cell.

On the other hand, pretreatment of cells with 10 mM NAC for 1 hour and removal of NAC prior to exposure of cells to Ag NP was unable to protect against Ag NP-induced inhibition of cell proliferation and did not affect the induction of oxidative stress in cells

(Figure 5.18). This is because there was not a high enough concentration of NAC available

+ to form (Ag-SR)n gels, and thus only molecular complexes of NAC with Ag , which can release the Ag+, would be formed. Addition of Trolox (Figure 5.19) also had no protective

222 effect on cell proliferation after treatment with Ag NP. Trolox is a water-soluble form of vitamin E that functions as a ROS scavenger and is particularly protective against lipid peroxidation. Based on these experiments, protection by NAC and Trolox via their antioxidant properties was minimal.

The rescue of Ag NP-mediated inhibition of cell proliferation is thus due to complexation of NAC with Ag ions and formation of (Ag-SR)n gels, preventing release of

Ag+ and thereby alleviating toxicity against cells. The fate of Ag+ held via these thiol gels within the cell is unclear, but they could end up bound by metallothioneins.323 When cells are pretreated with NAC, there is not enough extracellular NAC or a high enough concentration of intracellular NAC to form colloidal gel complexes with dissolved Ag+ to prevent them from interacting with other intracellular thiol-containing proteins and mediating toxicity. This contrast between the toxic response in the absence and presence of NAC is depicted in the scheme in Figure 5.23.

Impact of Ag NP upon stationary (confluent) versus proliferating cells

For confluent cells (>100,000 cells/cm2), Sytox Red staining, Annexin V FITC staining, the LDH assay, and the MTT assay revealed no cell death or decrease in metabolic activity induced by 10 µg/cm2 (40 µg/mL) Ag NP treatment (Figure 5.20A-D). We also did not observe a change in GSH/GSSG ratio in confluent cells after treatment with 0.25

µg/cm2 (1.25 µg/mL) Ag NP (Figure 5.20E). There are several studies in intestinal epithelial cell models consistent with our results that do not demonstrate toxicity at doses of 10 µg/cm2 (40 µg/mL) Ag NP.49, 95, 129 Other studies in Caco-2 cells have shown toxicity at Ag NP doses lower than those we used.52, 67, 68, 171 However, it is likely that these

223

Figure 5.23. Mechanism of Ag NP toxicity. A) Ag NP are internalized by cells. Ag ion dissolution both intracellularly and extracellularly produces Ag colloids which will readily release ions that can interact with cells. Within cells, Ag NP induce oxidative stress and inhibition of cell proliferation. B) In the presence of NAC, stable complexes are formed with Ag+ which keeps the Ag from inducing toxicity.

224 differences are due to studies being performed at lower cell densities (between 10,000 and

100,000 cells/cm2)67, 171 and differences in the Ag NP used in the studies.52, 68 Also, the species adsorbed to the Ag NP surface can play a role in toxicity. In one of these studies cells were treated with NP suspended in media without serum, which largely eliminates protein from the medium.171 Serum has been shown in previous studies of Ag NP and other

NP to play a significant role in NP toxicity and thus this may significantly affect Ag NP toxicity.124, 324 Exposure of cells to Ag NP for 48 hours did increase toxicity slightly (20% induction of apoptosis and 40% loss in mitochondrial activity, Figure 5.21), indicating that the length of Ag NP treatment will be important in addition to dose. This is consistent with the results of other studies.129

The lack of toxicity in confluent cells is due to their decreased sensitivity to cell stressors (because of overall decreased metabolic activity) and a decreased per-cell exposure to NP. Also, confluent cells are more tightly packed into cell culture plates so that each cell has a smaller available surface area to interact with Ag NP. These cells form intercellular junctions that prevent transport of Ag NP between and underneath cells. Since proliferating cells are not surrounded by cells on all sides, each cell has a greater surface area exposed to Ag NP and is also being exposed to more NP per cell than confluent cultures receiving the same total dose of NP. These conditions could increase cellular sensitivity to NP treatment. Although little literature exists on the effects of Ag NP on proliferating cells at < 10,000 cells/cm2, it has been shown that proliferating Caco-2 cells were more sensitive to Ag NP-induced toxicity and differentiated Caco-2 cells required higher doses of Ag NP to induce the same cytotoxic response.21, 129

225

Also, because of the many cell processes needed for replication, proliferating cells are more metabolically active and any damage induced by Ag NP may be more apparent in proliferating cells than it would be in confluent cells. It is well known in models of biological stress such as ionizing radiation that proliferating cells are more sensitive to toxic agents. The radiosensitivity of cells generally decreases with increasing cell differentiation and associated decreasing rate of proliferation. Thus, the cells that are actively proliferating are much more sensitive to irradiation.325 This is due to induction of

DNA damage which leads to either cell cycle arrest to allow time for DNA repair or induction of apoptosis.326 Thus, proliferating cells are expected to be more sensitive to toxic agents due to their greater metabolic activity.

Slight induction of IL-8 was detected by ELISA after treatment of confluent cells with 5 µg/cm2 Ag NP (Figure 5.22). Although the amount of IL-8 detected was so small as to be negligible (18.4 pg/mL), it may indicate a slight inflammatory response induced by

Ag NP treatment. This is in agreement with previous studies that have observed IL-8 production in Caco-2 cells treated with Ag NP.49, 52

In vivo implications

The current study has implications for proliferation of the intestinal epithelium in vivo. The stem cells at the base of intestinal crypts are constantly proliferating in order to renew the intestinal epithelium. If ingested Ag NP are able to slow this proliferation, this may seriously impair the epithelial barrier function. To date, in vivo studies with orally administered Ag NP have shown damage to the microvilli of intestinal epithelial cells and intestinal glands in mice,189 indications of slight liver damage,191 and changes in intestinal

226 mucus composition and greater release of mucus granules from Goblet cells along with cell shedding at the tips of villi in the ileum,190 but no studies known to us have reported impairment of the epithelial barrier. Inhibition of stem cell proliferation by Ag NP may have more serious consequences in diseased intestines in which the epithelial barrier is already compromised such as can be the case in inflammatory bowel diseases. Therefore,

Ag NP ingestion needs to be investigated in intestinal disease models.

227

Chapter 6: Conclusions and Future Directions

6.1 Summary of Findings

The increasing use of nanoparticles (NP) in foods and food packaging has necessitated research into the safety of NP ingestion. Although multiple in vitro studies have already been performed to investigate toxicity of inorganic food-relevant NP including SiO2, TiO2, ZnO, and Ag, the intent of this research was to further clarify toxicity of these NP in an in vitro intestinal epithelial cell model. The mild toxicity induced by ZnO and Ag NP and absence of toxicity by SiO2 and TiO2 at doses above the anticipated exposure in vivo suggest that ingestion of any of these NP will not cause acute toxicity to the intestinal epithelium. The toxic potential of NP in vitro was not changed upon repeated exposure of cell populations to the NP. Subconfluent proliferating cell populations were found to be more sensitive to NP toxicity, particularly to Ag NP. In these cells, Ag NP induced an increase in oxidative stress which likely contributed to the increased cell death, decreased mitochondrial activity, cell cycle arrest, and inhibition of cell proliferation observed. This Ag NP toxicity may have implications for proliferation of stem cells in intestinal crypts. It was also demonstrated that silica NP with fluorescent cores can be tracked by fluorescence after oral administration of NP to mice which will be useful in determining NP distribution, accumulation, and toxicity in vivo.

228

6.2 Importance of nanoparticle physicochemical characteristics

This work thoroughly characterized NP in order to draw conclusions about the physicochemical characteristics that dictate NP toxicity. When attempting to formulate a complete understanding of NP toxicity based on the current literature, it is difficult to compare studies because of differences in the NP used without thorough characterization of these NP. A strength of the work presented here is the NP characterization that was performed. Transmission electron microscopy (TEM) was used to assess NP shape and primary size. Dynamic light scattering measured the hydrodynamic radius of NP in cell culture media as well as after incubation in the various digestive solutions used to simulate digestion in vitro. Zeta potential measurements identified particle surface charge in cell culture media and digestive solutions as well as over a range of pH values which was used to determine the isoelectric point for the NP. X-ray diffraction was used to determine the

NP crystal structures. Dissolution of ZnO in the pepsin solution at pH 2 and Ag NP in culture media was measured by atomic absorption spectroscopy. Difference infrared spectra (diffuse reflectance infrared Fourier transform spectroscopy) was used to determine the differences in species adsorbed to the particle surface in cell culture media and water or after NP digestion. This analysis revealed proteins adsorbed to NP in media as well as after NP digestion and subtraction of these spectra allowed analysis of the species specific to the NP digestion. Raman spectroscopy also provided a way to differentiate species adsorbed to Ag NP both before and after digestion and revealed citrate adsorption to the

Ag NP which were not subjected to in vitro digestion. The elemental composition of the surface species adsorbed to pristine and digested Ag NP was determined by X-ray

229 photoelectron spectroscopy. This helped to further identify the adsorbed surface species as citrate on pristine Ag NP and a greater proportion of proteins after Ag NP digestion.

UV/Visible absorption and fluorescence spectra were used to characterize the excitation and emission spectra of the fluorescent NP. Fluorescence lifetime data was used to further characterize the synthesized QD and compare them to previously reported QD. Nuclear magnetic resonance spectroscopy was used to identify the coordination of the silicon atoms on the surface of Rhodamine 6G/silica and QD/silica particles. Together, this thorough characterization of each of the NP used makes it possible to compare this work to previous and future studies and will help to determine whether observed biological effects are specific to a certain NP physicochemical property.

It is known that the physicochemical properties of NP contribute to their biological interactions including efficiency of cell internalization, biopersistence and bioavailability, and association with biological molecules.90, 111, 119, 124, 126 This research clearly demonstrates the importance of material type and surface properties in determining toxicity by direct side-by-side comparison of several different NP both as-purchased or as-prepared and after simulated in vitro digestion using multiple concurrent assays. Although little acute NP toxicity was detected in confluent cells, there were differences observed between the NP types. All NP were internalized by cells, but only ZnO NP induced slight toxicity after 24-hour treatment. In proliferating (subconfluent) cells, Ag NP inhibited cell proliferation at much lower doses than the other NP, suggesting that this toxicity is mediated by a different mechanism than that observed after acute ZnO NP exposure.

230

Surface chemistry is also important in determining NP behavior. Proteins and other molecules that adsorb to the NP surface can change the pathways by which NP are internalized by cells and NP can sequester important proteins within cells.124, 302 It is known that ZnO and Ag NP toxicity is largely mediated by dissolved ions94, 188 and this work confirmed the importance of Ag ions in Ag NP toxicity. Changes in species adsorbed to the NP surface may also affect NP dissolution. In these studies, simulated in vitro digestion of NP changed the composition of the adsorbed surface species based on Raman, infrared, and XPS analysis, but this largely did not alter NP toxicity in confluent cells. Digested

TiO2 NP were observed to slightly decrease mitochondrial activity and this was attributed to TiO2 serving as a carrier to transport bile salts or other digestive solution components into cells which are then responsible for toxicity. The ability for NP to interact with proteins or toxins intracellularly and extracellularly and possibly transport these molecules into or out of cells may be able to increase toxicity of seemingly safe NP such as SiO2 and TiO2.

In proliferating cells, which were most sensitive to Ag NP-mediated toxicity, a higher dose of digested Ag NP (0.5 µg/cm2) was required to induce the same toxicity as 0.25 µg/cm2 pristine Ag NP, likely due to slower dissolution of Ag because of adsorbed surface species.

This once again highlights the importance of the NP surface chemistry in determining NP behavior.

The food matrix into which NP are incorporated was not accounted for in this work but is a major determinant of NP surface chemistry and the composition of the NP corona.

The acute toxicity data of NP from the current research can provide a basis for future studies with foods that contain NP. Experiments that perform digestion on NP-containing

231 foods are necessary to understand the availability of NP in these foods and their uptake and subsequent toxicity to cells. In addition to the in vitro digestion models that have been used for NP, there are models developed for digestion of food which incorporate the mechanical aspects of digestion using peristaltic pumps.327, 328 Studies performed with digested foods in vitro may be informative for understanding the kinetics of NP release during digestion and their potential for internalization by cells. However, it is difficult to accurately replicate the food matrix, the digestive tract, and all the in vivo components that will determine the surface chemistry of NP using in vitro cell models. Use of in vivo models will help to form a more comprehensive picture of ingested NP toxicity. Although such studies are often done in rodents, the digestive tract of pigs is more similar to that of humans329 and thus may serve as a good model for some of these studies.

6.3 Silver nanoparticle toxicity and disease

Ag NP were found to induce toxic effects in proliferating cells including increased necrotic cell death, inhibition of mitochondrial activity, increased oxidative stress in cells,

G2/M phase cell cycle arrest, and inhibition of cell proliferation. This suggests that Ag NP are able to increase cellular oxidative stress leading to initial death of a certain proportion of cells (~20%) and cell cycle arrest and inhibition of proliferation in the remaining cells.

This proposed mechanism is in agreement with the oxidative stress-mediated mechanism of Ag NP toxicity which has been shown by other researchers in multiple cell types.67, 127,

128, 130 Since oxidative stress-induced DNA damage has been shown to be responsible for cell cycle arrest and inhibition of cell proliferation in some models,127, 302-304, 310, 311 determining whether Ag NP induce DNA damage in our model would allow us to directly

232 link oxidative stress to the cell proliferation effects. Based on the experiments showing that

N-acetylcysteine was able to protect cells from Ag NP-mediated inhibition of cell proliferation while the antioxidant Trolox was not (Figures 5.17 and 5.19), it was concluded that NAC is protecting cells by forming stable complexes with toxic Ag ions and thus Ag NP toxicity is being mediated through dissolved Ag ions.

The observed toxicity and oxidative stress induced by Ag NP in proliferating cells was largely absent in confluent cells. Although stem cells in intestinal crypts are constantly proliferating to renew the intestinal epithelium, tight junctions between cells are retained during mitosis in order to maintain the barrier function of the epithelium.330, 331 Transient gaps are created when cells are extruded, but the normal epithelium is composed of an intact cell monolayer. Thus, normal intact epithelium more closely resembles a confluent cell monolayer in vitro than the subconfluent proliferating cell model used in this research in which significant gaps exist in the cell layer. One of the differences in the proliferating cell model is that a greater cell surface area is exposed to NP than in a confluent monolayer where cells are tightly packed together, and this difference would not exist in vivo.

However, subconfluent proliferating cells in vitro can serve as a model for a diseased intestine where there are gaps in the epithelium. Thus, since we have observed significant

Ag NP toxicity in these proliferating cells at relatively low doses (0.25 µg/cm2), it is necessary to further investigate whether Ag NP are particularly toxic in disease models.

Several in vitro models have been described that co-culture intestinal epithelial cells with immune cells.51, 85 These cultures can be treated with an inflammatory cytokine and used as a model of an inflammatory state, often one of the hallmarks of intestinal diseases.

233

Performing experiments with proliferating intestinal epithelial cells in one of these co- culture models would help to further elucidate whether Ag NP will be toxic in a disease state. In vivo experiments using a model of disease will also be necessary to determine the safety of Ag NP for consumers with intestinal diseases. Models such as dextran sulfate sodium-induced colitis in mice332 will allow the examination of NP toxicity in vivo in an inflamed model with gaps in the epithelium. If Ag NP treatment slows cell proliferation, this may exacerbate disease by leading to more epithelial gaps. Greater transport of NP through these gaps may also increase systemic absorption and lead to greater toxicity throughout the body. Ag NP toxicity can also be investigated in a more mild leaky gut disease model using the intracellular protein zonulin, which reversibly regulates tight junction permeability,64 or the peptide AT1002, a synthetic peptide fragment of zonula occludens toxin which opens tight junctions,333 to increase tight junction permeability. This leaky gut model is relevant to diseases including Celiac disease, type 1 diabetes, and inflammatory bowel diseases (Crohn’s disease and ulcerative colitis). This model will elucidate whether NP can accelerate disease progression and will be relevant for patients suffering from many of these diseases.

6.4 Safety of nanoparticles for oral ingestion in foods

Although toxicity experiments have to start somewhere, these studies have been performed using NP doses in vitro that are far greater (for instance, 10 µg/cm2) than what cells would be exposed to in vivo and thus these experiments may not be relevant for human exposure. Since minimal NP toxicity was observed at these doses in vitro, it is unlikely that toxicity would be observed in vivo. A recent study determined physiologically relevant NP

234 doses based on an estimated small intestinal total surface area of 2×106 cm2 and a duodenal surface area of ~900 cm2.43 The duodenum was chosen because the most nutrient absorption takes place there. These surface area values can be used along with human NP exposure estimates to calculate more physiologically relevant NP distribution through the

2 small intestine. Using the 0.2-0.7 mg/kg body weight/day estimate of adult TiO2 intake and assuming a 70 kg adult, the daily exposure based on these surface area estimates would be 0.007-0.0245 µg/cm2 spread across the small intestine and 15.6-54.4 µg/cm2 in the duodenum. For the 1.8 mg/kg body weight/day estimation of silica NP uptake,1 assuming a 70 kg person, the exposure translates to 0.063 µg/cm2 throughout the small intestine and

140 µg/cm2 in the duodenum alone. These estimates are summed up in Table 6.1. Although the stomach contents will pass through the digestive tract to some extent en masse, these are estimates for the entire day and not what would be eaten in one meal, and thus the dose estimate for the small intestine as a whole is likely more realistic. In addition, only a subset of ingested NP will migrate through the mucus layer to reach the cell surface. Based on this knowledge and these estimates of NP intake, this suggests that the doses used in this work (often 10 µg/cm2) are, as expected, likely much higher than what will cells will be exposed to in vivo. The dose at which toxicity was observed with Ag NP in proliferating cells (0.25 µg/cm2) is approaching a more realistic in vivo exposure. However, knowing the doses at which toxicity is observed in vitro can help us to design more relevant in vivo studies. In addition, as use of NP in foods increases, it is necessary to determine potential toxicity at high doses in order to set appropriate exposure limits.

235

Estimated human NP exposure In vitro NP doses for comparison Nanoparticle Duodenum Small intestine Nanoparticle Dose (~900 cm2) (2×106 cm2) 2 2 2 SiO2 140 µg/cm 0.063 µg/cm All NP (confluent 10 µg/cm cells) 2 TiO2 15.6-54.4 0.007-0.0245 Ag (proliferating 0.25 µg/cm µg/cm2 µg/cm2 cells) Table 6.1. Estimates of human nanoparticle exposure to intestinal epithelial cells to compare to doses used for in vitro studies. Calculations were based on a duodenal surface area of ~900 cm2 and a small intestinal 6 2 43 2 surface area of 2×10 cm , daily exposure estimates of 0.2-0.7 mg/kg TiO2 and 1.8 mg/kg 1 SiO2, and a 70 kg adult.

It is also possible that NP could accumulate within stem cells of the intestinal epithelium after repeated oral exposure and be passed to cell progeny as they renew the intestinal epithelium. If enough NP accumulated, they could eventually cause toxicity in cells. However, since the exposure is so low, it is unlikely that enough NP will accumulate before cell turnover to cause toxicity. When cells were repeatedly exposed to NP (once a week for 24 hours), no changes were observed in toxicity induced by 24-hour NP treatment or growth of the repeatedly treated cell populations. This suggests that repeated treatment is not sensitizing cells to NP treatment and it is likely that NP are largely being exocytosed out of cells if they do not eventually dissolve within cells. It is also worth noting that repeated NP treatment did not seem to desensitize cells to NP, at least for ZnO NP. Similar toxicity to ZnO NP treatment was observed after one exposure and after 26 total exposures

(Figures 3.11 and 3.15).

236

6.5 In vivo studies

As previously discussed, in vivo studies will be useful in determining NP effects in a more complex system that better models the human GI tract. This will also allow the detection of any systemic NP effects depending on whether NP are absorbed through the intestines and will not confine the investigation to the intestinal epithelium. In order to study NP effects in vivo, it is helpful to have a simple way to detect and observe the NP of interest. Since the food-relevant NP of interest do not have readily trackable properties such as magnetism, intrinsic fluorescence, or radioactivity, silica NP with a fluorescent core were synthesized. Other researchers have previously labeled NP with fluorescent probes, but having the fluorophore on the outside of the particle will change the surface chemistry of the particle and may change how it behaves in vivo. Thus, a silica shell was hydrolyzed around fluorescent cores so that the outer surface of the particle to which cells are exposed is silica. Administration of these NP for only 4 days showed NP localization to non-gastrointestinal tract tissues including kidney, spleen, lung, brain, and liver, showing that these NP are being transported across the intestinal epithelium, through the portal circulation, and into the systemic circulation.

Determining the accumulation and functional consequences of NP exposure throughout these tissues will be critical for determining potential NP toxicity after ingestion. Long-term exposure studies at NP doses similar to estimated human exposures will be especially important to determine whether NP can have long-term consequences despite the apparent lack of acute toxicity. In vivo studies can also be used to answer important questions of NP toxicity in different populations including children. It has been

237 estimated that children are exposed to higher doses of TiO2 because of its use in sweet foods (1-2 mg/kg body weight/day in children vs. 0.2-0.7 mg/kg body weight/day in adults).2 Based on the greater toxicity of Ag NP to proliferating intestinal epithelial cells shown here (Figure 5.12), and particularly if this toxicity translates to other cell populations, NP exposure and accumulation during growth and development of children may have more serious consequences than exposure in adults. Experiments in juvenile mice can help to address the question of NP toxicity in children. It is also possible that NP will be passed through the placenta and experiments in pregnant mice can be used to determine whether this is possible and the consequences for the pups.

6.6 Concluding Remarks

These studies suggest that food-relevant NP acute intestinal epithelial cell toxicity is minimal at the current levels of NP ingestion but leaves open questions associated with long-term consumption and potential accumulation effects. Hence, many more studies are needed before NP ingestion can be assumed to be safe for all populations. NP ingestion will likely continue to increase through novel incorporation of NP in foods and increasing contamination by NP used in agriculture and other industries. This will necessitate reassessment of NP toxicity at the resulting exposure levels.

238

References

1. Dekkers, S.; Krystek, P.; Peters, R. J.; Lankveld, D. P.; Bokkers, B. G.; van Hoeven-Arentzen, P. H.; Bouwmeester, H.; Oomen, A. G., Presence and risks of nanosilica in food products. Nanotoxicology 2011, 5, 393-405.

2. Weir, A.; Westerhoff, P.; Fabricius, L.; Hristovski, K.; von Goetz, N., Titanium dioxide nanoparticles in food and personal care products. Environ. Sci. Technol. 2012, 46, 2242-2250.

3. The Project on Emerging Nanotechnologies Consumer Products Inventory. http://www.nanotechproject.org/inventories/consumer/ (March 25, 2015).

4. Desai, M. P.; Labhasetwar, V.; Amidon, G. L.; Levy, R. J., Gastrointestinal uptake of biodegradable microparticles: effect of particle size. Pharm. Res. 1996, 13, 1838-1845.

5. Awaad, A.; Nakamura, M.; Ishimura, K., Imaging of size-dependent uptake and identification of novel pathways in mouse Peyer's patches using fluorescent organosilica particles. Nanomedicine 2012, 8, 627-636.

6. Freestone, I.; Meeks, N.; Sax, M.; Higgitt, C., The Lycurgus Cup — A Roman nanotechnology - Springer. Gold Bull. 2007, 40, 270-277.

7. Sciau, P. Nanoparticles in Ancient Materials: The Metallic Lustre Decorations of Medieval Ceramics. In The Delivery of Nanoparticles, Hashim, A. A., Ed.; InTech: Rijeka, Croatia, 2012; Chapter 25, pp 525-540.

8. Heiligtag, F. J.; Niederberger, M., The fascinating world of nanoparticle research. Mater. Today 2013, 16, 262-271.

9. Edwards, P. P.; Thomas, J. M., Gold in a metallic divided state--from Faraday to present-day nanoscience. Angew Chem. Int. Ed. Engl. 2007, 46, 5480-5486.

10. Roduner, E., Size matters: why nanomaterials are different. Chem. Soc. Rev. 2006, 35, 583-592.

239

11. Initiative, N. N. What's So Special about the Nanoscale? | Nano. http://www.nano.gov/nanotech-101/special (March 25, 2015).

12. McNeil, S. E., Nanotechnology for the biologist. J. Leukoc. Biol. 2005, 78, 585- 594.

13. Chaudhry, Q.; Scotter, M.; Blackburn, J.; Ross, B.; Boxall, A.; Castle, L.; Aitken, R.; Watkins, R., Applications and implications of nanotechnologies for the food sector. Food Addit. Contam., Part A 2008, 25, 241-258.

14. Chen, X. X.; Cheng, B.; Yang, Y. X.; Cao, A.; Liu, J. H.; Du, L. J.; Liu, Y.; Zhao, Y.; Wang, H., Characterization and preliminary toxicity assay of nano-titanium dioxide additive in sugar-coated chewing gum. Small 2013, 9, 1765-1774.

15. Gatti, A. M.; Tossini, D.; Gambarelli, A.; Montanari, S.; Capitani, F., Investigation of the presence of inorganic micro- and nanosized contaminants in bread and biscuits by environmental scanning electron microscopy. Crit. Rev. Food Sci. Nutr. 2009, 49, 275-282.

16. Goodman, B. E., Insights into digestion and absorption of major nutrients in humans. Advances in Physiology Education 2010, 34, 44-53.

17. Barrett, K. E., In Lange Medical Books, Barrett, K. E., Ed.; McGraw-Hill: New York, NY, 2014.

18. Ming, S.-C.; Goldman, H., Pathology of the Gastrointestinal Tract. W.B. Saunders Company (Harcourt Brace Jovanovich, Inc.): Philadelphia, PA, 1992.

19. Peters, R.; Kramer, E.; Oomen, A. G.; Rivera, Z. E.; Oegema, G.; Tromp, P. C.; Fokkink, R.; Rietveld, A.; Marvin, H. J.; Weigel, S.; Peijnenburg, A. A.; Bouwmeester, H., Presence of nano-sized silica during in vitro digestion of foods containing silica as a food additive. ACS Nano 2012, 6, 2441-2451.

20. Gerloff, K.; Pereira, D. I.; Faria, N.; Boots, A. W.; Kolling, J.; Forster, I.; Albrecht, C.; Powell, J. J.; Schins, R. P., Influence of simulated gastrointestinal conditions on particle-induced cytotoxicity and interleukin-8 regulation in differentiated and undifferentiated Caco-2 cells. Nanotoxicology 2013, 7, 353-366.

21. Bohmert, L.; Girod, M.; Hansen, U.; Maul, R.; Knappe, P.; Niemann, B.; Weidner, S. M.; Thunemann, A. F.; Lampen, A., Analytically monitored digestion of silver nanoparticles and their toxicity on human intestinal cells. Nanotoxicology 2014, 8, 631-642.

22. Hansson, G. C., Role of mucus layers in gut infection and inflammation. Curr. Opin. Microbiol. 2012, 15, 57–62.

240

23. Atuma, C.; Strugala, V.; Allen, A.; Holm, L., The adherent gastrointestinal mucus gel layer: thickness and physical state in vivo. Am. J. Physiol.: Gastrointest. Liver Physiol. 2001, 280, G922-G929.

24. Rescigno, M.; Urbano, M.; Valzasina, B.; Francolini, M.; Rotta, G.; Bonasio, R.; Granucci, F.; Kraehenbuhl, J. P.; Ricciardi-Castagnoli, P., Dendritic cells express tight junction proteins and penetrate gut epithelial monolayers to sample bacteria. Nat. Immunol. 2001, 2, 361-367.

25. Peterson, L. W.; Artis, D., Intestinal epithelial cells: regulators of barrier function and immune homeostasis. Nat. Rev. Immunol. 2014, 14, 141-153.

26. McConnell, R. E.; Higginbotham, J. N.; David A. Shifrin, J.; Tabb, D. L.; Coffey, R. J.; Tyska, M. J., The enterocyte microvillus is a vesicle-generating organelle. J. Cell Biol. 2009, 185, 1285-1298.

27. Smith, M. W.; Peacock, M. A., "M" cell distribution in follicle-associated epithelium of mouse Peyer's patch. Am. J. Anat. 1980, 159, 167-175.

28. Jepson, M. A.; Simmons, N. L.; Savidge, T. C.; James, P. S.; Hirst, B. H., Selective binding and transcytosis of latex microspheres by rabbit intestinal M cells. Cell Tissue Res. 1993, 271, 399-405.

29. Mabbott, N. A.; Donaldson, D. S.; Ohno, H.; Williams, I. R.; Mahajan, A., Microfold (M) cells: important immunosurveillance posts in the intestinal epithelium. Mucosal Immunol. 2013, 6, 666-677.

30. Jung, C.; Hugot, J. P.; Barreau, F., Peyer's Patches: The Immune Sensors of the Intestine. Int. J. Inflam. 2010, 2010, 823710.

31. Dharmsathaphorn, K.; McRoberts, J. A.; Mandel, K. G.; Tisdale, L. D.; Masui, H., A human colonic tumor cell line that maintains vectorial electrolyte transport. Am. J. Physiol. 1984, 246, G204-G208.

32. Polak-Charcon, S.; Shoham, J.; Ben-Shaul, Y., Junction formation in trypsinized cells of human adenocarcinoma cell line. Exp. Cell Res. 1978, 116, 1-13.

33. Hashimoto, K.; Shimizu, M., Epithelial properties of human intestinal Caco-2 cells cultured in a serum-free medium. Cytotechnology 1993, 13, 175-184.

34. Pinto, M.; Robine-Leon, S.; Appay, M.-D.; Kedinger, M.; Triadou, N.; Dussaulx, E.; Lacroix, B.; Simon-Assmann, P.; Haffen, K.; Fogh, J.; Zweibaum, A., Enterocyte-like differentiation and polarization of the human colon carcinoma cell line Caco-2 in culture. Biol. Cell 1983, 47, 323-330.

241

35. Peterson, M. D.; Mooseker, M. S., Characterization of the enterocyte-like brush border cytoskeleton of the C2BBe clones of the human intestinal cell line, Caco-2. J. Cell Sci. 1992, 102 (Pt 3), 581-600.

36. Peterson, M. D.; Bement, W. M.; Mooseker, M. S., An in vitro model for the analysis of intestinal brush border assembly. II. Changes in expression and localization of brush border proteins during cell contact-induced brush border assembly in Caco-2BBe cells. J. Cell Sci. 1993, 105 (Pt 2), 461-472.

37. Bourgine, J.; Billaut-Laden, I.; Happillon, M.; Lo-Guidice, J. M.; Maunoury, V.; Imbenotte, M.; Broly, F., Gene expression profiling of systems involved in the metabolism and the disposition of xenobiotics: comparison between human intestinal biopsy samples and colon cell lines. Drug Metab. Dispos. 2012, 40, 694-705.

38. Lesuffleur, T.; Barbat, A.; Dussaulx, E.; Zweibaum, A., Growth adaptation to methotrexate of HT-29 human colon carcinoma cells is associated with their ability to differentiate into columnar absorptive and mucus-secreting cells. Cancer Res. 1990, 50, 6334-6343.

39. Guri, A.; Gulseren, I.; Corredig, M., Utilization of solid lipid nanoparticles for enhanced delivery of curcumin in cocultures of HT29-MTX and Caco-2 cells. Food Funct. 2013, 4, 1410-1419.

40. Gullberg, E.; Leonard, M.; Karlsson, J.; Hopkins, A. M.; Brayden, D.; Baird, A. W.; Artursson, P., Expression of specific markers and particle transport in a new human intestinal M-cell model. Biochem. Biophys. Res. Commun. 2000, 279, 808-813.

41. Masuda, K.; Kajikawa, A.; Igimi, S., Establishment and Evaluation of an in vitro M Cell Model using C2BBe1 Cells and Raji Cells. Biosci. Microflora 2011, 30, 37-44.

42. Kadiyala, I.; Loo, Y.; Roy, K.; Rice, J.; Leong, K. W., Transport of chitosan- DNA nanoparticles in human intestinal M-cell model versus normal intestinal enterocytes. Eur. J. Pharm. Sci. 2010, 39, 103-109.

43. Mahler, G. J.; Esch, M. B.; Tako, E.; Southard, T. L.; Archer, S. D.; Glahn, R. P.; Shuler, M. L., Oral exposure to polystyrene nanoparticles affects iron absorption. Nat. Nanotechnol. 2012, 7, 264-271.

44. Mantis, N. J.; Rol, N.; Corthésy, B., Secretory IgA's Complex Roles in Immunity and Mucosal Homeostasis in the Gut. Mucosal Immunol. 2011, 4, 603-611.

45. Brown, E. M.; Sadarangani, M.; Finlay, B. B., The role of the immune system in governing host-microbe interactions in the intestine. Nat. Immunol. 2013, 14, 660-667.

242

46. Yu, Y.; Sitaraman, S.; Gewirtz, A. T., Intestinal epithelial cell regulation of mucosal inflammation. Immunol. Res. 2004, 29, 55-68.

47. Tarantini, A.; Lanceleur, R.; Mourot, A.; Lavault, M. T.; Casterou, G.; Jarry, G.; Hogeveen, K.; Fessard, V., Toxicity, genotoxicity and proinflammatory effects of amorphous nanosilica in the human intestinal Caco-2 cell line. Toxicol. In Vitro 2015, 29, 398-407.

48. Gerloff, K.; Pereira, D. I.; Faria, N.; Boots, A. W.; Kolling, J.; Forster, I.; Albrecht, C.; Powell, J. J.; Schins, R. P., Influence of simulated gastrointestinal conditions on particle-induced cytotoxicity and interleukin-8 regulation in differentiated and undifferentiated Caco-2 cells. Nanotoxicology 2013, 7, 353-366.

49. Abbott Chalew, T. E.; Schwab, K. J., Toxicity of commercially available engineered nanoparticles to Caco-2 and SW480 human intestinal epithelial cells. Cell Biol. Toxicol. 2013, 29, 101-116.

50. De Angelis, I.; Barone, F.; Zijno, A.; Bizzarri, L.; Russo, M. T.; Pozzi, R.; Franchini, F.; Giudetti, G.; Uboldi, C.; Ponti, J.; Rossi, F.; De Berardis, B., Comparative study of ZnO and TiO(2) nanoparticles: physicochemical characterisation and toxicological effects on human colon carcinoma cells. Nanotoxicology 2013, 7, 1361- 1372.

51. Susewind, J.; de Souza Carvalho-Wodarz, C.; Repnik, U.; Collnot, E. M.; Schneider-Daum, N.; Griffiths, G. W.; Lehr, C. M., A 3D co-culture of three human cell lines to model the inflamed intestinal mucosa for safety testing of nanomaterials. Nanotoxicology 2015, 1-10.

52. Kaiser, J. P.; Roesslein, M.; Diener, L.; Wick, P., Human health risk of ingested nanoparticles that are added as multifunctional agents to paints: an in vitro study. PLoS One 2013, 8, e83215.

53. De Berardis, B.; Civitelli, G.; Condello, M.; Lista, P.; Pozzi, R.; Arancia, G.; Meschini, S., Exposure to ZnO nanoparticles induces oxidative stress and cytotoxicity in human colon carcinoma cells. Toxicol. Appl. Pharmacol. 2010, 246, 116-127.

54. Brown, D. M.; Donaldson, K.; Stone, V., Effects of PM10 in human peripheral blood monocytes and J774 macrophages. Respir. Res. 2004, 5, 29.

55. Park, E. J.; Park, K., Oxidative stress and pro-inflammatory responses induced by silica nanoparticles in vivo and in vitro. Toxicol. Lett. 2009, 184, 18-25.

56. Nogueira, C. M.; de Azevedo, W. M.; Dagli, M. L. Z.; Toma, S. H.; Leite, A. A.; Lordello, M. L.; Nishitokukado, I.; Ortiz-Agostinho, C. L.; Duarte, M. I. S.; Ferreira, M.

243

A.; Sipahi, A. M., Titanium dioxide induced inflammation in the small intestine. World J. Gastroenterol. 2012, 18, 4729-4735.

57. Williams, K.; Milner, J.; Boudreau, M. D.; Gokulan, K.; Cerniglia, C. E.; Khare, S., Effects of subchronic exposure of silver nanoparticles on intestinal microbiota and gut-associated immune responses in the ileum of Sprague-Dawley rats. Nanotoxicology 2014, 1-11.

58. Turner, J. R., Intestinal mucosal barrier function in health and disease. Nat. Rev. Immunol. 2009, 9, 799-809.

59. Suzuki, T., Regulation of intestinal epithelial permeability by tight junctions. Cell Mol. Life Sci. 2013, 70, 631-659.

60. Neunlist, M.; Van Landeghem, L.; Mahe, M. M.; Derkinderen, P.; des Varannes, S. B.; Rolli-Derkinderen, M., The digestive neuronal-glial-epithelial unit: a new actor in gut health and disease. Nat. Rev. Gastroenterol. Hepatol. 2013, 10, 90-100.

61. Furuse, M., Molecular basis of the core structure of tight junctions. Cold Spring Harb. Perspect. Biol. 2010, 2, a002907.

62. Cunningham, K. E.; Turner, J. R., Myosin light chain kinase: pulling the strings of epithelial tight junction function. Ann. N. Y. Acad. Sci. 2012, 1258, 34-42.

63. Wang, W.; Uzzau, S.; Goldblum, S. E.; Fasano, A., Human zonulin, a potential modulator of intestinal tight junctions. J. Cell Sci. 2000, 113 Pt 24, 4435-4440.

64. El Asmar, R.; Panigrahi, P.; Bamford, P.; Berti, I.; Not, T.; Coppa, G. V.; Catassi, C.; Fasano, A., Host-dependent zonulin secretion causes the impairment of the small intestine barrier function after bacterial exposure. Gastroenterology 2002, 123, 1607- 1615.

65. Fasano, A., Zonulin and its regulation of intestinal barrier function: the biological door to inflammation, autoimmunity, and cancer. Physiol. Rev. 2011, 91, 151-175.

66. Koeneman, B. A.; Zhang, Y.; Westerhoff, P.; Chen, Y.; Crittenden, J. C.; Capco, D. G., Toxicity and cellular responses of intestinal cells exposed to titanium dioxide. Cell Biol. Toxicol. 2010, 26, 225-238.

67. Aueviriyavit, S.; Phummiratch, D.; Maniratanachote, R., Mechanistic study on the biological effects of silver and gold nanoparticles in Caco-2 cells--induction of the Nrf2/HO-1 pathway by high concentrations of silver nanoparticles. Toxicol. Lett. 2014, 224, 73-83.

68. Sahu, S. C.; Roy, S.; Zheng, J.; Yourick, J. J.; Sprando, R. L., Comparative genotoxicity of nanosilver in human liver HepG2 and colon Caco2 cells evaluated by 244 fluorescent microscopy of cytochalasin B-blocked micronucleus formation. J. Appl. Toxicol. 2014, 34, 1200-1208.

69. Schübbe, S.; Schumann, C.; Cavelius, C.; Koch, M.; Müller, T.; Kraegeloh, A., Size-Dependent Localization and Quantitative Evaluation of the Intracellular Migration of Silica Nanoparticles in Caco-2 Cells. Chem. Mater. 2011, 24, 914-923.

70. Faust, J. J.; Doudrick, K.; Yang, Y.; Westerhoff, P.; Capco, D. G., Food grade titanium dioxide disrupts intestinal brush border microvilli in vitro independent of sedimentation. Cell Biol. Toxicol. 2014, 30, 169-188.

71. Kadiyala, I.; Loo, Y.; Roy, K.; Rice, J.; Leong, K. W., Transport of chitosan- DNA nanoparticles in human intestinal M-cell model versus normal intestinal enterocytes. Eur. J. Pharm. Sci. 2010, 39, 103-109.

72. Powell, J. J.; Ainley, C. C.; Harvey, R. S.; Mason, I. M.; Kendall, M. D.; Sankey, E. A.; Dhillon, A. P.; Thompson, R. P., Characterisation of inorganic microparticles in pigment cells of human gut associated lymphoid tissue. Gut 1996, 38, 390-395.

73. Matricon, J.; Barnich, N.; Ardid, D., Immunopathogenesis of inflammatory bowel disease. Self Nonself 2010, 1, 299-309.

74. Fasano, A., Intestinal permeability and its regulation by zonulin: diagnostic and therapeutic implications. Clin. Gastroenterol. Hepatol. 2012, 10, 1096-1100.

75. Lammers, K. M.; Lu, R.; Brownley, J.; Lu, B.; Gerard, C.; Thomas, K.; Rallabhandi, P.; Shea-Donohue, T.; Tamiz, A.; Alkan, S.; Netzel-Arnett, S.; Antalis, T.; Vogel, S. N.; Fasano, A., Gliadin induces an increase in intestinal permeability and zonulin release by binding to the chemokine receptor CXCR3. Gastroenterology 2008, 135, 194-204 e3.

76. Zeissig, S.; Burgel, N.; Gunzel, D.; Richter, J.; Mankertz, J.; Wahnschaffe, U.; Kroesen, A. J.; Zeitz, M.; Fromm, M.; Schulzke, J. D., Changes in expression and distribution of claudin 2, 5 and 8 lead to discontinuous tight junctions and barrier dysfunction in active Crohn's disease. Gut 2007, 56, 61-72.

77. Vetrano, S.; Rescigno, M.; Cera, M. R.; Correale, C.; Rumio, C.; Doni, A.; Fantini, M.; Sturm, A.; Borroni, E.; Repici, A.; Locati, M.; Malesci, A.; Dejana, E.; Danese, S., Unique role of junctional adhesion molecule-a in maintaining mucosal homeostasis in inflammatory bowel disease. Gastroenterology 2008, 135, 173-184.

78. Camilleri, M.; Gorman, H., Intestinal permeability and irritable bowel syndrome. Neurogastroenterol. Motil. 2007, 19, 545-552.

245

79. Mooradian, A. D.; Morley, J. E.; Levine, A. S.; Prigge, W. F.; Gebhard, R. L., Abnormal intestinal permeability to sugars in diabetes mellitus. Diabetologia 1986, 29, 221-224.

80. Parlesak, A.; Schafer, C.; Schutz, T.; Bode, J. C.; Bode, C., Increased intestinal permeability to macromolecules and endotoxemia in patients with chronic alcohol abuse in different stages of alcohol-induced liver disease. J. Hepatol. 2000, 32, 742-747.

81. Ventura, M. T.; Polimeno, L.; Amoruso, A. C.; Gatti, F.; Annoscia, E.; Marinaro, M.; Di Leo, E.; Matino, M. G.; Buquicchio, R.; Bonini, S.; Tursi, A.; Francavilla, A., Intestinal permeability in patients with adverse reactions to food. Dig. Liver Dis. 2006, 38, 732-736.

82. Clayburgh, D. R.; Shen, L.; Turner, J. R., A porous defense: the leaky epithelial barrier in intestinal disease. Lab. Invest. 2004, 84, 282-291.

83. van Elburg, R. M.; Uil, J. J.; Mulder, C. J.; Heymans, H. S., Intestinal permeability in patients with coeliac disease and relatives of patients with coeliac disease. Gut 1993, 34, 354-357.

84. Hollander, D.; Vadheim, C. M.; Brettholz, E.; Petersen, G. M.; Delahunty, T.; Rotter, J. I., Increased intestinal permeability in patients with Crohn's disease and their relatives. A possible etiologic factor. Ann. Intern. Med. 1986, 105, 883-885.

85. Leonard, F.; Collnot, E. M.; Lehr, C. M., A three-dimensional coculture of enterocytes, monocytes and dendritic cells to model inflamed intestinal mucosa in vitro. Mol. Pharm. 2010, 7, 2103-2119.

86. Lomer, M. C.; Harvey, R. S.; Evans, S. M.; Thompson, R. P.; Powell, J. J., Efficacy and tolerability of a low microparticle diet in a double blind, randomized, pilot study in Crohn's disease. Eur. J. Gastroenterol. Hepatol. 2001, 13, 101-106.

87. Lomer, M. C.; Grainger, S. L.; Ede, R.; Catterall, A. P.; Greenfield, S. M.; Cowan, R. E.; Vicary, F. R.; Jenkins, A. P.; Fidler, H.; Harvey, R. S.; Ellis, R.; McNair, A.; Ainley, C. C.; Thompson, R. P.; Powell, J. J., Lack of efficacy of a reduced microparticle diet in a multi-centred trial of patients with active Crohn's disease. Eur. J. Gastroenterol. Hepatol. 2005, 17, 377-384.

88. Bellmann, S.; Carlander, D.; Fasano, A.; Momcilovic, D.; Scimeca, J. A.; Waldman, W. J.; Gombau, L.; Tsytsikova, L.; Canady, R.; Pereira, D. I.; Lefebvre, D. E., Mammalian gastrointestinal tract parameters modulating the integrity, surface properties, and absorption of food-relevant nanomaterials. Wiley Interdiscip. Rev. Nanomed. Nanobiotechnol. 2015, epub online Jan. 30, 2015.

246

89. nanoComposix Nanotoxicology: Particle Selection. http://nanocomposix.com/pages/nanotoxicology-particle-selection (April 20, 2015).

90. Borm, P.; Klaessig, F. C.; Landry, T. D.; Moudgil, B.; Pauluhn, J.; Thomas, K.; Trottier, R.; Wood, S., Research strategies for safety evaluation of nanomaterials, part V: role of dissolution in biological fate and effects of nanoscale particles. Toxicol. Sci. 2006, 90, 23-32.

91. Morey, G. W.; Fournier, R. O.; Rowe, J. J., The solubility of amorphous silica at 25°C. J. Geophys. Res. 1964, 69, 1995-2002.

92. Kittler, S.; Greulich, C.; Diendorf, J.; Köller, M.; Epple, M., Toxicity of silver nanoparticles increases during storage because of slow dissolution under release of silver ions. Chem. Mater. 2010, 22, 4548-4554.

93. Mwilu, S. K.; El Badawy, A. M.; Bradham, K.; Nelson, C.; Thomas, D.; Scheckel, K. G.; Tolaymat, T.; Ma, L.; Rogers, K. R., Changes in silver nanoparticles exposed to human synthetic stomach fluid: effects of particle size and surface chemistry. Sci. Total Environ. 2013, 447, 90-98.

94. Vandebriel, R. J.; De Jong, W. H., A review of mammalian toxicity of ZnO nanoparticles. Nanotechnol. Sci. Appl. 2012, 5, 61-71.

95. Bouwmeester, H.; Poortman, J.; Peters, R. J.; Wijma, E.; Kramer, E.; Makama, S.; Puspitaninganindita, K.; Marvin, H. J.; Peijnenburg, A. A.; Hendriksen, P. J., Characterization of translocation of silver nanoparticles and effects on whole-genome gene expression using an in vitro intestinal epithelium coculture model. ACS Nano 2011, 5, 4091-4103.

96. Cho, W. S.; Duffin, R.; Thielbeer, F.; Bradley, M.; Megson, I. L.; Macnee, W.; Poland, C. A.; Tran, C. L.; Donaldson, K., Zeta potential and solubility to toxic ions as mechanisms of lung inflammation caused by metal/metal oxide nanoparticles. Toxicol. Sci. 2012, 126, 469-477.

97. Stebounova, L. V.; Guio, E.; Grassian, V. H., Silver nanoparticles in simulated biological media: a study of aggrega. J. Nanopart. Res. 2011, 13, 233-244.

98. Gliga, A. R.; Skoglund, S.; Wallinder, I. O.; Fadeel, B.; Karlsson, H. L., Size- dependent cytotoxicity of silver nanoparticles in human lung cells: the role of cellular uptake, agglomeration and Ag release. Part. Fibre Toxicol. 2014, 11, 11.

99. Thibodeau, M. S.; Giardina, C.; Knecht, D. A.; Helble, J.; Hubbard, A. K., Silica- induced apoptosis in mouse alveolar macrophages is initiated by lysosomal enzyme activity. Toxicol. Sci. 2004, 80, 34-48.

247

100. Persson, H. L., Iron-dependent lysosomal destabilization initiates silica-induced apoptosis in murine macrophages. Toxicol. Lett. 2005, 159, 124-133.

101. Cho, W. S.; Duffin, R.; Howie, S. E.; Scotton, C. J.; Wallace, W. A.; MacNee, W.; Bradley, M.; Megson, I. L.; Donaldson, K., Progressive severe lung injury by zinc oxide nanoparticles; the role of Zn2+ dissolution inside lysosomes. Part. Fibre Toxicol. 2011, 8, 27.

102. Navarro, E.; Piccapietra, F.; Wagner, B.; Marconi, F.; Kaegi, R.; Odzak, N.; Sigg, L.; Behra, R., Toxicity of silver nanoparticles to Chlamydomonas reinhardtii. Environ. Sci. Technol. 2008, 42, 8959-8964.

103. Braydich-Stolle, L. K.; Lucas, B.; Schrand, A.; Murdock, R. C.; Lee, T.; Schlager, J. J.; Hussain, S. M.; Hofmann, M. C., Silver Nanoparticles Disrupt GDNF/Fyn kinase Signaling in Spermatogonial Stem Cells. Toxicol. Sci. 2010, 116, 577-589.

104. Kwak, J. I.; Lee, W. M.; Kim, S. W.; An, Y. J., Interaction of citrate-coated silver nanoparticles with earthworm coelomic fluid and related cytotoxicity in Eisenia andrei. J. Appl. Toxicol. 2014, 34, 1145-1154.

105. Setyawati, M. I.; Yuan, X.; Xie, J.; Leong, D. T., The influence of lysosomal stability of silver nanomaterials on their toxicity to human cells. Biomaterials 2014, 35, 6707-6715.

106. Szentkuti, L., Light microscopical observations on luminally administered dyes, dextrans, nanospheres and microspheres in the pre-epithelial mucus gel layer of the rat distal colon. J. Controlled Release 1997, 46, 233–242.

107. Gao, H.; Shi, W.; Freund, L. B., Mechanics of receptor-mediated endocytosis. Proc. Natl. Acad. Sci. U. S. A. 2005, 102, 9469-9474.

108. Zhang, S.; Li, J.; Lykotrafitis, G.; Bao, G.; Suresh, S., Size-Dependent Endocytosis of Nanoparticles. Adv. Mater. 2009, 21, 419-424.

109. Chithrani, B. D.; Ghazani, A. A.; Chan, W. C., Determining the size and shape dependence of gold nanoparticle uptake into mammalian cells. Nano Lett. 2006, 6, 662- 668.

110. Win, K. Y.; Feng, S. S., Effects of particle size and surface coating on cellular uptake of polymeric nanoparticles for oral delivery of anticancer drugs. Biomaterials 2005, 26, 2713-2722.

111. Shang, L.; Nienhaus, K.; Jiang, X.; Yang, L.; Landfester, K.; Mailander, V.; Simmet, T.; Nienhaus, G. U., Nanoparticle interactions with live cells: Quantitative

248 fluorescence microscopy of nanoparticle size effects. Beilstein J. Nanotechnol. 2014, 5, 2388-2397.

112. Schleh, C.; Semmler-Behnke, M.; Lipka, J.; Wenk, A.; Hirn, S.; Schäffler, M.; Schmid, G.; Simon, U.; Kreyling, W. G., Size and surface charge of gold nanoparticles determine absorption across intestinal barriers and accumulation in secondary target organs after oral administration. Nanotoxicology 2012, 6, 36-46.

113. Chithrani, B. D.; Chan, W. C., Elucidating the mechanism of cellular uptake and removal of protein-coated gold nanoparticles of different sizes and shapes. Nano Lett. 2007, 7, 1542-1550.

114. Qiu, Y.; Liu, Y.; Wang, L.; Xu, L.; Bai, R.; Ji, Y.; Wu, X.; Zhao, Y.; Li, Y.; Chen, C., Surface chemistry and aspect ratio mediated cellular uptake of Au nanorods. Biomaterials 2010, 31, 7606-7619.

115. Bartneck, M.; Keul, H. A.; Singh, S.; Czaja, K.; Bornemann, J.; Bockstaller, M.; Moeller, M.; Zwadlo-Klarwasser, G.; Groll, J., Rapid uptake of gold nanorods by primary human blood phagocytes and immunomodulatory effects of surface chemistry. ACS Nano 2010, 4, 3073-3086.

116. Villanueva, A.; Canete, M.; Roca, A. G.; Calero, M.; Veintemillas-Verdaguer, S.; Serna, C. J.; Morales Mdel, P.; Miranda, R., The influence of surface functionalization on the enhanced internalization of magnetic nanoparticles in cancer cells. Nanotechnology 2009, 20, 115103.

117. He, C.; Hu, Y.; Yin, L.; Tang, C.; Yin, C., Effects of particle size and surface charge on cellular uptake and biodistribution of polymeric nanoparticles. Biomaterials 2010, 31, 3657-3666.

118. Fröhlich, E., The role of surface charge in cellular uptake and cytotoxicity of medical nanoparticles. Int. J. Nanomedicine 2012, 7, 5577-5591.

119. Nagy, A.; Zane, A.; Cole, S. L.; Severance, M.; Dutta, P. K.; Waldman, W. J., Contrast of the biological activity of negatively and positively charged microwave synthesized CdSe/ZnS quantum dots. Chem. Res. Toxicol. 2011, 24, 2176-2188.

120. Abdulkarim, M.; Agullo, N.; Cattoz, B.; Griffiths, P.; Bernkop-Schnurch, A.; Borros, S. G.; Gumbleton, M., Nanoparticle diffusion within intestinal mucus: Three- dimensional response analysis dissecting the impact of particle surface charge, size and heterogeneity across polyelectrolyte, pegylated and viral particles. Eur. J. Pharm. Biopharm. 2015, epub online Feb. 4, 2015.

121. Lai, S. K.; Wang, Y. Y.; Hanes, J., Mucus-penetrating nanoparticles for drug and gene delivery to mucosal tissues. Adv. Drug Deliv. Rev. 2009, 61, 158-171. 249

122. Subbiah, R.; Veerapandian, M.; Yun, K. S., Nanoparticles: functionalization and multifunctional applications in biomedical sciences. Curr. Med. Chem. 2010, 17, 4559- 4577.

123. Monopoli, M. P.; Aberg, C.; Salvati, A.; Dawson, K. A., Biomolecular coronas provide the biological identity of nanosized materials. Nat. Nanotechnol. 2012, 7, 779- 786.

124. Lesniak, A.; Fenaroli, F.; Monopoli, M. P.; Åberg, C.; Dawson, K. A.; Salvati, A., Effects of the Presence or Absence of a Protein Corona on Silica Nanoparticle Uptake and Impact on Cells. ACS Nano 2012, 6, 5845-5857.

125. Drescher, D.; Orts-Gil, G.; Laube, G.; Natte, K.; Veh, R. W.; Osterle, W.; Kneipp, J., Toxicity of amorphous silica nanoparticles on eukaryotic cell model is determined by particle agglomeration and serum protein adsorption effects. Anal. Bioanal. Chem. 2011, 400, 1367-1373.

126. Tedja, R.; Lim, M.; Amal, R.; Marquis, C., Effects of serum adsorption on cellular uptake profile and consequent impact of titanium dioxide nanoparticles on human lung cell lines. ACS Nano 2012, 6, 4083-4093.

127. AshaRani, P. V.; Mun, G. L. K.; Hande, M. P.; Valiyaveettil, S., Cytotoxicity and genotoxicity of silver nanoparticles in human cells. ACS Nano 2009, 3, 279-290.

128. Chairuangkitti, P.; Lawanprasert, S.; Roytrakul, S.; Aueviriyavit, S.; Phummiratch, D.; Kulthong, K.; Chanvorachote, P.; Maniratanachote, R., Silver nanoparticles induce toxicity in A549 cells via ROS-dependent and ROS-independent pathways. Toxicol. In Vitro 2013, 27, 330-338.

129. Bohmert, L.; Niemann, B.; Thunemann, A. F.; Lampen, A., Cytotoxicity of peptide-coated silver nanoparticles on the human intestinal cell line Caco-2. Arch. Toxicol. 2012, 86, 1107-1115.

130. Martirosyan, A.; Bazes, A.; Schneider, Y. J., In vitro toxicity assessment of silver nanoparticles in the presence of phenolic compounds--preventive agents against the harmful effect? Nanotoxicology 2014, 8, 573-582.

131. Wiese, A. G.; Pacifici, R. E.; Davies, K. J., Transient adaptation of oxidative stress in mammalian cells. Arch. Biochem. Biophys. 1995, 318, 231-240.

132. Davies, K. J. A.; Ethel Percy Andrus Gerontology Center and Division of Molecular Biology, U. o. S. C., 3715 McClintock Avenue, Room 306, Los Angeles, CA, 90089‐0191, U.S.A., The Broad Spectrum of Responses to Oxidants in Proliferating Cells: A New Paradigm for Oxidative Stress. IUBMB Life 1999, 48, 41-47.

250

133. Davies, K. J., Oxidative stress, antioxidant defenses, and damage removal, repair, and replacement systems. IUBMB Life 2000, 50, 279-289.

134. Martindale, J. L.; Holbrook, N. J., Cellular response to oxidative stress: signaling for suicide and survival. J. Cell Physiol. 2002, 192, 1-15.

135. Jones, D. P., Redefining Oxidative Stress. Antioxid. Redox Signaling 2006, 8, 1865-1879.

136. Circu, M. L.; Aw, T. Y., Intestinal redox biology and oxidative stress. Semin. Cell Dev. Biol. 2012, 23, 729-737.

137. Schafer, K. A., The cell cycle: a review. Vet. Pathol. 1998, 35, 461-478.

138. Dahm, L. J.; Jones, D. P., Secretion of cysteine and glutathione from mucosa to lumen in rat small intestine. Am. J. Physiol. 1994, 267, G292-G300.

139. Manke, A., Wang, L., Rojanasakul, Y, Mechanisms of Nanoparticle-Induced Oxidative Stress and Toxicity. BioMed Res. Int. 2013, 942916.

140. Buzea, C.; Pacheco, I. I.; Robbie, K., Nanomaterials and nanoparticles: Sources and toxicity. Biointerphases 2007, 2, MR17-MR71.

141. Piao, M. J.; Kang, K. A.; Lee, I. K.; Kim, H. S.; Kim, S.; Choi, J. Y.; Choi, J.; Hyun, J. W., Silver nanoparticles induce oxidative cell damage in human liver cells through inhibition of reduced glutathione and induction of mitochondria-involved apoptosis. Toxicol. Lett. 2011, 201, 92-100.

142. Park, H. J.; Kim, J. Y.; Kim, J.; Lee, J. H.; Hahn, J. S.; Gu, M. B.; Yoon, J., Silver-ion-mediated reactive oxygen species generation affecting bactericidal activity. Water Res. 2009, 43, 1027-1032.

143. Li, N.; Sioutas, C.; Cho, A.; Schmitz, D.; Misra, C.; Sempf, J.; Wang, M.; Oberley, T.; Froines, J.; Nel, A., Ultrafine particulate pollutants induce oxidative stress and mitochondrial damage. Environ. Health Perspect. 2003, 111, 455-460.

144. van Kesteren, P. C.; Cubadda, F.; Bouwmeester, H.; van Eijkeren, J. C.; Dekkers, S.; de Jong, W. H.; Oomen, A. G., Novel insights into the risk assessment of the nanomaterial synthetic amorphous silica, additive E551, in food. Nanotoxicology 2014, epub online Jul. 18, 2014, 1-10.

145. Foldbjerg, R.; Wang, J.; Beer, C.; Thorsen, K.; Sutherland, D. S.; Autrup, H., Biological effects induced by BSA-stabilized silica nanoparticles in mammalian cell lines. Chem. Biol. Interact. 2013, 204, 28-38.

251

146. Hamilton, R. F.; Thakur, S. A.; Holian, A., Silica binding and toxicity in alveolar macrophages. Free Radic. Biol. Med. 2008, 44, 1246-1258.

147. Yang, Y. X.; Song, Z. M.; Cheng, B.; Xiang, K.; Chen, X. X.; Liu, J. H.; Cao, A.; Wang, Y.; Liu, Y.; Wang, H., Evaluation of the toxicity of food additive silica nanoparticles on gastrointestinal cells. J. Appl. Toxicol. 2014, 34, 424-435.

148. Gerloff, K.; Albrecht, C.; Boots, A. W.; Förster, I.; Schins, R. P. F., Cytotoxicity and oxidative DNA damage by nanoparticles in human intestinal Caco-2 cells. Nanotoxicology 2009, 3, 355-364.

149. Sergent, J. A.; Paget, V.; Chevillard, S., Toxicity and genotoxicity of nano-SiO2 on human epithelial intestinal HT-29 cell line. Ann. Occup. Hyg. 2012, 56, 622-630.

150. Gehrke, H.; Fruhmesser, A.; Pelka, J.; Esselen, M.; Hecht, L. L.; Blank, H.; Schuchmann, H. P.; Gerthsen, D.; Marquardt, C.; Diabate, S.; Weiss, C.; Marko, D., In vitro toxicity of amorphous silica nanoparticles in human colon carcinoma cells. Nanotoxicology 2013, 7, 274-293.

151. Moos, P. J.; Olszewski, K.; Honeggar, M.; Cassidy, P.; Leachman, S.; Woessner, D.; Cutler, N. S.; Veranth, J. M., Responses of human cells to ZnO nanoparticles: a gene transcription study. Metallomics 2011, 3, 1199-1211.

152. Docter, D.; Bantz, C.; Westmeier, D.; Galla, H. J.; Wang, Q.; Kirkpatrick, J. C.; Nielsen, P.; Maskos, M.; Stauber, R. H., The protein corona protects against size- and dose-dependent toxicity of amorphous silica nanoparticles. Beilstein J. Nanotechnol. 2014, 5, 1380-1392.

153. Jacobsen, N. R.; Wallin, H.; de Jong, W.; Oomen, A.; Brandon, E.; Krystek, P.; Apostolova, M.; Karadjova, I.; Cubadda, F.; Aureli, F.; Maranghi, F.; Dive, V.; Taran, F.; Czarny, B. Deliverable 7: Identification of target organs and biodistribution including ADME parameters.; Nanogenotox: 2013; pp 1-110.

154. Kim, Y. R.; Lee, S. Y.; Lee, E. J.; Park, S. H.; Seong, N. W.; Seo, H. S.; Shin, S. S.; Kim, S. J.; Meang, E. H.; Park, M. K.; Kim, M. S.; Kim, C. S.; Kim, S. K.; Son, S. W.; Seo, Y. R.; Kang, B. H.; Han, B. S.; An, S. S.; Lee, B. J.; Kim, M. K., Toxicity of colloidal silica nanoparticles administered orally for 90 days in rats. Int. J. Nanomedicine 2014, 9 Suppl 2, 67-78.

155. Yoshida, T.; Yoshioka, Y.; Takahashi, H.; Misato, K.; Mori, T.; Hirai, T.; Nagano, K.; Abe, Y.; Mukai, Y.; Kamada, H.; Tsunoda, S.; Nabeshi, H.; Yoshikawa, T.; Higashisaka, K.; Tsutsumi, Y., Intestinal absorption and biological effects of orally administered amorphous silica particles. Nanoscale Res. Lett. 2014, 9, 532.

252

156. Lee, C. M.; Lee, T. K.; Kim, D. I.; Kim, Y. R.; Kim, M. K.; Jeong, H. J.; Sohn, M. H.; Lim, S. T., Optical imaging of absorption and distribution of RITC-SiO2 nanoparticles after oral administration. Int. J. Nanomedicine 2014, 9 Suppl 2, 243-250.

157. So, S. J.; Jang, I. S.; Han, C. S., Effect of micro/nano silica particle feeding for mice. J. Nanosci. Nanotechnol. 2008, 8, 5367-5371.

158. van der Zande, M.; Vandebriel, R. J.; Groot, M. J.; Kramer, E.; Herrera Rivera, Z. E.; Rasmussen, K.; Ossenkoppele, J. S.; Tromp, P.; Gremmer, E. R.; Peters, R. J.; Hendriksen, P. J.; Marvin, H. J.; Hoogenboom, R. L.; Peijnenburg, A. A.; Bouwmeester, H., Sub-chronic toxicity study in rats orally exposed to nanostructured silica. Part. Fibre Toxicol. 2014, 11, 8.

159. Tarantini, A.; Huet, S.; Jarry, G.; Lanceleur, R.; Poul, M.; Tavares, A.; Vital, N.; Louro, H.; Joao Silva, M.; Fessard, V., Genotoxicity of synthetic amorphous silica nanoparticles in rats following short-term exposure. Part 1: Oral route. Environ. Mol. Mutagen. 2015, 56, 218-227.

160. Cargnello, M.; Gordon, T. R.; Murray, C. B., Solution-Phase Synthesis of Titanium Dioxide Nanoparticles and Nanocrystals. Chem. Rev. 2014, 114, 9319-9345.

161. Sayes, C. M.; Wahi, R.; Kurian, P. A.; Liu, Y.; West, J. L.; Ausman, K. D.; Warheit, D. B.; Colvin, V. L., Correlating nanoscale titania structure with toxicity: a cytotoxicity and inflammatory response study with human dermal fibroblasts and human lung epithelial cells. Toxicol. Sci. 2006, 92, 174-185.

162. Gerloff, K.; Fenoglio, I.; Carella, E.; Kolling, J.; Albrecht, C.; Boots, A. W.; Forster, I.; Schins, R. P., Distinctive toxicity of TiO2 rutile/anatase mixed phase nanoparticles on Caco-2 cells. Chem. Res. Toxicol. 2012, 25, 646-655.

163. Shi, H.; Magaye, R.; Castranova, V.; Zhao, J., Titanium dioxide nanoparticles: a review of current toxicological data. Part. Fibre Toxicol. 2013, 10, 15.

164. Gitrowski, C.; Al-Jubory, A. R.; Handy, R. D., Uptake of different crystal structures of TiO(2) nanoparticles by Caco-2 intestinal cells. Toxicol. Lett. 2014, 226, 264-276.

165. Chen, X. X.; Cheng, B.; Yang, Y. X.; Cao, A.; Liu, J. H.; Du, L. J.; Liu, Y.; Zhao, Y.; Wang, H., Characterization and preliminary toxicity assay of nano-titanium dioxide additive in sugar-coated chewing gum. Small 2013, 9, 1765-1774.

166. Janer, G.; Mas del Molino, E.; Fernández-Rosas, E.; Fernández, A.; Vázquez- Campos, S., Cell uptake and oral absorption of titanium dioxide nanoparticles. Toxicol. Lett. 2014, 228, 103-110.

253

167. Cho, W. S.; Kang, B. C.; Lee, J. K.; Jeong, J.; Che, J. H.; Seok, S. H., Comparative absorption, distribution, and excretion of titanium dioxide and zinc oxide nanoparticles after repeated oral administration. Part. Fibre Toxicol. 2013, 10, 9.

168. Wang, J.; Zhou, G.; Chen, C.; Yu, H.; Wang, T.; Ma, Y.; Jia, G.; Gao, Y.; Li, B.; Sun, J.; Li, Y.; Jiao, F.; Zhao, Y.; Chai, Z., Acute toxicity and biodistribution of different sized titanium dioxide particles in mice after oral administration. Toxicol. Lett. 2007, 168, 176-185.

169. Zhang, R.; Niu, Y.; Li, Y.; Zhao, C.; Song, B.; Zhou, Y., Acute toxicity study of the interaction between titanium dioxide nanoparticles and lead acetate in mice. Environ. Toxicol. Pharmacol. 2010, 30, 52-60.

170. Vaseem, M.; Umar, A.; Hahn, Y.-B. ZnO Nanoparticles: Growth, Properties, and Applications. In Metal Oxide Nanostructures and Their Applications, Umar, A.; Hahn, Y.-B., Eds.; American Scientific Publishers: 2010; Chapter 4, pp 1-36.

171. Song, Y.; Guan, R.; Lyu, F.; Kang, T.; Wu, Y.; Chen, X., In vitro cytotoxicity of silver nanoparticles and zinc oxide nanoparticles to human epithelial colorectal adenocarcinoma (Caco-2) cells. Mutat. Res. 2014, 769, 113-118.

172. Kang, T.; Guan, R.; Chen, X.; Song, Y.; Jiang, H.; Zhao, J., In vitro toxicity of different-sized ZnO nanoparticles in Caco-2 cells. Nanoscale Res. Lett. 2013, 8, 496.

173. Cao, Y.; Roursgaard, M.; Kermanizadeh, A.; Loft, S.; Moller, P., Synergistic effects of zinc oxide nanoparticles and Fatty acids on toxicity to caco-2 cells. Int. J. Toxicol. 2015, 34, 67-76.

174. Chung, H. E.; Yu, J.; Baek, M.; Lee, J. A.; Kim, M. S.; Kim, S. H.; Maeng, E. H.; Lee, J. K.; Jeong, J.; Choi, S. J., Toxicokinetics of zinc oxide nanoparticles in rats. J. Phys.: Conf. Ser. 2013, 429, 012037.

175. Pasupuleti, S.; Alapati, S.; Ganapathy, S.; Anumolu, G.; Pully, N. R.; Prakhya, B. M., Toxicity of zinc oxide nanoparticles through oral route. Toxicol. Ind. Health 2012, 28, 675-686.

176. Shokouhian, A.; Shokouhian, A.; Soheili, S.; Soheili, S.; Moradhaseli, S.; Moradhaseli, S.; Fazli, L.; Fazli, L.; Ardestani, M. S.; Ardestani, M. S.; Ghorbani, M.; Ghorbani, M., Toxicity of zinc oxide nanoparticles in lung tissue after repeated oral administration. Am. J. Pharmacol. Toxicol. 2013, 8, 148-154.

177. Kim, Y. R.; Park, J. I.; Lee, E. J.; Park, S. H.; Seong, N.; Kim, J. H.; Kim, G. Y.; Meang, E. H.; Hong, J. S.; Kim, S. H.; Koh, S. B.; Kim, M. S.; Kim, C. S.; Kim, S. K.; Son, S. W.; Seo, Y. R.; Kang, B. H.; Han, B. S.; An, S. S. A.; Yun, H. I.; Kim, M. K.,

254

Toxicity of 100 nm zinc oxide nanoparticles: a report of 90-day repeated oral administration in Sprague Dawley rats. Int. J. Nanomedicine 2014, 9, 109-126.

178. Li, C. H.; Shen, C. C.; Cheng, Y. W.; Huang, S. H.; Wu, C. C.; Kao, C. C.; Liao, J. W.; Kang, J. J., Organ biodistribution, clearance, and genotoxicity of orally administered zinc oxide nanoparticles in mice. Nanotoxicology 2012, 6, 746-756.

179. Abou El-Nour, K. M. M.; Eftaiha, A. a.; Al-Warthan, A.; Ammar, R. A. A., Synthesis and applications of silver nanoparticles. Arabian J. Chem. 2010, 3, 135-140.

180. Beer, C.; Foldbjerg, R.; Hayashi, Y.; Sutherland, D. S.; Autrup, H., Toxicity of silver nanoparticles - nanoparticle or silver ion? Toxicol. Lett. 2012, 208, 286-292.

181. Gaiser, B. K.; Fernandes, T. F.; Jepson, M. A.; Lead, J. R.; Tyler, C. R.; Baalousha, M.; Biswas, A.; Britton, G. J.; Cole, P. A.; Johnston, B. D.; Ju-Nam, Y.; Rosenkranz, P.; Scown, T. M.; Stone, V., Interspecies comparisons on the uptake and toxicity of silver and cerium dioxide nanoparticles. Environ. Toxicol. Chem. 2012, 31, 144-154.

182. Platonova, T. A.; Pridvorova, S. M.; Zherdev, A. V.; Vasilevskaya, L. S.; Arianova, E. A.; Gmoshinski, I. V.; Khotimchenko, S. A.; Dzantiev, B. B.; Popov, V. O.; Tutelyan, V. A., Identification of silver nanoparticles in the small intestinal mucosa, liver, and spleen of rats by transmission electron microscopy. Bull. Exp. Biol. Med. 2013, 155, 236-241.

183. van der Zande, M.; Vandebriel, R. J.; Van Doren, E.; Kramer, E.; Herrera Rivera, Z.; Serrano-Rojero, C. S.; Gremmer, E. R.; Mast, J.; Peters, R. J.; Hollman, P. C.; Hendriksen, P. J.; Marvin, H. J.; Peijnenburg, A. A.; Bouwmeester, H., Distribution, elimination, and toxicity of silver nanoparticles and silver ions in rats after 28-day oral exposure. ACS Nano 2012, 6, 7427-7442.

184. Loeschner, K.; Hadrup, N.; Qvortrup, K.; Larsen, A.; Gao, X.; Vogel, U.; Mortensen, A.; Lam, H. R.; Larsen, E. H., Distribution of silver in rats following 28 days of repeated oral exposure to silver nanoparticles or silver acetate. Part. Fibre Toxicol. 2011, 8, 18.

185. Kim, W. Y.; Kim, J.; Park, J. D.; Ryu, H. Y.; Yu, I. J., Histological study of gender differences in accumulation of silver nanoparticles in kidneys of Fischer 344 rats. J. Toxicol. Environ. Health A 2009, 72, 1279-1284.

186. Park, E. J.; Bae, E.; Yi, J.; Kim, Y.; Choi, K.; Lee, S. H.; Yoon, J.; Lee, B. C.; Park, K., Repeated-dose toxicity and inflammatory responses in mice by oral administration of silver nanoparticles. Environ. Toxicol. Pharmacol. 2010, 30, 162-168.

255

187. East, B. W.; Boddy, K.; Williams, E. D.; Macintyre, D.; McLay, A. L., Silver retention, total body silver and tissue silver concentrations in argyria associated with exposure to an anti-smoking remedy containing silver acetate. Clin. Exp. Dermatol. 1980, 5, 305-311.

188. Hadrup, N.; Lam, H. R., Oral toxicity of silver ions, silver nanoparticles and colloidal silver--a review. Regul. Toxicol. Pharmacol. 2014, 68, 1-7.

189. Shahare, B.; Yashpal, M., Toxic effects of repeated oral exposure of silver nanoparticles on small intestine mucosa of mice. Toxicol. Mech. Methods. 2013, 23, 161- 167.

190. Jeong, G. N.; Jo, U. B.; Ryu, H. Y.; Kim, Y. S.; Song, K. S.; Yu, I. J., Histochemical study of intestinal mucins after administration of silver nanoparticles in Sprague-Dawley rats. Arch. Toxicol. 2010, 84, 63-69.

191. Kim, Y. S.; Song, M. Y.; Park, J. D.; Song, K. S.; Ryu, H. R.; Chung, Y. H.; Chang, H. K.; Lee, J. H.; Oh, K. H.; Kelman, B. J.; Hwang, I. K.; Yu, I. J., Subchronic oral toxicity of silver nanoparticles. Part. Fibre Toxicol. 2010, 7, 20.

192. Lee, Y.; Choi, J.; Kim, P.; Choi, K.; Kim, S.; Shon, W.; Park, K., A transfer of silver nanoparticles from pregnant rat to offspring. Toxicol. Res. 2012, 28, 139-141.

193. Sawosz, E.; Binek, M.; Grodzik, M.; Zielinska, M.; Sysa, P.; Szmidt, M.; Niemiec, T.; Chwalibog, A., Influence of hydrocolloidal silver nanoparticles on gastrointestinal microflora and morphology of enterocytes of quails. Arch. Anim. Nutr. 2007, 61, 444-451.

194. Bhol, K. C.; Alroy, J.; Schechter, P. J., Anti-inflammatory effect of topical nanocrystalline silver cream on allergic contact dermatitis in a guinea pig model. Clin. Exp. Dermatol. 2004, 29, 282-287.

195. Bhol, K. C.; Schechter, P. J., Topical nanocrystalline silver cream suppresses inflammatory cytokines and induces apoptosis of inflammatory cells in a murine model of allergic contact dermatitis. Br. J. Dermatol. 2005, 152, 1235-1242.

196. Bhol, K. C.; Schechter, P. J., Effects of nanocrystalline silver (NPI 32101) in a rat model of ulcerative colitis. Dig. Dis. Sci. 2007, 52, 2732-2742.

197. Park, M. V.; Neigh, A. M.; Vermeulen, J. P.; de la Fonteyne, L. J.; Verharen, H. W.; Briede, J. J.; van Loveren, H.; de Jong, W. H., The effect of particle size on the cytotoxicity, inflammation, developmental toxicity and genotoxicity of silver nanoparticles. Biomaterials 2011, 32, 9810-9817.

256

198. McCracken, C.; Zane, A.; Knight, D. A.; Dutta, P. K.; Waldman, W. J., Minimal intestinal epithelial cell toxicity in response to short- and long-term food-relevant inorganic nanoparticle exposure. Chem. Res. Toxicol. 2013, 26, 1514-1525.

199. Sozer, N.; Kokini, J. L., Nanotechnology and its applications in the food sector. Trends Biotechnol. 2009, 27, 82-89.

200. Chaudhry, Q., Groves, K. Nanotechnology Applications for Food Ingredients, Additives and Supplements. In Nanotechnologies in Food, Chaudhry, Q., Castle, L., Watkins, R., Ed.; RSC Publishing: Cambridge, UK, 2010, pp 69-85.

201. Chaudhry, Q.; Scotter, M.; Blackburn, J.; Ross, B.; Boxall, A.; Castle, L.; Aitken, R.; Watkins, R., Applications and implications of nanotechnologies for the food sector. Food Addit. Contam., Part A 2008, 25, 241-258.

202. U.S. Food and Drug Administration Center for Food Safety and Applied Nutrition Ingredients, Additives, GRAS & Packaging - Guidance for Industry: Assessing the Effects of Significant Manufacturing Process Changes, Including Emerging Technologies, on the Safety and Regulatory Status of Food Ingredients and Food Contact Substances, Including Food Ingredients that Are Color Additives. http://www.fda.gov/Food/GuidanceRegulation/GuidanceDocumentsRegulatoryInformati on/IngredientsAdditivesGRASPackaging/ucm300661.htm (June 9, 2014).

203. Jani, P.; Halbert, G. W.; Langridge, J.; Florence, A. T., Nanoparticle uptake by the rat gastrointestinal mucosa: quantitation and particle size dependency. J. Pharm. Pharmacol. 1990, 42, 821-826.

204. Onishchenko, G. E.; Erokhina, M. V.; Abramchuk, S. S.; Shaitan, K. V.; Raspopov, R. V.; Smirnova, V. V.; Vasilevskaya, L. S.; Gmoshinski, I. V.; Kirpichnikov, M. P.; Tutelyan, V. A., Effects of titanium dioxide nanoparticles on small intestinal mucosa in rats. Bull. Exp. Biol. Med. 2012, 154, 265-270.

205. Warheit, D. B.; Webb, T. R.; Reed, K. L.; Frerichs, S.; Sayes, C. M., Pulmonary toxicity study in rats with three forms of ultrafine-TiO2 particles: differential responses related to surface properties. Toxicology 2007, 230, 90-104.

206. Waldman, W. J.; Kristovich, R.; Knight, D. A.; Dutta, P. K., Inflammatory properties of iron-containing carbon nanoparticles. Chem. Res. Toxicol. 2007, 20, 1149- 1154.

207. Nagy, A.; Harrison, A.; Sabbani, S.; Munson, R. S., Jr.; Dutta, P. K.; Waldman, W. J., Silver nanoparticles embedded in zeolite membranes: release of silver ions and mechanism of antibacterial action. Int. J. Nanomedicine 2011, 6, 1833-1852.

257

208. Mandalari, G.; Faulks, R. M.; Rich, G. T.; Lo Turco, V.; Picout, D. R.; Lo Curto, R. B.; Bisignano, G.; Dugo, P.; Dugo, G.; Waldron, K. W.; Ellis, P. R.; Wickham, M. S., Release of protein, lipid, and vitamin E from almond seeds during digestion. J. Agric. Food Chem. 2008, 56, 3409-3416.

209. Glahn, R. P.; Wien, E. M.; Van Campen, D. R.; Miller, D. D., Caco-2 cell iron uptake from meat and casein digests parallels in vivo studies: use of a novel in vitro method for rapid estimation of iron bioavailability. J. Nutr. 1996, 126, 332-339.

210. Reboul, E.; Richelle, M.; Perrot, E.; Desmoulins-Malezet, C.; Pirisi, V.; Borel, P., Bioaccessibility of carotenoids and vitamin E from their main dietary sources. J. Agric. Food Chem. 2006, 54, 8749-8755.

211. Connolly, M. L.; Lovegrove, J. A.; Tuohy, K. M., In vitro evaluation of the microbiota modulation abilities of different sized whole oat grain flakes. Anaerobe 2010, 16, 483-488.

212. Amin, M. R.; Orenuga, T.; Tyagi, S.; Dudeja, P. K.; Ramaswamy, K.; Malakooti, J., Tumor necrosis factor-alpha represses the expression of NHE2 through NF-kappaB activation in intestinal epithelial cell model, C2BBe1. Inflamm. Bowel Dis. 2011, 17, 720-731.

213. Fredenburgh, L. E.; Velandia, M. M.; Ma, J.; Olszak, T.; Cernadas, M.; Englert, J. A.; Chung, S. W.; Liu, X.; Begay, C.; Padera, R. F.; Blumberg, R. S.; Walsh, S. R.; Baron, R. M.; Perrella, M. A., Cyclooxygenase-2 deficiency leads to intestinal barrier dysfunction and increased mortality during polymicrobial sepsis. J. Immunol. 2011, 187, 5255-5267.

214. Suttiponparnit, K.; Jiang, J.; Sahu, M.; Suvachittanont, S.; Charinpanitkul, T.; Biswas, P., Role of Surface Area, Primary Particle Size, and Crystal Phase on Titanium Dioxide Nanoparticle Dispersion Properties. Nanoscale Res. Lett. 2010, 6, 27.

215. Kosmulski, M.; Hartikainen, J.; Maczka, E.; Janusz, W.; Rosenholm, J. B., Multiinstrument study of the electrophoretic mobility of fumed silica. Anal. Chem. 2002, 74, 253-256.

216. Mudunkotuwa, I. A.; Rupasinghe, T.; Wu, C. M.; Grassian, V. H., Dissolution of ZnO nanoparticles at circumneutral pH: a study of size effects in the presence and absence of citric acid. Langmuir 2012, 28, 396-403.

217. Bian, S. W.; Mudunkotuwa, I. A.; Rupasinghe, T.; Grassian, V. H., Aggregation and dissolution of 4 nm ZnO nanoparticles in aqueous environments: influence of pH, ionic strength, size, and adsorption of humic acid. Langmuir 2011, 27, 6059-6068.

258

218. Roach, P.; Farrar, D.; Perry, C. C., Surface tailoring for controlled protein adsorption: effect of topography at the nanometer scale and chemistry. J. Am. Chem. Soc. 2006, 128, 3939-3945.

219. Kasthuri, J.; Rajendiran, N., Functionalization of silver and gold nanoparticles using amino acid conjugated bile salts with tunable longitudinal plasmon resonance. Colloids Surf., B. 2009, 73, 387-393.

220. Fogh, J.; Fogh, J. M.; Orfeo, T., One hundred and twenty-seven cultured human tumor cell lines producing tumors in nude mice. J. Natl. Cancer Inst. 1977, 59, 221-226.

221. Peterson, M. D.; Mooseker, M. S., An in vitro model for the analysis of intestinal brush border assembly. I. Ultrastructural analysis of cell contact-induced brush border assembly in Caco-2BBe cells. J. Cell Sci. 1993, 105 (Pt 2), 445-460.

222. Schuerer-Maly, C. C.; Eckmann, L.; Kagnoff, M. F.; Falco, M. T.; Maly, F. E., Colonic epithelial cell lines as a source of interleukin-8: stimulation by inflammatory cytokines and bacterial lipopolysaccharide. Immunology 1994, 81, 85-91.

223. Baek, M.; Chung, H. E.; Yu, J.; Lee, J. A.; Kim, T. H.; Oh, J. M.; Lee, W. J.; Paek, S. M.; Lee, J. K.; Jeong, J.; Choy, J. H.; Choi, S. J., Pharmacokinetics, tissue distribution, and excretion of zinc oxide nanoparticles. Int. J. Nanomedicine 2012, 7, 3081-3097.

224. Preda, S.; Teodorescu, V. S.; Musuc, A. M.; Andronescu, C.; Zaharescu, M., Influence of the TiO2 precursors on the thermal and structural stability of titanate-based nanotubes. J. Mater. Res. 2013, 28, 294-303.

225. Oberdorster, G.; Maynard, A.; Donaldson, K.; Castranova, V.; Fitzpatrick, J.; Ausman, K.; Carter, J.; Karn, B.; Kreyling, W.; Lai, D.; Olin, S.; Monteiro-Riviere, N.; Warheit, D.; Yang, H., Principles for characterizing the potential human health effects from exposure to nanomaterials: elements of a screening strategy. Part. Fibre Toxicol. 2005, 2, 8.

226. Kroll, A.; Pillukat, M. H.; Hahn, D.; Schnekenburger, J., Interference of engineered nanoparticles with in vitro toxicity assays. Arch. Toxicol. 2012, 86, 1123- 1136.

227. Han, X.; Gelein, R.; Corson, N.; Wade-Mercer, P.; Jiang, J.; Biswas, P.; Finkelstein, J. N.; Elder, A.; Oberdorster, G., Validation of an LDH assay for assessing nanoparticle toxicity. Toxicology 2011, 287, 99-104.

228. Fisichella, M.; Dabboue, H.; Bhattacharyya, S.; Saboungi, M. L.; Salvetat, J. P.; Hevor, T.; Guerin, M., Mesoporous silica nanoparticles enhance MTT formazan exocytosis in HeLa cells and astrocytes. Toxicol. In Vitro 2009, 23, 697-703. 259

229. Fisichella, M.; Berenguer, F.; Steinmetz, G.; Auffan, M.; Rose, J.; Prat, O., Intestinal toxicity evaluation of TiO2 degraded surface-treated nanoparticles: a combined physico-chemical and toxicogenomics approach in caco-2 cells. Part. Fibre Toxicol. 2012, 9, 18.

230. Shukla, R. K.; Sharma, V.; Pandey, A. K.; Singh, S.; Sultana, S.; Dhawan, A., ROS-mediated genotoxicity induced by titanium dioxide nanoparticles in human epidermal cells. Toxicol. In Vitro 2011, 25, 231-241.

231. Blanchette, C. D.; Woo, Y. H.; Thomas, C.; Shen, N.; Sulchek, T. A.; Hiddessen, A. L., Decoupling internalization, acidification and phagosomal-endosomal/lysosomal fusion during phagocytosis of InlA coated beads in epithelial cells. PLoS One 2009, 4, e6056.

232. Plum, L. M.; Rink, L.; Haase, H., The essential toxin: impact of zinc on human health. Int. J. Environ. Res. Public Health 2010, 7, 1342-1365.

233. Brown, A. M.; Kristal, B. S.; Effron, M. S.; Shestopalov, A. I.; Ullucci, P. A.; Sheu, K. F.; Blass, J. P.; Cooper, A. J., Zn2+ inhibits alpha-ketoglutarate-stimulated mitochondrial respiration and the isolated alpha-ketoglutarate dehydrogenase complex. J. Biol. Chem. 2000, 275, 13441-13447.

234. Sheline, C. T.; Behrens, M. M.; Choi, D. W., Zinc-induced cortical neuronal death: contribution of energy failure attributable to loss of NAD(+) and inhibition of glycolysis. J. Neurosci. 2000, 20, 3139-3146.

235. Tan, Y.; Qi, J.; Lu, Y.; Hu, F.; Yin, Z.; Wu, W., Lecithin in mixed micelles attenuates the cytotoxicity of bile salts in Caco-2 cells. Toxicol. In Vitro 2013, 27, 714- 720.

236. Walczak, A. P.; Kramer, E.; Hendriksen, P. J.; Helsdingen, R.; van der Zande, M.; Rietjens, I. M.; Bouwmeester, H., In vitro gastrointestinal digestion increases the translocation of polystyrene nanoparticles in an in vitro intestinal co-culture model. Nanotoxicology 2015, epub online Feb. 12, 2015, 1-9.

237. Zane, A.; McCracken, C.; Knight, D. A.; Waldman, W. J.; Dutta, P. K., Spectroscopic Evaluation of the Nucleation and Growth for Microwave-Assisted CdSe/CdS/ZnS Quantum Dot Synthesis. J. Phys. Chem. C 2014, 118, 22258-22267.

238. Zane, A.; McCracken, C.; Knight, D. A.; Young, T.; Lutton, A. D.; Olesik, J. W.; Waldman, W. J.; Dutta, P. K., Uptake of bright fluorophore core-silica shell nanoparticles by biological systems. Int. J. Nanomedicine 2015, 10, 1547-1567.

239. Jiang, W.; Kim, B. Y. S.; Rutka, J. T.; Chan, W. C. W., Nanoparticle-mediated cellular response is size-dependent. Nat. Nanotechnol. 2008, 3, 145-150. 260

240. Brus, L. E., A simple model for the ionization potential, electron affinity, and aqueous redox potentials of small semiconductor crystallites. J. Chem. Phys. 1983, 79, 5566-5571.

241. Dabbousi, B. O.; Rodriguez-Viejo, J.; Mikulec, F. V.; Heine, J. R.; Mattoussi, H.; Ober, R.; Jensen, K. F.; Bawendi, M. G., (CdSe)ZnS Core−Shell Quantum Dots: Synthesis and Characterization of a Size Series of Highly Luminescent Nanocrystallites. J. Phys. Chem. B 1997, 101, 9463-9475.

242. Li, L.; Qian, H.; Ren, J., Rapid synthesis of highly luminescent CdTe nanocrystals in the aqueous phase by microwave irradiation with controllable temperature. Chem. Commun. (Cambridge, U. K.) 2005, 28, 528-530.

243. Bowers, M. J., 2nd; McBride, J. R.; Rosenthal, S. J., White-light emission from magic-sized cadmium selenide nanocrystals. J. Am. Chem. Soc. 2005, 127, 15378-15379.

244. Ma, Q.; Su, X., Near-infrared quantum dots: synthesis, functionalization and analytical applications. Analyst 2010, 135, 1867-1877.

245. Resch-Genger, U.; Grabolle, M.; Cavaliere-Jaricot, S.; Nitschke, R.; Nann, T., Quantum dots versus organic dyes as fluorescent labels. Nat. Methods 2008, 5, 763-775.

246. Hyldahl, M. G.; Bailey, S. T.; Wittmershaus, B. P., Photo-stability and performance of CdSe/ZnS quantum dots in luminescent solar concentrators. Sol. Energy 2009, 83, 566–573.

247. Lee, K. S.; Lee, D. U.; Choo, D. C.; Kim, T. W.; Ryu, E. D.; Kim, S. W.; Lim, J. S., Organic light-emitting devices fabricated utilizing core/shell CdSe/ZnS quantum dots embedded in a polyvinylcarbazole. J. Mater. Sci. 2011, 46, 1239-1243.

248. Jang, E.; Jun, S.; Jang, H.; Lim, J.; Kim, B.; Kim, Y., White-light-emitting diodes with quantum dot color converters for display backlights. Adv. Mater. 2010, 22, 3076- 3080.

249. de Mello Donega, C.; Liljeroth, P.; Vanmaekelbergh, D., Physicochemical evaluation of the hot-injection method, a synthesis route for monodisperse nanocrystals. Small 2005, 1, 1152-1162.

250. Hines, M. A.; Guyot-Sionnest, P., Synthesis and Characterization of Strongly Luminescing ZnS-Capped CdSe Nanocrystals. J. Phys. Chem. 1996, 100, 468-471.

251. Spanhel, L.; Haase, M.; Weller, H.; Henglein, A., Photochemistry of colloidal semiconductors. 20. Surface modification and stability of strong luminescing CdS particles. J. Am. Chem. Soc. 1987, 109, 5649-5655.

261

252. Pong, B.-K.; Trout, B. L.; Lee, J.-Y., Modified Ligand-Exchange for Efficient Solubilization of CdSe/ZnS Quantum Dots in Water: A Procedure Guided by Computational Studies. Langmuir 2008, 24, 5270-5276.

253. Rogach, A. L.; Kornowski, A.; Gao, M.; Eychmüller, A.; Weller, H., Synthesis and Characterization of a Size Series of Extremely Small Thiol-Stabilized CdSe Nanocrystals. J. Phys. Chem. B 1999, 103, 3065-3069.

254. Kapitonov, A. M.; Stupak, A. P.; Gaponenko, S. V.; Petrov, E. P.; Rogach, A. L.; Eychmüller, A., Luminescence Properties of Thiol-Stabilized CdTe Nanocrystals. J. Phys. Chem. B 1999, 103, 10109-10113.

255. He, Y.; Lu, H. T.; Sai, L. M.; Su, Y. Y.; Hu, M.; Fan, C. H.; Huang, W.; Wang, L. H., Microwave Synthesis of Water‐Dispersed CdTe/CdS/ZnS Core‐Shell‐Shell Quantum Dots with Excellent Photostability and Biocompatibility. Adv. Mater. 2008, 20, 3416- 3421.

256. Qian, H.; Qiu, X.; Li, L.; Ren, J., Microwave-Assisted Aqueous Synthesis: A Rapid Approach to Prepare Highly Luminescent ZnSe(S) Alloyed Quantum Dots. J. Phys. Chem. B 2006, 110, 9034-9040.

257. Schumacher, W.; Nagy, A.; Waldman, W. J.; Dutta, P. K., Direct Synthesis of Aqueous CdSe/ZnS-Based Quantum Dots Using Microwave Irradiation. J. Phys. Chem. C 2009, 113, 12132-12139.

258. Han, H.; Francesco, G. D.; Maye, M. M., Size Control and Photophysical Properties of Quantum Dots Prepared via a Novel Tunable Hydrothermal Route. J. Phys. Chem. C 2010, 114, 19270-19277.

259. Gerbec, J. A.; Magana, D.; Washington, A.; Strouse, G. F., Microwave-enhanced reaction rates for nanoparticle synthesis. J. Am. Chem. Soc. 2005, 127, 15791-15800.

260. Roy, M. D.; Herzing, A. A.; De Paoli Lacerda, S. H.; Becker, M. L., Emission- tunable microwave synthesis of highly luminescent water soluble CdSe/ZnS quantum dots. Chem. Commun. 2008, 28, 2106-2108.

261. Dekkers, S.; Bouwmeester, H.; Bos, P. M.; Peters, R. J.; Rietveld, A. G.; Oomen, A. G., Knowledge gaps in risk assessment of nanosilica in food: evaluation of the dissolution and toxicity of different forms of silica. Nanotoxicology 2013, 7, 367-377.

262. Stöber, W.; Fink, A., Controlled growth of monodisperse silica spheres in the micron size range. J. Colloid Interface Sci. 1968, 26, 62–69.

262

263. Rahman, I. A.; Padavettan, V., Synthesis of silica nanoparticles by sol-gel: size- dependent properties, surface modification, and applications in silica-polymer nanocomposites — a review. J. Nanomater. 2012, 2012, e132424.

264. Osseo-Asare, K.; Arriagada, F. J., Preparation of SiO2 nanoparticles in a non- ionic reverse micellar system. Colloids and Surf. 1990, 50, 321–339.

265. Darbandi, M.; Thomann, R.; Nann, T., Single Quantum Dots in Silica Spheres by Microemulsion Synthesis. Chem. Mater. 2005, 17, 5720-5725.

266. Daberkow, T.; Meder, F.; Treccani, L.; Schowalter, M.; Rosenauer, A.; Rezwan, K., Fluorescence labeling of colloidal core-shell particles with defined isoelectric points for in vitro studies. Acta Biomater. 2012, 8, 720-727.

267. Chu, M.; Sun, Y.; Xu, S., Silica-coated quantum dots fluorescent spheres synthesized using a quaternary 'water-in-oil' microemulsion system. J. Nanopart. Res. 2008, 10, 613-624.

268. Ma, N.; Marshall, A. F.; Gambhir, S. S.; Rao, J., Facile synthesis, silanization, and biodistribution of biocompatible quantum dots. Small 2010, 6, 1520-1528.

269. Zhang, B.; Gong, X.; Hao, L.; Cheng, J.; Han, Y.; Chang, J., A novel method to enhance quantum yield of silica-coated quantum dots for biodetection. Nanotechnology 2008, 19, 465604.

270. Bae, S. W.; Tan, W.; Hong, J.-I., Fluorescent dye-doped silica nanoparticles: new tools for bioapplications. Chem. Commun. 2012, 48, 2270-2282.

271. Park, S.-J.; Kim, Y.-J.; Park, S.-J., Size-Dependent Shape Evolution of Silica Nanoparticles into Hollow Structures. Langmuir 2008, 24, 12134-12137.

272. Mahon, E.; Hristov, D. R.; Dawson, K. A., Stabilising fluorescent silica nanoparticles against dissolution effects for biological studies. Chem. Commun. 2012, 48, 7970-7972.

273. Yokoi, T.; Sakamoto, Y.; Terasaki, O.; Kubota, Y.; Okubo, T.; Tatsumi, T., Periodic arrangement of silica nanospheres assisted by amino acids. J. Am. Chem. Soc. 2006, 128, 13664-13665.

274. Frangioni, J. V., In vivo near-infrared fluorescence imaging. Curr. Opin. Chem. Biol. 2003, 7, 626-634.

275. Magde, D.; Wong, R.; Seybold, P. G., Fluorescence quantum yields and their relation to lifetimes of rhodamine 6G and fluorescein in nine solvents: improved absolute standards for quantum yields. Photochem. Photobiol. 2002, 75, 327-334.

263

276. Alessi, A.; Salvalaggio, M.; Ruzzon, G., Rhodamine 800 as reference substance for fluorescence quantum yield measurements in deep red emission range. J. Lumin. 2013, 134, 385–389.

277. Kawata, G.; Ogawa, Y.; Minami, F., Density dependence of photoluminescence lifetime of CdSe/ZnS core-shell colloidal quantum dots. J. Appl. Phys. 2011, 110, 064323.

278. Gong, H. M.; Zhou, Z. K.; Song, H.; Hao, Z. H.; Han, J. B.; Zhai, Y. Y.; Xiao, S.; Wang, Q. Q., The influence of surface trapping and dark states on the fluorescence emission efficiency and lifetime of CdSe and CdSe/ZnS quantum dots. J. Fluoresc. 2007, 17, 715-720.

279. Baranov, A. V.; Rakovich, Y. P.; Donegan, J. F.; Perova, T. S.; Moore, R. A.; Talapin, D. V.; Rogach, A. L.; Masumoto, Y.; Nabiev, I., Effect of ZnS shell thickness on the phonon spectra in CdSe quantum dots. Phys. Rev. B 2003, 68, 165306.

280. Al-Oweini, R.; El-Rassy, H., Synthesis and characterization by FTIR spectroscopy of silica aerogels prepared using several Si(OR)4 and R′′Si(OR′)3 precursors. J. Mol. Struct. 2009, 919, 140–145.

281. El Rassy, H.; Pierre, A. C., NMR and IR spectroscopy of silica aerogels with different hydrophobic characteristics. J. Non-Cryst. Solids 2005, 351, 1603–1610.

282. Leadbeater, N. E., In situ reaction monitoring of microwave-mediated reactions using IR spectroscopy. Chem. Commun. (Cambridge, U. K.) 2010, 46, 6693-6695.

283. Ziegler, J.; Merkulov, A.; Grabolle, M.; Resch-Genger, U.; Nann, T., High- quality ZnS shells for CdSe nanoparticles: rapid microwave synthesis. Langmuir 2007, 23, 7751-7759.

284. Jiang, F.; Muscat, A. J., Ligand-Controlled Growth of ZnSe Quantum Dots in Water during Ostwald Ripening. Langmuir 2012, 28, 12931-12940.

285. Wang, Y.; Tang, Z.; Correa-Duarte, M. A.; Pastoriza-Santos, I.; Giersig, M.; Kotov, N. A.; Liz-Marzán, L. M., Mechanism of Strong Luminescence Photoactivation of Citrate-Stabilized Water-Soluble Nanoparticles with CdSe Cores. J. Phys. Chem. B 2004, 108, 15461-15469.

286. Jones, M.; Nedeljkovic, J.; Ellingson, R. J.; Nozik, A. J.; Rumbles, G., Photoenhancement of Luminescence in Colloidal CdSe Quantum Dot Solutions. J. Phys. Chem. B 2003, 107, 11346-11352.

264

287. Torimoto, T.; Nishiyama, H.; Sakata, T.; Mori, H.; Yoneyama, H., Characteristic Features of Size‐Selective Photoetching of CdS Nanoparticles as a Means of Preparation of Monodisperse Particles. J. Electrochem. Soc. 1998, 145, 1964-1968.

288. Bakalova, R.; Zhelev, Z.; Jose, R.; Nagase, T.; Ohba, H.; Ishikawa, M.; Baba, Y., Role of free cadmium and selenium ions in the potential mechanism for the enhancement of photoluminescence of CdSe quantum dots under ultraviolet irradiation. J. Nanosci. Nanotechnol. 2005, 5, 887-894.

289. Wang, L.-L.; Jiang, J.-S., Optical performance evolutions of reductive glutathione coated CdSe qu. J. Nanopart. Res. 2011, 13, 1301-1309.

290. Gaponik, N.; Talapin, D. V.; Rogach, A. L.; Hoppe, K.; Shevchenko, E. V.; Kornowski, A.; Alexander Eychmüller, a.; Weller, H., Thiol-Capping of CdTe Nanocrystals: An Alternative to Organometallic Synthetic Routes. J. Phys. Chem. B 2002, 106, 7177-7185.

291. Costa, C. A. R.; Leite, C. A. P.; Galembeck, F., Size Dependence of Stöber Silica Nanoparticle Microchemistry. J. Phys. Chem. B 2003, 107, 4747-4755.

292. Gerion, D.; Pinaud, F.; Williams, S. C.; Parak, W. J.; Zanchet, D.; Weiss, S.; Alivisatos, A. P., Synthesis and Properties of Biocompatible Water-Soluble Silica-Coated CdSe/ZnS Semiconductor Quantum Dots. J. Phys. Chem. B 2001, 105, 8861-8871.

293. Buiculescu, R.; Hatzimarinaki, M.; Chaniotakis, N. A., Biosilicated CdSe/ZnS quantum dots as photoluminescent transducers for acetylcholinesterase-based biosensors. Anal. Bioanal. Chem. 2010, 398, 3015-3021.

294. Chen, H.; Cui, S.; Tu, Z.; Gu, Y.; Chi, X., In vivo monitoring of organ-selective distribution of CdHgTe/SiO2 nanoparticles in mouse model. J. Fluoresc. 2012, 22, 699- 706.

295. Alexander, J. W., History of the medical use of silver. Surg. Infect. (Larchmt.) 2009, 10, 289-292.

296. Kim, J. S.; Kuk, E.; Yu, K. N.; Kim, J. H.; Park, S. J.; Lee, H. J.; Kim, S. H.; Park, Y. K.; Park, Y. H.; Hwang, C. Y.; Kim, Y. K.; Lee, Y. S.; Jeong, D. H.; Cho, M. H., Antimicrobial effects of silver nanoparticles. Nanomedicine 2007, 3, 95-101.

297. Eom, H. J.; Choi, J., p38 MAPK activation, DNA damage, cell cycle arrest and apoptosis as mechanisms of toxicity of silver nanoparticles in Jurkat T cells. Environ. Sci. Technol. 2010, 44, 8337-8342.

298. Doty, R. C.; Tshikhudo, T. R.; Brust, M.; Fernig, D. G., Extremely stable water- soluble Ag nanoparticles. Chem. Mater. 2005, 17, 4630-4635.

265

299. Biggs, S.; Scales, P. J.; Leong, Y.-K.; Healy, T. W., Effects of citrate adsorption on the interactions between zirconia surfaces. J. Chem. Soc., Faraday Trans. 1995, 91, 2921-2928.

300. Canamares, M. V.; Garcia-Ramos, J. V.; Gomez-Varga, J. D.; Domingo, C.; Sanchez-Cortes, S., Comparative study of the morphology, aggregation, adherence to glass, and surface-enhanced Raman scattering activity of silver nanoparticles prepared by chemical reduction of Ag+ using citrate and hydroxylamine. Langmuir 2005, 21, 8546- 8553.

301. Oh, S. J.; Kim, H.; Liu, Y.; Han, H. K.; Kwon, K.; Chang, K. H.; Park, K.; Kim, Y.; Shim, K.; An, S. S.; Lee, M. Y., Incompatibility of silver nanoparticles with lactate dehydrogenase leakage assay for cellular viability test is attributed to protein binding and reactive oxygen species generation. Toxicol. Lett. 2014, 225, 422-432.

302. AshaRani, P.; Sethu, S.; Lim, H. K.; Balaji, G.; Valiyaveettil, S.; Hande, M. P., Differential regulation of intracellular factors mediating cell cycle, DNA repair and inflammation following exposure to silver nanoparticles in human cells. Genome Integr. 2012, 3, 2.

303. Lee, Y. S.; Kim, D. W.; Lee, Y. H.; Oh, J. H.; Yoon, S.; Choi, M. S.; Lee, S. K.; Kim, J. W.; Lee, K.; Song, C. W., Silver nanoparticles induce apoptosis and G2/M arrest via PKCζ-dependent signaling in A549 lung cells. Arch. Toxicol. 2011, 85, 1529-1540.

304. Peng, H.; Zhang, X.; Wei, Y.; Liu, W.; Li, S.; Yu, G.; Fu, X.; Cao, T.; Deng, X., Cytotoxicity of silver nanoparticles in human embryonic stem cell-derived fibroblasts and an L-929 cell line. J. Nanomater. 2012, 2012, 9.

305. Baek, M.; Kim, M. K.; Cho, H. J.; Lee, J. A.; Yu, J.; Chung, H. E.; Choi, S. J., Factors influencing the cytotoxicity of zinc oxide nanoparticles: particle size and surface charge. J. Phys.: Conf. Ser. 2011, 304, 7.

306. Wang, J. J.; Sanderson, B. J.; Wang, H., Cyto- and genotoxicity of ultrafine TiO2 particles in cultured human lymphoblastoid cells. Mutat. Res. 2007, 628, 99-106.

307. Komatsu, T.; Tabata, M.; Kubo-Irie, M.; Shimizu, T.; Suzuki, K.; Nihei, Y.; Takeda, K., The effects of nanoparticles on mouse testis Leydig cells in vitro. Toxicol. In Vitro 2008, 22, 1825-1831.

308. Barone, F.; Berardis, B. D.; Bizzarri, L.; Degan, P.; Andreoli, C.; Zijno, A.; Angelis, I. D., Physico-chemical characteristics and cyto-genotoxic potential of ZnO and TiO2 nanoparticles on human colon carcinoma cells. J. Phys.: Conf. Ser. 2011, 304.

309. Pavelic, K.; Hadzija, M.; Bedrica, L.; Pavelic, J.; Dikic, I.; Katic, M.; Kralj, M.; Bosnar, M. H.; Kapitanovic, S.; Poljak-Blazi, M.; Krizanac, S.; Stojkovic, R.; Jurin, M.; 266

Subotic, B.; Colic, M., Natural zeolite clinoptilolite: new adjuvant in anticancer therapy. J. Mol. Med. (Berl.) 2001, 78, 708-720.

310. AshaRani, P.; Hande, M.P.; Valiyaveettil, S., Anti-proliferative activity of silver nanoparticles. BMC Cell Biol. 2009, 10, 65.

311. Ramirez-Lee, M. A.; Rosas-Hernandez, H.; Salazar-Garcia, S.; Gutierrez- Hernandez, J. M.; Espinosa-Tanguma, R.; Gonzalez, F. J.; Ali, S. F.; Gonzalez, C., Silver nanoparticles induce anti-proliferative effects on airway smooth muscle cells. Role of nitric oxide and muscarinic receptor signaling pathway. Toxicol. Lett. 2014, 224, 246- 256.

312. Miethling-Graff, R.; Rumpker, R.; Richter, M.; Verano-Braga, T.; Kjeldsen, F.; Brewer, J.; Hoyland, J.; Rubahn, H. G.; Erdmann, H., Exposure to silver nanoparticles induces size- and dose-dependent oxidative stress and cytotoxicity in human colon carcinoma cells. Toxicol. In Vitro 2014, 28, 1280-1289.

313. Kang, K.; Jung, H.; Lim, J. S., Cell Death by Polyvinylpyrrolidine-Coated Silver Nanoparticles is Mediated by ROS-Dependent Signaling. Biomol. Ther. 2012, 20, 399- 405.

314. Cortese-Krott, M. M.; Munchow, M.; Pirev, E.; Hessner, F.; Bozkurt, A.; Uciechowski, P.; Pallua, N.; Kroncke, K. D.; Suschek, C. V., Silver ions induce oxidative stress and intracellular zinc release in human skin fibroblasts. Free Radical Biol. Med. 2009, 47, 1570-1577.

315. Walczak, A. P.; Fokkink, R.; Peters, R.; Tromp, P.; Herrera Rivera, Z. E.; Rietjens, I. M.; Hendriksen, P. J.; Bouwmeester, H., Behaviour of silver nanoparticles and silver ions in an in vitro human gastrointestinal digestion model. Nanotoxicology 2013, 7, 1198-1210.

316. Yang, X.; Gondikas, A. P.; Marinakos, S. M.; Auffan, M.; Liu, J.; Hsu-Kim, H.; Meyer, J. N., Mechanism of silver nanoparticle toxicity is dependent on dissolved silver and surface coating in Caenorhabditis elegans. Environ. Sci. Technol. 2012, 46, 1119- 1127.

317. Saptarshi, S. R.; Duschl, A.; Lopata, A. L., Interaction of nanoparticles with proteins: relation to bio-reactivity of the nanoparticle. J. Nanobiotechnol. 2013, 11, 26.

318. Adams, N.; Kramer, J., Potentiometric determination of silver thiolate formation constants using a Ag2S electrode. Aquat. Geochem. 1999, 5, 1-11.

319. Bell, R. A.; Kramer, J. R., Structural chemistry and geochemistry of silver‐sulfur compounds: Critical review. Environ. Toxicol. Chem. 1999, 18, 9-22.

267

320. Alekseev, V.; Semenov, A.; Pakhomov, P., Complexation of Ag+ ions with L- cysteine. Russ. J. Inorg. Chem. 2012, 57, 1041-1044.

321. Komarov, P. V.; Mikhailov, I. V.; Alekseev, V. G.; Khizhnyak, S. D.; Pakhomov, P. M., Self-assembly and gel formation processes in an aqueous solution of L-cysteine and silver nitrate. J. Struct. Chem. 2012, 53, 988-1005.

322. Pakhomov, P. M.; Abramchuk, S. S.; Khizhnyak, S. D.; Ovchinnikov, M. M.; Spiridonova, V. M., Formation of nanostructured hydrogels in L-cysteine and silver nitrate. Nanotechol. Russ. 2010, 5, 209-213.

323. Miyayama, T.; Arai, Y.; Suzuki, N.; Hirano, S., Mitochondrial electron transport is inhibited by disappearance of metallothionein in human bronchial epithelial cells following exposure to silver nitrate. Toxicology 2013, 305, 20-29.

324. Powers, C. M.; Badireddy, A. R.; Ryde, I. T.; Seidler, F. J.; Slotkin, T. A., Silver nanoparticles compromise neurodevelopment in PC12 cells: critical contributions of silver ion, particle size, coating, and composition. Environ. Health Perspect. 2011, 119, 37-44.

325. NATO Handbook on the Medical Aspects of NBC Defensive Operations AMedP- 6(B), In Army Field Manual 8-9; 2015.

326. Jaklevic, B.; Uyetake, L.; Lemstra, W.; Chang, J.; Leary, W.; Edwards, A.; Vidwans, S.; Sibon, O.; Tin Su, T., Contribution of Growth and Cell Cycle Checkpoints to Radiation Survival in Drosophila. Genetics 2006, 174, 1963-1972.

327. Arnold, J. G.; Dubois, A., In vitro studies of intragastric digestion. Dig. Dis. Sci. 1983, 28, 737-741.

328. Hur, S. J.; Lim, B. O.; Decker, E. A.; McClements, D. J., In vitro human digestion models for food applications. Food Chem. 2011, 125, 1–12.

329. Guilloteau, P.; Zabielski, R.; Hammon, H. M.; Metges, C. C., Nutritional programming of gastrointestinal tract development. Is the pig a good model for man? Nutr. Res. Rev. 2010, 23, 4-22.

330. Baker, J.; Garrod, D., Epithelial cells retain junctions during mitosis. J. Cell Sci. 1993, 104 ( Pt 2), 415-425.

331. Ragkousi, K.; Gibson, M. C., Cell division and the maintenance of epithelial order. J. Cell Biol. 2014, 207, 181-188.

332. Chassaing, B.; Aitken, J. D.; Malleshappa, M.; Vijay-Kumar, M., Dextran Sulfate Sodium (DSS)-Induced Colitis in Mice. Curr. Protoc. Immunol. 104, Unit-15.25.

268

333. Song, K. H.; Fasano, A.; Eddington, N. D., Effect of the six-mer synthetic peptide (AT1002) fragment of zonula occludens toxin on the intestinal absorption of cyclosporin A. Int. J. Pharm. 2008, 351, 8-14.

269