ABSTRACT HILTON, GINA MARIE. Proteomic Investigation of Various Methods Used for Carbon Nanotube Exposure

ABSTRACT

HILTON, GINA MARIE. Proteomic Investigation of Various Methods used for Carbon Nanotube Exposure (Under the direction of Dr. Michael Bereman and Dr. James Bonner).

Nanotechnology is a rapidly emerging field that has produced several types of nanomaterials, such as carbon nanotubes, for both industrial and medical applications. Carbon nanotubes contain highly desirable unique physical properties for use in electronics. However, some types of carbon nanotubes have been shown to initiate inflammatory response and induce fibrosis in the lung upon inhalation exposure. The specific biological response is dependent on the functionalization (i.e. coating) and the physical properties of the individual nanotube.

Thus, toxicologists are faced with a significant problem because carbon nanotubes are evolving quickly and being manufactured faster than researchers can provide traditional acute and chronic toxicity testing. Thus, there is a need to evaluate alternative methods for higher- throughput toxicity screening for emerging nanomaterials.

One of the major limitations of traditional toxicity testing is the time, cost, and data extrapolation associated with non-human mammal-based chronic exposures. Fortunately, these issues are now being alleviated by advances in bio-engineering and tissue/cell culture methods. Researchers are now able to conduct toxicity testing with more physiologically relevant cell culture based methods, using technologies such as 3D bioprinting and organs-on- a-chip.

An additional limitation to traditional toxicity testing is the collection of a few select data endpoints. While this approach was viable in previous decades, advances in technology now lend opportunity to assess global changes in gene expression upon toxicant exposure.

Using a global approach provides novel mediators that have not been previously considered for their potential contributions to toxicant response. Thus, a global approach to toxicity testing can provide invaluable insight into cellular response that was not previously considered.

Herein, a mass spectrometry based proteomics approach was used to evaluate global changes to proteins in order to assess several methods for carbon nanotube toxicity testing.

Proteomic carbon nanotube toxicity testing was conducted by in-vitro mono-culture, in-vitro co-culture, and by in-vivo models in order to compare merit for each methods. We hypothesize that inter-cell signaling (i.e. cytokines and growth factors) from in-vitro co-cultures will express a proteomic response similar to the proteomic signaling observed in in-vivo studies compared to in-vitro mono-cultures. The results from each acute exposure study overall indicate an oxidative stress response, consistent with previous findings. However, each method had its own merit and drawbacks that are important for the nanotoxicology community to consider moving forward with nanomaterial toxicity testing.

by Gina Marie Hilton

A dissertation submitted to the Graduate Faculty of North Carolina State University in partial fulfillment of the requirements for the degree of Doctorate in Philosophy

Toxicology

Raleigh, North Carolina

2017

APPROVED BY:

______Dr. Michael Bereman Dr. James Bonner Committee Co-Chair Committee Co-Chair

______Dr. Jane Hoppin Dr. Rob Smart

DEDICATION

In dedication to my family, friends, and Sweetpea, the greatest friend I’ve ever known.

BIOGRAPHY

Gina Hilton was born in Annapolis Maryland in 1985. She received her high school diploma in Maryland in 2003, and subsequently earned an undergraduate degree in biological sciences at the University of Maryland, Baltimore County (2007). During her undergraduate research she learned basic molecular biology methodologies that she applied to the complex genetics of energy allocation in life history traits. During Gina’s undergraduate studies, she successfully conducted undergraduate research and maintained a part time job to help fund her education.

After receiving her Bachelor of Science degree, she chose to further her research experience in genetics by accepting a research technologist position at The Johns Hopkins University,

Mckusick-Nathans Institute of Genetic Medicine. Gina’s primary research focus was to conduct Genome Wide Association Studies (GWAS) in order to identify single nucleotide polymorphisms in a variety of human complex genetic disorders. This research provided her with valuable expertise in generating and organizing high-throughput genomic data, in addition to learning basic statistical analysis appropriate for genetic association studies. She received invaluable experience working at JHU by learning how to thrive with high pressure/fast paced research intensive groups that successfully produced significant results with collaborators world-wide. Not only did she conduct effective research, but she also trained several lab groups to use multiple genotyping platforms and helped medical doctors conduct statistical data analysis, both of which helped develop my strong communication skills. After her time researching in genetics, she decided to broaden her research skill set by pursuing a master’s degree in chemistry at Wake Forest University (WFU) in 2010. At WFU, Gina gained significant experience researching how synthetic organic compounds can be used to label biological molecules (i.e. proteins) via ‘click’ chemistry. Throughout her time researching in

iii

chemistry, she acquired a knowledge base to better predict and understand chemical reactivity in a biological system. Gina successfully earned her Masters of Science degree in 2013, and decided to pursue her PhD at North Carolina State University (NCSU). Gina’s research at

NCSU was highly interdisciplinary; transcending fields including analytical chemistry, proteomics, and toxicology. The combination of her previous and current research experience has made her well-suited to research the development of a proteomic in vitro screen for carbon nanotube toxicity. Her diverse background in the biological and chemical sciences has provided a strong foundation for her to be a successful toxicologist in order to ultimately elucidate mechanisms of toxicity and prevent adverse effects on human health. Gina anticipates earning her PhD in 2017, and will pursue research in alternative testing to reduce animal use in regulatory screening.

ACKNOWLEDGMENTS

I would like to thank my friends and family for their unconditional support. I would also like to thank my peers and research advisors for their guidance and mentorship, without which I

would not be a successful PhD candidate.

TABLE OF CONTENTS

LIST OF TABLES ...... ix

LIST OF FIGURES ...... xii

LIST OF EQUATIONS ...... xvii

LIST OF SYMBOLS AND ABBREVIATIONS ...... xviii

CHAPTER 1: Introduction ...... 1

1.1 Introduction to Carbon Nanotubes ...... 1

1.1.1 Carbon Nanotube Characteristics ...... 1

1.1.2 Electronic Properties of Carbon Nanotubes...... 2

1.1.3 Carbon Nanotube Synthesis ...... 3

1.1.4Applications of Carbon Nanotubes ...... 4

1.2 Carbon Nanotubes as a Potential Emerging Toxicant ...... 5

1.2.1 CNT Toxicity Overview ...... 5

1.2.2 In-vivo CNT Toxicity ...... 6

1.2.3 In-vitro CNT Toxicity ...... 9

1.3 Proteomic Toxicity Testing ...... 11

1.3.1 Proteomics ...... 11

1.3.2 Mass Spectrometry ...... 13

1.3.3 Data Analysis ...... 18

1.4 Synopsis of Completed Research ...... 20

CHAPTER 2: Mapping Differential Cellular Protein Response of Mouse Alveolar

Epithelial Cells to Multi-Walled Carbon Nanotubes as a Function of Atomic Layer

Deposition Coating ...... 23

2.1 Introduction ...... 25

2.2 Methods ...... 28

2.3 Results ...... 40

2.4 Discussion...... 52

2.5 Conclusion ...... 58

CHAPTER 3: Proteomic Cellular Response Comparison of a 3D Lung Model Exposed to MWCNT Under Submerged versus Air-Liquid Interface Conditions ...... 60

3.1 Introduction ...... 61

3.2 Methods ...... 64

3.3 Results ...... 73

3.4 Discussion...... 81

3.5 Conclusion ...... 85

CHAPTER 4: Toxicoproteomic Analysis of Pulmonary Carbon Nanotube Exposure using LC-MS/MS ...... 87

4.1 Introduction ...... 88

4.2 Methods ...... 90

4.3 Results ...... 97

4.4 Discussion...... 105

4.5 Conclusion ...... 108

vii

CHAPTER 5: Multi-Walled Carbon Nanotubes as an Abundant Protein Depletion

Material for Proteomic Sample Digestion ...... 109

5.1 Introduction ...... 109

5.2 Methods ...... 112

5.3 Results ...... 120

5.4 Discussion...... 125

5.5 Conclusion ...... 128

CHAPTER 6: Conclusions ...... 129

6.1 General Conclusions ...... 129

6.2 Potential Applications ...... 131

APPENDICES ...... 134

REFERENCES ...... 195

viii

LIST OF TABLES

CHAPTER 2: Mapping Differential Cellular Protein Response of Mouse Alveolar Epithelial Cells to Multi-Walled Carbon Nanotubes as a Function of Atomic Layer Deposition Coating

Table 2.1: ANOVA results for regression model in Equation 2.1 with E10 dataset ...... 45

APPENDIX A: CHAPTER 2 SUPPLEMENTAL TABLES ...... 134

Table A1: Table detailing hydrodynamic diameter and zeta potential for MWCNT ..136

Table A2: LC methods for sample run on the orbitrap (method A), and triple quadrupole (method B) ...... 137

Table A3: Comparison of protein expression trends for the DDA mode orbitrap peak area results versus the SRM mode triple quadrupole results ...... 138

Table A4: List of significant two-sample t-test p values and log2 fold change for each MWCNT ‘high’ dose exposure compared to control...... 139

Table A5: List of significant two-sample t-test p values and log2 fold change for each MWCNT ‘low’ dose exposure compared to control ...... 151

Table A6: Data of pathway analysis used to generate heat maps of top 30 enriched pathways upon exposure to MWCNTs. The data are separated by increased and suppressed enrichment relative to control ...... 157

Table A7: Log2 fold change values for each heat map created for the following pathways: Nrf-2 mediated oxidative stress response, IL-1 signaling, Inhibition of angiogenesis by TSP1, mTOR signaling, eIF4/p70S6K signaling, and Oxidative phosphorylation...... 159

CHAPTER 3: Proteomic Cellular Response Comparison of a 3D Lung Model Exposed to MWCNT Under Submerged versus Air-Liquid Interface Conditions

APPENDIX B: CHAPTER 3 SUPPLEMENTAL TABLES ...... 166

Table B1: LC method gradients. Method A used to process cell samples, and Method B used to process media and wash samples ...... 167

Table B2: Regression model used to fit the media sample data set ...... 168

Table B3: Number of significant protein (p-value < 0.05) count for MWCNT versus control ...... 169

Table B4: Upstream mediator z-score for the apical and basolateral 96 hour media proteome. M-7 MWCNT exposure versus BSA control data was used to general enrichment analysis ...... 170

Table B5: Upstream mediator z-score for the apical and basolateral 96 hour M-7 MWCNT exposed proteome. ALI versus submerged exposure data was used to general enrichment analysis ...... 171

Table B6: Log2 Fold change for proteins associated with upstream mediators listed in Figure 5. M-7 MWCNT exposure / BSA control data was used to calculate fold change...... 172

Table B7: Log2 Fold change for proteins associated with upstream mediators listed in Figure 6. ALI / Submerged M-7 exposure data was used to calculate fold change ...... 177

CHAPTER 4: Toxicoproteomic Analysis of Pulmonary Carbon Nanotube Exposure using LC-MS/MS

APPENDIX C: CHAPTER 4 SUPPLEMENTAL TABLES ...... 181

Table C1: Method for MCX sample clean up ...... 183

Table C2: Gradient method used for LC ...... 184

Table C3: Peak area analysis t-test p-values and protein regulation. The regulation of the exposed groups with control are reported by comparison to control. The regulation of the exposed groups by each other are reported by comparison to U-MWCNT ...... 185

Table C4: Protein count data with Benjamini corrected p-value for GO enrichment analysis terms ...... 189

CHAPTER 5: Multi-Walled Carbon Nanotubes as an Abundant Protein Depletion Material for Proteomic Sample Digestion

Table 5.1: Method overview summarizing the starting amount of protein and experimental conditions for each experiment ...... 116

APPENDIX D: CHAPTER 5 SUPPLEMENTAL TABLES ...... 191

Table D1: Top 10 protein PSM for control and corona reported for each experiment ...... 191

APPENDIX E: CHAPTER 6 SUPPLEMENTAL TABLES ...... 194

Table E1: List of candidate proteins that could be used for the development of a targeted mass spectrometry assay ...... 194

LIST OF FIGURES

CHAPTER 1: Introduction

Figure 1.1: Common CNT chiral indices, including: zigzag (n1 = 0, or n2 = 0), armchair (n1 = n2), and chiral (n1 ≠ 0, n1 ≠ n2) ...... 2

Figure 1.2: Schematic illustration of the planar graphene sp2 hybridized orbital configuration ...... 2

Figure 1.3: Differences in band gap energy depicted for metal, semi-conductor, and insulator. Metals more better conductors due their small band gap energy ...... 3

Figure 1.4: Schematic representing possible airway CNT clearance. A) Upper airway clearance through upward mucus movement, and B) Lower airway clearance via macrophage engulfment ...... 7

Figure 1.5: Possible mechanisms for the induction of fibrosis; including injury to fibroblasts, epithelial cells, and macrophages ...... 10

Figure 1.6: Overview for how to use proteomics to answer biological questions ...... 12

Figure 1.7: Mass spectrometry instrument overview highlighting sample separation, ionization, and mass spectrometer. An orbitrap mass analyzer is depicted in this schematic ...... 14

Figure 1.8: Summary of tandem mass spectrometry peptide ionization representing MS1 and MS2 scans ...... 17

Figure 1.9: General schematic overview of proteomic workflow for label-free data analysis ...... 19

CHAPTER 2: Mapping Differential Cellular Protein Response of Mouse Alveolar Epithelial Cells to Multi-Walled Carbon Nanotubes as a Function of Atomic Layer Deposition Coating

Figure 2.1: Transmission electron microscopy (TEM) of uncoated MWCNTs and ALD- coated MWCNTs. (A) Uncoated (U)-MWCNTs. (B) Al2O3-coated (A)-MWCNTs. (C) ZnO-coated (Z)-MWCNTs. 50 cycles of ALD were applied to A-MWCNTs and Z- MWCNTs ...... 30

Figure 2.2: (A) Atomic layer deposition with Al2O3 or ZnO applied to uncoated (U)- MWCNT to derive A-MWCNT or Z-MWCNT, respectively. (B) E10- mouse alveolar epithelial cell exposure via Latin square block design. For treatment groups a-g cells were exposed to PBS (control) or MWCNT (U, A, or Z) at the indicated doses (2.5, 5, 100)

xii

expressed in μg/ml. (C) Proteomic sample preparation through the following steps: SDC detergent treatment to isolate proteins, peptide generation from trypsin digestion, and sample cleanup with cation exchange MCX columns. All samples were processed on the following ThermoScientific LC-MS/MS instruments: Q-Exactive Plus (Discovery) and Quantiva (Targeted) ...... 33

Figure 2.3: Log10 LDH dose response of E10 cells to the following exposures: U-MWCNT, A-MWCNT, and Z-MWCNT ...... 41

Figure 2.4: TEM images of E10 cell exposure. (A) 5 µg/mL U-MWCNT exposure, (B) 5 µg/mL A-MWCNT, and (C) 1 µg/mL Z-MWCNT. Arrows indicate MWCNTs within the cytoplasm of E10 cells ...... 43

Figure 2.5: Venn diagram of significant proteins by expression. (A) All significant proteins with increased expression compared to control. Common proteins include: hemoglobin subunit beta, hemoglobin subunit gamma, and proteolipid protein 2. (B) All significant proteins decreased expression compared to control. The common protein for decreased expression is fatty acid synthase ...... 44

Figure 2.6: Heat map of top 30 enriched pathways upon exposure to MWCNTs. Increasing significance shown from blue to red (red being the most significant). (A) Significant pathways corresponding to proteins with up-regulated expression upon exposure to control, and (B) Significant pathways corresponding to proteins with down- regulated expression upon exposure to control. A-MWCNT (aluminum oxide MWCNT compared to control), Z-MWCNT (zinc oxide MWCNT compared to control), and U- MWCNT (uncoated MWCNT compared to control) ...... 47

Figure 2.7: Heat maps of log2 fold change for proteins of MWCNT exposed compared to control for the following pathways of interest: (A) Nrf-2 mediated oxidative stress response, (B) IL-1 signaling, (C) Inhibition of angiogenesis by TSP1, (D) mTOR signaling*, (E) eIF4/p70S6K signaling*, and (F) Oxidative phosphorylation. Log2 fold changes in protein abundance can be read as: increased expression compared to the control (Red), and decreased expression compared to the control (Blue). *Outliers removed for higher resolution of heat map scaling. Table A7 provides full list of proteins and log2 fold change values ...... 49

APPENDIX A: CHAPTER 2 SUPPLEMENTAL FIGURES ...... 134

Figure A1: Protein Log2 peak area fold change plotted for DDA collected data versus SRM collected data. Pearson correlation coefficient r = 0.9635 ...... 134

Figure A2: Boxplot for A) un-normalized and un-filtered peak area, and B) normalized and filtered peak area for each sample ...... 135

xiii

CHAPTER 3: Proteomic Cellular Response Comparison of a 3D Lung Model Exposed to MWCNT Under Submerged versus Air-Liquid Interface Conditions

Figure 3.1: A) General schematic of the small airway tri-culture cell model of small, including the following human cell types: THP1 Macrophages, A549 alveolar epithelial cells, and MRC5 fibroblasts. B) Experimental overview of exposure method and sample collection. Samples were collected as apical wash, media, and cell after a 96 hour time point ...... 74

Figure 3.2: TEM micrographs of M-7 MWCNTs in suspension (A, B) and deposited after the nebulization (C, D) ...... 75

Figure 3.3: Cell morphology comparing the air-liquid interface to the submerged M-7 MWCNT exposure. Apical, basolateral, and cross-section images were collected at the 24 and 96 hour time point ...... 76

Figure 3.4: Cytotoxicity quantification through an LDH assay ...... 77

Figure 3.5: Heat map of top 25 enriched upstream mediators for the apical and basolateral 96 hour media proteome. M-7 MWCNT exposure versus control data was used to general enrichment analysis ...... 80

Figure 3.6: Heat map of top 25 enriched upstream mediators for the apical and basolateral 96 hour M-7 exposed proteome. ALI versus submerged exposure data was used to general enrichment analysis ...... 81

APPENDIX B: CHAPTER 3 SUPPLEMENTAL FIGURES ...... 166

Figure B1: Bar graphs depicting average log10 complement component 3 protein response of MWCNT 96 hour exposure for: A) Submerged versus ALI basolateral samples, and B) Submerged versus ALI apical samples ...... 166

CHAPTER 4: Toxicoproteomic Analysis of Pulmonary Carbon Nanotube Exposure using LC-MS/MS

Figure 4.1: Transmission electron photomicrographs of (A) non-functionalized MWCNTs and (B) ALD Al2O3-functionalized MWCNTs ...... 91

Figure 4.2: General overview of sample preparation and data analysis. (A) Mouse exposure to MWCNTs, (B) lung lavage extraction, (C) tryptic digestion of lung lavage and SPE sample clean up, (D) LC-MS/MS, and (E) Label-free protein quantification by: (a) spectral counting, and (b) verification by MS1 peak area analysis of unique peptides...... 98

xiv

Figure 4.3: Volcano plots of log2 spectral count fold change versus –log10 pvalue calculated by Fisher’s exact test for: (A) U-MWCNT/control, (B) A-MWCNT/control, and (C) U-MWCNT/A-MWCNT. Proteins plotted as a log2 fold change of 3 and -3 represent protein detection in exposed groups, but not control, and vice versa (respectively). Positive fold change in spectral count data represent up-regulation in exposed groups versus the control. Shaded areas highlight significance of p < 0.05. Abundant proteins (i.e., human serum albumin) were included in calculations of the Fisher’s exact test, but not shown in plots. α = Pulmonary surfactant-associated protein- B. β = Myeloperoxidase. γ = Lactotransferrin...... 100

Figure 4.4: Peak area analysis was conducted on proteins that were discovered to be significant using the Fisher’s Exact test on the spectral count data. (A) Chromatogram overlay used to assess changes in peak area by group. (B) Peak area was tested for significance across every group using the two-sample t-test: Control versus U-MWCNT (p = 7.02E-08), Control versus A-MWCNT (p = 3.96E-05), and U-MWCNT versus A- MWCNT = 0.679). *Control peak area = 0, but imputed as a value of 10 to allow for log adjustment. Peaks were identified using a combination of MS2 identification (blue lines), mass measurement accuracy (< 3 ppm), dot products of theoretical isotope abundance, and retention time reproducibility ...... 102

Figure 4.5: GO analysis terms for (A) Biological Process, and (B) Molecular Function for significant proteins by peak area analysis both of the exposed groups versus the control. Significance (p < 0.05) was calculated by a modified Fisher’s exact test (EASE score) and adjusted by the Benjamini correction and reported as the –Log10 Benjamini p-value. Significant GO terms are illustrated to the right of the dashed line ...... 104

APPENDIX C: CHAPTER 4 SUPPLEMENTAL FIGURES ...... 181

Figure C1: Volcano plots for in-vitro experiment of log2 spectral count fold change versus –log10 pvalue calculated by Fisher’s exact test for: (A) U-MWCNT/control, (B) A- MWCNT/control, and (C) U-MWCNT/A-MWCNT. Proteins plotted as a log2 fold change of 3 and -3 represent protein detection in exposed groups, but not control, and vice versa (respectively). Positive fold change values represent up-regulation in exposed groups. Shaded areas highlight significance of p < 0.05 ...... 181

CHAPTER 5: Multi-Walled Carbon Nanotubes as an Abundant Protein Depletion Material for Proteomic Sample Digestion

Figure 5.1: Experimental workflow. A) Protein corona formation and isolation, and B) Protein digest followed by LC-MS/MS data collection and analysis ...... 115

Figure 5.2: Protein quantitation by BCA assay A) Standard curve, and B) Protein concentration measured in protein corona washes ...... 121

Figure 5.3: Number of protein groups identified in the protein corona versus the control lavage for each LC-MS/MS experiment ...... 122

Figure 5.4: Comparison of peptide gravy score for the control versus the protein corona for each LC-MS/MS experiment. Positive gravy score represents more hydrophobic peptides, and a negative gravy score represent a more hydrophilic peptide ...... 123

Figure 5.5: Ratio of trypsin digest peptide missed cleavages by total peptide count for the protein corona versus control peptides ...... 124

Figure 5.6: Top 10 most abundant proteins found in the control versus the protein corona ...... 125

CHAPTER 6: Conclusion

Figure 6.1: Schematic heat map representing each CNT exposure method tested by proteomics and their associated enriched signaling pathways. Increasing color intensity correlates to increasing significance in pathway enrichment ...... 130

xvi

LIST OF EQUATIONS

CHAPTER 2: Mapping Differential Cellular Protein Response of Mouse Alveolar Epithelial Cells to Multi-Walled Carbon Nanotubes as a Function of Atomic Layer Deposition Coating

Equation 2.1: Peak Area = Intercept + Dose + Coating ...... 40

CHAPTER 3: Proteomic Cellular Response Comparison of a 3D Lung Model Exposed to MWCNT Under Submerged versus Air-Liquid Interface Conditions

Equation 3.1: Protein Intensity = Intercept + Exposure Method + Exposure ...... 78

CHAPTER 5: Multi-Walled Carbon Nanotubes as an Abundant Protein Depletion Material for Proteomic Sample Digestion

Equation 5.1: MWCNT bound protein (µg) = Starting protein (µg) – Σ wash1-n (µg)

...... 116

xvii

LIST OF SYMBOLS AND ABBREVIATIONS

4',6-diamidino-2-phenylindole - DAPI Adverse outcome pathway - AOP Air-liquid interface - ALI Aluminum oxide - Al2O3 Aluminum oxide MWCNT - A-MWCNT American type culture collection - ATCC Ammonium hydroxide - NH4OH Atomic layer deposition - ALD Bicinchoninic acid - BCA Bovine serum albumin - BSA Bronchoalveolar lavage fluid - BALF Carbon nanotubes - CNTs C-C motif chemokine ligand 5 - CCL5 Chemical vapor deposition - CVD Collision induced dissociation - CID Complement component protein - C3, C4b, and C9 Deionized - DI Dependent acquisition - DDA Diethylzinc - DEZ Dithiothreitol - DTT Dulbecco’s modified eagle’s medium - DMEM Electrospray ionization - ESI Enzyme-linked immunosorbent assay - ELISA Ethylenedinitrilotetraacetic acid - EDTA Extracellular signal–regulated kinase ½ - ERK1/2 Fast atom bombardment - FAB Fetal bovine serum - FBS Fibroblast growth factor - FGF Filter aided sample preparation - FASP Formic acid - FA Full width half max - FWHM Fungizone antimycotic - Fz Gene ontology - GO Feme oxygenase 1 - HO-1 Hemoglobin - Hb Heparan sulfate proteoglycans - HSPG Highest occupied molecular orbital - HOMO High-pressure liquid chromatography - HPLC Hydrochloric acid - HCl Ingenuity pathway analysis - IPA Interleukin - IL Interleukin 1β - IL1β Interleukin 6 - IL6

xviii

Intratracheal instillation - IT Intravenous - IV Iodoacetamide - IAM Lactate dehydrogenase - LDH Lactotransferrin - LTF Liquid chromatography tandem mass spectrometry - LC-MS/MS Lowest unoccupied molecular orbital - LUMO Mammalian targeting of rapamycin - mTOR Mass spectrometry - MS Mass-to-charge ratio - m/z Minimum essential medium - MEM Mitogen-activated protein kinase 14 - MAPK14 Mitsui-7 - M-7 Mixed-mode reversed-phase/strong cation exchange - MCX Molecular weight cut off - MWCO Mothers against decapentaplegic homolog 3 - SMAD3 Multi-walled CNT - MWCNT Myeloperoxidase - MPO Nanoparticles - NPs National institute for occupational safety and health - NIOSH National institute of environmental health sciences - NIEHS Neutrophil gelatinase-associated lipocalin - NGAL Nuclear factor erythroid 2-related factor 2 - Nrf2 Oropharyngeal aspiration - OPA Oxidative phosphorylation – OXPHOS Paraformaldehyde - PFA Pathogen-associated molecular patterns - PAMPs Phorbol 12-myristate 13-acetate - PMA Phosphate-buffered saline - PBS Phosphatidylinositol-3-kinases - PI3K Platelet-derived growth factor - PDGF Post translational modifications - PTMs Principal component analysis - PCA Protein-protein interaction - PPI Pulmonary surfactant protein - SP Quantitative structure activity relationship - QSAR Quarz crystal microbalance - QCM Reactive oxygen species - ROS Recommended exposure limit - REL Roswell Park Memorial Institute medium - RPMI Selected reaction monitoring - SRM Single-walled CNT - SWCNT Sodium chloride - NaCl Sodium deoxycholate - SDC Tandem mass spectrometry - MS/MS

xix

Thrombospondin - TSP Transforming growth factor beta - TGFβ Transmission electron microscopy - TEM Trimethylaluminum - TMA Trizma® hydrochloride - Tris-HCl Tumor necrosis factor alpha - TNFα Two-dimensional polyacrylamide gel electrophoresis - 2D PAGE Tyrosin-protein kinase fyn - FYN Uncoated MWCNT - U-MWCNT Vascular endothelial growth factor - VEGF Extracellular matrix - ECM Zinc oxide MWCNT - Z-MWCNT

CHAPTER 1

Introduction

1.1 Introduction to Carbon Nanotubes

1.1.1 Carbon Nanotube Characterization

Carbon nanotubes (CNTs) have been commonly described as seamless sheets of graphene that have been rolled to make either one layer, single-walled CNT (SWCNT), or more than one layer, multi-walled CNT (MWCNT) [1]. The diameter of CNTs generally range from 1 – 20 nm, and the lengths can range from 100 nm to several centimeters, owing to their

‘needle-like’ characterization. CNTs contain unique physical properties depending on the direction in which the graphene sheet is wrapped. The direction of the wrapping, also known as chirality, is characterized by a pair of indices (n1, n2) which represent the number of unit vectors of two directions on the graphene lattice. The most common types of CNT chiral indices include: zigzag (n1 = 0, or n2 = 0), armchair (n1 = n2), and chiral (n1 ≠ 0, n1 ≠ n2) (Figure

1.1) [2]. Additionally, each of these types of CNTs possess unique physical features that are determined by their individual chirality. Armchair CNTs are metallic, whereas zigzag and chiral CNTs are generally on a spectrum of metallic or semiconducting, depending on the degree of their individual chirality [3].

(n,0) zigzag

(n,n) armchair Armchair Zigzag Chiral

Figure 1.1: Common CNT chiral indices, including: zigzag (n1 = 0, or n2 = 0), armchair (n1 = n2), and chiral (n1 ≠ 0, n1 ≠ n2) [4].

1.1.2 Electronic Properties of Carbon Nanotubes

In order to better understand how differential CNT chirality harnesses unique metallic properties, the electronics of the original graphene structure must be considered. In a planar sheet of graphene, each hexagonal unit contains six carbons with four valence electrons each where three of the electrons participate in σ-bonding with its neighbor in the sp2 configuration, and the fourth valence electron occupies a π orbital (Figure 1.2). This configuration leaves an unfilled π orbital which is perpendicular to the graphene sheet, thus allowing a delocalized π network across the CNT [5].

σ-orbital

π-orbital

Figure 1.2: Schematic illustration of the planar graphene sp2 hybridized orbital configuration.

The delocalized electrons in the π network create the opportunity for flow of charge, thus creating metal-like properties. Electronic properties of metals can be explained through band gap (or energy gap), where in solid-state physics the band gap represents the energy difference between the valence and conduction band. Metals contain the greatest overlap, thus allowing for the ease of electronic flow, following by an increasing band gap for semi- conductors and then insulators (Figure 1.3). While band gap is generally used to describe solid-state metal chemistry, it is highly analogous to the highest occupied molecular orbital

(HOMO) / lowest unoccupied molecular orbital (LUMO) gap used to describe non-solid-state chemistry, thus helping to explain the behavior of electron flow in carbon nanotubes.

Overlap Conduction Band

Band Gap

Valence energy Increasing Band Metal Semi- Insulator conductor

Figure 1.3: Differences in band gap energy depicted for metal, semi-conductor, and insulator.

Metals are better conductors due their small band gap energy.

1.1.3 Carbon Nanotube Synthesis

CNT synthesis is primarily achieved through gas phase processes, including: chemical vapor deposition (CVD) [6, 7], laser ablation technique [8-10], and carbon arc discharge [11-

13]. Early synthesis of CNTs were primarily conducted through high temperature techniques like laser ablation (>1,200 °C) and arc-discharge (>1,700 °C); however, CVD has become the prominent technique because synthesis can be successfully achieved at lower

temperatures (< 800 °C). Lower temperature synthesis offers significant advantage because the CNT length, diameter, purity, and orientation can be more precisely controlled compared to synthesis at higher temperatures. CNT synthesis via CVD starts with a substrate containing a layer of metal catalyst, usually nickel, cobalt, iron, or some combination of those metals [14].

Next, CNT growth starts to occur once the substrate is heated to 700 °C in a reactor containing a blend of non-carbon gases (ammonia, hydrogen, or nitrogen), and carbon based gases

(acetylene, ethanol, ethylene, or methane). The CNT growth occurring within the reactor are a product of the carbon breakdown of the carbon based gases at the surface of the catalyst on the substrate [15]. Furthermore, surface modification of CNTs has been of significant interest to researchers due to the ability of functionalized CNTs to harness unique physical properties.

Surface modification has been traditionally achieved through both covalent and non-covalent modifications. Recent advances in surface modification have been achieved through atomic layer deposition (ALD). ALD is a powerful method used to apply conformal nanoscale coatings on various types of CNTs. The ALD reaction is self-limiting, thus allowing for precise control over thin-film thickness on MWCNTs [16-18]. One of the greatest benefits of

ALD CNT synthesis is the surface modifications can be organic, inorganic, or a hybrid of organic and inorganic molecules, thus allowing for broader flexibility in CNT synthesis and application [19].

1.1.4 Applications of Carbon Nanotubes

Since the discovery of CNTs in 1991, a global market has emerged to take advantage of their unique physical properties for countless applications. MWCNTs have been reported to have 100 GPa tensile strength, making them > 100 times stronger than steel at a fraction of

the weight, thus allowing them to be used for applications in construction [20]. Additionally,

MWCNT are metallic in nature and are capable of carrying currents up to 109 A cm-2, making them viable options for electronics [21]. MWCNTs and SWCNTs both have extremely high thermal conductivity of 3000 - 3500 W m-1 K-1 at room temperature (respectively), exceeding the thermal conductivity of a diamond [22]. Some of the most common applications of CNT include: electrically conductive fillers, polymers, and resins, coatings, and conducting films.

Ultimately, these applications allow for CNTs to be commercialized for use in composite materials, microelectronics, energy storage, and various applications in biotechnology [23].

1.2 Carbon Nanotubes as a Potential Emerging Toxicant

1.2.1 CNT Toxicity Overview

While CNTs possess highly desirable physical properties, they also have at least four properties that are linked to pathogenicity of particles: fiber-like shape, biopersistence, high surface area per unit mass, and trace metal catalysts from the manufacturing process such as nickel, cobalt, and iron [24-27]. Some CNTs bare physical properties similar to asbestos, thus creating a great concern for exposure due to the development of fibrosis and mesothelioma following asbestos exposure. The most likely human exposure to CNTs is thought to occur through aerosolization during production in an occupational setting [28-30]. Due to the aerosolization of CNTs, as well as their deleterious properties to human health, there is great concern that inhalation of these materials will cause chronic lung disease. One such chronic lung disease associated with exposure to fiber-like materials is pulmonary fibrosis [31].

Pulmonary fibrosis is characterized by scaring of the lung with an average human survival of

6 to 24 months after symptom onset (i.e. shortness of breath, increased cough, and worsening pulmonary function on tests) [32]. The prevalence of pulmonary fibrosis in the United States from 1996-2000 was estimated to be approximately 42 in 100,000 people [33]. Fibrosis is thought to occur from injury that results in a runaway wound healing response. More specifically, inflammatory mediators signal to fibroblasts to proliferate and transition to myofibroblasts, ultimately leading to excess deposition of collagen and extracellular matrix that are characteristic of scar tissue [34]. In light of the potential for CNTs to cause fibrosis, the National Institute for Occupational Safety and Health (NIOSH) has a CNT recommended exposure limit (REL) of approximately 1 µg/m3 respirable carbon (8-hour time-weighted average), which is based on estimated levels in occupational settings and animal dose-response studies [35]. Whether CNTs will cause adverse health effects in humans is unknown, since widespread exposures have not yet occurred and clinical signs of lung disease will take years to manifest. Thus, there is now a significant effort in the nanotoxicology community to try and assess potential risks of emerging nanomaterials through both in-vivo and in-vitro assays.

1.2.2 In-vivo CNT Toxicity

CNTs can be administered during in-vivo studies through various methods, including: oral, intravenous (IV) injections, intratracheal instillation (IT), oropharyngeal aspiration

(OPA), inhalation, transdermal, subcutaneous injection, and intraperitoneally [24]. While

CNTs can enter through various routes of exposure, the greatest concern for occupational exposure is through inhalation. Toxicity studies in rodents have primarily focused on pulmonary exposure through OPA, IT, and inhalation due to the resulting increased inflammatory response compared to oral or dermal exposure. In rodents, inhaled MWCNTs

deposit and interact directly with the respiratory epithelium of the lower lung. However, studies have shown variable deposition depending on exposure technique [36]. The ability of the lungs to clear CNTs is highly dependent on the size and shape of the material, which will also lead to variability in deposition within the lungs. If the CNTs deposit in the conducting airway, they are able to be rapidly cleared (Figure 1.4.A). However, if they deposit in the alveolar region, they can damage the epithelial lining to reach the interstitium, leading to a fibrotic response [35, 37]. One of the ways the body tries to clear CNT materials in the lower airway is by macrophage engulfment (Figure 1.4.B). However, due to the long fibrous properties of some CNTs, the macrophages cannot fully engulf the material, leading to the macrophage becoming ‘frustrated’, which initiates pro-inflammatory cytokines production as well as innate immune response [38, 39].

Figure 1.4: Schematic representing possible airway CNT clearance. A) Upper airway clearance through upward mucus movement, and B) Lower airway clearance via macrophage engulfment.

In general, in-vivo studies have shown evidence of lung inflammation, oxidative stress, and fibrosis upon OPA or IT exposure to MWCNTs [25, 40, 41]. Several endpoints are generally measured as markers for inflammation, immune response, as well as fibrotic response. Some of the most common protein endpoints measured for in-vivo CNT exposure include: chemokines, interleukins (IL-1β, IL-6, IL-8), platelet-derived growth factor (PDGF), transforming growth factor beta (TGF-β) and tumor necrosis factor alpha (TNF-α). Additional data are usually collected through histopathology where researchers report: neutrophil count, collagen deposition, granuloma formation, and DNA synthesis. Most in-vivo studies report a combination of histopathology data in addition to levels of select proteins. For example, an early study using MWCNT IT exposure in rats showed increase inflammatory responses 15 days after exposure marked by increased levels of lactate dehydrogenase (LDH) and TNF-α, followed by increased levels of collagen and granuloma formation 60 days post exposure [40].

Additional studies have shown fibrotic effects of low dose CNT exposure in rodents that develop as early as 28 days post exposure, by measuring neutrophil, lymphocyte, and macrophage accumulation, as well as elevation in proinflammatory cytokines (TNF-α, IL-1β) and a fibrogenic mediator (TGF-β) [42-46, 37, 47, 48].

Ultimately, in-vivo studies possess some advantage because they can provide information on CNT distribution, metabolism, cell-cell interaction (i.e. tumor or scar tissue formation), and researchers are more easily able to obtain subchronic and chronic exposure data. However, in-vivo studies also have limitations, including: cost, interspecies data extrapolation, and require the sacrifice of animals, thus limiting the use of in-vivo studies for much needed high-throughput screening [49].

1.2.3 In-vitro CNT Toxicity

Advances in-vitro toxicity testing are rapidly being developed in response to pressure to eliminate animal testing [50], reduce cost [51], as well as generate human models to alleviate concern about inter-species data extrapolation [52]. Similarly to in-vivo studies, in-vitro CNT testing has been focused on pulmonary exposure due to the greater potential of inhalation in an occupational environment. Some of the most common lung cell types used for CNT exposures include: bronchial or alveolar epithelial cells, endothelial cells, monocytes, and fibroblasts. Cell lines are typically derived from either normal or cancerous tissue from mouse or human, and then immortalized for the ease of cell culture maintenance. Traditional in-vitro

CNT exposure have been conducted using submerged mono-culture lung based cell models

[49]. Results from mono-culture MWCNT exposure studies, in general, have shown an increase in reactive oxygen species (ROS) in a variety of different cell lines [53-57, 19].

Increases in ROS from fibroblasts, macrophages, and epithelial cells can ultimately lead to increased expression levels of pro-inflammatory mediators, such as several interleukins (IL-

1β, IL-6, IL-8) as well as TGFβ and TNFα (Figure 1.5) [58, 53, 59-61].

In addition to mono-culture studies, significant advances have been achieved in respiratory tissue engineering that better represent the human pulmonary system, including the development of an epithelial tissue barrier applied to 3D cell culture models that contain various cells types and supports [62-65]. Utilizing the advances in bioengineering, a 3D model of the human alveolar region was constructed and tested at an air-liquid interface

(ALI) with an aerosolized MWCNT exposure [66]. Although the aerosolized MWCNT

ALI exposure method more closely mimics true inhalation, most small airway cell culture

exposures are conducted under more traditional submerged conditions [66]. Exposure method is a critical variable to consider when executing exposures to MWCNT due to their physical properties that lead to variation in dosimetry [67, 68]. However, there is a need to investigate potential differences in exposure method on emerging 3D cell culture models.

Thus, cellular response should be evaluated for the 3D cell culture exposures to MWCNT by

ALI and by submerged exposure methods in order to try and adopt a consistent and physiologically relevant in-vitro exposure method that could be uniformly adopted by the nanotoxicology community. Ultimately, a thorough investigation of global cellular response can be achieved by a systems biology, or ‘omics’ approach.

CNT

Fibroblasts Epithelial Cells Macrophages

ROS ROS Inflammasome MAPK ROS

p-38 MAPK Inflammasome IL-1β IL-8 IL-6

TGF-β IL-1β PDGF TGF-β Smad ERK 1/2 PDGF TGF-β

FIBROSIS

Collagen Myofibroblast production differentiation Proliferation

Figure 1.5: Possible mechanisms for the induction of fibrosis; including injury to fibroblasts, epithelial cells, and macrophages [58].

1.3 Proteomic Toxicity Testing

1.3.1 Proteomics

The field of proteomics, large scale studies of proteins, rapidly emerged out of the significant advances achieved in DNA sequencing during the 1990s [69]. Automated DNA sequencing and data collection set the stage for the sequencing of whole genomes [70-72], which ultimately gave rise to ‘omic’ experimentation, including: genomics, transcriptomics, and proteomics (global study of the genome, transcriptome, and proteome, respectively) [73-

75]. The primary difference between the ‘omics’ data is that genomics provides sequence information of genes, whereas transcriptomics and proteomics represent the gene products, or gene expression. While the transcriptome and the proteome both serve to investigate gene expression, RNA transcript levels are not necessarily a good proxy for protein abundance [76].

Some of the most notable differences between RNA and proteins include: 1) reduced half-life of mRNA (i.e. protein is on average 5 times more stable than mRNA) [77, 78], 2) codon-bias

(i.e. highly expressed genes have a bias toward more frequent codons which impacts rate of translational elongation) [79], and 3) variation in synthesis and degradation of proteins (i.e. differences in mRNA and protein abundance turnover) [78]. Despite these differences, transcriptomics has been a preferred method for many scientists for the interrogation of gene expression due to ease of sample preparation and data collection. Transcriptomic data is generally collected using gene expression arrays or robust sequencing instruments, while proteomic experimentation generally entails the use of a mass spectrometer, thus deterring scientists that are not familiar with mass spectrometry.

Proteomics ultimately serves to elucidate molecular function that cannot be explained by genomics or transcriptomics, including: protein-protein interaction (PPI), post translational modifications (PTMs), changes in protein abundance, and the presence protein isoforms [80,

81]. The advantages of proteomics provide powerful implication for researchers to be able to assess global changes of protein abundance and/or modification in response to changes in a cellular state (Figure 1.6). Over 10,000 publications have emerged from application of proteomics in biological sciences; however, limited studies have been published reporting biochemical toxicities that lead to an undesirable phenotype [82]. Thus, there is a great potential to explore proteomic application is exposure science.

Biological Question DAVID IPA

STRING KEGG Cells/T Functional issue Analysis Cytoscape

Proteins Protein Assembly

Peptides Peptide ID

LC-MS/MS Sequence Data HPLC

ESI Orbitrap MS

Figure 1.6: General overview for how to use proteomics to answer biological questions [83].

1.3.2 Mass Spectrometry

Mass spectrometry, in its simplest form, is an analytical technique that uses a mass spectrometer to measure ions based on a mass-to-charge ratio (m/z). The use of mass spectrometry for protein identification is classically conducted by either ‘top-down’ or

‘bottom-up’ methodology. Top-down proteomics measures intact proteins, whereas bottom- up proteomics identifies proteins through the analysis of peptide fragments [84]. While both methods are still used in mass spectrometry based proteomic research, bottom-up proteomics is more commonly used due to numerous factors including the difficulty of solubilizing intact proteins for liquid chromatography tandem mass spectrometry (LC-MS/MS) compatible solutions. Furthermore, the use of mass spectrometry for proteomics is complex and requires the combination of: 1) separation, 2) ionization, and 3) mass spectrometer (Figure 1.7) [81].

Significant advances have been made in each of these areas and will be briefly reviewed herein, followed by a summary of advantages of mass spectrometry based proteomic applications in molecular biology.

Separation Ionization

Column ESI Ion guide

HPLC Quadrupole

C-Trap Mass Spectrometer HCD cell Orbitrap

Figure 1.7: Mass spectrometry instrument overview highlighting sample separation, ionization, and mass spectrometer. An orbitrap mass analyzer is depicted in this schematic.

Separation

The use of mass spectrometry for proteomic analysis is highly dependent on separation of complex biological mixtures. In order to achieve identification of peptides in bottom-up proteomics, peptide separation must be sufficient enough to allow for sampling by the instrument on an LC time scale. The most common traditional method used for protein separation involves gel-based separation method through the use of two-dimensional polyacrylamide gel electrophoresis (2D PAGE) [85, 86]. Samples run through 2D PAGE would be excised, digested, and analyzed by a mass spectrometer. The 2D gel separation method was more commonly used before advances in high-pressure liquid chromatography

(HPLC) became available, which combined with other developments ultimately gave rise to coupling of the liquid chromatography / mass spectrometry (LC/MS) set up commonly used today. One of the greatest advantages of the LC/MS coupling was the ability to achieve

continuous sample separation through the HPLC which is directly attached to a mass spectrometer, thus producing continuous protein/peptide identification [87]. Significant advances in instrumentation have been made since the mid-1990s in order to couple HPLCs with a wide verity of mass analyzers, overall providing greater sample separation and detection.

Ionization

The evolution of ionization for mass spectrometers started in 1918 with the development of the first electron impact source, and was improved upon through the 1960s for use in measuring small volatile hydrocarbons [88]. However, the mass spectrometry analysis of proteins and peptides was not attainable until the early 1980s due to the fact that their low volatility and large size was not compatible with the pre-existing technology. In 1981 a significant development was made in creation of fast atom bombardment (FAB) which provided a ‘soft’ ionization procedure that was capable of ionization of large peptide species without extensive fragmentation [89]. Following the advent of FAB came a discovery that revolutionized proteomics called electrospray ionization (ESI), which was developed by Nobel laureate John B. Fenn [90]. In ESI analyte in solution is pumped through a tapered capillary in which a high voltage is applied to a liquid junction. A Taylor cone is formed and charge analyte containing droplets are emitted from this cone. The mechanism of ESI is still under active investigation but it is believed that the electric field on the charged droplets surface is sufficient to “lift” ions from solution into the gas phase. [91].

Mass spectrometer

Mass spectrometers manufactured in the 21st century normally contain the following parts: ion source (details above), optics, mass analyzer, and data processing electronics.

Optics are used to guide ions to the mass analyzer where ions are separated based on mass-to- charge ratio. Several mass analyzers are commercially available, including the ion trap,

Orbitrap, ion cyclotron resonance, quadrupoles, and time-of-flight [92-97]. While each mass analyzer provide their own merit, a hybrid Orbitrap-quadrupole technology (Thermo

ScientificTM Q-Exactive PlusTM) was used for this research and will be the focus of discussion herein.

Orbitrap technology, pioneered by Alexander Makarov in 2000 [98], revolutionized the field of mass spectrometry by achieving high mass measurement accuracy and high resolving power without the need for expensive superconducting magnets [99, 100]. The orbitrap technology works by trapping ions in an orbital of electrostatic fields. The ions oscillate in an axial direction around a central electrode where a fast Fourier transform algorithm is used to convert time-domain signal into frequency which can then be converted to m/z based on calibration equations (i.e. ions with different m/z will have different oscillatory frequency) [101]. In the Q-Exactive Plus, ions are guided through a series of lenses that focus the ions beam and neutral species are removed before passing through a quadrupole mass filter and entering the Orbitrap [102]. Furthermore, tandem mass spectrometry (MS/MS) can be achieved using the Q-Exactive Plus by sending ions through two MS scans. MS1 scan passes all ions from the quadrupole through to the orbitrap, followed by MS2 where only a preselected ion from the MS1 scan will be allowed to pass through the quadrupole and undergo fragmentation into product ions before entering the orbitrap (Figure 1.8). The number of MS2

scans can be chosen by the researcher allowing for flexibility in data collection. Ultimately,

Orbitrap technology is one of the most advantageous mass analyzers to use for global proteomic investigation because it can identify 1000s of proteins with high mass accuracy in a small amount of time [102].

Figure 1.8: Summary of tandem mass spectrometry peptide ionization representing MS1 and

MS2 scans.

Applications

Overall, mass spectrometry based proteomics is a powerful ‘omics’ method used to investigate global changes on the protein level in limitless applications [103]. The current instruments used to conduct global proteomics, namely liquid LC–MS/MS, are a prevailing analytical tool used to investigate the proteome due to their high sensitivity and molecular specificity [102]. Traditional investigation of protein endpoints for an exposure study would generally involve the use of antibodies based methods like Western Blot and enzyme-linked immunosorbent assay (ELISA). While these methods are very commonly used in molecular biology, they are both limited by availability of antibodies, which eliminates the assessment of novel mediators [104]. Thus, one of the greatest advantages of LC-MS/MS based proteomics

over traditional methods is to be able to accurately characterize and quantify 1000s of proteins in a cost and time effective manner.

1.3.3 Data Analysis

Considering the ability for LC-MS/MS based proteomics to identify 1000s of proteins in one sample run, a cumbersome amount of data is generated and needs to be carefully curated for proper analysis and interpretation [105]. The general work flow for label-free proteomics data analysis involves the following steps: searching raw data for protein/peptide identification, check data quality controls, data normalization, statistical analysis, as well as functional analysis (Figure 1.9).

Intensity m/z LC-MS/MS Data Collection Identification / Data Extraction Quantification

Row_ID A-1 A-2 A-3 B-1 B-2 B-3 P14209 7.08 6.94 7.01 7.03 6.85 6.79 P0C0L4 4.63 4.63 4.63 6.39 6.36 6.96 Q6EMK4 6.49 6.81 7.01 6.50 4.63 5.87 P04080 4.63 4.63 4.63 4.63 4.63 4.63 Q9P258 6.88 7.04 7.07 7.08 6.94 7.01 Data Normalization and Imputation Data Quality Check

Statistical Analysis Functional Analysis (IPA)

Figure 1.9: General schematic overview of proteomic workflow for label-free data analysis.

The searching of raw bottom-up proteomics data can be achieved through various search engines, most notably: MASCOT [106], SEQUEST [107], PEAKS [108], X!TANDAM [109], and Andromeda [110]. For example, SEQUEST is commonly used to search tandem mass spectrometry data by evaluating each peptide mass spectrum individually and then searching the sequence in a database of known peptides. Once data is searched, it is exported as a text file which contains both the protein and peptide information, and the quality of the data is check before further analysis. Data quality is can be screened by evaluating analytical, technical, and biological replications for variation. Furthermore, quality control samples must

also be checked to ensure proper function of the instruments during data collection [111]. After data quality is ensured, the samples are then further normalized and prepared for statistical analysis [112]. The statistical analysis ultimately serves to quantify magnitude of effect between sample sets, and several methods can used to generate p-value, including Fishers exact test, t-test, and regression models [113-115]. Additionally, a fold change is generally reported with a p-value in order to inform on the magnitude and direction of the protein change in expression. The combination of these data (i.e. protein ID, p-value, and fold change) can then be used to evaluate sample protein enrichment in that give rise to specific functions, also known as functional or enrichment analysis. Several methods are available to general functional analysis, including Cytoscape [116], DAVID [117], STING [118], and ingenuity pathway analysis (IPA) [119]. Protein enrichment analysis is often seen in proteomics as the most valuable interpretation of the data due to its ability to provide a purpose for the interplay of proteins within a biological complex, signaling pathway, or network module [120].

1.4 Synopsis of Completed Research

The research presented in this dissertation describes the proteomic investigation of various methods and types of MWCNT exposure. Chapter 2 describes the investigation of differential cellular protein response of mouse alveolar epithelial cells to MWCNTs as a function of ALD coating. The focus of this chapter is to elucidate potential global proteome changes in response to different types of MWCNT functionalized metal oxides, including: zinc-oxide and aluminum oxide. The experimental setup included a mono-culture mouse alveolar epithelial

cell for the ease of conducting a higher-throughput proteomic toxicity screen of functionalized

MWCNTs.

Conversely, Chapter 3 is focused on the proteomic comparison of cellular response of a

3D lung model exposed to MWCNT under submerged versus air-liquid interface conditions.

The cell culture used in this chapter is a human based tri-culture, thus limiting throughput compared to results shown in Chapter 2. However, conducting co-cultures are more advantageous then mono cultures because they represent a more physiologically relevant system (i.e. cell-cell communication). Furthermore, this chapter explores a critical issue in the nanotoxicology community of methods of exposure for toxicity testing. Due to the physical properties of the MWCNTs, they will exhibit differential deposition depending on exposure method. To date, there is no uniform protocol for nanomaterial exposure method, thus these results provide a significant contribution to the community to help them interpret differences in cellular response.

Chapter 4 provides results of an in-vivo proteomic analysis of pulmonary MWCNT exposure. This chapter explores a different type of MWCNT exposure method by examining the proteome of bronchoalveolar lavage fluid (BALF) collected from mice exposed to ALD coated MWCNTs similar to those studies in Chapter 2. Thus results from this study provided unique insight of global proteomic data not previously reported for a commonly studies exposure model.

Lastly, the research reported in Chapter 5 provide a proteomic investigation of protein corona formation in uncoated MWCNTs. These results show evidence that MWCNTs can be used as an abundant protein depletion material for proteomic sample digestion. A total of five experiments were reported in this chapter, and each experiment showed a greater number of

proteins identified in a MWCNT protein corona compared to controls. These results are likely due to the depletion of abundant proteins in the preparation of the protein corona. These results will be advantageous to both the nanotoxicology community, as well as the proteomics community, by introducing a cost effective method for sample enrichment.

CHAPTER 2

Mapping Differential Cellular Protein Response of Mouse Alveolar Epithelial Cells to

Multi-Walled Carbon Nanotubes as a Function of Atomic Layer Deposition Coating

Adopted from publication: Nanotoxicology, Volume 11, 2017

Gina M. Hilton1, Alexia J. Taylor1, Salik Hussain2, Erinn C. Dandley4, Emily H. Griffith3, Stavros Garantziotis2, Gregory N. Parsons4, James C. Bonner1, Michael S. Bereman1

1Toxicology Program, Department of Biological Sciences, North Carolina State University, Raleigh, NC

2 Clinical Research Unit, National Institute of Environmental Health Sciences/National Institute of Health, Research Triangle Park, NC

3Department of Statistics, North Carolina State University, Raleigh, NC

4Department of Chemical and Biomolecular Engineering, North Carolina State University, Raleigh, NC

*Author for Correspondence Michael S. Bereman, Ph.D. Department of Biological Sciences Center for Human Health and the Environment North Carolina State University Raleigh, NC Phone: 919.515.8520 Email: [email protected]

Abstract

Carbon nanotubes (CNTs), a prototypical engineered nanomaterial, have been increasingly manufactured for a variety of novel applications over the past two decades.

However, since CNTs possess fiber-like shape and cause pulmonary fibrosis in rodents, there is concern that mass production of CNTs will lead to occupational exposure and associated pulmonary diseases. The aim of this study was to use contemporary proteomics to investigate the mechanisms of cellular response in E10 mouse alveolar epithelial cells in-vitro after exposure to multi-walled CNTs (MWCNTs) that were functionalized by atomic layer deposition (ALD). ALD is a method used to generate highly uniform and conformal nanoscale thin-film coatings of metals to enhance novel conductive properties of CNTs. We hypothesized that specific types of metal oxide coatings applied to the surface of MWCNTs by ALD would determine distinct proteomic profiles in mouse alveolar epithelial cells in-vitro that could be used to predict oxidative stress and pulmonary inflammation. Uncoated (U)-MWCNTs were functionalized by ALD with zinc oxide (ZnO) to yield Z-MWCNTs or aluminum oxide (Al2O3) to yield A-MWCNTs. Significant differential protein expression was found in the following critical pathways: mTOR/eIF4/p70S6K signaling and Nrf-2 mediated oxidative stress response increased following exposure to Z-MWCNTs, interleukin-1 signaling increased following U-

MWCNT exposure, and inhibition of angiogenesis by thrombospondin-1, oxidative phosphorylation, and mitochondrial dysfunction increased following A-MWCNT exposure.

This study demonstrates that specific types of metal oxide thin film coatings applied by ALD produce distinct cellular and biochemical responses related to lung inflammation and fibrosis compared to uncoated MWCNT exposure in-vitro.

2.1 Introduction

Carbon nanotubes (CNTs), nanomaterials resembling rolled sheets of graphene, are quickly emerging in the field of nanotechnology due to extraordinary applications in electronics, engineering, and medicine [121]. Multi-walled (MW) CNTs are used primarily to increase the tensile strength of a variety polymers in the electronics and semi-conductor industry [14]. MWCNTs are being developed for a wide range of applications including electronics, energy storage and incorporation into polymers [122, 123]. For some applications, surface modification or thin film coatings on the MWCNTs can add enhanced functionality to improve electronic or physical performance. Atomic layer deposition, ALD, is a novel process to generate highly uniform and conformal nanoscale thin-film coatings, including: metal oxides, metals, and hybrid metal/organic materials [124-126]. While CNTs are quickly evolving for numerous applications, the fact still remains that they possess fiber-like physical characteristics similar to asbestos [127], a material that has resulted in hundreds of thousands of cases of pulmonary fibrosis and mesothelioma [128]. In addition to their fiber-like structure,

CNTs have been reported to exhibit varying degrees of toxicity depending on factors including: length, width, residual metal content, agglomeration status, and surface functionalization, which are thought to contribute to pulmonary inflammation and disease [24].

Pulmonary fibrosis is a fatal disease that is characterized by scaring of the lung tissue, which ultimately results in impaired lung function [129]. Rodent studies have shown that pulmonary exposure to SWCNTs or MWCNTs by inhalation, instillation, or oropharyngeal aspiration (OPA) results in pulmonary fibrosis [37]. In addition to in vivo studies of fibrogenesis in experimental animals, in vitro studies have also shown that MWCNT exposure

induces the production of growth factors and cytokines involved in the fibrogenic response, which is largely initiated through oxidative stress mechanisms [130, 53]. In particular, the alveolar epithelium is the primary target of CNT deposition in the distal lung and therefore alveolar epithelial cells are an appropriate cell type to elucidate mechanisms of CNT-induced lung disease in vitro [25]. We previously investigated the pulmonary toxicity of ALD- functionalized MWCNTs, coated with either aluminum oxide (A-MWCNT) or zinc oxide (Z-

MWCNT), in mice in vivo after delivery to the lungs by oropharyngeal aspiration [57, 131].

In these studies, we compared in vivo induction of pro-inflammatory and pro-fibrogenic cytokines in the bronchoalveolar lavage fluid (BALF) from mice with production of cytokines by human THP-1 monocytic cells. A-MWCNT caused less pulmonary fibrosis in mice compared to uncoated MWCNTs (U-MWCNTs) and caused reduced levels of pro- inflammatory and pro-fibrogenic cytokines (interleukin-6, tumor necrosis factor-alpha, osteopontin) in THP-1 cells in vitro [57]. Z-MWCNTs caused a similar degree of pulmonary fibrosis compared to U-MWCNT, but caused marked acute lung and systemic inflammation in mice with high levels of interleukin-6 that corresponded to exaggerated levels of interleukin-6 induced by Z-MWCNTs in THP-1 cells in vitro [131]. These studies highlighted vastly different pathologic and molecular responses to different ALD-MWCNTs in mice that could be partly predicted by cytokine profiles from THP-1 cells, but were limited by the measurement of only a few cytokine biomarkers of inflammation and fibrosis. To better understand underlying cellular mechanisms of response to various ALD-coated MWCNTs, cutting-edge tools emerging in measurement science, i.e. liquid chromatography tandem mass spectrometry

(LC-MS/MS), offer superior advantages to identify a large number of proteins in an unbiased manner to rapidly elucidate toxicity of functionalized MWCNTs.

Mass spectrometry based proteomics is a powerful ‘omics’ method used in measurement science to evaluate global changes in proteins; examples include: analysis of post-translational modifications (PTMs), identification of protein-protein interaction (PPI), and changes in protein abundance due to system perturbation. Recent reports have suggested that the proteome serves as a direct mediator between toxicants and the resulting cellular response to insult [132]. Currently, LC-MS/MS is a prevailing analytical tool used in proteomics due to its high sensitivity and unparalleled molecular specificity. Fortunately, proteomics can be used to generate large amounts of data that represent the cellular state by examining changes in protein expression upon toxicant exposure. Due to the large amount of data generated in proteomic experimentation, enrichment analyses, such as pathway analysis, are helpful to find biological changes that result from differential protein expression [83].

More specifically, enrichment analysis serves to identify over-represented groups of proteins that can be further associated with a specific pathway or function. Changes in biological pathways that are associated with groups of differentially expressed proteins can serve as a signature to specific perturbations in a biological system. Enriched pathways help to highlight proteins of interest for further examination, and ultimately identify markers that can help predict toxicant response.

In this study, we postulated that specific types of metal oxide coatings applied to the surface of MWCNTs by ALD would determine distinct proteomic profiles in mouse alveolar epithelial cells in vitro that could be used to predict oxidative stress and pulmonary inflammation. Herein, we investigated changes in protein expression as a function of MWCNT coating by using a combination of shotgun and targeted proteomic methods. The E10 cell line, isolated from normal mouse alveolar epithelial tissue, was used to create an in vitro model of

MWCNT exposure [133]. The following pathways were enriched for significant differences in protein expression as a function of MWCNT coating type: mTOR signaling from Z-

MWCNT exposure, mitochondrial dysfunction and oxidative phosphorylation signaling from

A-MWCNT exposure, and interleukin-1 signaling from uncoated-MWCNT exposure. These studies provide key insight into the mechanisms of cellular response upon in vitro exposure to functionalized MWCNT.

2.2 Methods

Materials

CMRL cell medium 1066-1x, fungizone antimycotic (Fz), fetal bovine serum (FBS), glutamine, penicillin-streptomycin, Trump’s transmission electron microscopy (TEM) fixative, noble agar, and the pierce lactate dehydrogenase (LDH) assay kit were purchased from ThermoFisher Scientific (Waltham, MA). Acetic acid, ammonium bicarbonate, sodium deoxycholate (SDC), dithiothreitol (DTT), iodoacetamide (IAM), formic acid (FA), ammonium hydroxide, hydrochloric acid (HCl), and bovine serum albumin (BSA) were obtained from Sigma Aldrich (St. Louis, MO). Diethylzinc (DEZ) and trimethylaluminum

(TMA) were purchased through Strem Chemicals at a minimum 98% purity (Newburyport,

MA). Multi-walled carbon nanotubes (MWCNTs) were purchased at Helix Materials

Solutions, Inc. (Richardson, TX) at 0.5-40 μm in length. P-type (<100>) silicon substrates were acquired through University Wafers (Boston, MA). High purity nitrogen gas was purchased from Machine & Welding Supply Co. Sequencing grade trypsin was purchased from Promega (Madison, WI). HPLC grade water, methanol, and acetonitrile were purchased

from VWR International (Morrisville, NC). Oasis MCX 30 µm particle size solid phase extraction cartridges were obtained from Waters (Milford, MA).

Nanomaterials

MWCNTs 0.5 – 40 m in length were synthesized by chemical vapor deposition.

Characterization of the MWCNTs was provided by the manufacturer and verified by

Millennium Research Laboratories (Woburn, MA) [134]. Some of the MWCNTs were coated with conformal nanoscale thin films of aluminum oxide or zinc oxide by atomic layer deposition (ALD) (Figure 2.1). Zinc oxide coating was achieved by co-reacting DEZ and deionized (DI) water. The aluminum oxide layer was achieved using sequential saturated exposures of TMA (Al(CH3)3) and water. Both reactions were conducted in a custom made, viscous-flow, hot-walled, vacuum reactor and purged with high purity nitrogen gas, and then further purified with an Entegris GateKeeper upstream from the reactor input [135, 136, 18].

TEM and mass gain were used to monitor the growth rate for the aluminum oxide and zinc oxide ALD coating process on MWCNTs; both types of nanotubes used in this study had a coating of roughly 10 nm. The details of ALD coating of carbon nanotubes have been previously described [131, 16, 57].

Figure 2.1: Transmission electron microscopy (TEM) of uncoated MWCNTs and ALD- coated MWCNTs. (A) Uncoated (U)-MWCNTs. (B) Al2O3-coated (A)-MWCNTs. (C) ZnO- coated (Z)-MWCNTs. 50 cycles of ALD were applied to A-MWCNTs and Z-MWCNTs.

Preparation of MWCNTs

Uncoated MWCNTs (U-MWCNT), aluminum oxide coated (A-MWCNT), and zinc oxide coated (Z-MWCNT) were weighed using a milligram scale (Mettler, Toledo OH) suspended in a sterile 0.1% pluronic F-68 (Sigma-Aldrich, St. Louis MO) phosphate buffer solution to achieve the final concentration of 10 mg/mL. Vials containing the suspended

nanomaterials were dispersed using a cuphorn sonicator (Qsonica, Newton CT) at room temperature for 1 minute prior to dosing. The A-MWCNT and Z-MWCNT concentrations were normalized to the U-MWCNT nanoparticle number in order to account for the mass increase caused by the surface modification of the CNT. The A-MWCNT were dosed at 2.5 times the U-MWCNT dose and the Z-MWCNT were dosed at 2.85 times the U-MWCNT dose.

A limulus amebocyte lysate chromogenic assay (Lonza Inc., Walkersville MD) was used to test the nanomaterials for endotoxin contamination. All MWCNTs tested negative (< 0.3

EU/mL) for endotoxin.

TEM Imaging

Methods used for TEM preparation E10 cells were plated in 4 x 60mm plates and grown to 100% confluency using the same media as described in the cell culture section above. Once cells reached 100% confluency they were exposed to U-MWCNT, A-MWCNT, and Z-

MWCNT, and one plate remained un-exposed as a control. All of the MWCNTs were prepared for dosing as described in the preparation of MWCNTs section above. Following a 24 hour exposure, cells were washed with Phosphate-buffered saline (PBS), the supernatant was removed, and 0.5 mL of Trump’s TEM fixative was added and cells were stored at 4 °C. Next, cells were placed in a 3% water agar suspension, and TEM imaging was processed through the

Center for Electron Microscopy facility at North Carolina State University [137].

Dynamic Light Scattering Analyses

U-MWCNT, A-MWCNT and Z-MWCNT suspensions were made as described in the preparations of MWCNTs section. Hydrodynamic diameter, size distribution and zeta potential

of the freshly prepared suspensions in E10 cell culture media were determined using dynamic light scattering (ZetaSizer Nano, Malvern Instruments, Westborough, MA) as described previously [61]. Electrophoretic mobility was converted into zeta potential using the

Helmholtz-Smoluchowski equation (Table A1).

Cell Culture

E10 alveolar epithelial cells were provided as a kind gift from Dr. Michael Fessler at the National Institute of Environmental Health Sciences (NIEHS) and were originally derived from the laboratory of Dr. Joseph Mizgerd at Boston University School of Medicine

[133]. The E10 cell culture was maintained in CMRL 1066 medium containing: 10% (v/v)

FBS, 0.5mM Glutamine, 100U/mL penicillin, and 100ug/mL streptomycin. Once cells were approximately 80% confluent, they were trypsinized and then plated into 6 x 6-well plates (36 wells total), and a separate set of 4 x 60mm plates. Each well of the 6-well plate received a seeding density of 0.166 x 106 cells, and the 60 mm plates were seeded with a density of 0.375 x 106 cells. The 6-well plates were used for the proteomic experiment, and the 60mm plates were used to collect cells for electron microscope imaging (further described in the TEM imaging section). Cell cultures were incubated at 37°C in a 5% CO2 atmosphere.

Experimental Design

A Latin square block design was used for the experimental setup of the MWCNT E10 cell dosing in order to control for plate affect, thus reducing experimental bias (Figure 2.2).

Every treatment (i.e. dose) was assigned a random number using a random number generator, and was then dosed accordingly as a Latin square design. Each dose and coating type had 4

replicates to account for biological variability; also known as biological replicates. 6-well plates contained 2 control wells (no MWCNT exposure) and 4 treatments wells (MWCNT exposure) [138]. A total of 36 cell culture samples were collected, 1 of the samples was randomly picked and was digested twice to generate technical replicates, and 10 samples were run as analytical replicates on the orbitrap LC-MS/MS (48 injections total).

Figure 2.2: (A) Atomic layer deposition with Al2O3 or ZnO applied to uncoated (U)-MWCNT to derive A-MWCNT or Z-MWCNT, respectively. (B) E10- mouse alveolar epithelial cell exposure via Latin square block design. For treatment groups a-g cells were exposed to PBS

(control) or MWCNT (U, A, or Z) at the indicated doses (2.5, 5, 100) expressed in μg/ml [139].

(C) Proteomic sample preparation through the following steps: SDC detergent treatment to isolate proteins, peptide generation from trypsin digestion, and sample cleanup with cation exchange MCX columns. All samples were processed on the following ThermoScientific LC-

MS/MS instruments: Q-Exactive Plus (Discovery) and Quantiva (Targeted).

Exposure / Isolation

Once the E10 cells in the 6-well plates reached 100% confluency, they were dosed with

MWCNTs. The MWCNTs were prepared as described in the preparation of MWCNTs section.

After dosing with MWCNTs, the 6-well cell culture plates were then placed in a 37°C at 5%

CO2 atmosphere incubator for 24 hours. Isolation of cells was carried out one 6-well plate at time to reduced stress response. Briefly, media was removed and cells were quickly rinsed with 1 mL of sterile PBS to help wash off MWCNTs from the cells and to remove excess media. The PBS wash was discarded and then another 1 mL of PBS was added to each well, the cells were gently scraped, and isolated. Finally, the PBS supernatant was removed and the cells were flash frozen in liquid nitrogen and stored at -20 °C overnight. The same exposure and cell isolation was conducted on the 4 x 60mm plates used for TEM imaging as the 6x6- well plates used for the proteomic experiment. The cells used for each set up came from the same stock, and the exposure and isolation was carried out at the same time.

Cytotoxicity Assay

An LDH assay was conducted to evaluate cytotoxicity in order to ensure appropriate dose concentration by coating type [140]. The E10 cells were exposed to U-, A-, and Z-

MWCNT in the following concentrations to establish a dose response curve: 0, 5, 10, 25, 50,

100 µg/mL (see above sections for how MWCNT were prepared for dosing). The LDH assay was conducted on 50 µL samples of media from MWCNT cell culture exposure with 2 replicates per sample. Absorbance was measured using a multiskanTM microplate photometer

(ThermoFisher), and the percent cytotoxicity was calculated by the manufacture’s protocol.

Protein Digestion

Each cell sample was suspended in a solution of 50 mM ammonium bicarbonate (pH

8.0) with 1% SDC. Probe sonication was applied to each sample in 2 pulses for 20 seconds per pulse, and an amplitude setting of 20%. The cell debris was centrifuged down for 2 minutes at 10,000 rpm. The supernatant was retained and protein quantitation was achieved using a nanodrop to measure the absorbance at 280 nm. The amount of protein in each sample was then adjusted with the 1% SDC solution in 50 mM ammonium bicarbonate such that the final amount of protein was 50 μg in 100 µL (i.e., 0.5 µg/µL). DTT was added to each sample to make a final concentration of 5 mM and then incubated at 60ºC for 30 minutes in order to reduce disulfide bonds. Following the reduction, samples were cooled to room temperature and IAM was added to make a final concentration of 15 mM, and incubated in the dark for 20 minutes at room temperature. Tryptic digestion was achieved by hydrating lyophilized trypsin to a stock solution of 1 µg/µL with 0.01% acetic acid in water. The trypsin solution was added to the protein mixture (i.e. 20 µg protein) in a 1:50 ratio (~0.4 µg trypsin), and then incubated at 37 °C for 4 hours. Following the digestion, samples were acidified with 6 M HCl to make a final concentration of 250 mM (pH ≤ 3) [141]. Sample purification and concentration was achieved using mixed-mode reversed-phase/strong cation exchange (MCX) cartridges. After the sample was added to the column, salts were removed with water (0.1% formic acid), neutrals and negatively charged species were removed with 1 mL of methanol (0.1% formic acid), and then peptides were eluted in 10% NH4OH in methanol. Finally, samples were concentrated down in vacuo (10 Torr) at 45 °C for 3 hours (speedvac concentrator, Thermo

Fisher Scientific), and then reconstituted in mobile phase A (98 % water, 2 % acetonitrile, and

0.1% formic acid) to yield a final concentration 0.5 µg/µL.

Liquid Chromatography

All of the samples were processed by 2 methods: A) Discovery proteomics method using a quadrupole orbitrap (Q Exactive Plus, Bremen Germany), and B) A targeted proteomics method using a triple quadrupole MS (Quantiva, ThermoFisher, San Jose, CA) operating in selected reaction monitoring mode (Table A2). Pico-frit columns were purchased from New Objective (Woburn, MA) and packed to a length of 20 cm for method A, and 15 cm for method B with reverse phase ReproSil-Pur 120 C-18-AQ 3 µm particles (Dr. Maisch,

Germany). The trap was packed in house to a final length of 3 cm. A 2 µL injection of 0.5

µg/µL peptide in mobile phase A (98 % water, 2 % acetonitrile, and 0.1 % formic acid) was washed onto the trap at a flow of 2.0 µL/min for 4 minutes. Peptide separation was achieved on the LC using a gradient of mobile phase A and mobile phase B (100 % acetonitrile, 0.1 % formic acid). The LC method A consisted of a 225 minute gradient with a linear ramp from 0

% B to 40 % B across 180 minutes (2-182 minutes), a ramp and wash at 80% B (182-193 minutes), followed by equilibration of the column at 0% B (194-225). Method B consisted of a 75 minute gradient with a linear ramp from 0 % B to 40 % B across 60 minutes (2-62 minutes), a ramp and wash at 80% B (62-69 minutes), followed by equilibration of the column at 0% B (70-75).

Mass Spectrometry

Orbitrap

Orbitrap tandem mass spectrometry was performed using a Thermo Scientific Q-

Exactive Plus (Bremen, Germany) in a top 12 data dependent acquisition mode (DDA), where the 12 most abundant preursors were selected for fragmentation per full scan. MS1 and MS2

scans were performed at a resolving power of 70,000 and 17,500 at m/z 200, respectively. A dynamic exclusion window of 30 seconds was used to avoid repeated interrogation of abundant species. Automatic gain control was 1e6 and 5e4 for MS1 and MS2 scans, respectively.

Samples were run in random order, and a quality control BSA digest was run every fifth injection to ensure proper LC-MS/MS reproducibility. Metrics were monitored in using the

Statistical Process Control in Proteomics algorithm [111]. The E10 data set contained 48 LC-

MS/MS runs, including: biological, technical, and analytical replicates. An average of 2200 protein groups were identified by LC-MS/MS in data dependent acquisition (DDA) mode.

After filtering, normalizing, and imputing (as described in the data analysis section), 1576 proteins were retained for further expression analysis.

Triple Quadrupole

Triple quadrupole mass spectrometry was conducted using the Thermo Scientific TSQ

Quantiva via selected reaction monitoring (SRM). The SRM method was created in Skyline-

Daily version 3.1.1.8884 [142] by importing the proteins, selecting one unique peptide to represent the protein, and then choosing the 5 top ranked transitions based on the spectral libraries created from the discovery data. The following parameters were set for SRM: dwell time 10 ms, use calibrated RF lens, Q1 and Q3 resolution full width half max (FWHM) is 0.7, collision induced dissociation (CID) is 1.5 mTorr,. The samples were run in a random order with a QC BSA digest every fifth injection [111]. While DDA is a powerful method to assess global proteomics, the nature of the method yields the potential for incomplete sampling of peptides [143], thus verification of relative protein abundance is ideal for confirming accurate quantitation. We verified differential protein abundance from the discovery proteomics

experiment using a triple quadrupole mass spectrometer operating in selected reaction monitoring (SRM) mode. Proteins were screened for the verification to represent high and low abundance. A single unique peptide was chosen as a proxy for protein abundance and the method was exported using Skyline. Raw data were imported into the created Skyline template and each individual peptide was manually evaluated to ensure retention time reproducibility, high dot product (> 0.8, match between discovery and targeted data), and proper integration boundaries. A pairwise comparison of peptide abundance for each MWCNT exposure versus control (i.e. A-MWCNT versus control) was conducted in order to represent expression fold change for validation of the DDA data with targeted SRM data (Figure A1). A Pearson’s correlation coefficient indicated the discovery differential proteomics data strongly correlated with the targeted proteomics data (r = 0.9635) (Table A3).

Database Search

Database searches were conducted using Proteome Discoverer 1.4 and the Sequest hyper-threaded algorithm. Data were searched against the Mus Musculus Swiss Prot protein database (number of sequences: 16657, date accessed: 06/30/2015) [144]. Peptide spectrum matches were post processed using percolator [145] to enforce a peptide spectral match threshold of less than 0.01 q value, the minimal false discovery threshold to which a spectral identification is accepted as correct. The law of strict parsimony was used for protein inference and grouping [146].

Data Analysis

Peak area data generated from Proteome Discoverer label free node was exported as a text file, and further analyzed using R version 3.2.2. Protein peak area was first log10 transformed and plotted as box-and-whisker plots to ensure all samples maintained roughly the same minimum, median, and maximum peak area values (Figure A2). In order to remove proteins that generated inconsistent results, the data were filtered by retaining proteins that had detection in at least 3 out of 4 biological replicates (i.e. protein maintained signal in 75% of the biological replicates within control or exposure groups). Additionally, proteins were retained if they did not have signal across all biological replicates in one group (i.e. all of the A-

MWCNT 100 μg/mL dosed samples did not have expression for protein x, but all of the U-

MWCNT 100 μg/mL dosed samples had expression for protein x). Thus, proteins were removed from the protein list if peak area signal was inconsistent within a class of samples.

Data were then central tendency median normalized, followed by imputation of missing values with the minimum peak area across entire data set [112]. Principal component analysis (PCA) was used to screen for potential confounding effects; such as plate effects. A two-sample

Welch t-test was conducted by pairwise comparison of protein peak area across each exposure compared to control. The regression model represented in Equation 2.1 was run on the data set after filtering and normalizing in order to evaluate independent effects that might contribute to the prediction of peak area. Ingenuity pathway analysis was used for enrichment analysis

[119]. Targeted peak area data was examined and exported from skyline-daily for further analysis.

Peak Area = Intercept + Dose + Coating

Equation 2.1: Analysis of variance model of peak area for proteins in common across each coating type.

2.3 Results

Exposure Dose-Response

A general overview of the experimental design is illustrated in Figure 2.2. Dose- response curves were generated for the E10 cell line by exposing cells to the following doses of U-MWCNT, A-MWCNT, and Z-MWCNT: 0, 1, 5, 10, 50, and 100 (µg/mL) (Figure 2.3).

Both U-MWCNT and A-MWCNT showed no cytotoxicity according to the LDH assay; however, the Z-MWCNT showed 100% cytotoxicity at the 10 µg/mL dose. The dose-response results were used to set a ‘low’ and ‘high’ exposure range for the dosing of E10 cells. The following doses were given in biological replicates of 4: U-MWCNT dosed at 5 and 100

µg/mL, A-MWCNT dosed at 5 and 100 µg/mL, and Z-MWCNT was dosed at 2.5 and 5 µg/mL

(lower dose of Z-MWCNT to adjust for cytotoxicity).

Figure 2.3: Log10 LDH dose response of E10 cells to the following exposures: U-MWCNT,

A-MWCNT, and Z-MWCNT.

Characterization of MWCNTs in Cell Culture Media

Dry U-MWCNTs have been previously characterized in terms of length, width, residual metals and agglomeration status [134]. U-MWCNTs are 30 to 50 nm in diameter and have a heterogeneous range of 0.3 to 50 micrometers in length, and a surface area of 40 to 300 m2/g.

A-MWCNTs and Z-MWCNTs produced by ALD coating of U-MWCNTs have also been thoroughly characterized width, length, and thickness of metal oxide coating and the details of these ALD-functionalized tubes are previously published [57, 131]. We also sought to characterize these nanomaterials after addition to the cell culture media used for maintaining

E10 cells. Specifically, we measured hydrodynamic diameter as an indication of MWCNT agglomeration size in media and zeta potential as an index of surface charge (Table A1). A-

MWCNTs and Z-MWCNTs had reduced hydrodynamic diameters (416 nm and 192 nm respectively) compared to U-MWCNTs (567 nm), indicating less agglomeration and better dispersion. A-MWCNTs and Z-MWCNTs did not have significantly different zeta potential

or polydispersity indices, indicating that the ALD coatings applied to MWCNTs did not alter surface charge.

TEM Imaging

TEM images of the E10 cells dosed in this experiment clearly show each type of

MWCNT were taken up by the alveolar epithelial cells (Figure 2.4 A-C). The Al2O3 coating on A-MWCNTs remained intact after cellular uptake (Figure 2.4 B). Interestingly, U-

MWCNT were similar to Z-MWCNT in appearance after uptake by epithelial cells (Figure

2.4 A and 4 C respectively), indicating that the Z-MWCNT coating was lost. Our previous work demonstrated that Zn+2 ions are released after Z-MWCNTs are added to cell culture media, indicating at least partial dissolution of the ALD coating [131]. Other studies have indicated that partial dissolution of ZnO nanoparticles occurs in the cell media but complete dissolution of the nanoparticles likely occurs within the cell [147, 148].

Figure 2.4: TEM images of E10 cell exposure. (A) 5 µg/mL U-MWCNT exposure, (B) 5

µg/mL A-MWCNT, and (C) 1 µg/mL Z-MWCNT. Arrows indicate MWCNTs within the cytoplasm of E10 cells.

E10 Proteomic Expression Changes

Of the 1576 proteins retained for global expression analysis, the following number of proteins were found to be significant by t-test (p < 0.05) at the highest dose of each exposure

(i.e. 100 µg/mL for U- and A-MWCNT, and 5 µg/mL Z-MWCNT): 138 U-MWCNT versus control, 210 A-MWCNT versus control, 103 Z-MWCNT versus control (Table A4). Fewer proteins were found to be significant in each ‘low’ dose exposure, and results can be found in

Table A5. Due to the limited significance found in the low dose exposure, the results will not be discussed herein. Further analysis was conducted on the ‘high’ dose exposure to evaluate

if the same proteins shared significance across each exposure type versus control to possibly indicate shared cellular response mechanisms. Interestingly, Venn diagrams plotted by categories of ‘increased’ and ‘decreased’ regulation of significantly differentially expressed proteins by t-test show little overlap in commonly significant proteins (Figure 2.5). Common proteins found for increased expression by exposure compared to control include: hemoglobin subunit beta, hemoglobin subunit gamma, and proteolipid protein 2. The common protein found for the decreased expression is fatty acid synthase.

Regression Model

Regression model results are reported in Table 2.1 which demonstrate that dose does not have significant effect on the prediction of peak area, but coating type is a marginally significant contribution to peak area (p = 0.06). These results illustrate that coating type has a larger contribution to cellular response (i.e., variance in protein expression) than dose. The dose was optimized to not cause cytotoxicity in order to better examine mechanisms of biological response to differentially functionalized MWCNTs.

Table 2.1: ANOVA results for modeling regression in Equation 2.1 with E10 dataset.

DF Sum Sq Mean Sq F value Pr(>F) Dose 2 2 0.9197 1.605 0.2009

Coating 1 2 1.993 3.478 0.0622

Residuals 25212 14445 0.573

Pathway Analysis

Pathway analysis was conducted using ingenuity pathway analysis (IPA) on the significant proteins for each MWCNT exposure compared to control. First, all of the significantly differentially expressed proteins, separated by increased or decreased expression relative to the control, were imported into IPA one MWCNT exposure group at a time.

Pathways for each MWCNT exposure were generated in IPA using the core analysis search, and the search species was specified as mouse. Comparison analyses were then conducted on each set of enriched pathways for exposure compared to control. Results shown in Figure 2.6 45

illustrate heat maps generated in R using the data output from IPA to compare pathways with increased pathway enrichment (Figure 2.6.A), and suppressed pathway enrichment (Figure

2.6.B) for exposure relative to control. The heat maps were re-created in R by transforming the –log10 p value output from IPA into Z scores, with dark red being the most significant p value and dark blue being the least significant. Table A6 lists the –log10 p value associated with each pathway across every MWCNT exposure. While several pathways exhibit significant differential protein expression, the following pathways will be discussed herein to better highlight the unique proteomic responses as a function of MWCNT coating type: inhibition of angiogenesis by thrombospondin-1, mTOR/eIF4/p70S6K signaling, oxidative phosphorylation, interleukin-1 signaling, and Nrf-2 mediated oxidative stress response.

Furthermore, the protein peak area log2 fold change associated with each exposure versus control were plotted as heat maps to illustrate specific changes in protein expression related to pathways of interest (Figure 2.7, Table A7).

Significant pathways corresponding to proteins with down-regulated expression upon exposure to control. A-MWCNT (aluminum oxide MWCNT compared to control), Z-

MWCNT (zinc oxide MWCNT compared to control), and U-MWCNT (uncoated MWCNT compared to control).

Figure 2.7: Heat maps of log2 fold change for proteins of MWCNT exposed compared to control for the following pathways of interest: (A) Nrf-2 mediated oxidative stress response,

(B) IL-1 signaling, (C) Inhibition of angiogenesis by TSP1, (D) mTOR signaling*, (E) eIF4/p70S6K signaling*, and (F) Oxidative phosphorylation. Log2 fold changes in protein abundance can be read as: increased expression compared to the control (Red), and decreased

expression compared to the control (Blue). *Outliers removed for higher resolution of heat map scaling. Table A7 provides full list of proteins and log2 fold change values.

2.4 Discussion

The aim of this study was to investigate how the alveolar epithelial cell proteome might be influenced by exposure to different MWCNT ALD coating types through a shotgun proteomics approach. Differential expression patterns obtained in cell culture can identify mechanisms of cellular response, which may help to predict toxicity and biological outcome in the lungs of mice and humans. To better understand mechanisms of the alveolar epithelial cell response to MWCNT in vitro exposures, pathway analysis was used to map cellular function to statistically differentially expressed proteins. Results from the pathway analysis demonstrate that either aluminum oxide-coated MWCNT (A-MWCNT) or zinc oxide-coated MWCNT (Z-

MWCNT) had both common and unique pathway enrichment that represents significant differences in cellular response compared to uncoated MWCNT (U-MWCNT) exposure. The following pathway enrichments will be discussed in further detail: Nrf-2 mediated oxidative stress response, interleukin-1 signaling, inhibition of angiogenesis by thrombospondin-1, mTOR/eIF4/p70S6K signaling, and oxidative phosphorylation.

Nrf-2 Mediated Oxidative Stress Response

The nuclear factor erythroid 2-related factor 2 (Nrf2)- mediated oxidative stress response pathway has been established as a significant survival response upon exposure to various environmental toxicants that are known to cause oxidative stress [149]. More

specifically, the activation of the Nrf2 signaling pathway is inversely proportional to inflammatory and pro-fibrotic cytokines. Studies have shown the importance of Nrf2 activation by exposing Nrf2 knockout mice to MWCNTs, which yielded excessive oxidative stress relative to the control exposure [150]. Of the exposures given in this study, Z-MWCNT had the most significant enrichment for the Nrf2 mediated oxidative stress response relative to control with 8 proteins showing significant increased enrichment relative to control (pathway enrichment p-value = 2.80E-07).

Our previous investigation revealed that Z-MWCNT exposure to THP-1 cells in vitro stimulate pro-inflammatory cytokine expression [131]. This pro-inflammatory pulmonary stress response has been routinely quantified by measuring antioxidant gene expression, such as heme oxygenase 1 (HO-1), and other proteins downstream of Nrf2 signaling (Figure 2.7.A)

[151]. Interestingly, zinc oxide was the only MWCNT coating to cause a significant increase in HO-1 (p-value = 0.001), thus indicating that the Z-MWCNT exposure is causing an oxidative stress response leading to Nrf2 activation and significantly increased HO-1 expression.

Interleukin-1 Signaling Pathway

The interleukin-1 (IL-1) family is a group of 11 cytokines which are known to mediate inflammatory response [152]. Activation of the IL-1 signaling pathway occurs as a stress response in order to produce various pro-inflammatory mediators [153]. Several studies have shown that MWCNT exposure causes acute inflammation via inflammasome activation and release of IL-1β that binds to specific receptors on a variety of lung cells to mediate acute inflammation [154-156, 61]. While the IL-1 proteins were not detected in this study due to

potential limitation in dynamic range, downstream proteins showed enrichment for IL-1 signaling activation. Significant enrichment of 5 proteins downstream of the IL-1 signaling pathway were found in the U-MWCNT exposure (pathway enrichment p-value = 1.620E-06).

Of the proteins enriched through the U-MWCNT exposure for IL-1 signaling pathway, mitogen-activated protein kinase 14 (MAPK14) is perhaps the most critical protein due to its essential role in inflammatory cytokine induction (Figure.7.B) [157]. MAPK14 was significantly upregulated in the U-MWCNT exposure (p-value = 0.043), but not the coated exposures to A- or Z-MWCNT, thus indicating that the alveolar epithelial cells present a stronger downstream IL-1 signaling response to U-MWCNT exposure.

Inhibition of Angiogenesis by Thombospondin-1

Angiogenesis, the process by which new blood vessels are formed, plays an integral role in the regulation of promotors that contribute to pulmonary hypertension and pulmonary fibrosis [158]. The regulation of angiogenesis is primarily controlled by fibroblast growth factor (FGF), vascular endothelial growth factor (VEGF), and heparan sulfate proteoglycans

(HSPG) proteins [159]. HSPGs can act to inhibit angiogenesis through signaling from thrombospondin-1 (TSP-1) [160]. TSP-1 is a matricellular glycoprotein that plays an influential role in the structure of cellular matrix, and can have direct and indirect inhibition of angiogenesis. The direct effects on inhibition occur via TSP-1 signaling to HSPG which inhibit angiogenesis, and indirect effects by TSP-1 binding to and activating transforming growth factor beta (TGF-beta) [161].

In addition to significant contributions to angiogenesis and pulmonary hypertension,

TSP-1 has been reported to have increased expression in malignant mesotheliomas caused by

asbestos exposure [162], thus suggesting MWCNT exposure may induce a similar mechanism of cellular response as exposure to asbestos due to similarities in physical properties. The results from this in vitro study show that TSP-1 was significantly upregulated in A-MWCNT

(p-value = 2.53E-05) and U-MWCNT (p-value = 0.003) exposures (Figure.7.C). The following proteins were significantly increased in expression upon A-MWCNT exposure and enriched for in the inhibition of angiogenesis signaling pathway: TSP-1 and tyrosin-protein kinase fyn (FYN). Additionally, U-MWCNT exposure drove significant upregulation in TSP-

1, FYN, and MAPK14. Ultimately, the A- and U-MWCNT exposures generated significant proteins that were enriched in the inhibition of angiogenesis pathway, thus indicating their contributions to inhibit angiogenesis.

mTOR/eIF4/p70S6K Signaling Pathway

Signaling from mammalian targeting of rapamycin (mTOR), like angiogenesis, has been reported to play a critical role in vascular remodeling and has been implicated for its contributions to pulmonary disease through the progression of pulmonary hypertension [163]. mTOR signaling can be induced by growth factors, and has also been well established for its contribution to regulating autophagy [164]. Under cellular stress and limited energy, mTOR signaling is repressed and autophagic signaling can be initiated [165]. Additional mediators of autophagy that are regulated by mTOR signaling include p70S6k and eIF4 proteins, which are reported to lead to translation of proteins that mediate cell cycle activators, ribosome biogenesis, and angiogenesis [166]. The results from this study show mTOR/eIF4/p70S6K signaling pathways have significant enrichment upon Z-MWCNT exposure (p-value = 1.46E-

09 mTOR signaling, and p-value = 5.59124E-08 eIF4/p70S6K signaling), but not A- and U-

MWCNT exposure (Figure 2.7.D and E). Moreover, 10 proteins are significantly upregulated in the Z-MWCNT exposure that are downstream of the mTOR/eIF4/p70S6K signaling pathway, thus indicating activation of signaling by Z-MWCNT compared to control. These results suggest that the Z-MWCNT drive increased expression of protein mediators that contribute to cell proliferation that were not significant in the U-MWCNT and A-MWCNT exposures.

Oxidative phosphorylation

Oxidative phosphorylation (OXPHOS) is the metabolic pathway comprised of protein complexes that make up the mitochondrial electron transport chain, and function to generate

ATP [167]. OXPHOS signaling is generally associated with mitochondrial function and ultimately effect the oxidative state of the cell, which can serve a critical marker for injury.

Several pathways can be initiated in response to a change in oxidative state in the cell, including cell survival or cell death via apoptosis or necrosis [168]. More specifically, there is a sensitive intracellular balance in response to oxidative stress between mitochondrial biogenesis and mitochondrial dysfunction [169]. Several studies have shown mitochondrial dysfunction as a result of excessive oxidative stress during hyperoxia [170], as well as asbestos exposure [171] and MWCNT exposure [172].

The most significant increased enrichment for the OXPHOS signaling response was found in the A-MWCNT (pathway enrichment p-value = 4.26E-10). Proteins associated with the mitochondrial dysfunction pathway were also significantly enriched for in the A-MWCNT exposure (p-value = 8.50E-11). Of the 14 proteins enriched for mitochondrial dysfunction, 11 were increased in A-MWCNT exposure, and 5 proteins were increased for U-MWCNT

(Figure 2.7.F). Most of the upregulated proteins enriched across both exposures were identified as proteins in the electron transport chain. Overall, the increased expression for the proteins enriched in OXPHOS and mitochondrial dysfunction pathways associated with aluminum oxide coated and uncoated MWCNT exposure may indicate metabolic adaptation to help counter the pro-fibrogenic effect of MWCNT exposure observed in vivo and in vitro

[57].

Common Protein Expression

While most of the proteins upregulated upon exposure to ALD coated and uncoated

MWCNT produced differential expression related to different pathways, one protein showed significant upregulation across every exposure relative to control: Hemoglobin (Hb). Alveolar

Type II epithelial cells have been reported to express Hb in several cell lines, both primary and transformed [173]. There have also been studies reporting increased Hb expression in response to oxidative stress, thus implicating the pulmonary epithelium may play a role in protection against oxidative/nitrosative stress [174, 175]. Both coated and uncoated MWCNT exposure exhibited increased beta-Hb and gamma-Hb expression compared to the corresponding Hb subunit levels expressed in the control samples (Table A4). Note, the alpha-Hb sub-unit was also highly expressed in each MWCNT exposure, but that protein was removed during data filtering due to variable low expression in the control samples. Therefore, a common protective mechanism in response to MWCNT exposure may be through increased Hb expression in order to scavenge free oxygen and nitric oxide species.

In vitro versus In vivo Comparison

The proteomic results from the E10 mouse alveolar epithelial cells in vitro revealed mechanisms of cellular response that were not characterized in our previous mouse in vivo studies using the same ALD coated MWCNTs [131, 57]. For example, mice exposed to A-

MWCNTs by oropharyngeal aspiration had elevated levels of IL-1β in BALF compared to animals treated with U-MWCNTs [57]. Mice exposed to Z-MWCNTs by oropharyngeal aspiration exhibited high levels of the acute phase reactant protein IL-6 in BALF [131].

However, the endpoints for both of the previous studies were limited to only a few selected cytokines that were measured by ELISA and did not yield a strong bases for comparison for proteomic results. Also, the cytokines measured in the previous in vivo studies could have been produced by cell types other than alveolar epithelial cells such as alveolar macrophages or fibroblasts. Finally, our results in the present study with E10 cells focused only on intracellular proteins since evaluation of secreted proteins would have been confounded by serum proteins present in the cell culture medium. Therefore, there are some limitations of in vitro cell culture experiments for predicting disease outcomes in vivo. Nonetheless, the results of the proteomic analysis of E10 cells in the present study is an important step towards identifying new molecular targets and biomarkers of disease that can be further investigated in future mouse exposure studies.

2.5 Conclusion

Given that CNT functionalization will yield a diversity of nanomaterials that have unknown potential to cause pulmonary diseases, more sensitive and high-throughput toxicity

screening needs be developed to raise awareness of unique nanomaterial hazards. Markers for toxicity can be found in differential protein response as a function of CNT coating type. Tools like pathway enrichment can be used to aid the screening process by mapping statistically significant differentially expressed proteins to a biological function. Unique pathway regulation was found in this study through the following results: increased autophagy signaling from Z-MWCNT exposure, increased mitochondrial dysfunction and oxidative phosphorylation from A-MWCNT exposure, and increased interluekin-1 signaling from U-

MWCNT exposure. Ultimately, this study demonstrates the use of proteomics as a powerful sensitive measurement technique that can unravel differential cellular protein expression in cultured lung epithelial cells as a function of CNT coating type. Moreover, these differential cellular protein expression profiles may be useful towards screening carbon nanotubes for toxicity and predicting hazard for human exposure.

Acknowledgments

The authors acknowledge support from the National Institute of Environmental Health

Sciences (NIEHS) Training Grant: T32 ES007046 (GSH), NIEHS R01ES020897 (JCB, GNP,

ECD, AJT), and NIEHS P30ES025128 (MSB, GSH, EHG). This work was also supported

(in part) by the NIEHS Intramural Research Program (SH, SG)

CHAPTER 3

Proteomic Cellular Response Comparison of a 3D Lung Model Exposed to MWCNT

Under Submerged versus Air-Liquid Interface Conditions

Adopted from manuscript in preparation for publication, July 2017

G. Hilton1, H. Barosova2, B. Rothen-Rutishauser2, M. Bereman1

1Toxicology Program, North Carolina State University, Raleigh, NC 27606 2Adolphe Merkle Institute, Université de Fribourg, Fribourg, Switzerland

Abstract

With the emerging concern over the potential toxicity associated with carbon nanotube inhalation exposure, several in vitro methods have been developed to evaluate cellular response. Since the major concern for adverse effects by carbon nanotubes is inhalation, various lung cell culture models have been established for toxicity testing, thus creating a wide variation of methodology used for nanomaterial testing. Limited studies have conducted side- by-side comparisons of common methods used for carbon nanotube toxicity testing. Thus, the aim of this study was to use global proteomics to evaluate cellular response including pro- inflammatory and pro-fibrotic mediators of a 3D lung model mimicking the human alveolar epithelial tissue barrier and composed of macrophages, epithelial cells as well as fibroblasts.

The cells were exposed to multi-walled carbon nanotubes under submerged and air-liquid interface conditions. The proteomics data produced minimally significant differences in the multi-walled carbon nanotube exposure compared to controls for each of the exposure condition. However, the comparison of multi-walled carbon nanotube exposure method, submerged versus air-liquid interface, produced a very significant differential response. This study demonstrates a side-by-side comparison of commonly deployed carbon nanotube

exposure methods, which ultimately do not produce comparable results. These data should be considered by the nanotoxicology community when interpreting or cross comparing in vitro exposure results.

3.1 Introduction

The development of the nanotechnology field has led to an increasing production of various types of nanomaterials, including carbon nanotubes (CNTs), which harness several unique physicochemical properties. CNTs are hollow nanofibers formed from carbon and have extraordinary properties; such as: thermal stability, electrical conductivity and remarkable mechanical durability, and they are one of the most widely used nanomaterials [176, 177].

Increasing demand for CNT production for industrial and biomedical applications leads to the potential for greater risk of natural eco-system contamination as well as human exposure.

Primary concerns for CNTs exposure include biopersistence, biodurability, and their asbestos fiber-like structure [178-180]. A recent human health study showed CNT workers (with at least six months direct contact with aerosolized multi-walled CNTs (MWCNT)) had significant changes in gene expression associated with cell-cycle regulation, apoptosis and proliferation, with potential increased pulmonary, cardiovascular and carcinogenic disease risk [181].

Additionally, in vivo studies report the adverse effect of CNTs is increased in the case of long

(10 – 30 µm) and thin fibers of high rigidity are able to penetrate to deeper lung regions, and persist there due to an impaired macrophage clearance [182-184]. A growing number of animal studies demonstrate that exposure to CNTs via inhalation potentially triggers airway injury, inflammation, fibrosis and granuloma [185, 186]. Many studies addressing possible

adverse effects of CNTs were performed on animal-based approaches; however, there is a clear need to avoid such time-consuming, cost-intensive and ethically compromising in vivo studies.

In order to alleviate demands for reduced animal use, human lung cell co-culture models have recently emerged as a rapid alternative toxicity test for the primary screening of large array of materials in order to investigate the interaction of inhaled particles with the cells and subsequent cellular responses [187, 188].

When conducting in vitro inhalation toxicity testing, exposure method and frequency is a critical variable to consider due to the CNT physical properties that lead to variation in dosimetry [67, 68]. The majority of the in vitro studies do not consider the realistic concentrations in an occupational setting [189], nor the nature of such a exposure, i.e. exposure in manufacturing facilities most probably occurs repeatedly over a long timeframe. It is crucial to obtain an insight into the pulmonary toxicity of prolonged (subchronic) CNT exposure to repeated doses at a low and realistic concentration. In addition, most existing in vitro studies focusing on acute or long-term exposures are usually carried out using relatively high doses and under submerged conditions [190-192], despite such exposure conditions having less physiological relevance. In order to investigate potential adverse effect of CNTs as realistic as possible, various Air-Liquid interface (ALI) exposure equipment have been developed [193-

196] allowing the homogenous spreading of nebulized material onto the cell model surface with online measurement of deposited material (using quartz crystal microbalance). Such an approach allows investigators to monitor the deposited material concentration online. With the development of new exposure technologies, there is a need to investigate potential differences compared to traditional submerged exposure method on cell culture models.

Preliminary data collected from CNT co-culture exposures, under submerged or ALI conditions, have been focused on measuring a few specific endpoints; including inflammatory and pro-fibrogenic protein markers. However, a systems biology approach lends a more viable option to thoroughly assess the cellular response to CNTs beyond pre-selected endpoints.

Mass spectrometry (MS) based proteomics is a powerful systems biology method used to investigate global changes on the protein level [103]. The instruments employed for global proteomic experimentation, namely liquid chromatography tandem mass spectrometry (LC–

MS/MS), are a prevailing analytical tool due to their high sensitivity and molecular specificity

[102]. Traditional investigation of protein endpoints for an exposure study would generally involve Western blot and/or an enzyme-linked immunosorbent assay (ELISA). However,

Western blots have low dynamic range, are low throughput, and are limited to the availability of a specific antibody. An ELISA, although extremely sensitive, is also dependent on the availability of an antibody, thus limiting the assessment of novel protein mediators [104]. To circumvent the issue of sensitivity and throughput, LC-MS/MS based proteomics will be used in this study to elucidate global differences in cellular response to difference methods of

MWCNT exposure [197].

The aim of this study was to apply a novel in vitro co-culture model consisting of three human cell lines types, i.e. macrophages (activated THP-1), epithelial cells (A549) and fibroblasts (MRC-5), and to compare the prolonged repeated exposure of Mitsui-7 (M-7)

MWCNT under submerged and ALI exposures. Proteomic investigation was used to evaluate samples collected apical and basolateral of the epithelial cells in the tri-culture at a 96 hour time point. In this study, we postulate that the LC-MS/MS based proteomics method will elucidate a polarized response in apical and basolateral samples; as well as, demonstrate

differential proteome expression unique to exposure method. The overarching goal of this study is to provide potential novel mediators that can be used to further the 3D cell culture method development for regulatory toxicity testing.

3.2 Methods

Materials

The cell lines (A549, MRC-5, THP-1) were purchased from American Type Culture

Collection (ATCC, Virginia, USA). Roswell Park Memorial Institute medium (RPMI 1640),

Minimum essential medium (MEM), MEM Non-Essential Amino Acids solution (NEAA),

Fetal Bovine Serum (FBS), Penicillin/streptomycin and L-Glutamine were all purchased from

Gibco, Massachusets, USA. Bovine serum albumin (BSA), 2-mercaptoethanol, Phorbol 12- myristate 13-acetate (PMA), Triton-X, paraformaldehyde (PFA) and nucleus DAPI staining were purchased from Sigma Aldrich (Switzerland). Sodium chloride for exposure purposes

(NAAPREP® physiological saline) was purchased from GlaxoSmithKline (France). BD

Falcon 12-well cell culture inserts (high pore density transparent PET membrane, with 0.4 µm diameter pore size and 0.9 cm2 effective growth area) and culture 12-well plates with flat bottom were purchased from BD biosciences (Switzerland). Chicken polyclonal antibody to vimentin, goat polyclonal antibody to chicken IgY Alexa647 and goat anti-mouse IgG H&L

Alexa488 antibody were purchased from Abcam (UK), while Rhodamine Phalloidin was purchased from Life Technologies (Switzerland). Mitsui-7 Multi-Walled Carbon Nanotubes

(Mitsui & Co, Japan, M-7 MWCNTs) were a kind gift from professor Vicki Stone, Heriot-

Watt University, Edinburg, UK. Acetic acid, ammonium bicarbonate, sodium deoxycholate

(SDC), dithiothreitol (DTT), iodoacetamide (IAM), formic acid (FA), ammonium hydroxide

(NH4OH), and hydrochloric acid (HCl) were obtained from Sigma Aldrich (St. Louis, MO).

High purity nitrogen gas was purchased from Machine & Welding Supply Co. Sequencing grade trypsin was purchased from Promega (Madison, WI). HPLC grade water, methanol, and acetonitrile were purchased from VWR International (Morrisville, NC). Vivacon500® 10K and 30K molecular weight cut off (MWCO) spin filters were purchased from ThermoFisher

Scientific (Waltham, MA). Pico-frit columns were purchased from New Objective (Woburn,

MA), and reverse phase ReproSil-Pur 120 C-18-AQ 3 µm particles were purchased from Dr.

Maisch (Entringen ,Germany).

Preparation of MWCNTs

MWCNTs (M-7 MWCNTs) were dispersed in 0.1 % BSA to obtain a well-dispersed homogenous suspension for the exposure experiments. Briefly, M-7 MWCNTs dry powder was weighed and heat sterilized at 100°C overnight and subsequently cooled down at room temperature. Sterile filtered 0.1% BSA was added to the vial to reach the final concentration of 50 μg/mL M-7 MWCNTs and sonicated with continuous shaking for 3h. The M-7

MWCNTs suspension was stored at 4 °C, until further usage. Prior to the exposure, the stock suspension was sonicated for 1h to disperse sediment and agglomerated M-7 MWCNTs. This suspension was then either diluted in complete cell culture medium for suspension experiment

(working concentration 10 μg/mL) or applied to the nebulizing head to perform Air-Liquid

Interface (ALI) exposures.

Characterization of M-7 MWCNTs solution

Samples deposited onto a copper grid (200 mesh and single grid as well) were analyzed via Transmission electron microscope (TEM) FEI Tecnai Spirit (Oregon, USA) operating at

120 kV. Images were recorded with a Veleta CCD camera (Olympus, Japan).

Endotoxin Content

The endotoxin concentration in M-7 MWCNTs stock solution was analyzed using the

PierceTM LAL Chromogenic Endotoxin Quantitation kit (ThermoFisher Scientific), following the manufacturer’s instructions.

Cell culture

The triple cell co-culture model designed of human epithelial type II cells (A549 cell line) with human macrophages (THP-1 cell line) on the top of epithelial layer and human lung fibroblasts (MRC-5 cell line) on the bottom part of the model. The detailed triple cell co-culture protocol was modified for the recently described co-culture described by Lehmann et al. [198] with the adaption that instead of dendritic cells fibroblasts have been seeded. Briefly, MRC-5 cells were cultured prior assembling the co-culture model in MEM supplemented with 10%

FBS, 1% penicillin/streptomycin, 1% L-Glutamine and 1% NEAA) at 37 °C, 5% CO2 and subsequently seeded on the bottom of the BD Falcon cell culture inserts at the inverted position at a density of 104 cells/cm2. Then, the inserts with the cells were incubated in this position for

3 h at 37 °C, 5% CO2 in a covered sterile Petri dish, then turned up-side down and placed into

12-well plates and A549 cells (previously cultured in Roswell Park Memorial Institute medium

(RPMI 1640) supplemented with 10% FBS, 1% penicillin/streptomycin and 1% L-Glutamine)

5 2 at 37 °C, 5% CO2) were seeded onto the apical insert side at a density of 2.9x10 cells/cm .

This co-culture model was cultured for 4 days under submerged conditions (i.e. covered with

1 mL of the RPMI in upper chamber and 1.5 mL of the MEM in the lower chamber of the insert). In parallel, the THP-1 cells were cultured in RPMI supplemented with 10% FBS, 1% penicillin/streptomycin, 1% L-Glutamine and 0.05M mercaptoethanol at 37 °C, 5% CO2. In order to assemble the triple cell co-culture model, the THP-1 cells were activated by addition of 20 μg/mL of PMA in supplemented RPMI overnight and finally seeded on top of the epithelial cells. After 24 h under submerged conditions the triple co-cultures were either exposed under submerged conditions, or transferred to the ALI conditions, by removing the medium in the upper chamber and replacing the medium in the lower transwell chamber with

0.6 mL of fresh culture medium mixed in ratio 1:1 RPMI:MEM. The cells were subsequently exposed to air for additional 24 h before ALI particle exposure was performed.

Exposure Methods

Cells were transferred to serum-free medium 18±2 h prior sample collection in order to minimalize background protein signal from FBS and avoid the cell starving. The M-7

MWCNT exposures were conducted by two methods, including: submerged and air-liquid interface.

Submerged exposures:

Three individual cell culture inserts were exposed to tested material per each exposure time point. Cell cultures were exposed to 10μg/mL M-7 MWCNTs (stock solution was diluted in cell culture medium to final concentration) and as a negative control 0.02% BSA was used

in order to investigate the influence of the dispersant on the cell culture model. Cells were exposed to the tested materials for 96h and medium was exchanged after 48h with fresh media containing i.e. M-7 MWCNTs or BSA.

Air-Liquid Interface exposures:

Aerosolization of M-7 MWCNTs or BSA alone was performed using the

VITROCELL© Cloud system. Briefly, the exposure system consists of a nebulizer, an exposure chamber as well as a quarz crystal microbalance (QCM) (operated at 5 MHz, detection limit: 0.1 μg/cm2) for online measurements of the deposited dose. For each aerosolization, 200 μL of suspended sample with 0.09% sodium chloride (NaCl) was added to the nebulizer (Aeroneb® Lab, Dangal, Galway, Ireland). The vibrating perforated membrane at the neck of the nebulizer generates the aerosol, which is transported into the exposure chamber. Inside the chamber, it gently deposits the aerosolized sample onto the co-culture models that are maintained at the ALI. The selected flow rate (5L/min) is ideal for the aerosol to sufficiently mix to all sides of the chamber, hence resulting in uniform droplet deposition.

Three individual inserts of cell co-cultures cultured at ALI were exposed to aerosolized samples (M-7 MWCNTs or BSA). The cells were collected 96h post exposure were exposed every day to 5 doses of M-7 MWCNTs in resulting in a total deposition 10 μg/cm2 of the composite M-7 MWCNTs-BSA. The medium was exchanged after 48h with fresh medium.

Cytotoxicity

Cell culture medium was collected for cytotoxicity evaluation and stored at 4°C until further use. The release of lactate dehydrogenase (LDH) into the supernatant, a well-known

indicator of membrane impairment [199], was evaluated using the commercially available

LDH detection kit (Roche Applied Science, Mannheim, Germany), according to the manufacturer’s protocol. Each sample was tested in triplicates and LDH values are presented relative to the negative control (BSA-treated cells). Determination of the enzyme activity was measured by the absorbance at 490nm (reference wavelength at 630nm). Cells cultures exposed apically to 0.2% Triton X-100 for 24h were used as positive control.

Cell morphology

Cell cultures were fixed for 15 mins in 4% PFA to assess cellular morphology at 24 and 96 hour. The fixed cells were treated with 0.1M glycine for 15min and subsequently permeabilized in 0.2% Triton X-100 for another 15min. All antibodies were diluted in 0.3%

Triton X-100 and 1% BSA in PBS and incubated for 2h. Cells were stained first with mouse anti-CD68 and chicken anti-vimentin (both 1:100 dilution) and with anti-mouse Alexa 488, anti-chicken Alexa 647, phalloidin-rhodamine and DAPI (all dilution 1:100) afterwards. Anti-

CD68 is used for macrophage staining, anti-vimentin stains fibroblasts, Phalloidin-rhodamine stains F-actin cytoskeleton and DAPI stains nucleus. Following antibody incubation, membranes were embedded in Glycergel. Representative z-stack images were performed using an inverted laser scanning confocal microscope (LSM) 710 (Axio Observer.Z1, Carl Zeiss,

Germany). Image processing was conducted with IMARIS 3D restoration software (Bitplane

AG, Zurich, Switzerland).

Sample Preparation for Digestion

Cell samples were suspended in 100 µL of a 1% SDC 50 mM ammonium bicarbonate solution (pH 8.0). An amplitude of 20% probe sonication was applied to each cell sample in

2 pulses for 20 seconds per pulse. The cell debris was centrifuged down for 5 minutes at

10,000 rpm. Protein quantitation was achieved using a nanodrop to measure absorbance

(280nm) of each retained supernatant.

Serum Free media samples were prepared by concentrating 500 µL of sample onto a

10K MWCO Vivacon spin filters, followed by centrifugation at 10,000 rpm for 10 minutes.

The volume of 500 µL media concentrated down to 25 µL, and the filters were washed with an additional 75 µL of 1% SDC 50 mM ammonium bicarbonate solution (pH 8.0). The total volume of 100 µL was retained from the top portion of the filter. Protein quantitation was achieved using a nanodrop to measure absorbance (280nm) of each retained supernatant.

Wash samples were prepared by first checking protein amount using a nanodrop to measure absorbance (280nm). Protein content of wash samples was high enough that sample concentration was not needed.

Filter Aided Sample Preparation- Protein Digestion

Each sample was adjusted to contain the same starting concentration of protein using

1% SDC solution in 50 mM ammonium bicarbonate such that the final amount of protein was

30 μg in 100 µL (i.e., 0.30 µg/µL). Disulfide bonds were reduced by the addition of DTT to make a final concentration of 5 mM and then incubated at 60ºC for 30. Samples were then cooled to room temperature. IAM was added next to make a final concentration of 15 mM and incubated in the dark for 20 minutes at room temperature. Each sample was then carried

through a modified filter aided sample preparation (FASP) [200]. Briefly, 30K MWCO

Vivacon spin filters were conditioned with 20 µL of 1% SDC solution in 50 mM ammonium bicarbonate for 20 minutes before adding sample. The samples were then centrifuged for 15 minutes at 12,000 rpm, and then washed twice with 200 µL 8 M urea and centrifuged for 15 minutes at 12,000 rpm between each wash. Flow through was discarded between each centrifugation step. The samples were then washed twice with 200 µL 50 mM ammonium bicarbonate following the same centrifugation steps detailed in the urea wash. The filters containing the protein samples were then placed in a new collection tube, and a 30 µL 1:50 trypsin solution (µg trypsin:µg total protein) in 50 mM ammonium bicarbonate buffer was added to each filter. Samples were then incubated at 37 °C for 4 hours. Following the trypsin digestion, samples were centrifuged for 15 minutes at 12,000 rpm. An additional 30 µL of 50 mM ammonium bicarbonate was added to the spin filter, and centrifuged for 15 minutes at

12,000 rpm. Finally, the samples were acidified by the addition of 6 M HCl to a final concentration of 250 mM (< 3.0 pH). The final concentration of each sample was diluted to

0.25 µg/µL [141].

Nanoflow LC

Pico-frit columns were packed to a length of 30 cm with reverse phase ReproSil-Pur

120 C-18-AQ 3 µm particles. The trap was packed in house with reverse phase packing material to a final length of 3 cm. A 2 µL injection of 0.25 µg/µL peptide sample was washed onto the trap at a flow of 2.0 µL/min for 4 minutes. Peptide separation was achieved on the

LC using a gradient of mobile phase A (98 % water, 2 % acetonitrile, and 0.1 % formic acid) and mobile phase B (100 % acetonitrile, 0.1 % formic acid). Due to differences in the dynamic

range of the sample sets, two LC methods were used to separate peptides (Table B1). Method

A was used to process the cell samples, and Method B was used to process media and wash samples. Method A consisted of a 225 minute gradient with a linear ramp from 0 % B to 40

% B across 180 minutes (2-182 minutes), a ramp and wash at 80% B (182-193 minutes), followed by equilibration of the column at 0% B (194-225). Method B consisted of a 90 minute gradient with a linear ramp from 0 % B to 40 % B across 70 minutes (2-72 minutes), a ramp and wash at 80% B (73-78 minutes), followed by equilibration of the column at 0% B

(79-90).

Orbitrap LC-MS/MS

Tandem mass spectrometry was performed using a Thermo Scientific Q-Exactive Plus

(Bremen, Germany). A top 12 data dependent acquisition mode (DDA) was used for every full scan where the 12 most abundant precursors were selected for fragmentation. A resolving power of 70,000 and 17,500 at m/z 200 were used for MS1 and MS2 scans, respectively. A dynamic exclusion window of 30 seconds was used to avoid repeated interrogation of abundant species. Automatic gain control was 1e6 and 5e4 for MS1 and MS2 scans, respectively.

Samples were run in a randomized fashion, and a BSA digest was run every fifth injection as a quality control to ensure proper LC-MS/MS reproducibility. Instrument performance metrics were monitored in using the Statistical Process Control in Proteomics algorithm [111, 201].

Database Search

Database searches were conducted using the MaxQuant (v1.5.6.0) [202]. Data were searched against the Homo sapiens Swiss Prot protein database (number of sequences: 20183,

date accessed: 10/17/2013) [144], with the following defined parameters: carbamidomethylation of cysteine as a fixed modification, oxidation of methionine and acetylation of protein N-terminal set as variable modifications, trypsin/P digestion cleavage with a maximum of two missed cleavages, and label-free quantitation (LFQ) mode.

Data Analysis

Protein intensity generated from the MaxQuant LFQ search was exported as a text file, and further analyzed using R Studio version 3.2.2. Data was first log10 transformed and plotted as box-and-whisker plots to ensure samples maintained similar minimum, median, and maximum values. To ensure proper normalization, each data set was uniquely central tendency normalized by sample group (i.e. cell, media and wash samples were all separately normalized), and then missing values were imputed with minimum protein intensity [112]. Principal component analysis (PCA) was used to screen for sample outliers. Data was modeled for regression analysis, and the p-values for significant factors were calculated by a pairwise comparison two-sample Welch t-test of protein intensity. Regression modeling was conducted using SAS JMP Pro 13® (Cary, NC). Enrichment analysis was conducted using ingenuity pathway analysis (IPA) [203]. All heat maps were generated using the R Studio gplots package.

3.3 Results

A schematic overview depicted in Figure 3.1 represents the lower airway tri-cell culture experimental design. Samples were exposed to M-7 MWCNT by two different lung

exposure methods: submerged or ALI. Proteomic experimentation was conducted on the collected samples, and the results are detailed herein.

Figure 3.1: A) General schematic of the 3D human epithelial tissue barrier, including the following human cell types: THP1 Macrophages, A549 alveolar epithelial type II cells, and

MRC5 fibroblasts. B) Experimental overview of exposure method and sample collection.

Cells were exposed to the air at ALI at day 0. ALI exposures were conducted every day for 5 days prior to collection on day 5. The submerged exposure were given as a bolus dose on the first day of the experiment with media/sample change on the third day, and collected after 96 hours. C) Samples were collected as apical wash, cells, and media in the lower chamber for

ALI cultures, and media in the upper as well as lower chamber for submerged co-cultures at the 96 hour exposure time point.

Characterization of MWCNT

TEM micrographs show M-7 MWCNTs used for suspension experiments sediment onto a TEM grid (Figure 3.2.A and B), and the M-7 MWCNTs structure after the nebulization

(Figure 3.2.C and D). Figure 3.2.D illustrates the presence of salt crystals, which are necessary an optimal nebulization of the material, can be seen as black dots. The images represent the concentration of M-7 MWCNT after one exposure, i.e. two days-dose in case of suspension exposure and daily dose in case of ALI exposure.

A B

C D

Figure 3.2: TEM micrographs of M-7 MWCNTs in suspension (A, B) and deposited after the nebulization (C, D).

Characterization of Cell Morphology

The cell morphology is not impaired after exposure to M-7 MWCNT under both exposure conditions (Figure 3.3). The cells exposed to M-7 MWCNTs suspension for 96h showed a weaker F-actin staining, however, this might be due to the high concentration of M-

7 MWCNTs covering the majority of the sample surface. In addition, it can be observed for both exposure scenarios that the epithelial cell layer thickness increased upon prolonged culture time, which indicates an overgrow of the cells.

Figure 3.3: Cell morphology comparing the air-liquid interface to the submerged M-7

MWCNT exposure. Apical, basolateral, and cross-section images were collected at the 24 and

94h time points. F-actin is shown in magenta, the cell nuclei in cyan and the intermediate filaments on the basolateral side in yellow. Scale bar 20 μm.

Cytotoxicity

No statistically significant increase in cytotoxicity was observed for any of the exposure method. Triton-X (0.2%) was applied as positive control resulting in a statistically significant increase in LDH release (Figure 3.4).

Figure 3.4: Cytotoxicity quantification through an LDH assay.

Proteomic Response

The following number of proteins groups were identified for global expression analysis: 2751 proteins in cell samples, 1224 proteins in media samples, and 1896 in wash samples. Of the samples processed, outliers were identified in passage 3, thus only results from passage 5 will be described herein. Due to the multifactorial nature of the experimental design, a regression model was run first to help identify which factors had a significant influence on the resulting proteome expression. Next, a Welch t-test was conducted on the significant factors found in the regression model, followed by enrichment analysis.

Regression Modeling

A regression model was used to identify the key factors driving changes in the proteome in the multifactorial study. Potential effects were examined using basolateral media samples in the regression model Equation 3.1 (Table B2). The following variables were considered: exposure (M-7 versus BSA control), and exposure method (submerged versus

ALI). Exposure method proved to be the only significant effect (p-value < 0.0001) while the

M-7 MWCNT exposure was not significant (p-value = 0.4468).

Equation 3.1: Protein Intensity = Intercept + Exposure Method + Exposure

Welch t-test

According to the regression model, exposure method significantly influenced the global proteome; however, exposure to M-7 MWCNT did not produce a significant effect. To better understand the results of the regression model, the Welch t-test was used to calculate significance for exposures (MWCNT versus control) and exposure method (ALI versus submerged). Table B3 illustrates a breakdown of the limited number of proteins that were found to be significant when testing for exposure. Conversely, a greater number of proteins were found to be significantly different between exposure methods. The common proteins found to be significant for both the MWCNT and control samples for ALI versus submerged are reported by sample collection: 472 apical, 104 cell, and 58 basolateral. The increased number of significant proteins found in the apical comparison are confounded by the fact that samples are comprised from two different sample types (i.e. wash samples from air-liquid ALI exposure, and media samples from the submerged exposure). When only common proteins

are considered from the apical wash and media samples, 173 proteins are found to be significant. These results overall demonstrate a greater amount of significant found between proteomes based on cell culture method (i.e. submerged versus ALI) at low dose MWCNT exposure.

Enrichment Analysis

Differential protein response was examined by calculating differences in protein intensities by M-7 compared to control, and by ALI compared to submerged exposure method.

An example comparison of protein response by differences in exposure method is highlighted in Figure B1. The culmination of these data were next used for enrichment analysis which was analyzed using ingenuity pathway analysis (IPA). Each sample set was uploaded into IPA containing the following data: protein ID, p-value (calculated from t-test), and log2 fold change

(M-7 exposure / matched BSA control, ALI / submerged). Sample sets were processed individually using the Core Analysis search function, and human was specified as the search species. The comparison analysis function was then used to examine enrichment trends for variables found to be significant in the regression models. The results from the comparison analyses were used to generate heat maps in order to show trends upstream mediator enrichment. The color key of the heat map is associated with Z scores, with red being the most significantly enriched p-value, and blue being the least significant [197]. Figure 3.5 shows enrichment analysis demonstrating the apical and basolateral proteome of the submerged media samples upon M-7 MWCNT compared to BSA exposure. The z-score values for the upstream mediators illustrated in Figure 3.5 are reported in Table B4. Figure 3.6 illustrates comparison analysis of ALI versus submerged exposures. The z-score values for the upstream mediators

illustrated in Figure 3.6 are reported in Table B5. Due to the multifactorial nature of the experiment, several combinations of comparisons were examined; however, the discussion herein will focus on significant results found in the enrichment results exhibiting polarized response (Figure 3.5), and exposure method (Figure 3.6).

Figure 3.5: Heat map of top 25 enriched upstream mediators for the apical and basolateral 96 hour media proteome. M-7 MWCNT exposure versus control data was used to general enrichment analysis.

Figure 3.6: Heat map of top 25 enriched upstream mediators for the apical and basolateral 96 hour M-7 exposed proteome. ALI versus submerged exposure data was used to general enrichment analysis.

3.4 Discussion

The aim of this study was to examine various aspects of a 3D tri-culture lung model proteome, comprised of human-originating cells. Differences in proteomic response can be useful in understanding changes in cellular state in response to exposure [204], which are particularly valuable when assessing method development of potential models for regulatory testing. To better examine features of the small airway tri-culture ALI MWCNT exposure, the following variables of the global proteomic experiment will be discussed herein: polarized expression (apical versus basolateral), and exposure method (submerged versus ALI).

Polarized Expression

Epithelium, a continuous layer of cells connected by tight and adheren junctions, are well known to function as a polarized unit that exhibits unique apical (i.e. towards the exterior) and basolateral (i.e. away from the exterior) signaling (Figure 3.1.A) [205]. Polarity is established in epithelial cells by a series of events controlled by protein complexes, which lead to the asymmetric distribution of intracellular organelle, as well as cytoskeleton and membrane trafficking [206-208]. Studies have shown that the epithelial cell interaction with extracellular matrix (ECM) contributes to the basolateral surface configuration, which in turn signals the creation of an apical face on the opposing membrane surface [209, 210]. Furthermore, the dynamic interaction of the cells found in the human alveolar region have been well characterized to include polarized signaling between: epithelial and fibroblast cells [211], as well as epithelial and macrophage cells [212].

A classic example of this polarized response is found in the cellular mechanism of pulmonary fibrosis, a disease characterized by the fatal accumulation of fibroblasts and ECM in the alveolus [213]. A recent adverse outcome pathway (AOP) detailing the mechanisms of pulmonary fibrosis was generated for CNT exposure, which carefully outlined signaling of known mediators between macrophages, epithelial cells, and fibroblasts [58]. Thus, a key feature of the 3D tri-culture, including macrophages, epithelial cells, and fibroblasts, is to better investigate potential enhanced polarized signaling upon MWCNT exposure.

In order to evaluate the tri-culture model as a prototype for the human small airway, samples from apical and basolateral sides of the epithelial cells were processed by global proteomics. A comparison analysis was conducted on enriched upstream mediators which

illustrate a distinct polarized response of the apical versus basolateral media samples collected after 96 hours of MWCNT exposure (Figure 3.5). The upstream mediators found to be enriched in the basolateral comparison of M-7 MWCNT exposure versus control include:

Tumor necrosis factor (TNF), Mothers against decapentaplegic homolog 3 (SMAD3), C-C motif chemokine ligand 5 (CCL5), Transforming growth factor β1 (TGFβ1), and Interleukin

1β (IL1β). Additional details regarding the log2 fold change of proteins associated with these upstream mediators are detailed in Table B6. Furthermore, the upstream mediators found to be significantly enriched are commonly associated with basement membrane stress response signaling to fibroblasts. While there are increased signaling for TGFβ1 and IL1β associated with both apical and basolateral sides, most of the other signaling pathways demonstrate a polarized response, thus indicating differential proteome expression by sample location.

Submerged versus ALI Expression

MWCNT exhibit physical characteristics that require careful consideration when choosing an exposure method. Traditional methods used for in-vitro dosing with MWCNT are performed under submerged conditions [214-218]. Submerged exposures offer some benefit as they are more high-throughput and do not require aerosolizing equipment; however, they do not mimic inhalations as accurately as the aerosolized MWCNT ALI exposure. Differences in

MWCNT physical characteristics, including size and density, will affect how the material deposits under submerged conditions. However, the variability in dosimetry is significantly reduced when MWCNTs are delivered as a dry solid (dust) or in a liquid partible suspension aerosol [68]. Fortunately, recent advances in aerosol technology have successfully aerosolized

MWCNT onto an in-vitro ALI model, thus creating a prototype for more physiologically relevant exposures [219, 68, 220].

Due to increasing awareness of the variation that can arise during inhalation toxicity testing, focus has been given to investigate differences between in-vitro exposure methods

(aerosol exposure of ALI versus submerged culture exposures) [221]. The results from this side-by-side repeat exposure study using submerged and ALI conditions show a significant disparity between the methods by a global proteome response. These differences in exposure method were found in the regression model run on media samples at 96 hours post exposure.

The results indicate there are significant differences in the proteome response based on how the cells are maintained and exposed (p-value < 0.0001). Ultimately, the results indicate there are significant differences in global proteome expression that are unique to how the cell cultures are exposure and maintained.

To better understand the differential proteomic response by exposure method, a heat map of enriched upstream mediators was created using M-7 exposed samples comparing ALI and submerged exposure (Figure 3.6). Interestingly, both apical and basolateral locations exhibit similar trends of expression when comparing exposure methods (ALI versus submerged). Figure 3.6 illustrates enriched signaling pathways that exhibit greater stress response by ALI exposure, including: Phosphatidylinositol-3-kinases (PI3K), Interleukin 1β

(IL1β), Interleukin 6 (IL6), C-C motif chemokine ligand 5 (CCL5), Extracellular signal– regulated kinase ½ (ERK1/2). The details regarding the log2 fold change of proteins associated with these upstream mediators are detailed in Table B7. While these findings are interesting, it is important to consider why these signaling pathways are enriched. The regression model showed limited significance for the MWCNT exposure, thus indicating the results of the

exposure method comparison may be due to general cell culture maintenance. Additionally, variation could arise from the differences in the collected sample matrix. The ALI apical samples were collected in PBS, whereas the submerged apical samples were collected in media. Media is a more complex matrix compared to PBS, resulting in impaired protein identification. Nonetheless, these results provide important implications that should be considered when comparing in vitro methods used for inhalation toxicity testing moving forward.

3.5 Conclusions

Advances in cell culture technology have yielded innovative in vitro assays allowing for the maintenance and exposure of pulmonary cells at an air-liquid interface. Despite emerging technology, researchers still conduct more traditional exposures under submerged conditions. Thus, the aim of this study was to investigate global differential protein response for co-cultures exposure to M-7 MWCNT under ALI or submerged conditions. The proteomics data show a polarized response upon exposure compared to controls, as was expected for the tri-culture experimental setup. Furthermore, differential protein expression was found to be unique to samples collected from either ALI or submerged conditions.

Overall, this study demonstrates proteomic differences in the samples collected from ALI and submerged tri-culture exposures. These data indicant careful consideration should be used when comparing results of in vitro studies conducted under different exposure methods.

Acknowledgments

GH would like to acknowledge Dr. Emily Griffith at NC State University Statistics Department for statistical analysis counsel. The authors acknowledge support from the National Institute of Environmental Health Sciences (NIEHS) Training Grant T32 ES007046 (GH), and NIEHS

P30ES025128 (GH and MB). The authors also acknowledge support from the Peta

International Science Consortium and Adolphe Merkle Foundation (HB and BRH).

CHAPTER 4

Toxicoproteomic Analysis of Pulmonary Carbon Nanotube Exposure using LC-MS/MS

Adopted from publication: Toxicology, Volume 329, pages 80-87, 2015

Gina M. Hilton1, Alexia J. Taylor1, Christina D. McClure2, Gregory N. Parsons2, James C. Bonner1, Michael S. Bereman1

1Department of Biological Sciences, Environmental and Molecular Toxicology, North Carolina State University, Raleigh, NC

2Department of Chemical and Biomolecular Engineering, North Carolina State University, Raleigh, NC

Abstract

Toxicoproteomics is a developing field that utilizes global proteomic methodologies to investigate the physiological response as a result of adverse toxicant exposure. The aim of this study was to compare the protein secretion profile in lung bronchoalveolar lavage fluid (BALF) from mice exposed to non-functionalized multi-walled carbon nanotubes (U-MWCNTs) or

MWCNTs functionalized by nanoscale Al2O3 coatings (A-MWCNT) formed using atomic layer deposition (ALD). Proteins were identified using liquid chromatography tandem mass spectrometry (LC-MS/MS), and quantified using a combination of two label-free proteomic methods: spectral counting and MS1 peak area analysis. On average 465 protein groups were identified per sample and proteins were first screened using spectral counting and the Fisher’s exact test to determine differentially regulated species. Significant proteins by Fisher’s exact test (p<0.05) were then verified by integrating the intensity under the extracted ion chromatogram from a single unique peptide for each protein across all runs. A two sample t- test based on integrated peak intensities discovered differences in 27 proteins for control versus

U-MWCNT, 13 proteins for control versus A-MWCNT, and 2 proteins for U-MWCNT versus

A-MWCNT. Finally, an in-vitro binding experiment was performed yielding 4 common proteins statistically different (p<0.05) for both the in-vitro and in-vivo study. Several of the proteins found to be significantly different between exposed and control groups are known to play a key role in inflammatory and immune response. A comparison between the in-vitro and in-vivo CNT exposure emphasized a true biological response to CNT exposure.

4.1 Introduction

Carbon Nanotubes (CNTs) have been increasingly investigated for a wide range of applications since their characterization in 1991 [222]. Sumio Iijima first described CNTs as

“needle-like tubes” that range from 4 to 30 nm in diameter and up to 1 µm in length [1]. These tubes are now more commonly described as rolled sheets of hexagonal graphite, being either single-walled carbon nanotubes (SWCNTs), one layer, or multi-walled carbon nanotubes

(MWCNTs), greater than one layer. CNTs have been tested and reported to be stronger than steel and a fraction of the weight [223], thus making them ideal for material additives [224].

They are also being investigated in the medical field to serve as a targeted drug delivery vehicle, offering a larger drug pay load and greater bio-availability [225]. CNTs unique physical properties make them ideal for various structural support and medical applications; however, the combination of increased use of CNTs, lack of knowledge of the risks from CNT exposure [46], and the absence of proper safety requirements have all led to a greater potential for adverse occupational or environmental exposure [226].

One of the greatest concerns over the mass production and use of CNTs is that they possess fiber-like characteristics similar to asbestos, thus creating the potential for induction of pulmonary fibrosis and lung cancer upon exposure. Several studies have shown that mice or rats exposed to CNTs by inhalation, oropharyngeal aspiration (OPA), or intratracheal instillation (IT) develop lung inflammation and pulmonary fibrosis [25, 41]. CNTs delivered to the lungs of mice are avidly engulfed by macrophages within the alveolar region [227] which produce a variety of soluble proteins (e.g., cytokines, growth factors, proteinases, extracellular matrix) that play roles in lung homeostasis, tissue repair, or disease pathogenesis [228].

Additionally, the lung epithelium produces critical proteins, such as pulmonary surfactant proteins [229] and myeloperoxidase [230], as an innate defense immune response to foreign bodies. Fibrotic response to CNTs in the alveolar interstitium is caused by the over-production of collagen and other extracellular matrix proteins by myofibroblasts that result in an increased thickness of the connective tissue [37]. Both the fibrotic and immune responses are likely due to the generation of reactive oxygen species from residual metal content from the CNTs and the persistence of CNTs endowed by their fiber-like shape [225].

A variety of post-synthesis chemical engineering modifications have been developed to enhance the unique mechanical and electronic properties of CNTs [122, 123]. Atomic layer deposition (ALD) is a novel process used to apply a highly conformal thin-film coating of oxides, metals, and hybrid metal/organic materials to CNTs to enhance conductivity, photovoltaic or catalytic applications, and attachment of biomolecules [124, 126, 125]. For example, aluminum oxide (Al2O3) and titanium oxide change surface functionality of organic fibers, increase surface hydrophilic properties, and enhance biomolecule attachment [126,

231]. Zinc oxide or titanium oxide coating imparts increased photosensitivity of CNTs for

photovoltaic or catalytic applications [231]. ALD was initially developed for use in the semiconductor industry and has become increasingly used for work at the nano-scale because of the highly uniform thin film coatings that are achieved via a sequence of self-limiting reactions [17, 232, 126]. ALD modified CNTs are being explored in microelectronics for enhancing conductivity, and energy storage applications [232]. In assessing the potential hazards of CNTs on human health, it is therefore important to consider CNTs that are modified by ALD.

Herein, we investigate the variations in protein secretion from mouse lungs following a 24 hour acute exposure to non-functionalized or ALD-functionalized MWCNTs and control via liquid chromatography-tandem mass spectrometry (LC-MS/MS). LC-MS/MS provides a powerful platform to evaluate global changes in the proteome as a response to injury or disease.

In this study, protein abundance was quantified by label-free proteomics using both spectral counts and peak intensities [233, 234]. Finally, we demonstrate the changes in protein expression are the result of specific biological processes through gene ontology enrichment analysis.

4.2 Methods

Materials

Acetic acid, ammonium bicarbonate, sodium deoxycholate (SDC), dithiothreitol

(DTT), iodoacetamide (IAM), formic acid (FA), ammonium hydroxide, hydrochloric acid

(HCl), and bovine serum albumin (BSA) were obtained from Sigma Aldrich (St. Louis, MO).

Sequencing grade trypsin was purchased from Promega (Madison, WI). HPLC grade water,

methanol, and acetonitrile were purchased from VWR International (Morrisville, NC). The

Pierce bicinchoninic acid (BCA) protein assay kit was purchased through ThermoFisher

Scientific (Waltham, MA). Oasis MCX 30 µm particle size solid phase extraction cartridges were obtained from Waters (Milford, MA).

Nanomaterials

MWCNT 0.5 – 40 m in length were synthesized by chemical vapor deposition and purchased from Helix Materials Solutions (Richardson, TX). Characterization of the

MWCNTs was provided by the manufacturer and verified by Millennium Research

Laboratories (Woburn, MA) [134]. MWCNTs were surface coated with Al2O3 via atomic layer deposition (ALD). The aluminum oxide layer is achieved using sequential saturation exposures of trimethylaluminum (Al(CH3)3) and water, separated by inert gas purging steps. Based on the known growth rate for the Al2O3 ALD process on MWCNTs, the nanotubes used in this study had a coating estimated to be ~50 ALD layers (Figure 4.1). The details of ALD coating of carbon nanotubes have been previously described [16].

Figure 4.1: Transmission electron photomicrographs of (A) non-functionalized MWCNTs and (B) ALD Al2O3-functionalized MWCNTs.

Preparation of MWCNTs

Uncoated MWCNTs (U-MWCNT) and aluminum oxide coated (A-MWCNT) were weighed using a milligram scale (Mettler, Toledo OH) suspended in a sterile, 0.1% pluronic

F-68 (Sigma-Aldrich, St. Louis MO) in phosphate buffer solution to achieve the final concentration of 10 mg/ml. Vials containing the suspended nanomaterials were dispersed using a cuphorn sonicator (Qsonica, Newton CT) at room temperature for 1 minute prior to dosing. A limulus amebocyte lysate (LAL) chromogenic assay (Lonza Inc., Walkersville MD) was used to test the nanomaterials for endotoxin contamination. All MWCNTs, both U-

MWCNT and A-MWCNT, tested negative (< 0.3 EU/ml) for endotoxin.

Exposure of Mice to MWCNTs and Processing of Lung Tissue

Mice (C57BL6, Jackson Laboratories) approximately 6 to 8 weeks old were exposed to U-MWCNT and A-MWCNT at 4 mg/kg in 0.1% pluronic surfactant solution or 0.1% pluronic alone for control, via oropharyngeal aspiration while under isoflurane anesthesia.

Three mice per group were evaluated. Mice were euthanized via intraperitoneal injection of

Fatal Plus (Vortech Pharmaceuticals, Dearborn, MI) on day 1 after MWCNT exposure [235].

At necropsy, the lungs were serially lavaged two times with 0.5ml Dulbecco’s Phosphate

Buffered Saline and bronchoalveolar lavage fluids (BALFs) were collected. Lung lavages were centrifuged at 14,000 rpm for 2 minutes, and isolated away from MWCNTs.

In Solution Protein Digestion

Samples were quantified by a BCA protein assay, and a final concentration of 50 mM total ammonium bicarbonate (pH 8.0) was achieved by adding the appropriate amount of

ammonium bicarbonate solution to the BALFs. SDC was added to generate a final concentration of 1% detergent. Disulfides were reduced upon addition of 5 mM DTT and incubated for 30 minutes at 60ºC. Samples were then cooled to room temperature and IAM was added to make a final concentration of 15 mM, and incubated in the dark for 20 minutes at room temperature. Tryptic digestion was achieved by hydrating lyophilized trypsin to a stock solution of 1 µg/µL with 0.01% acetic acid in water. The trypsin solution was added to the protein mixture (i.e. 20 µg protein) in a 1:50 ratio (~0.4 µg trypsin), and then incubated at

37 °C for 4 hours. Following the digestion, samples were acidified with 6 M HCl to make a final concentration of 250 mM (pH ≤ 3) [141]. Sample purification and concentration was achieved using MCX cartridges. Samples were added to the conditioned column after steps 1-

4 (Table C1). After the sample was added to the column, salts were removed with water (0.1% formic acid), neutrals and negatively charged species were removed with 1 mL of methanol

(0.1% formic acid), and then peptides were eluted in 10% NH4OH in methanol. Finally, samples were concentrated down in vacuo (10 Torr) at 45 °C for 2 hours (speedvac concentrator, ThermoFisher Scientific), and then reconstituted in mobile phase A (98 % water,

2 % acetonitrile, and 0.1% formic acid) to yield a final concentration 100 ng/µL.

In-vivo Exposure of Mouse Lung Lavage to MWCNTs

Each of the 9 lung lavage samples (3 controls, 3 U-MWCNT exposures, and 3 A-

MWCNT exposures) were isolated from mice as described in the ‘Exposure of Mice to

MWCNTs and Processing of Lung Tissue’, and then diluted to 175 µL with 50 mM ammonium bicarbonate solution. Sample digestion and purification was achieved as described in the ‘In

solution Protein Digestion’ section. A final sample concentration was 100 ng/µL peptide was prepared for LC-MS/MS.

In-vitro Exposure of Mouse Lung Lavage to MWCNTs

A single lung lavage sample from a control mouse (C57BL6, Jackson Laboratories) was isolated, as described in the ‘Exposure of Mice to MWCNTs and Processing of Lung

Tissue’, and divided into 3 separate aliquots. The sample was quantified and recorded to be

0.2 µg/µL protein concentration. Each aliquot contained 150 µL of 0.1% pluronic solution to yield 30 µg of protein. The designated U- and A-MWCNT samples received 4 mg/kg of respective MWCNTs in 0.1% pluronic surfactant solution, and the control received 0.1% pluronic alone to ensure equal concentrations. Samples were incubated and shaken for 24 hours at 37ºC and 180 rpm (MaxQ4000, ThermoFisher Scientific). After incubation, samples were centrifuged at 14,000 rpm for 2 minutes, and isolated away from MWCNTs. Lung lavages were further diluted with 400 mM ammonium bicarbonate to achieve a final concentration of 50 mM in 200 µL total volume. Each sample pH was checked and recorded to be 8.0. Sample digestion was achieved as described in the ‘In solution Protein Digestion’ section. A final sample concentration was 0.5 µg/µL peptide was prepared for LC-MS/MS.

LC-MS/MS

Nanoflow liquid chromatography (LC) was conducted using the Thermo Scientific

Easy-nLC 1000 Liquid Chromatography system. Pico-frit columns were purchased from New

Objective (Woburn, MA) and packed to a length of 20 cm with reverse phase ReproSil-Pur

120 C-18-AQ 3 µm particles (Dr. Maisch, Germany). The trap was packed in house to a final

length of 3 cm. A 2 µL injection (200 ng total protein) was washed onto the trap at a flow of

2.0 µL/min for 4 minutes. Peptide separation was achieved on the LC using a gradient of mobile phase A (98 % water, 2 % acetonitrile, and 0.1 % formic acid) and mobile phase B (100

% acetonitrile, 0.1 % formic acid). The 90 minute method consisted of an LC gradient with a linear ramp from 2 % B to 40 % B across 70 minutes (2-72 minutes), a ramp and wash at 80%

B (72-78 minutes), followed by equilibration of the column at 0% B (Table C2). Tandem mass spectrometry was performed using a Thermo Scientific Q-Exactive Plus in a top 12 mode where the 12 most abundant precursors were selected for fragmentation per full scan. MS1 and MS2 scans were performed at a resolving power of 70k and 17.5k at m/z 200, respectively.

A dynamic exclusion window of 30 seconds was used to avoid repeated interrogation of abundant species. Automatic gain control was 1e6 and 5e4 for MS1 and MS2 scans, respectively. A quality control BSA digest was run every fifth injection to ensure proper LC-

MS/MS reproducibility. Metrics were monitored in using the Statistical Process Control in

Proteomics algorithm [111].

Database Search

Database searches were conducted using Proteome Discoverer 1.4 and the Sequest hyper-threaded algorithm. Data were searched against the Mus Musculus Swiss Prot protein database (number of sequences: 16657, date accessed: 02/11/2014) [236]. Peptide spectrum matches were post processed using percolator to enforce a peptide spectral match threshold of q value <0.01. The law of strict parsimony was used for protein inference and grouping [146].

Data analysis

The experimental data were exported from Proteome Discoverer as an inclusive tab delimited file containing each protein group ID and spectral count. A protein’s spectral count is the total sum of the number of times peptides belonging to that protein were identified in the experiment. It is a fast method for relative quantification and identification of putative proteins of interest [237]. Missing spectral count data for any protein group (i.e., not identified) was assigned a zero. Each experimental group’s spectral count data was summed by protein group

ID to generate 3 groups of 3 samples each: control, U-MWCNT, and A-MWCNT. We first screened the list of proteins identified using the Fisher’s exact test to calculate p-values based on the combined spectral count of each protein for each pairwise comparison: control versus

U-MWCNT, control versus A-MWCNT, and U-MWCNT versus A-MWCNT. Next, these results were further verified by integrating the summed area under the extracted ion chromatogram of the M, M+1, and M+2 signals of a single unique peptide from each putative protein of interest identified from the Fisher’s exact test using MS1 full scan filtering in Skyline

[238, 142]. Spectral libraries were created directly in Skyline from the .msf files and used in combination with mass measurement accuracy, retention time reproducibility, and isotope dot products to integrate signals corresponding to peptides of interest. Unique peptides for each protein were determined using the background proteome database function in Skyline. The peak area data were log transformed, all zero values were imputed with a value of 10, and p- values were calculated by using a two-sample t-test for each group (Table C3). Proteins that were identified as significant by both spectral count (p<0.05) and peak intensity integration

(p<0.05) were further analyzed by gene ontology enrichment analysis using the DAVID

Bioinformatics Resources 6.7 database [117].

In-vitro experimental data was processed similarly to the in-vivo experimental data, as described above. Analytical replicate spectral count data was summed for each of the 3 samples: control, U-MWCNT, and A-MWCNT. Fisher’s exact test was run for each pairwise comparison.

4.3 Results

In-vivo study

The experimental design for in-vivo mouse exposure to A-MWCNT and U-MWCNT, followed by lung lavage isolation, protein digestion, LC-MS/MS, and data analysis is shown in Figure 4.2.

Figure 4.3 displays volcano plots in which the log2 fold change is plotted against -log10 p-value generated from the Fisher’s exact test. Of the protein groups identified by SEQUEST across the sample set, 68 were significantly different by the Fisher’s exact test (p-value < 0.05) in control versus U-MWCNT, 47 significant by control versus A-MWCNT, and 21 significant by U- versus A-MWCNT. Thus, a greater than 2 fold difference in number of protein groups significant by Fisher’s exact test were detected between the control and each MWCNT- exposed group versus the differences between the both of the exposed groups.

Figure 4.3: Volcano plots of log2 spectral count fold change versus –log10 pvalue calculated by Fisher’s exact test for: (A) U-MWCNT/control, (B) A-MWCNT/control, and (C) U-

MWCNT/A-MWCNT. Proteins plotted as a log2 fold change of 3 and -3 represent protein detection in exposed groups, but not control, and vice versa (respectively). Positive fold change in spectral count data represent up-regulation in exposed groups versus the control.

Shaded areas highlight significance of p < 0.05. Abundant proteins (i.e., human serum

100

albumin) were included in calculations of the Fisher’s exact test, but not shown in plots. α =

Pulmonary surfactant-associated protein-B. β = Myeloperoxidase. γ = Lactotransferrin.

Further examination of the significant proteins identified by spectral counting between the control and U-MWCNT groups were investigated by MS1 peak area analysis in Skyline.

A total of 26 proteins found to be significant by the Fisher’s exact test across all combinations of comparison did not have a unique peptide that could be confidently integrated using the rules outlined in the methods section due largely to low signal to noise, and thus were not included in the analysis of the comparisons. Figure 4.4 illustrates an example of peak area analysis and subsequent t-test results for myeloperoxidase. Of the significant proteins found by Fisher’s exact test that were further analyzed by peak area analysis, 27 proteins were found to be significant (p-value < 0.05) between control versus U-MWCNT, 13 proteins were significant between control versus A-MWCNT, and 2 proteins were significant between U-

MWCNT versus A-MWCNT (Table C3). These results illustrate the more conservative p- value results by peptide peak area analysis compared to spectral counting, which is attributed to a combination of accounting for the variance amongst the biological replicates inherent in the t-statistic, raw spectral counts include possible non-unique peptides, and not every protein had a unique peptide that could be confidently integrated. Figure 4.4.A illustrates peak area analysis of myeloperoxidase, a protein associated with innate immune response, where the protein was not detected in the control, but was detected in both of the exposed groups.

Myeloperoxidase levels across control and exposed groups were not significant in the in-vitro study, and therefore, are more likely to be a true biological response versus a physical interaction of the MWCNTs adhering to protein.

101

Figure 4.4: Peak area analysis was conducted on proteins that were discovered to be significant using the Fisher’s Exact test on the spectral count data. (A) Chromatogram overlay used to assess changes in peak area by group. (B) Peak area was tested for significance across every group using the two-sample t-test: Control versus U-MWCNT (p = 7.02E-08), Control versus A-MWCNT (p = 3.96E-05), and U-MWCNT versus A-MWCNT = 0.679). *Control peak area = 0, but imputed as a value of 10 to allow for log adjustment. Peaks were identified using a combination of MS2 identification (blue lines), mass measurement accuracy (< 3 ppm), dot products of theoretical isotope abundance, and retention time reproducibility.

In-vitro study

To confirm a true biological response, as oppose to a physical interaction, we conducted an in-vitro study to probe for potential protein binding with MWCNTs. Of the approximately

500 proteins groups identified, 11 proteins were statistically significant by Fisher’s exact test

(p-value < 0.05) for control versus U-MWCNT, and 15 proteins for the control versus A- 102

MWCNT as illustrated in the volcano plot in Figure C1. However, only 4 proteins were identified as significant by the Fisher’s exact test in both the in-vitro and in-vivo study. These results indicate that the remaining proteins identified as being statistically significant in the in- vivo study are due to a true biological response and not physical binding of MWCNTs to proteins.

Gene Ontology analysis

Gene ontology (GO) enrichment analysis was used to further classify the biological function of the significant proteins calculated by t-test for the control versus U-MWCNT, and the control versus A-MWCNT in the in-vivo study. Figure 4.5 illustrates the GO analysis results for both biological process and molecular function. Biological processes are 1) considered molecular events with a beginning and end, and 2) have a vital role in living units

(i.e. cells, tissues, organs, and organisms). Molecular function GO term defines the functions of gene product in accordance with their abilities. Of the biological process GO terms identified in the comparison of control versus U-MWCNT, 58 % were identified in the category of response to stimulus, and the remaining 42 % were identified in the category of immune response (Table C4). The significant biological process GO terms identified in the control versus A-MWCNT comparison comprised 75 % related to a response to stimulus, and

25 % an immune response. The results of the GO analysis for the control versus MWCNT exposed groups show enrichment for proteins in both immune response, and response to stimuli.

103

Figure 4.5: GO analysis terms for (A) Biological Process, and (B) Molecular Function for significant proteins by peak area analysis both of the exposed groups versus the control.

Significance (p < 0.05) was calculated by a modified Fisher’s exact test (EASE score) and

104

adjusted by the Benjamini correction and reported as the –Log10 Benjamini p-value.

Significant GO terms are illustrated to the right of the dashed line [117].

4.4 Discussion

The aim of this study was to investigate proteomic changes upon acute pulmonary exposure to coated and uncoated MWCNTs. Of the proteins that were found to be statistically significant by comparing control versus MWCNT exposed groups, several were identified to be associated with an immune response and/or a response to stimuli, including: myeloperoxidase (MPO), lactotransferrin (LTF), neutrophil gelatinase-associated lipocalin

(NGAL), histone H4, pulmonary surfactant-associated protein B (SP-B), and complement (C3,

C4b, and C9) proteins. It has been well established that neutrophils use MPO, a lysosomal enzyme, as an innate immune response to kill microbes by generating hydrogen peroxide [239,

230]. In this study we observed significantly increased levels of MPO expression in both of the exposed groups and no detectable expression in the control, thus indicating a pulmonary immune response to injury upon MWCNT exposure. Interestingly, MPO has recently been shown to degrade MWCNTs and therefore MPO represents a potentially important defense to break down otherwise persistent nanotubes [240]. Moreover, mice deficient in MPO had reduced clearance of CNTs and enhanced inflammation and fibrosis [241]. Our data show that either uncoated or Al2O3-coated MWCNT both induce MPO protein expression equally in the lungs of mice. However, it is possible that the Al2O3 coating might alter MPO-induced degradation of CNTs and this is an important issue to address in future studies.

105

In addition to MPO, LTF, also known as lactoferrin, is present in neutrophil granules and plays an important role to bridge the innate and adaptive immune response by aiding cellular regulation of oxidative stress, as well as, control excess inflammatory response [242].

Like MPO, LTF protein expression was also significantly increased upon both U-MWCNT and A-MWCNT exposure which indicates an up regulated process to control for inflammatory response in the lung. Additional evidence of innate immune response and inflammatory induction by MWCNT exposure was found by the increased expression of NGAL in both U-

MWCNT and A-MWCNT exposure compared to the control, and an increase in histone H4 in the U-MWCNT exposed groups compared to control. NGAL is known to be expressed by immune cells and serves to modulate oxidative stress, and when reactive oxygen species are generated upon exposure, DNA damage can occur and release histone H4 and serve as a marker for inflammatory response [243, 244]. Along with NGAL and histone H4, immune defense systems, such as the complement component system, have been detected for up regulation upon MWCNT exposure. The complement system, containing more than 30 proteins, is considered to play a vital role in innate immune response, and can be activated through three major pathways: classical, lectin, and alternative [245]. Initiation of the classical pathway occurs through antigen/antibody immune complexes, the lectin pathway occurs by lectins recognition of pathogen-associated molecular patterns (PAMPs), and the alternative pathway is activated by interaction with pathogenic surfaces. Our data analysis showed statistically significant differences in complement components C9, C3, and C4b in the U-MWCNT exposed group compared to the control. C9 is a part of the terminal complement component system that serves to initiate membrane attack on certain pathogens and cells. C3 protein acts through the alternative pathway and is broken down by C3 convertase to C3b, a protein that

106

binds complement receptors on phagocytes, and C3a, a mediator of inflammation. Lastly, C4b acts as an opsonin on the surface of targeted cells. All three complement components were up regulated in U-MWCNT exposure compared to the control, thus indicating activation of the complement component system as an innate immune response.

Another key protein group known to induce immune response are the pulmonary surfactant proteins (SP), including: SP-A, SP-B, and SP-D. Surfactant proteins compose a major component of pulmonary surfactant which is essential for maintaining lung homeostasis as it prevents alveolar collapse [246]. Induction of immune response has been shown to be enhanced by SP-B containing antigen vesicles within the airways [247]. Our proteomic analysis demonstrated that of the 3 common surfactant proteins, only SP-B was statistically significant in the comparison of U-MWCNT versus control. Interestingly, SP-B was detected in lower levels for both of the exposed groups compared to the control, thus following the trend previously observed with SP-A and SP-D where the proteins were bound to the CNTs

[229]. The reduction in SP-B levels could be indicative of an immune response whereby the secreted proteins are adhering to the CNTs. However, none of the surfactant proteins were significant by the in-vitro study, indicating that the binding of the surfactant proteins to the

CNTs may be concentration and/or environment dependent.

Comparing the different MWCNT (Al2O3-coated versus uncoated) exposure groups yielded little differences in protein abundances, thus indicating the major response in this specific study was due to CNT exposure. However, more precise absolute quantitative LC-

MS/MS methods are needed to screen a variety of different coatings based on the markers identified herein and will be the subject of future studies.

107

4.5 Conclusion

This study examined changes in protein abundance associated with immune response and inflammation were detected in mice lung BALF after 24 hour exposure to functionalized and non-functionalized MWCNTs. These experiments demonstrated the ability to use LC-

MS/MS to probe for global changes in lung lavage fluid upon pulmonary insult to carbon nanotubes. Differences in protein expression were screened by examining spectral count data and further verified by peak area analysis of a single unique peptide. Investigation of the potential protein interaction with CNTs was conducted as an in-vitro study to examine whether biological changes in protein expression occur upon toxicant exposure versus protein binding with the CNTs. Further protein binding studies are needed to examine potential protein interaction with various functionalized CNTs. Future studies will use protein cleavage isotope dilution mass spectrometry to validate the absolute changes in protein amounts and examine the differences in protein expression across multiple types of MWCNT functionalization. In addition, we will use global proteomic approaches to investigate perturbations of intracellular pathways in relation to carbon nanotube functionalization using model cell lines. Ultimately, putative markers of interest will be used to develop targeted peptide assays based on selective reaction monitoring to determine degree of toxicity of a variety of carbon nanotube coatings.

Acknowledgments

The authors acknowledge support from the NIEHS Training Grant: Molecular Pathways to

Pathogenesis in Toxicology T32 ES007046, the NIEHS R01ES020897, and the NIEHS

RC2ES018772.

108

CHAPTER 5

Multi-Walled Carbon Nanotubes as an Abundant Protein Depletion Material for

Proteomic Sample Digestion

Abstract

Protein coronas readily form in a biological fluid due to the high surface free energy of nanomaterials. Studies have shown that several parameters of the nanomaterial will affect the conformation for the protein corona; including: size, shape, surface charge, solubility, and surface modification. Thus, the effects of protein adsorption onto various nanomaterials has potential implications for enhancing proteomics sample preparation by depleting abundant proteins in order to improve dynamic range. The aim of this proteomics study was to characterize the physiochemical properties of the multi-walled carbon nanotube (MWCNT) protein corona, which have not been previously well characterized. The results show overall increased protein identification by on MWCNT digestion compared to controls. The increase in the number of proteins identified is likely due to a depletion of the top most abundant proteins. Ultimately, abundant protein depletion by MWCNT protein corona formation is a significant application in proteomic sample preparation to better identify more protein species.

5.1 Introduction

Carbon nanotubes (CNTs), first characterized in 1991 as “needle-like-tubes” [1], are now commonly distinguished as single-walled CNTs (SWCNT) or multi-walled CNTs

(MWCNT). Due to the extraordinary electrical conductivity properties of MWCNT, there has

109

been an exponential increase in their manufacturing for applications in a wide array of electronics and engineering [14]. Additionally, CNTs are known to interact with biological molecules, and have been implicated for use in drug delivery and cancer treatment [248-251].

An important feature of CNT use for drug delivery is their ability to be surface functionalized in order to have specific interaction with target tissues and longer residence time in circulation

[252]. The combination of these effects make functionalized nanomaterials an attractive tool to help enhance drug delivery and chemotherapeutic treatments.

Equally important to the use of CNT for drug delivery are the potential consequences nanotoxicity from CNT exposure. CNT are known to potentially cause systemic toxicity primarily by inducing immune response and/or forming aggregates that can create clots [253].

Factors that contribute to the toxicity of CNTs include: dose, SWCNT versus MWCNT, length, catalyst residues from synthesis, degree of aggregation, and synthetic functionalization [217].

Furthermore, biological molecules, such as proteins, bound on the material will also influence the CNTs in-vivo toxicity. Previous toxicity studies on CNTs were primarily focused on the induction of stress response without considering the how the material changes under physiological conditions [254]. For example, the coating of nanomaterial with surfactant proteins resident in the lung are more likely to interact with lipids, thus causing increased susceptibility to respiratory diseases [255]. Thus, nanotoxicity needs to be considered for nanomaterials alone, as well as nanomaterials that are coated with resident biomolecules, i.e. the protein corona.

A protein corona, in its simplest form, is the adsorption of proteins and other small molecules onto the surface of a nanomaterial through non-covalent bonding [256]. Protein coronas readily form in a biological fluid due to the high surface free energy of nanomaterials.

110

The binding forces responsible for the protein adsorption onto the material include: van der

Waals interactions, hydrogen bonds, hydrophobic interactions, electrostatic interactions, and π

– π stacking [257]. Studies have shown that several parameters of the nanomaterial will affect the conformation for the protein corona; including: size, shape, surface charge, solubility, and surface modification [258-261]. To best understand the interaction between nanoparticles and cells, it is necessary to characterize the proteins that make up the protein corona. A comprehensive review from KE Sapsford et al describes several techniques that have been utilized for protein corona characterization; including: UV-vis absorption, X-ray diffraction, microscopy, and mass spectrometry [262]. While each of the techniques listed have their own unique merit, mass spectrometry (MS) is perhaps the most powerful tool that can be used to characterize and quantify the proteins present in the protein corona. Advances in liquid chromatography tandem mass spectrometry (LC-MS/MS) technology have led to investigation of protein corona formation on nanoparticles (NPs) [263-267, 254].

In addition to the significant effects protein corona formation can exhibit in toxicity studies, the effects of protein adsorption onto various nanomaterials has potential implications for enhancing proteomics sample preparation. LC-MS/MS based proteomic sample preparation generally includes depletion of the top abundant proteins to better identify low abundant protein species of interest. The most common method used to deplete top abundant proteins is by immunoaffinity depletion with antibodies [268]. While the immunoaffinity method is generally capable of removing the top 7 – 14 abundant plasma proteins, the depletion kits are costly and non-specifically bind proteins of interest, thus reducing overall protein abundance. Due to these limitations, other methods should be considered for potential abundant protein depletion; such as, protein corona formation.

111

While several studies have used LC-MS/MS to characterize the NP protein corona, limited LC-MS/MS studies have been conducted to evaluate MWCNTs [269, 270]. Of the protein corona experiments conducted on MWCNTs, only cell culture based proteins extracted from media or cell lysate were used to form the corona, hence limiting the degree of how representative the protein mixture is to a true physiological state. Thus, the goal of this study was to characterize the protein corona formed on MWCNTs when incubated with mouse bronchoalveolar lavage fluid (BALF) for the potential depletion of high abundant proteins.

Herein, we report an LC-MS/MS based label-free proteomic investigation of the protein corona physiochemical properties, as well as amount of proteins identified in a MWCNT protein corona compared to BALF control.

5.2 Methods

Materials

Multi-walled carbon nanotubes (MWCNT) were purchased from Helix Material

Solutions, Inc. (Richardson, TX). 129 mice were purchased from Jackson Laboratories (Bar

Harbor, ME). Fatal Plus was purchased from Vortech Pharmaceuticals (Dearborn, MI). The

Pierce bicinchoninic acid (BCA) protein assay kit was purchased through ThermoFisher

Scientific (Waltham, MA). Dulbecco’s phosphate buffered saline (PBS),

Trizma® hydrochloride (Tris-HCl), sodium chloride (NaCl), Ethylenedinitrilotetraacetic acid

(EDTA), sterile pluronic F-68, acetic acid, ammonium bicarbonate, sodium deoxycholate

(SDC), dithiothreitol (DTT), iodoacetamide (IAM), formic acid (FA), ammonium hydroxide, and hydrochloric acid (HCl) were obtained from Sigma Aldrich (St. Louis, MO). High purity

112

nitrogen gas was purchased from Machine & Welding Supply Co. Sequencing grade trypsin was purchased from Promega (Madison, WI). HPLC grade water, methanol, and acetonitrile were purchased from VWR International (Morrisville, NC). Vivacon500® 10K and 30K molecular weight cut off (MWCO) spin filters were purchased from ThermoFisher Scientific

(Waltham, MA). Pico-frit columns were purchased from New Objective (Woburn, MA), and reverse phase ReproSil-Pur 120 C-18-AQ 3 µm particles were purchased from Dr. Maisch

(Entringen ,Germany).

MWCNT Characterization

MWCNTs were synthesized by carbon vapor deposition (CVD) to a “standard” length of (0.5 – 40 µm) as reported by the manufacturer, Helix Material Solutions. The material characterizations were subsequently verified by independent analysis from Millennium

Research Laboratories Inc., Woburn, MA. Detailed analysis and characterization of this specific MWCNT was reported by Ryman-Rasmussen et.al. [134].

Preparation of MWCNTs

MWCNTs were weighed using a milligram scale (Mettler, Toledo OH) suspended in a sterile 0.1% pluronic F-68 phosphate buffer solution to achieve the final concentration of 10 mg/mL. Vials containing the suspended nanomaterials were dispersed using a cuphorn sonicator (Qsonica, Newton CT) at room temperature for 1 minute prior to use for protein corona formation.

113

Lavage Collection

Adult male mice were euthanized via intraperitoneal injection of Fatal Plus. At necropsy, the lungs were serially lavaged two times with 0.5ml PBS and bronchoalveolar lavage fluids (BALFs) were collected. Immediately following BALF collection, all samples were flash frozen using liquid nitrogen and stored at -20 °C until further use.

Protein Corona Formation and Isolation

Prior to mixing lavage proteins with MWCNT, protein quantitation was achieved using the Pierce BCA assay according to manufacturer’s protocol. Equal amounts of lavage protein and MWCNT (50 - 500 µg) were subsequently mixed at 600 rpm for 1 hour at 37 °C using low protein binding 1.5 mL centrifuge tubes. Samples were then cooled to room temperature, and centrifuged for 5 minutes at 12 G. Following centrifugation, the supernatant was carefully removed as to not disturb the MWCNT pellet. The supernatant was placed in a fresh 1.5 mL vile and saved for protein quantitation. Next, the MWCNT pellet was vigorously washed with

500 µL PBS until solution was cloudy without apparent MWCNT aggregates. Then the

MWCNT wash was centrifuged for 5 minutes at 12 G. The wash supernatant was stored in a fresh 1.5 mL tubes, and labeled as ‘wash 1’. The wash step was repeated 5 – 15 times, and the collected supernatants were quantified for their protein amount using the BCA assay.

Following the last wash collection, the MWCNTs were suspended in 100 uL of a 1 % SDC 50 mM ammonium bicarbonate solution (pH 8.0). Overview of experimental design is illustrated in Figure 5.1.

114

Figure 5.1: Experimental workflow. A) Protein corona formation and isolation, and B)

Protein digest followed by LC-MS/MS data collection and analysis.

Experimental Conditions

Experimental conditions for the protein corona isolation were optimized over the course of five separate proteomic experiments. The following parameters were optimized over the course of the five experiments (respectively): 1) starting protein/MWCNT amount, 2) variation in number of washes, 3) variation of wash solution and 4) protein corona formation from lavage proteins versus house dust mite proteins, and 5) repeat testing of experiment 4 using less starting amount of protein (Table 5.1).

115

Table 5.1: Method overview summarizing the starting amount of protein and experimental conditions for each experiment.

Experiment Protein Conditions number Amount (µg) 1 20 Helix MWCNT. PBS wash 3x.

2 20 Helix MWCNT. Increase PBS washing 15x.

3 20 Helix MWCNT. New wash buffer containing: 25 mM Tris- HCL, 1 mM EDTA, 100 mM NaCl, 500 mL HPLC grade water [269]. Washed corona 15x. 4 35 Helix MWCNT. Test corona formation with house dust mite (HDM) proteins and lavage proteins. Wash 10x using PBS. 5 Range of 7 Repeat testing of Helix MWCNT protein corona formation - 15 with HDM proteins and lavage proteins. Wash 5x using PBS.

Sample Preparation for Digestion

The data collected from the BCA assay protein quantitation was used to calculate the sum of the supernatant and wash protein amounts, and then subtracted from the starting amount of protein added to the MWCNT in order to calculate amount of MWCNT bound protein

(Equation 5.1). A control lavage was separately prepared to be the same starting amount of protein in the same concentration with a final volume of 100 µL of a 1 % SDC 50 mM ammonium bicarbonate solution (pH 8.0).

Equation 5.1: MWCNT bound protein (µg) = Starting protein (µg) – Σ wash1-n (µg) 116

Filter Aided Sample Preparation- Protein Digestion

Each sample was adjusted to contain the same starting concentration of protein using

1% SDC solution in 50 mM ammonium bicarbonate. Due to variability in starting amount of protein in the protein corona, all samples were diluted to the smallest amount of protein quantified to ensure equal starting amount of protein for each experiment. For example, if the smallest amount of protein quantified was 10 μg, then each sample would be diluted to 10 μg in 100 µL (i.e., 0.10 µg/µL). Disulfide bonds were reduced by the addition of DTT to make a final concentration of 5 mM and then incubated at 60 ºC for 30 minutes. Samples were then cooled to room temperature, followed by IAM addition to the sample to make a final concentration of 15 mM, and was incubated in the dark for 20 minutes at room temperature.

Each sample was then carried through a modified filter aided sample preparation (FASP) [200].

Briefly, 30K MWCO Vivacon spin filters were conditioned with 20 µL of 1% SDC solution in 50 mM ammonium bicarbonate for 20 minutes before adding sample. The samples were then centrifuged for 15 minutes at 12,000 rpm, and then washed twice with 200 µL 8 M urea and centrifuged for 15 minutes at 12,000 rpm between each wash. Flow through was discarded between each centrifugation step. The samples were then washed twice with 200 µL 50 mM ammonium bicarbonate following the same centrifugation steps detailed in the urea wash. The filters containing the protein samples were then placed in a new collection tube, and a 20 µL

1:50 trypsin solution (µg trypsin:µg total protein) in 50 mM ammonium bicarbonate buffer was added to each filter. Samples were then incubated at 37 °C for 4 hours. Following the trypsin digestion, samples were centrifuged for 15 minutes at 12,000 rpm. An additional 30

µL was added to the spin filter, and centrifuged for 15 minutes at 12,000 rpm. Finally, the

117

samples were acidified using 6 M HCl to a final concentration of 250 mM (< 3.0 pH). The final concentration of each sample was diluted to 0.25 µg/µL [141].

Nanoflow LC

All samples were processed by a discovery based proteomics method using a quadrupole orbitrap (Q Exactive Plus, Bremen Germany). Pico-frit columns were packed to a length of 30 cm with reverse phase ReproSil-Pur 120 C-18-AQ 3 µm particles. The trap was packed in house with reverse phase packing material to a final length of 3 cm. A 2 µL injection of 0.25 µg/µL peptide sample was washed onto the trap at a flow of 2.0 µL/min for 4 minutes.

Peptide separation was achieved on the LC using a gradient of mobile phase A (98 % water, 2

% acetonitrile, and 0.1 % formic acid) and mobile phase B (100 % acetonitrile, 0.1 % formic acid). A 90 minute LC method was used to process all samples. This method consisted of a

90 minute gradient with a linear ramp from 0 % B to 40 % B across 70 minutes (2-72 minutes), a ramp and wash at 80% B (73-78 minutes), followed by equilibration of the column at 0% B

(79-90).

Orbitrap LC-MS/MS

Orbitrap tandem mass spectrometry was performed using a Thermo Scientific Q-

Exactive Plus (Bremen, Germany). A top 12 data dependent acquisition mode (DDA) was used for every full scan where the 12 most abundant precursors were selected for fragmentation. A resolving power of 70,000 and 17,500 at m/z 200 were used for MS1 and

MS2 scans, respectively. A dynamic exclusion window of 30 seconds was used to avoid repeated interrogation of abundant species. Automatic gain control was 1e6 and 5e4 for MS1

118

and MS2 scans, respectively. Samples were run in a randomized fashion, and a BSA digest was run every fifth injection as a quality control to ensure proper LC-MS/MS reproducibility.

Instrument performance metrics were monitored in using the Statistical Process Control in

Proteomics algorithm [111].

Database Search

Database searches were conducted using Proteome Discoverer 1.4 and the Sequest hyper- threaded algorithm. Data were searched against the Mus Musculus Swiss Prot protein database

(number of sequences: 16657, date accessed: 02/11/2014) [236]. Peptide spectrum matches were post processed using percolator to enforce a peptide spectral match threshold of q value <0.01. The law of strict parsimony was used for protein inference and grouping [146].

Data Analysis

The experimental data were exported from Proteome Discoverer as a tab delimited files containing separate documents for peptide and protein group. The peptide file contained the sequence information for each protein, and the protein group file contained information about protein amount; i.e. protein spectral count. A protein’s spectral count is the total sum of the number of times peptides belonging to that protein were identified in the experiment. It is a fast method for relative quantification and identification of putative proteins of interest [237].

Missing spectral count data for any protein group (i.e., not identified) was assigned a zero. The data for each experiment were analyzed for differences between corona and control samples in the following metrics: total number of protein groups, molecular weight, gravy score, and top abundant proteins. All data were analyzed using R version 3.2.2, and Microsoft Excel 2013.

119

The gene ontology (GO) enrichment analysis was generated using the DAVID Bioinformatics

Resources 6.7 database [117].

5.3 Results

Several experiments were conducted in an attempt to isolate and characterize the protein that formed a corona on Helix MWCNTs. Overall, five separate experiments were executed under different experimental conditions with the goal of optimizing a protocol for protein corona isolation. Despite variation in experimental conditions, every experiment identified a greater number of proteins in the corona sample compared to the lavage control.

The results described herein provide a detailed comparison of proteins identified in the corona versus lavage control.

BCA Assay

The BCA assay was used to quantify protein amount by first creating a standard curve using the manufactures protocol, and then extrapolating protein amount based of the measured protein absorbance (560 nm). Figure 5.2 shows an example BCA assay for each wash isolated from experiment number 5. There was no detectible protein in any of the samples after the 3rd wash. This trend was similarly found for each experiment. Thus, conducting 5-10 washes ensures the proteins found in the corona samples are from the proteins on the MWCNT, and not residual proteins in the test tube.

120

Figure 5.2: Protein quantitation by BCA assay A) Standard curve, and B) Protein concentration measured in protein corona washes.

Amount of Proteins Identified

The data for each experiment was first analyzed by comparing the total protein group count for the protein corona compared to its respective lavage control. On average, each protein corona digestion on the MWCNT produced 200-400 more protein groups compared to the lavage control (Figure 5.3). These preliminary results suggest more efficient digestion of proteins on a MWCNT compared to a simple lavage solution.

121

Figure 5.3: Number of protein groups identified in the protein corona versus the control lavage for each LC-MS/MS experiment.

Gravy score

Gravy scores were calculated and plotted to better characterize potential differences in peptides between the protein corona compared to the control samples (Figure 5.4). The gravy score is a measure of a peptides hydrophobicity, with a positive score meaning the peptide is more hydrophobic, and a negative score meaning the peptide is more hydrophilic. As shown in Figure 5.4, there are no significant differences in gravy score comparing the protein corona peptides by the control peptides for each experiment. This indicates that the overall physical properties of the peptides bound to the MWCNT are not contributing to the amount of proteins being identified in each experiment.

122

Digestion Efficiency

The protein digestion in each experiment was achieved using trypsin, which is an enzyme known to cleave at the amino acids arginine and lysine. To assess digestion efficiency, missed cleavages were compared for all peptides in the protein corona and the control lavage

(Figure 5.5). The results show that on average, the protein corona peptides have a higher ratio of missed cleavages (number of missed cleavages / total number of peptides) compared to the control. While there are increased missed cleavages in the protein corona digestion, both data

123

sets of missed cleavages are within an acceptable range of known digestion efficiency of the trypsin enzyme [271].

Figure 5.5: Ratio of trypsin digest peptide missed cleavages by total peptide count for the protein corona versus control peptides.

Top Abundant Proteins

The top abundant proteins found within proteomic samples can be potentially problematic because they inhibit the instrument from detecting less abundant species [268].

Thus, the top 10 most abundant proteins were compared between the lavage control and the protein corona data (Figure 5.6). These plots highlight a significant feature of the protein corona samples in that their top 10 most abundant proteins are lower abundant than the control lavage across every experiment. For example, Figure 5.6.D shows the serum albumin protein abundance is 396 spectral counts in the control sample compared to 108 in the protein corona sample. Serum albumin and serotransferrin both exhibit a 2 to 4 fold increase in abundance in

124

control compared to corona samples across each experiment (Table D1). The overall trend in the top 10 abundant protein were similar across every experiment.

Figure 5.6: Top 10 most abundant proteins found in the control versus the protein corona.

5.4 Discussion

Multiwalled carbon nanotubes (MWCNT) are being increasingly investigated for their potential lung toxicity; however, limited proteomic studies of the MWCNT protein corona have been reported. The aim of this study was to mimic physiological conditions of MWCNT inhalation exposure by incubating mouse lung lavage with MWCNT, followed by the subsequent evaluation of protein corona formation compared to a lavage control. The following measures were collected to characterize and quantify the proteins that comprise the

125

protein corona: protein amount by the number of protein group IDs, gravy score, digestion efficiency, abundant protein comparison, and gene ontology (GO). One of the most striking trends of these results is the increased protein identification in each protein corona experiment compared to their respective controls (Figure 5.3). After finding this result upon repeated experimentation under varying conditions, further analysis was conducted to try and ascertain why more proteins are routinely being identified in the protein corona.

Naturally, the physicochemical properties of the peptides associated with the protein corona were first analyzed in order to evaluate trends in hydrophobicity. The MWCNTs used are uncoated, and thus generate a more hydrophobic surface, thus creating the potential for protein corona formation of more hydrophobic proteins. The degree of hydrophobicity of peptides is measured in gravy score, with a more positive number being more hydrophobic.

The histogram of gravy score shown in Figure 5.4 demonstrate similar distribution patterns for the protein corona proteins compared to the control lavage. Thus, the protein corona is being formed by proteins that are both hydrophobic and hydrophilic. Ultimately, there is a possibility that the interior proteins interacting with the material are more hydrophobic, and the outer shell of the corona in comprised of more hydrophilic proteins. However, our experimental design was to simply digest all of the proteins off of the MWCNT without distinguishing outer and inner corona proteins (i.e. soft or hard corona, respectively).

The evaluation of digestion efficiency was also a primary interest for this experiment because the protein corona proteins were digested off of the MWCNT, thus potentially hindering or enhancing digestion. Digestion efficiency can be estimated as a ratio of peptides with 1 or more missed cleavages to total number of peptides. The overall results shown in

Figure 5.5 demonstrate that on average, proteins digested off the MWCNT have a higher ratio

126

of missed cleavages compared to the control. While the protein corona peptides have a higher ratio of missed cleavages, the rate of missed cleavages is still within an acceptable range.

Trypsin is known to miss cleave under the following conditions: proline precedes lysine, lysine and arginine start in position 1 (of the peptide sequence), and if negatively charged amino acids like glutamate and aspartate surround the cleavage site; thus leading to 10 – 30% missed cleavages [271]. Therefore, these results do not demonstrate a significant contribution of digestion efficiency to the overall trend of increased number of protein groups in the protein corona samples compared to controls.

An additional consideration for factors leading to increased protein identification in the protein corona data was to evaluate the top abundant proteins compared to the controls.

Abundant proteins have long been a deleterious factor in proteomic data collection due to their influence on dynamic range. Samples that contain abundant protein, such as serum albumin, hinder protein identification of lower abundant proteins because LC-MS/MS sampling is bias towards higher abundant proteins. Furthermore, albumin is highly abundant in biological fluids, thus presenting a common problem for preparation of these types of samples. Thus, the depletion of abundant proteins is highly desirable because it will ultimately lead to increased identification of lower abundant proteins that are more likely to serve as biomarkers of exposure or disease. The results shown in Figure 5.6 demonstrate that each experiment consistently produced a reduction in serum albumin and serotransferrin, proteins that researchers routinely deplete in order to enhance the dynamic range for low abundant protein detection in LC-MS/MS based proteomic experiments (Table D1) [272]. The depletion of albumin likely occurred through the Vroman effect, which states that sequential binding patterns can occur such that the more abundant and mobile proteins may initially adsorb to a

127

surface, and then subsequently be replaced by other proteins that have a higher binding affinity

[273, 274]. Therefore, highly abundant and mobile proteins may have initially bound to the

MWCNT and then were displaced into the PBS wash, resulting in a protein depletion via on material digestion.

5.5 Conclusion

Overall, the results of this proteomics study served to characterize the physiochemical properties of the MWCNT protein corona, which have not been previously well characterized.

The results show overall increased protein identification by on MWCNT digestion compared to controls. The increase in the number of proteins identified is likely due to a depletion of the top most abundant proteins. Depletion of abundant proteins will lead to enhanced identification of lower abundant species that are likely of interest for biomarker discovery.

Ultimately, abundant protein depletion by MWCNT protein corona formation is a significant application in proteomic sample preparation. Additional studies are underway to more carefully characterize the both the proteins found in the corona and the washes.

128

CHAPTER 6

Conclusion

6.1 General Conclusions

In summary, the research detailed in this dissertation set out to compare different methods used for CNT exposure through the use of global proteomics. More specifically, LC-

MS/MS based global proteomics was used to conduct toxicity testing through in-vivo, mono- culture in-vitro, and co-culture in-vitro under submerged and air-liquid interface (ALI) conditions. Our original hypothesis stated that the in-vitro co-culture method would likely produce a proteomic response to CNT exposure that was more similar to in-vivo models compared to in-vitro mono-culture protein expression. The results for each of these CNT exposure methods suggest there is some validation to the original hypothesis when comparing the ALI in-vitro co-culture to the in-vivo mouse model due to the increased proteomic response of mediators involved in cytokine and chemokine signaling; including: IL-1β, IL-6, IL-8 and

CCL5. While the ALI in-vitro study produces signaling similar to commonly measured cytokines for in-vivo studies, the results found in the co-culture study were likely due to the cell-culture setup and maintenance. Increased cell-cell signaling is likely to be measured when the cultures are grown in the presence of more than one cell type (similar to what is found in the in-vivo models). The increased cytokine signaling can also be attributed to the co-cultures being maintained at an ALI, conversely to the cultures being exposed under submerged conditions (i.e. the epithelial cells will have increased signaling towards differentiation compared to the submerged cultures). There was limited significance found in the MWCNT

129

exposure, which is likely due to low levels of material dosing. Thus the similarities in cell expression between the in-vitro co-culture and the in-vivo models can be possibly be attributed to the re-creation of a similar physiological environment, including an ALI culture of macrophages, epithelial cells, and fibroblasts. Furthermore, when comparing all three methods, the overarching trend found to be in common across the CNT exposure proteomic response was the induction of oxidative stress and inflammatory response (i.e. increased expression of mediators found in the following signaling pathways: Interleukin-1 beta, NRF2- mediated Oxidative Stress Response, thrombospondin-1, Mitochondrial Dysfunction,

Apoptosis, Acute Phase Response, Complement System Activation, and Integrin signaling)

(Figure 6.1).

130

While some mediators were found to have similar trends of expression across each exposure method, the cellular response to CNT exposure was highly dependent on exposure material, and the sample matrix to be analyzed (i.e. media, PBS, cells, or lavage). One of the greatest limitations in the comparison of these experiments is the variability in materials used for MWCNT exposure. Futures studies to cross compare exposure methods should be ideally conducted using the same batch of CNT from one manufacturing source. In addition to the variability found from differences in MWCNT used for exposure, there are likely other source of variation in global proteomic data, including: A) differential protein expression and secretion by cell type, B) interspecies extrapolation, C) differences in cell-cell communication,

D) differences in functional and mechanical supports, E) differential sample matrix composition, F) instrument stability, and G) exposure technique. Considering the vast sources of variability in nanomaterial exposure and data collection, it will be critical for the nanotoxicology community moving forward to adopt a standardized exposure method and exposure model.

6.2 Potential Applications

Despite the variability found in proteome response to CNT exposure across different exposure methods, the proteins repeatedly identified in each experiment could be used to develop a multiplexed targeted mass spectrometry assay to enhance traditional protein measurement techniques (i.e. Western blot and ELISA). A targeted mass spectrometry based proteomics method could be used to develop a multiplexed assay that is both sensitive, higher- throughput (i.e. measure 25-50 proteins in one assay compared to 1 protein per assay) than

131

traditional protein quantification methods in order to more rapidly measure cellular response to CNT exposure. Candidate proteins that could be used for a targeted multiplexed assay to assess early pulmonary oxidative stress and inflammatory response are shown in Table E1.

Additionally, the global protein endpoint data collection from these different CNT exposure methods provides valuable information on potential novel mediators that have not been previously considered through traditional toxicity testing. Thus, a global approach to toxicity testing can provide greater insight into cellular response that was not previously considered. These data are critical for the development of new non-test approaches. Non lab- based test approaches (i.e. computational algorithms), are quickly gaining interest in the toxicology community due to their toxicity testing benefits, including: high-throughput, low cost, and minimize animal testing. The most notable non-test approaches being developed are read-across and quantitative structure activity relationship (QSAR). QSAR utilizes known physical properties of a chemical structure and establishes a correlation with experimentally collected biological response data. Systems are built to assess correlation for known chemical structures and toxicity data that are then used to predict exposure response to chemicals with similar physical characteristics. In addition to QSARs, read-across is a powerful data filling technique that serves to fill data gaps in toxicity testing. Read-across technology utilizes data collected from chemicals with known structure and function to impute missing data for chemicals with similar physical properties.

Thus, systems data (i.e. transcriptomic and proteomic) collected from exposures to thoroughly characterized chemicals and materials are highly advantageous to fill data gaps in high-throughput data curation in regulatory testing. Ultimately, the data reported in this dissertation would be of significant use is a QSARs and read-across application due to the well

132

characterized material and exposure methodologies, as well as the global data generated to highlight known and novel mediators.

133

APPENDIX A

Supplemental Figures and Tables: Chapter 2

Figure A1: Protein Log2 peak area fold change plotted for DDA collected data versus SRM collected data. Pearson correlation coefficient r = 0.9635.

134

Figure A2: Boxplot for A) un-normalized and un-filtered peak area, and B) normalized and filtered peak area for each sample.

135

Table A1: Table detailing hydrodynamic diameter and zeta potential for each MWCNT.

Hydrodynamic pDi Zeta Potential Diameter (nm) (mV)

U-MWCNT 567 ± 120 0.58 9.2 ± 1

A-MWCNT 416 ± 19 0.4 10.1 ± 2

Z-MWCNT 192 ± 44 0.57 10.3± 1

136

Table A2: LC methods for sample run on the orbitrap (method A), and triple quadrupole

(method B).

Method Time Duration Flow (nL/min) %B A 0 N/A 300 0 A 2 2 300 0 A 180 180 300 40

A 183 1 300 80 A 193 10 300 80 A 194 1 300 0 A 195 1 300 0 A 225 30 300 0

B 0 N/A 300 0 B 2 2 300 0 B 62 60 300 40 B 64 2 300 80 B 69 5 300 80

B 70 1 300 0 B 75 5 300 0

137

Table A3: Comparison of protein expression trends for the DDA mode orbitrap peak area results versus the SRM mode triple quadrupole results.

Comparison ID Fold_SRM Fold_DDA pvalue_SRM pvalue_DDA A100vsControl P35441 5.202158173 7.319114857 0.000441025 2.52994E-05 A100vsControl P26040 0.526619128 0.562015768 0.005270549 0.02480655 A100vsControl Q06185 1.104553967 1.59072873 0.204541591 0.030449986 A100vsControl P62897 0.817317796 0.679531549 0.132624318 0.066388406 A100vsControl Q9CQA3 1.197332503 1.20167562 0.056885838 0.007509511 A100vsControl P19096 0.968878133 0.851630095 0.364695163 0.008671923 A100vsControl P32261 1.293445993 2.358265265 0.127086565 0.140681559 U100vsControl P19096 0.956033233 0.693175812 0.299122435 0.002873122 U100vsControl Q9JII6 0.884154161 0.542154849 0.032114604 0.023491967 U100vsControl P51174 0.933036918 0.702650926 0.322615017 0.056310435 U100vsControl Q61753 0.825947895 0.780445969 0.070403048 0.167530099 U100vsControl Q9JLZ6 1.227798368 1.26242133 0.010826247 0.007844849

U100vsControl P12382 1.362534069 1.495013123 0.021971169 0.011791488 U100vsControl Q93092 0.894772718 0.876365455 0.137869537 0.716330911

U100vsControl Q64435 0.903597827 0.716026605 0.191823208 0.577561499 U100vsControl O09131 0.792518743 0.743815716 0.024720653 0.6023575

Z2.5vsControl P14901 4.291721858 9.120140069 1.70988E-07 2.53943E-05 Z2.5vsControl Q61699 1.698008991 1.764286451 2.15023E-05 0.00549269

Z2.5vsControl Q61881 1.268585208 1.636452186 0.000283168 0.001732942 Z2.5vsControl P11276 0.494496432 0.502634005 0.000229057 0.005144942

Z2.5vsControl Q6ZQM8 0.792021566 0.748852 0.016893703 0.014020487 Z2.5vsControl P51174 0.95183901 0.765782347 0.519448352 0.030550026 Z2.5vsControl Q6URW6 0.930799559 1.443938534 0.161363136 0.141789805 Z2.5vsControl Q9CPY7 0.692666392 0.714507795 0.019494977 0.014048865 Z2.5vsControl P47738 0.890218335 0.822874796 0.049235639 0.025582211 Z2.5vsControl Q9DC69 0.883559867 0.621160274 0.084610214 0.058764268 Z5vsControl P14901 3.702880412 6.879653947 0.017948036 0.001488353 Z5vsControl Q61699 1.673089634 1.732922401 0.049531921 0.022713651 Z5vsControl Q9CPY7 0.446609757 0.463763081 0.067373611 0.044591074 Z5vsControl P11276 0.459962144 0.501489957 0.087463976 0.103224821 Z5vsControl P51174 0.942242957 0.801012782 0.610486479 0.227938847 Z5vsControl P47738 0.877522989 0.806726229 0.146934642 0.166066684 Z5vsControl Q6ZQM8 0.772863984 0.859948197 0.023509147 0.611808755 Z5vsControl Q61881 1.174232233 1.204531223 0.055791972 0.477511644 Z5vsControl Q9DC69 0.848689422 0.880244661 0.130789352 0.948334788 Z5vsControl Q6URW6 0.890600093 0.961155314 0.358599337 0.643364908

r = 0.963457932227356

138

Table A4: List of significant two-sample t-test p values and log2 fold change for each

MWCNT ‘high’ dose exposure compared to control.

ID Gene Pvalue A-MWCNT vs C Log2_Fold_Change Q8BK67 Rcc2 3.03714E-07 -9.1541059 P84104 Srsf3 3.12182E-07 -8.920104274 Q6PDM2 Srsf1 1.51476E-06 -9.604612677 Q02105 C1qc 2.79063E-06 7.870151884 Q05D44 Eif5b 2.88434E-06 -6.992767387 Q923D5 Wbp11 7.02668E-06 -5.400572422 P26369 U2af2 7.21236E-06 -8.143601375 P02104 Hbb-y 1.14712E-05 11.34365293 Q9Z1T1 Ap3b1 1.52136E-05 -7.071559826 O35326 Srsf5 1.5666E-05 -8.566315493 P02088 Hbb-b1 1.94706E-05 10.96710996 P25976 Ubtf 2.12048E-05 -6.283188028 O08784 Tcof1 2.25025E-05 -7.412244688 P35441 Thbs1 2.52994E-05 2.731460287 Q8BL97 Srsf7 2.57316E-05 -8.344960866 Q9JIX8 Acin1 2.73677E-05 -6.482623953 Q3TEA8 Hp1bp3 3.97776E-05 -7.298142658 Q04750 Top1 5.41025E-05 -5.550867393 Q8BTI8 Srrm2 5.9604E-05 -6.403636411 Q9CQ75 Ndufa2 8.97723E-05 0.394569608 Q05793 Hspg2 9.29867E-05 -1.254978965 Q9DAW9 Cnn3 9.56023E-05 -0.228301628 P62996 Tra2b 0.000134638 -7.333742976 Q9D6Z1 Nop56 0.000146486 -2.579007603 P11276 Fn1 0.000148624 -1.220622552 Q91VD9 Ndufs1 0.000151632 0.923189472 P70698 Ctps1 0.000160837 -0.743254166 Q9CPX8 Uqcr11 0.000199185 2.237284661 Q4KML4 Abracl 0.00038829 -0.522278004 Q569Z5 Ddx46 0.000407101 -5.216403953 Q99NB9 Sf3b1 0.000593021 0.434247966 Q60865 Caprin1 0.00068852 0.468660568 Q9D8T2 Gsdmdc1 0.000818172 -0.300535669 P47911 Rpl6 0.000827187 -0.724830227 P24369 Ppib 0.000894071 -2.629114933

139

Q03265 Atp5a1 0.000913184 0.167500461 Q8BMF4 Dlat 0.000938701 0.500700232 Q920B9 Supt16h 0.000947821 -3.752477533 Q60817 Naca 0.001046895 -1.085100611 P31001 Des 0.001112403 0.592320828 Q62093 Srsf2 0.001242089 -9.107807081 Q8K019 Bclaf1 0.001491761 -6.809399417 P02802 Mt1 0.001567196 -0.469770709 P15864 Hist1h1c 0.001591698 -2.962572421 P43277 Hist1h1d 0.001591698 -2.962572421 P62754 Rps6 0.001693462 -0.609148571 P43274 Hist1h1e 0.00169749 -2.979215117 Q08943 Ssrp1 0.001803768 -5.320506955 Q9QYR6 Map1a 0.001822152 -6.681886748 Q9D883 U2af1 0.002318444 -6.321329277 Q6NVF9 Cpsf6 0.002369215 -1.299790305 O54774 Ap3d1 0.002403089 -5.794008606 Q8VE97 Srsf4 0.002433922 -7.795167544 P83940 Tceb1 0.002988331 -0.342673809 Q52KI8 Srrm1 0.003202888 -5.925118825 Q00612 G6pdx 0.003427891 -0.271953194 Q80UM3 Naa15 0.003436453 -1.587646065 Q80XI3 Eif4g3 0.00350215 2.049730197 B2RY56 Rbm25 0.003690476 -4.240387839 P83882 Rpl36a 0.003696614 -0.68302128 P02798 Mt2 0.003799069 -0.485227303 Q8BG05 Hnrnpa3 0.003973069 0.264606189 Q60875 Arhgef2 0.00404761 -1.185588872 P62962 Pfn1 0.004366383 -0.14996027 Q61576 Fkbp10 0.004725386 0.608265733 Q99JY0 Hadhb 0.00475802 1.780864422 Q9CRB2 Nhp2 0.005353673 -5.866706342 Q6DID3 Scaf8 0.005627923 -4.379117933 P15331 Prph 0.005848328 0.358038038 Q9Z130 Hnrnpdl 0.006080211 0.257106982 Q9JIK5 Ddx21 0.006334596 -3.165690968 Q9JLZ6 Hic2 0.006354617 0.33138049 Q5SUF2 Luc7l3 0.006770076 -3.686357992 P20152 Vim 0.006970712 0.163917279 P12388 Serpinb2 0.007080186 -0.681484097 Q9EPU0 Upf1 0.007214337 0.108197641 P62849 Rps24 0.007323101 -0.578506629 140

O88569 Hnrnpa2b1 0.007456627 0.243530806 Q9CQA3 Sdhb 0.007509511 0.221058511 Q9DBG6 Rpn2 0.007605833 0.449939413 O08709 Prdx6 0.007907714 -0.435734572 P10649 Gstm1 0.00826285 -0.310648341 O54734 Ddost 0.008326721 0.367644878 Q61543 Glg1 0.00844739 0.403609185 P70349 Hint1 0.008635084 -0.284862014 P19096 Fasn 0.008671923 -0.294319463 Q9CPR4 Rpl17 0.008914267 -0.679321442 Q922Q8 Lrrc59 0.008931284 -1.172263743 Q7TNC4 Luc7l2 0.009003582 -6.155024681 P62317 Snrpd2 0.009139031 -0.484626887 Q8BT60 Cpne3 0.009492596 -5.908568359 Q9JKB3 Ybx3 0.009578719 -0.800129692 P70372 Elavl1 0.009723667 1.417057123 P62242 Rps8 0.009790509 -0.547214046 Q99KI0 Aco2 0.00991721 0.18978425 Q8CI51 Pdlim5 0.010515026 -0.391045589 P68369 Tuba1a 0.010684982 -0.220164077 P68373 Tuba1c 0.010684982 -0.220164077 P27659 Rpl3 0.011777749 -0.352146641 Q9JMH6 Txnrd1 0.011814021 -0.31944531 P61358 Rpl27 0.011933521 -0.394090802 P10605 Ctsb 0.01205163 0.353753645 P26443 Glud1 0.012284119 0.243505276 Q8QZY1 Eif3l 0.012705653 0.228797787 P43276 Hist1h1b 0.012851167 -3.133681749 Q8K4Z3 Naxe 0.013292031 -1.647832643 P62960 Ybx1 0.013502259 -0.766790542 Q91YN5 Uap1 0.013506347 -0.283267606 P45376 Akr1b1 0.013592545 -0.562727104 O09167 Rpl21 0.014028895 -0.184557558 Q00519 Xdh 0.0140377 -0.253661853 P48036 Anxa5 0.014082838 -0.177844874 P62830 Rpl23 0.014191809 -0.286450232 Q9Z2N8 Actl6a 0.014288812 2.219882123 Q60716 P4ha2 0.014292983 0.703942526 Q02819 Nucb1 0.014343313 0.518724306 P54227 Stmn1 0.014464742 -0.507558016 Q61074 Ppm1g 0.014811234 -0.967262727 P17710 Hk1 0.015002292 0.245129276 141

Q64511 Top2b 0.015040547 -4.543935742 P48678 Lmna 0.015309496 0.24497907 O70493 Snx12 0.015505005 -0.474733347 Q78XF5 Ostc 0.016090082 0.386710122 Q7M6Y3 Picalm 0.016290101 1.14728714 Q9D0J4 Arl2 0.016488711 -0.46011405 Q99PT1 Arhgdia 0.01764865 -0.407645618 Q5SSL4 Abr 0.017684524 -4.014499668 P51863 Atp6v0d1 0.018455918 0.623052255 P50544 Acadvl 0.018593274 0.496830305 P99024 Tubb5 0.018734377 -0.221110862 Q9D1M0 Sec13 0.018829434 0.349185879 P62983 Rps27a 0.019421044 -0.404176302 Q6DFW4 Nop58 0.019647348 -2.46793959 P47963 Rpl13 0.019686673 -0.373041295 Q91V92 Acly 0.02035018 -0.289611238 Q91V61 Sfxn3 0.020388105 0.266828029 Q922D8 Mthfd1 0.020677451 -0.346228244 P17225 Ptbp1 0.021213174 0.15670549 P46978 Stt3a 0.021399616 0.504063307 Q8R180 Ero1a 0.02141526 0.620664054 O88447 Klc1 0.022002906 -0.607555822 Q9WV54 Asah1 0.022323876 0.858795769 Q9DB77 Uqcrc2 0.022700602 0.273792712 Q08509 Eps8 0.023026207 0.865866344 Q60932 Vdac1 0.023722676 0.21738327 Q8BP67 Rpl24 0.023867506 -0.500315073 Q64727 Vcl 0.024200538 -0.148258178 O08539 Bin1 0.024276546 0.605299483 P62918 Rpl8 0.024331937 -0.390445077 Q6ZWV7 Rpl35 0.024417266 -0.766304033 Q922K7 Nop2 0.024419156 -1.363022602 P26043 Rdx 0.024804802 -0.89348496 P26040 Ezr 0.02480655 -0.893792834 P58281 Opa1 0.024898808 0.719902097 Q9DBJ1 Pgam1 0.025054834 -0.567231312 Q8BH59 Slc25a12 0.02506883 0.380250572 Q3TW96 Uap1l1 0.025810084 -0.348352765 Q9Z0K8 Vnn1 0.025851352 0.393149259 P39688 Fyn 0.026143443 0.838043275 Q91VH6 Memo1 0.026201418 -1.164054231 P35278 Rab5c 0.026271782 0.242439213 142

Q7TQI3 Otub1 0.02653563 -1.670982782 Q9DBR7 Ppp1r12a 0.027193695 -0.521235707 O35382 Exoc4 0.027510398 -0.45696927 P68368 Tuba4a 0.028555449 -0.171324716 Q64433 Hspe1 0.028913993 0.175710228 P21981 Tgm2 0.029286995 -0.601427797 P12787 Cox5a 0.029801639 0.441329473 Q99PL5 Rrbp1 0.029840861 -1.699697355 Q60715 P4ha1 0.030036243 0.318922673 Q06185 Atp5i 0.030449986 0.632668674 Q9DCU6 Mrpl4 0.030797543 1.649055113 Q9R0X4 Acot9 0.030953891 0.332459628 Q80WS3 Fbll1 0.031325957 -0.380241375 Q99LC5 Etfa 0.032246304 0.161360984 P62267 Rps23 0.032414442 -0.643557163 Q91WK0 Lrrfip2 0.032670244 1.440704957 Q91ZX7 Lrp1 0.032672527 0.260071287 Q9WTM5 Ruvbl2 0.033084033 0.285745175 Q99K48 Nono 0.03347476 0.285809349 Q60710 Samhd1 0.033597289 -0.90663613 P26041 Msn 0.033656854 -0.843159118 P45591 Cfl2 0.034272187 -0.186237518 P68372 Tubb4b 0.035214567 -0.177119049 Q9DB20 Atp5o 0.03545883 0.498542026 Q9CR68 Uqcrfs1 0.036732059 0.690607963 Q62351 Tfrc 0.036735557 1.328376775 Q9CQX2 Cyb5b 0.036816523 2.115826719 P45952 Acadm 0.037860152 1.607207254 P61294 Rab6b 0.037950692 -0.621618779 P62259 Ywhae 0.038377179 -0.113283176 O35350 Capn1 0.038509294 -0.476395604 O70325 Gpx4 0.038992106 -0.313735246 Q9QXX4 Slc25a13 0.040013678 0.427531149 Q3V3R1 Mthfd1l 0.040271663 0.247914351 P26039 Tln1 0.040918472 -0.129135997 Q6NZJ6 Eif4g1 0.041003792 -0.3413431 P12970 Rpl7a 0.04183092 -0.47086332 Q9R1Q7 Plp2 0.042562072 2.675253649 Q922W5 Pycr1 0.043828607 0.34163411 Q9WTI7 Myo1c 0.044478849 -0.298460657 P38647 Hspa9 0.044528812 0.139933736 P07901 Hsp90aa1 0.044890622 -0.075776358 143

Q9Z2W0 Dnpep 0.046670692 0.698310861 Q9CYR0 Ssbp1 0.046733675 0.755133661 P15626 Gstm2 0.046914119 -0.265586123 Q8VIJ6 Sfpq 0.048269451 0.255919098 O08917 Flot1 0.048317726 0.646078317 P62320 Snrpd3 0.048501907 -0.427220647 P27046 Man2a1 0.049252728 0.257702143

ID Gene Pvalue Z-MWCNT vs C Log2_Fold_Change P02088 Hbb-b1 1.50678E-07 12.07784859 P01902 H2-K1 3.16727E-05 -5.599349911 P02104 Hbb-y 0.000139544 11.82496261 Q920B9 Supt16h 0.000947821 -3.752477533 P97310 Mcm2 0.001225238 0.415011232 P14901 Hmox1 0.001488353 2.844241665 Q8VC30 Tkfc 0.002045909 -0.471981095 Q80XI3 Eif4g3 0.002317679 2.240745538 P97493 Txn2 0.002773729 0.352486188 P70168 Kpnb1 0.004317782 0.199377367 P61514 Rpl37a 0.005581973 0.350019943 P47955 Rplp1 0.006153796 0.460880239 Q64337 Sqstm1 0.007078006 1.521643943 Q9CRB9 Chchd3 0.008096565 0.771401455 Q99K85 Psat1 0.008565677 0.360159662 Q3TXS7 Psmd1 0.009273031 -0.20213507 Q07076 Anxa7 0.0096467 0.273337405 P63037 Dnaja1 0.009674606 0.805671895 P52927 Hmga2 0.009821165 1.916733039 P53996 Cnbp 0.010031209 0.385586796 Q93092 Taldo1 0.010321858 0.194875047 P97351 Rps3a 0.010612226 0.153103294 Q9R0P3 Esd 0.011031366 0.336401077 P62311 Lsm3 0.011035255 0.248489109 P80315 Cct4 0.012398277 0.222385898 Q06185 Atp5i 0.012628233 0.351394258 Q61166 Mapre1 0.013036314 0.641206122 P62996 Tra2b 0.013313229 -1.403645055 P14869 Rplp0 0.014037033 0.232294209 Q8BG32 Psmd11 0.014459325 -0.205520802 Q60973 Rbbp7 0.014555653 0.551642118 Q9D7G0 Prps1 0.014601825 -0.242357511 144

P61294 Rab6b 0.014874168 -0.797090299 Q9Z2N8 Actl6a 0.015045943 2.176834963 P62082 Rps7 0.015080865 0.489636404 P80317 Cct6a 0.017201164 0.159952769 P45377 Akr1b8 0.017420028 -0.422421173 P14206 Rpsa 0.017424963 0.305658955 Q9DCL9 Paics 0.017917545 0.350291198 Q9EP71 Rai14 0.018030386 -0.942728729 Q9CZX8 Rps19 0.018068634 0.228577545 Q60865 Caprin1 0.018079448 0.284985193 P61202 Cops2 0.018744173 -0.256463483 P61164 Actr1a 0.019541344 -0.996587244 P10852 Slc3a2 0.019820867 0.313834298 Q9Z1E4 Gys1 0.01995933 -4.476239697 Q8VBW6 Nae1 0.020208516 1.342146334 Q6NZB0 Dnajc8 0.020979191 -3.851188264 Q8VDJ3 Hdlbp 0.021319131 -0.414180115 Q9QZF2 Gpc1 0.022070089 -0.792144726 Q8QZY1 Eif3l 0.022281372 0.276315628 Q60972 Rbbp4 0.02240855 0.678107857 Q91ZX7 Lrp1 0.022673357 -0.317784312 Q61699 Hsph1 0.022713651 0.869129065 P49718 Mcm5 0.023358163 0.458671312 P19096 Fasn 0.024111631 -0.318950069 Q9Z1Z2 Strap 0.024431253 0.207806288 P60843 Eif4a1 0.024542581 0.147943241 P59708 Sf3b6 0.024797978 0.682634533 P62317 Snrpd2 0.024861339 0.333869581 P83940 Tceb1 0.025260326 -0.370244365 Q6ZWM4 Lsm8 0.025347834 0.598665708 Q9D1A2 Cndp2 0.025433687 -0.268368244 P46935 Nedd4 0.026358685 -1.726259316 Q62318 Trim28 0.026474998 -0.672282946 P10630 Eif4a2 0.026711841 0.183972699 P40124 Cap1 0.028296171 -0.146401717 Q9CT10 Ranbp3 0.028576473 2.316615696 P35279 Rab6a 0.029136507 -0.409392479 Q91V41 Rab14 0.029395158 -0.228126355 Q6ZQ58 Larp1 0.03002805 0.671768736 Q8BHN3 Ganab 0.030393093 0.22069259 Q99JI6 Rap1b 0.032078341 0.590404102 Q8BGD9 Eif4b 0.032347987 0.183097381 145

P09602 Hmgn2 0.032354117 1.953116711 P62830 Rpl23 0.032532327 0.141872791 Q68FL6 Mars 0.033829681 0.169447884 Q3UHX2 Pdap1 0.034173658 0.416587277 O35658 C1qbp 0.034599164 0.26434076 P56480 Atp5b 0.03506055 0.162242147 P10126 Eef1a1 0.035899153 0.1148712 P80313 Cct7 0.037878578 0.171370574 P62835 Rap1a 0.038059708 0.696819294 P31324 Prkar2b 0.038774399 0.60733279 Q91WK2 Eif3h 0.039266913 0.503932722 P47754 Capza2 0.039346007 -0.204597088 Q5SSL4 Abr 0.039414379 2.927519655 Q9JMH6 Txnrd1 0.039924272 0.331110718 Q9WUM4 Coro1c 0.040409462 -0.430576702 Q9R1Q7 Plp2 0.040915471 2.689779537 Q61074 Ppm1g 0.04177973 -0.786180184 P14824 Anxa6 0.042124696 -0.3271111 P57759 Erp29 0.043273293 0.290247414 Q9CXW3 Cacybp 0.044181599 0.388776733 Q9CPY7 Lap3 0.044591074 -1.093762594 O55135 Eif6 0.045688466 -0.556409968 P63017 Hspa8 0.045697911 0.246243136 P11370 Fv4 0.04696939 1.379479726 P24369 Ppib 0.047060105 0.211973996 P80318 Cct3 0.047787123 0.161511189 P19157 Gstp1 0.048988473 0.297727168 Q8CAQ8 Immt 0.049727722 -0.658755469 P36552 Cpox 0.049853628 0.551295396

ID Gene Pvalue U-MWCNT vs C Log2_Fold_Change P02104 Hbb-y 2.95203E-08 13.87822896 P62862 Fau 7.72145E-06 -9.347753795 P02088 Hbb-b1 2.80208E-05 13.94399864 P47963 Rpl13 0.000145736 -0.621309686 Q9CQA3 Sdhb 0.000231198 0.319867508 O70250 Pgam2 0.000236052 0.283773343 Q99NB9 Sf3b1 0.00050878 0.441432777 Q9JIK5 Ddx21 0.000719446 -0.283454443 P07356 Anxa2 0.000786703 0.229279154 Q03265 Atp5a1 0.000997084 0.167184889 146

P62918 Rpl8 0.001141923 -0.460857263 P62960 Ybx1 0.001244582 -1.775259895 Q9JKB3 Ybx3 0.002106044 -1.453029336 Q9CPR4 Rpl17 0.002818066 -1.272008775 P19096 Fasn 0.002873122 -0.37104988 P20152 Vim 0.003322092 0.17091886 P35441 Thbs1 0.003549738 0.581266593 Q91VM5 Rbmxl1 0.003748876 -0.338033538 P17751 Tpi1 0.003881413 0.358463361 Q8VEK3 Hnrnpu 0.004020194 -0.534231243 Q99KI0 Aco2 0.004556758 0.180886691 P08752 Gnai2 0.004930542 0.576655162 Q60716 P4ha2 0.005074126 0.960627954 P68037 Ube2l3 0.005598826 -0.738102515 P62702 Rps4x 0.006332161 -0.206414774 Q8R3Q6 Ccdc58 0.006987898 0.357106031 P62911 Rpl32 0.007213848 -0.433319174 O09167 Rpl21 0.007371084 -0.165393267 Q99JY0 Hadhb 0.007533245 1.62121263 P27659 Rpl3 0.007565443 -0.38446275 Q9JLZ6 Hic2 0.007844849 0.478760175 P97351 Rps3a 0.007973377 -0.176890447 P62751 Rpl23a 0.00826328 -0.726526892 Q9DCC4 Pycrl 0.008454543 0.382849929 P16858 Gapdh 0.00867079 0.244354788 P63260 Actg1 0.008729183 0.136609421 Q8VDD5 Myh9 0.009842024 0.171721092 Q8VBW6 Nae1 0.009960939 1.673921073 P83882 Rpl36a 0.010708246 -0.720228112 Q99LC5 Etfa 0.010813049 0.258900672 Q922K7 Nop2 0.01083494 -1.291278785 P37889 Fbln2 0.010859091 0.332085622 Q99N92 Mrpl27 0.010936462 -3.749805361 O08997 Atox1 0.011014917 -0.321689032 P12382 Pfkl 0.011791488 0.642752057 P62830 Rpl23 0.011935279 -0.196235192 P25444 Rps2 0.011976053 -0.262139879 Q9D1M0 Sec13 0.012244251 -0.225834917 P30416 Fkbp4 0.012359446 -0.165441218 P39688 Fyn 0.013138247 1.150297583 P13020 Gsn 0.013763019 0.163461448 P60335 Pcbp1 0.014085253 0.154454396 147

Q91VR5 Ddx1 0.015423168 -0.420316627 Q61024 Asns 0.015484927 -0.600049118 P68033 Actc1 0.015566246 0.137262668 P35278 Rab5c 0.016450102 0.290655349 Q8R010 Aimp2 0.016656153 0.964620631 P62270 Rps18 0.016901808 -0.219916022 O54734 Ddost 0.017015737 0.292328187 P62874 Gnb1 0.017076335 0.295858546 Q60715 P4ha1 0.018006698 0.501034068 P70372 Elavl1 0.018534983 1.178131696 Q9D0F9 Pgm1 0.020824754 0.466552671 P99024 Tubb5 0.021074032 -0.187795213 P80313 Cct7 0.021183372 0.113232771 P12849 Prkar1b 0.021414783 2.50792626 P62242 Rps8 0.021596116 -0.394881315 P08249 Mdh2 0.021608294 0.215610502 Q9DC51 Gnai3 0.022134579 0.646878773 Q62095 Ddx3y 0.022535677 -0.495566953 Q62167 Ddx3x 0.023364397 -0.535134716 Q9JII6 Akr1a1 0.023491967 -0.713833782 Q9Z204 Hnrnpc 0.023497375 0.340041111 Q9Z1Z0 Uso1 0.024093679 0.614054237 P07901 Hsp90aa1 0.024186069 -0.05647782 Q01853 Vcp 0.024320943 0.200624053 Q6ZWM4 Lsm8 0.024497634 0.820019966 P14206 Rpsa 0.024939809 -0.252675811 P17710 Hk1 0.025842225 0.319394842 Q60932 Vdac1 0.026158626 0.211924388 Q9D2Y4 Mlkl 0.026391583 -0.193731451 P47915 Rpl29 0.026652156 -1.491950443 Q91VI7 Rnh1 0.026910678 -0.420803298 Q7TMY8 Huwe1 0.026923709 3.041641274 P68372 Tubb4b 0.027249299 -0.22240837 P47911 Rpl6 0.027376408 -0.568356183 P70698 Ctps1 0.02740231 -0.10649089 Q6ZWV7 Rpl35 0.027860473 -0.666802713 Q68FL6 Mars 0.029291616 0.180226148 Q6ZWN5 Rps9 0.029443955 -0.742552599 P52503 Ndufs6 0.029629675 1.056053092 Q3ULJ0 Gpd1l 0.029854205 1.55404792 Q9CYH6 Rrs1 0.029925064 -3.230766824 P62315 Snrpd1 0.029959531 -0.541748986 148

Q99JX4 Eif3m 0.030145216 -0.556019731 P17426 Ap2a1 0.030591475 -0.528579831 Q922D8 Mthfd1 0.03078922 0.547467385 Q9Z0P4 Palm 0.031011047 1.711376984 Q91VH2 Snx9 0.03136532 0.703764752 Q9EPL8 Ipo7 0.032104201 -0.330230808 P17182 Eno1 0.032291981 0.66937219 P62880 Gnb2 0.032425991 0.278075051 O35864 Cops5 0.032503259 0.509137472 Q80UG5 42622 0.032991598 0.365509847 Q920B9 Supt16h 0.034624518 1.016835138 Q8VC30 Tkfc 0.035750419 -0.461652408 P12970 Rpl7a 0.036178077 -0.454557149 Q62318 Trim28 0.037128198 -0.395967635 P32067 Ssb 0.037399823 -0.248843364 P45952 Acadm 0.038285263 1.680575697 P68368 Tuba4a 0.038470533 -0.128622211 Q9CQD1 Rab5a 0.03884145 0.377842891 Q9JLQ0 Cd2ap 0.039005309 0.24840581 O35639 Anxa3 0.039869487 -0.246129676 Q7TMM9 Tubb2a 0.040233032 -0.368077214 P12388 Serpinb2 0.040262664 -0.498438055 Q99JI6 Rap1b 0.040341252 0.553602149 Q99MN1 Kars 0.040455033 -0.470753569 P27773 Pdia3 0.040807064 0.179824752 P10605 Ctsb 0.04137899 0.275865087 Q9ERG0 Lima1 0.04158151 0.59085951 Q9Z1Z2 Strap 0.041917649 -0.201040118 Q64727 Vcl 0.041940344 0.140808126 Q9R1Q7 Plp2 0.04205561 2.650966903 Q8K009 Aldh1l2 0.042692645 -0.653132334 P17809 Slc2a1 0.042804152 1.473209761 P47811 Mapk14 0.043515691 2.160798063 Q3THS6 Mat2a 0.044052781 0.195353181 Q91WK0 Lrrfip2 0.044529456 1.269157159 O88543 Cops3 0.04523309 -1.027902497 Q9ERD7 Tubb3 0.046055134 -0.302260646 Q3V3R1 Mthfd1l 0.046220712 0.23868361 P11370 Fv4 0.047012805 1.416471598 P14869 Rplp0 0.047383224 0.123186597 Q60710 Samhd1 0.047948752 0.431899005 P27046 Man2a1 0.048954594 0.388589678 149

P62835 Rap1a 0.049076702 0.634012781 P62754 Rps6 0.049594075 -0.668958267

150

Table A5: List of significant two-sample t-test p values and log2 fold change for each

MWCNT ‘low’ dose exposure compared to control.

ID Description pvalue U5 vs C log2_FC_U5 Caprin-1 OS=Mus musculus GN=Caprin1 PE=1 SV=2 - Q60865 [CAPR1_MOUSE] 0.000574 0.489704 Superoxide dismutase [Cu-Zn] OS=Mus musculus GN=Sod1 P08228 PE=1 SV=2 - [SODC_MOUSE] 0.001073 0.236664 Macrophage-capping protein OS=Mus musculus GN=Capg P24452 PE=1 SV=2 - [CAPG_MOUSE] 0.001107 0.268558 60S ribosomal protein L23a OS=Mus musculus GN=Rpl23a P62751 PE=1 SV=1 - [RL23A_MOUSE] 0.001649 0.398565 Ras-related protein Rab-5C OS=Mus musculus GN=Rab5c P35278 PE=1 SV=2 - [RAB5C_MOUSE] 0.001858 0.382667 Mitochondrial import inner membrane translocase subunit Tim13 OS=Mus musculus GN=Timm13 PE=1 SV=1 - P62075 [TIM13_MOUSE] 0.003381 0.494663 Tripeptidyl-peptidase 2 OS=Mus musculus GN=Tpp2 PE=2 Q64514 SV=3 - [TPP2_MOUSE] 0.003409 0.400327 Isocitrate dehydrogenase [NADP], mitochondrial OS=Mus P54071 musculus GN=Idh2 PE=1 SV=3 - [IDHP_MOUSE] 0.004321 0.896017 Adenylate kinase isoenzyme 1 OS=Mus musculus GN=Ak1 Q9R0Y5 PE=1 SV=1 - [KAD1_MOUSE] 0.004848 0.345381 60S acidic ribosomal protein P1 OS=Mus musculus GN=Rplp1 P47955 PE=2 SV=1 - [RLA1_MOUSE] 0.004892 0.529830 Talin-1 OS=Mus musculus GN=Tln1 PE=1 SV=2 - P26039 [TLN1_MOUSE] 0.005678 0.204144 Fumarate hydratase, mitochondrial OS=Mus musculus GN=Fh P97807 PE=1 SV=3 - [FUMH_MOUSE] 0.005743 0.371473 Elongation factor Tu, mitochondrial OS=Mus musculus Q8BFR5 GN=Tufm PE=1 SV=1 - [EFTU_MOUSE] 0.006857 0.347418 PDZ and LIM domain protein 1 OS=Mus musculus GN=Pdlim1 O70400 PE=2 SV=4 - [PDLI1_MOUSE] 0.007726 0.307976 26S protease regulatory subunit 10B OS=Mus musculus P62334 GN=Psmc6 PE=1 SV=1 - [PRS10_MOUSE] 0.008059 0.561262 Serpin H1 OS=Mus musculus GN=Serpinh1 PE=1 SV=3 - P19324 [SERPH_MOUSE] 0.008135 0.271751 ATP synthase subunit e, mitochondrial OS=Mus musculus Q06185 GN=Atp5i PE=1 SV=2 - [ATP5I_MOUSE] 0.008937 0.338165 Transketolase OS=Mus musculus GN=Tkt PE=1 SV=1 - P40142 [TKT_MOUSE] 0.009059 0.284165 Glyoxalase domain-containing protein 4 OS=Mus musculus Q9CPV4 GN=Glod4 PE=2 SV=1 - [GLOD4_MOUSE] 0.009825 0.479064 Elongation factor 1-beta OS=Mus musculus GN=Eef1b PE=1 O70251 SV=5 - [EF1B_MOUSE] 0.010063 0.848239 Heterogeneous nuclear ribonucleoproteins A2/B1 OS=Mus O88569 musculus GN=Hnrnpa2b1 PE=1 SV=2 - [ROA2_MOUSE] 0.011223 0.230675 Testin OS=Mus musculus GN=Tes PE=1 SV=1 - P47226 [TES_MOUSE] 0.012142 0.384916 Tyrosine-protein kinase Fyn OS=Mus musculus GN=Fyn PE=1 P39688 SV=4 - [FYN_MOUSE] 0.012377 0.967910 Phosphoglycerate mutase 2 OS=Mus musculus GN=Pgam2 O70250 PE=1 SV=3 - [PGAM2_MOUSE] 0.013026 0.274631 Calreticulin OS=Mus musculus GN=Calr PE=1 SV=1 - P14211 [CALR_MOUSE] 0.013339 0.164085 Fructose-bisphosphate aldolase A OS=Mus musculus P05064 GN=Aldoa PE=1 SV=2 - [ALDOA_MOUSE] 0.013438 0.540396 Eukaryotic translation initiation factor 5A-1 OS=Mus musculus P63242 GN=Eif5a PE=1 SV=2 - [IF5A1_MOUSE] 0.015885 0.293852 Glutathione S-transferase Mu 5 OS=Mus musculus GN=Gstm5 P48774 PE=1 SV=1 - [GSTM5_MOUSE] 0.016321 -0.738258

151

Annexin A7 OS=Mus musculus GN=Anxa7 PE=2 SV=2 - Q07076 [ANXA7_MOUSE] 0.016357 0.362861 Macrophage migration inhibitory factor OS=Mus musculus P34884 GN=Mif PE=1 SV=2 - [MIF_MOUSE] 0.016878 0.214218 Ras-related protein Rab-21 OS=Mus musculus GN=Rab21 P35282 PE=1 SV=4 - [RAB21_MOUSE] 0.019327 0.241411 Lysine--tRNA ligase OS=Mus musculus GN=Kars PE=1 SV=1 Q99MN1 - [SYK_MOUSE] 0.019526 0.459502 Lysosome membrane protein 2 OS=Mus musculus O35114 GN=Scarb2 PE=1 SV=3 - [SCRB2_MOUSE] 0.019903 0.422559 Ras-related protein Rab-5A OS=Mus musculus GN=Rab5a Q9CQD1 PE=1 SV=1 - [RAB5A_MOUSE] 0.020443 0.420841 Calpastatin OS=Mus musculus GN=Cast PE=1 SV=2 - P51125 [ICAL_MOUSE] 0.020707 0.425841 Deoxynucleoside triphosphate triphosphohydrolase SAMHD1 OS=Mus musculus GN=Samhd1 PE=1 SV=2 - Q60710 [SAMH1_MOUSE] 0.020769 0.579124 Actin-related protein 3 OS=Mus musculus GN=Actr3 PE=1 Q99JY9 SV=3 - [ARP3_MOUSE] 0.020897 0.271442 Triosephosphate isomerase OS=Mus musculus GN=Tpi1 P17751 PE=1 SV=4 - [TPIS_MOUSE] 0.021622 0.248745 78 kDa glucose-regulated protein OS=Mus musculus P20029 GN=Hspa5 PE=1 SV=3 - [GRP78_MOUSE] 0.022099 0.207443 Vimentin OS=Mus musculus GN=Vim PE=1 SV=3 - P20152 [VIME_MOUSE] 0.023058 0.148030 Aconitate hydratase, mitochondrial OS=Mus musculus Q99KI0 GN=Aco2 PE=1 SV=1 - [ACON_MOUSE] 0.024913 0.235251 Granulins OS=Mus musculus GN=Grn PE=1 SV=2 - P28798 [GRN_MOUSE] 0.026104 0.450800 60S ribosomal protein L13a OS=Mus musculus GN=Rpl13a P19253 PE=1 SV=4 - [RL13A_MOUSE] 0.028341 0.980935 T-complex protein 1 subunit zeta OS=Mus musculus P80317 GN=Cct6a PE=1 SV=3 - [TCPZ_MOUSE] 0.029491 0.155078 60S ribosomal protein L12 OS=Mus musculus GN=Rpl12 P35979 PE=1 SV=2 - [RL12_MOUSE] 0.030155 0.159405 Heterogeneous nuclear ribonucleoprotein A3 OS=Mus Q8BG05 musculus GN=Hnrnpa3 PE=1 SV=1 - [ROA3_MOUSE] 0.030463 0.237270 Histone-binding protein RBBP7 OS=Mus musculus GN=Rbbp7 Q60973 PE=1 SV=1 - [RBBP7_MOUSE] 0.030563 0.410365 Rho GDP-dissociation inhibitor 1 OS=Mus musculus Q99PT1 GN=Arhgdia PE=1 SV=3 - [GDIR1_MOUSE] 0.030794 0.155493 RNA-binding protein Raly OS=Mus musculus GN=Raly PE=1 Q64012 SV=3 - [RALY_MOUSE] 0.030964 0.294805 Cellular nucleic acid-binding protein OS=Mus musculus P53996 GN=Cnbp PE=2 SV=2 - [CNBP_MOUSE] 0.031512 0.203990 Pre-mRNA-processing factor 19 OS=Mus musculus Q99KP6 GN=Prpf19 PE=2 SV=1 - [PRP19_MOUSE] 0.031726 0.485857 Prohibitin OS=Mus musculus GN=Phb PE=1 SV=1 - P67778 [PHB_MOUSE] 0.031955 0.195361 Peptidyl-tRNA hydrolase 2, mitochondrial OS=Mus musculus Q8R2Y8 GN=Ptrh2 PE=2 SV=1 - [PTH2_MOUSE] 0.032821 0.610780 60S ribosomal protein L13 OS=Mus musculus GN=Rpl13 P47963 PE=2 SV=3 - [RL13_MOUSE] 0.034186 0.258252 Cathepsin Z OS=Mus musculus GN=Ctsz PE=2 SV=1 - Q9WUU7 [CATZ_MOUSE] 0.034395 0.577081 Myosin-14 OS=Mus musculus GN=Myh14 PE=1 SV=1 - Q6URW6 [MYH14_MOUSE] 0.034446 0.568686 Protein transport protein Sec61 subunit alpha isoform 1 OS=Mus musculus GN=Sec61a1 PE=2 SV=2 - P61620 [S61A1_MOUSE] 0.036274 0.358140 UTP--glucose-1-phosphate uridylyltransferase OS=Mus Q91ZJ5 musculus GN=Ugp2 PE=2 SV=3 - [UGPA_MOUSE] 0.036513 0.425497 Phosphoglycerate mutase 1 OS=Mus musculus GN=Pgam1 Q9DBJ1 PE=1 SV=3 - [PGAM1_MOUSE] 0.036660 0.258264 Ornithine aminotransferase, mitochondrial OS=Mus musculus P29758 GN=Oat PE=1 SV=1 - [OAT_MOUSE] 0.037074 0.317400 Histone-binding protein RBBP4 OS=Mus musculus GN=Rbbp4 Q60972 PE=1 SV=5 - [RBBP4_MOUSE] 0.037255 0.521770 152

Coatomer subunit beta' OS=Mus musculus GN=Copb2 PE=2 O55029 SV=2 - [COPB2_MOUSE] 0.037847 0.532386 Destrin OS=Mus musculus GN=Dstn PE=1 SV=3 - Q9R0P5 [DEST_MOUSE] 0.038024 0.401234 Glutathione S-transferase Mu 6 OS=Mus musculus GN=Gstm6 O35660 PE=2 SV=3 - [GSTM6_MOUSE] 0.038672 0.342389 RNA-binding protein FUS OS=Mus musculus GN=Fus PE=2 P56959 SV=1 - [FUS_MOUSE] 0.038920 0.214719 Serine/threonine-protein phosphatase 2A 65 kDa regulatory subunit A alpha isoform OS=Mus musculus GN=Ppp2r1a Q76MZ3 PE=1 SV=3 - [2AAA_MOUSE] 0.039789 0.212176 Calpain-2 catalytic subunit OS=Mus musculus GN=Capn2 O08529 PE=2 SV=4 - [CAN2_MOUSE] 0.039807 0.465147 Myosin-11 OS=Mus musculus GN=Myh11 PE=1 SV=1 - O08638 [MYH11_MOUSE] 0.039872 0.312705 Eukaryotic translation initiation factor 3 subunit H OS=Mus Q91WK2 musculus GN=Eif3h PE=1 SV=1 - [EIF3H_MOUSE] 0.040851 0.610047 Dolichyl-diphosphooligosaccharide--protein glycosyltransferase subunit 2 OS=Mus musculus GN=Rpn2 Q9DBG6 PE=2 SV=1 - [RPN2_MOUSE] 0.041346 0.412608 40S ribosomal protein S13 OS=Mus musculus GN=Rps13 P62301 PE=2 SV=2 - [RS13_MOUSE] 0.042646 0.327349 Ubiquitin carboxyl-terminal hydrolase 14 OS=Mus musculus Q9JMA1 GN=Usp14 PE=1 SV=3 - [UBP14_MOUSE] 0.042841 0.530781 DnaJ homolog subfamily A member 1 OS=Mus musculus P63037 GN=Dnaja1 PE=1 SV=1 - [DNJA1_MOUSE] 0.043044 0.478438 Aspartate--tRNA ligase, cytoplasmic OS=Mus musculus Q922B2 GN=Dars PE=2 SV=2 - [SYDC_MOUSE] 0.045323 0.262342 Protein FAM49B OS=Mus musculus GN=Fam49b PE=2 SV=1 Q921M7 - [FA49B_MOUSE] 0.045520 0.843707 SUMO-activating enzyme subunit 2 OS=Mus musculus Q9Z1F9 GN=Uba2 PE=2 SV=1 - [SAE2_MOUSE] 0.045666 0.207652 Cullin-1 OS=Mus musculus GN=Cul1 PE=1 SV=1 - Q9WTX6 [CUL1_MOUSE] 0.045912 0.448581 RuvB-like 2 OS=Mus musculus GN=Ruvbl2 PE=2 SV=3 - Q9WTM5 [RUVB2_MOUSE] 0.046392 0.294051 Acid ceramidase OS=Mus musculus GN=Asah1 PE=1 SV=1 - Q9WV54 [ASAH1_MOUSE] 0.046669 0.594261 Myosin-10 OS=Mus musculus GN=Myh10 PE=1 SV=2 - Q61879 [MYH10_MOUSE] 0.046894 0.141872 Protein S100-A10 OS=Mus musculus GN=S100a10 PE=1 P08207 SV=2 - [S10AA_MOUSE] 0.048049 0.471115 Fibulin-2 OS=Mus musculus GN=Fbln2 PE=1 SV=2 - P37889 [FBLN2_MOUSE] 0.048152 0.352250 Neuronal proto-oncogene tyrosine-protein kinase Src OS=Mus P05480 musculus GN=Src PE=1 SV=4 - [SRC_MOUSE] 0.048535 0.426056 Sideroflexin-3 OS=Mus musculus GN=Sfxn3 PE=1 SV=1 - Q91V61 [SFXN3_MOUSE] 0.049689 0.395856 Eukaryotic translation initiation factor 3 subunit L OS=Mus Q8QZY1 musculus GN=Eif3l PE=1 SV=1 - [EIF3L_MOUSE] 0.049913 0.231251 T-complex protein 1 subunit epsilon OS=Mus musculus P80316 GN=Cct5 PE=1 SV=1 - [TCPE_MOUSE] 0.049927 0.210072

ID Description pvalue A5 vs C log2_FC_A5 Heme oxygenase 1 OS=Mus musculus GN=Hmox1 PE=1 P14901 SV=1 - [HMOX1_MOUSE] 0.003234 0.751298 4F2 cell-surface antigen heavy chain OS=Mus musculus P10852 GN=Slc3a2 PE=1 SV=1 - [4F2_MOUSE] 0.005804 0.327893 Macrophage-capping protein OS=Mus musculus GN=Capg P24452 PE=1 SV=2 - [CAPG_MOUSE] 0.007810 0.173564 Eukaryotic translation initiation factor 3 subunit G OS=Mus Q9Z1D1 musculus GN=Eif3g PE=1 SV=2 - [EIF3G_MOUSE] 0.013845 -0.221833 NADH dehydrogenase [ubiquinone] 1 alpha subcomplex subunit 9, mitochondrial OS=Mus musculus GN=Ndufa9 PE=1 Q9DC69 SV=2 - [NDUA9_MOUSE] 0.014581 0.583403

153

Coiled-coil-helix-coiled-coil-helix domain-containing protein 3, mitochondrial OS=Mus musculus GN=Chchd3 PE=1 SV=1 - Q9CRB9 [CHCH3_MOUSE] 0.016175 0.771555 40S ribosomal protein SA OS=Mus musculus GN=Rpsa PE=1 P14206 SV=4 - [RSSA_MOUSE] 0.021039 0.204892 Adenylate kinase isoenzyme 1 OS=Mus musculus GN=Ak1 Q9R0Y5 PE=1 SV=1 - [KAD1_MOUSE] 0.022491 0.295822 Filamin-C OS=Mus musculus GN=Flnc PE=1 SV=3 - Q8VHX6 [FLNC_MOUSE] 0.024350 0.232853 Endoplasmic reticulum resident protein 44 OS=Mus musculus Q9D1Q6 GN=Erp44 PE=1 SV=1 - [ERP44_MOUSE] 0.025220 0.335485 Polypyrimidine tract-binding protein 3 OS=Mus musculus Q8BHD7 GN=Ptbp3 PE=2 SV=1 - [PTBP3_MOUSE] 0.025491 0.478372 Dihydropyrimidinase-related protein 3 OS=Mus musculus Q62188 GN=Dpysl3 PE=1 SV=1 - [DPYL3_MOUSE] 0.025947 0.286219 Heterogeneous nuclear ribonucleoprotein L OS=Mus musculus Q8R081 GN=Hnrnpl PE=1 SV=2 - [HNRPL_MOUSE] 0.027691 0.558111 Apoptotic chromatin condensation inducer in the nucleus Q9JIX8 OS=Mus musculus GN=Acin1 PE=1 SV=3 - [ACINU_MOUSE] 0.028772 -0.798360 Metallothionein-1 OS=Mus musculus GN=Mt1 PE=1 SV=1 - P02802 [MT1_MOUSE] 0.029151 0.306774 Metallothionein-2 OS=Mus musculus GN=Mt2 PE=1 SV=2 - P02798 [MT2_MOUSE] 0.030472 0.185264 Cullin-1 OS=Mus musculus GN=Cul1 PE=1 SV=1 - Q9WTX6 [CUL1_MOUSE] 0.031316 0.407096 Caprin-1 OS=Mus musculus GN=Caprin1 PE=1 SV=2 - Q60865 [CAPR1_MOUSE] 0.033110 0.214313 Protein flightless-1 homolog OS=Mus musculus GN=Flii PE=1 Q9JJ28 SV=1 - [FLII_MOUSE] 0.033245 -0.203638 Transketolase OS=Mus musculus GN=Tkt PE=1 SV=1 - P40142 [TKT_MOUSE] 0.036034 0.199464 DnaJ homolog subfamily A member 1 OS=Mus musculus P63037 GN=Dnaja1 PE=1 SV=1 - [DNJA1_MOUSE] 0.037754 0.534591 Far upstream element-binding protein 2 OS=Mus musculus Q3U0V1 GN=Khsrp PE=1 SV=2 - [FUBP2_MOUSE] 0.039753 -0.416360 Thrombospondin-1 OS=Mus musculus GN=Thbs1 PE=1 SV=1 P35441 - [TSP1_MOUSE] 0.040094 0.847857 GMP synthase [glutamine-hydrolyzing] OS=Mus musculus Q3THK7 GN=Gmps PE=1 SV=2 - [GUAA_MOUSE] 0.040318 0.593311 Cathepsin Z OS=Mus musculus GN=Ctsz PE=2 SV=1 - Q9WUU7 [CATZ_MOUSE] 0.043739 0.522730 Transcription intermediary factor 1-beta OS=Mus musculus Q62318 GN=Trim28 PE=1 SV=3 - [TIF1B_MOUSE] 0.045770 -0.579998 Coatomer subunit beta OS=Mus musculus GN=Copb1 PE=1 Q9JIF7 SV=1 - [COPB_MOUSE] 0.049196 0.358978

ID Description pvalue Z2.5 vs C log2_FC_Z2.5 Calcyclin-binding protein OS=Mus musculus GN=Cacybp Q9CXW3 PE=1 SV=1 - [CYBP_MOUSE] 0.000468 0.410834 Transcription elongation factor B polypeptide 1 OS=Mus P83940 musculus GN=Tceb1 PE=1 SV=1 - [ELOC_MOUSE] 0.001853 -0.552605 Asparagine synthetase [glutamine-hydrolyzing] OS=Mus Q61024 musculus GN=Asns PE=2 SV=3 - [ASNS_MOUSE] 0.002534 -0.853011 DNA replication licensing factor MCM7 OS=Mus musculus Q61881 GN=Mcm7 PE=1 SV=1 - [MCM7_MOUSE] 0.002967 0.646830 Serine/threonine-protein phosphatase PP1-gamma catalytic subunit OS=Mus musculus GN=Ppp1cc PE=1 SV=1 - P63087 [PP1G_MOUSE] 0.003945 -0.609985 Estradiol 17-beta-dehydrogenase 12 OS=Mus musculus O70503 GN=Hsd17b12 PE=2 SV=1 - [DHB12_MOUSE] 0.004023 1.115271 Heme oxygenase 1 OS=Mus musculus GN=Hmox1 PE=1 P14901 SV=1 - [HMOX1_MOUSE] 0.004481 2.783627 Serine/threonine-protein phosphatase PP1-alpha catalytic subunit OS=Mus musculus GN=Ppp1ca PE=1 SV=1 - P62137 [PP1A_MOUSE] 0.005012 -0.636249 PDZ and LIM domain protein 1 OS=Mus musculus GN=Pdlim1 O70400 PE=2 SV=4 - [PDLI1_MOUSE] 0.006322 0.295902 154

26S proteasome non-ATPase regulatory subunit 1 OS=Mus Q3TXS7 musculus GN=Psmd1 PE=1 SV=1 - [PSMD1_MOUSE] 0.008936 -0.266877 Methionine aminopeptidase 2 OS=Mus musculus GN=Metap2 O08663 PE=1 SV=1 - [MAP2_MOUSE] 0.009032 -0.377595 Inosine-5'-monophosphate dehydrogenase 2 OS=Mus P24547 musculus GN=Impdh2 PE=1 SV=2 - [IMDH2_MOUSE] 0.010715 0.310175 Golgi reassembly-stacking protein 2 OS=Mus musculus Q99JX3 GN=Gorasp2 PE=1 SV=3 - [GORS2_MOUSE] 0.011070 0.568643 Coiled-coil-helix-coiled-coil-helix domain-containing protein 3, mitochondrial OS=Mus musculus GN=Chchd3 PE=1 SV=1 - Q9CRB9 [CHCH3_MOUSE] 0.013639 0.950419 40S ribosomal protein SA OS=Mus musculus GN=Rpsa PE=1 P14206 SV=4 - [RSSA_MOUSE] 0.013758 0.296062 Copper transport protein ATOX1 OS=Mus musculus O08997 GN=Atox1 PE=2 SV=1 - [ATOX1_MOUSE] 0.015550 -0.428961 Uridine 5'-monophosphate synthase OS=Mus musculus P13439 GN=Umps PE=2 SV=3 - [UMPS_MOUSE] 0.015755 0.382752 Plexin-B2 OS=Mus musculus GN=Plxnb2 PE=1 SV=1 - B2RXS4 [PLXB2_MOUSE] 0.017054 -0.566682 Disabled homolog 2 OS=Mus musculus GN=Dab2 PE=1 SV=2 P98078 - [DAB2_MOUSE] 0.017387 -0.739773 Histone-binding protein RBBP7 OS=Mus musculus GN=Rbbp7 Q60973 PE=1 SV=1 - [RBBP7_MOUSE] 0.018083 0.514156 Alanine aminotransferase 1 OS=Mus musculus GN=Gpt PE=2 Q8QZR5 SV=3 - [ALAT1_MOUSE] 0.019226 -0.654007 Mitogen-activated protein kinase 1 OS=Mus musculus P63085 GN=Mapk1 PE=1 SV=3 - [MK01_MOUSE] 0.020265 0.860472 14-3-3 protein theta OS=Mus musculus GN=Ywhaq PE=1 P68254 SV=1 - [1433T_MOUSE] 0.022353 -0.467952 UDP-glucose 6-dehydrogenase OS=Mus musculus GN=Ugdh O70475 PE=1 SV=1 - [UGDH_MOUSE] 0.024892 -0.237571 40S ribosomal protein S17 OS=Mus musculus GN=Rps17 P63276 PE=1 SV=2 - [RS17_MOUSE] 0.027689 0.249349 Histone-binding protein RBBP4 OS=Mus musculus GN=Rbbp4 Q60972 PE=1 SV=5 - [RBBP4_MOUSE] 0.027926 0.583239 Nucleosome assembly protein 1-like 1 OS=Mus musculus P28656 GN=Nap1l1 PE=1 SV=2 - [NP1L1_MOUSE] 0.028114 0.310067 Fatty acid synthase OS=Mus musculus GN=Fasn PE=1 SV=2 P19096 - [FAS_MOUSE] 0.028281 -0.252004 14-3-3 protein beta/alpha OS=Mus musculus GN=Ywhab Q9CQV8 PE=1 SV=3 - [1433B_MOUSE] 0.028656 -0.429821 Thrombospondin-1 OS=Mus musculus GN=Thbs1 PE=1 SV=1 P35441 - [TSP1_MOUSE] 0.028857 0.482896 Heat shock-related 70 kDa protein 2 OS=Mus musculus P17156 GN=Hspa2 PE=1 SV=2 - [HSP72_MOUSE] 0.029507 0.208849 14-3-3 protein eta OS=Mus musculus GN=Ywhah PE=1 SV=2 P68510 - [1433F_MOUSE] 0.030833 -0.392275 Vacuolar protein sorting-associated protein 29 OS=Mus Q9QZ88 musculus GN=Vps29 PE=1 SV=1 - [VPS29_MOUSE] 0.031794 -0.374797 Peroxiredoxin-2 OS=Mus musculus GN=Prdx2 PE=1 SV=3 - Q61171 [PRDX2_MOUSE] 0.033205 -0.236245 Cytosolic non-specific dipeptidase OS=Mus musculus Q9D1A2 GN=Cndp2 PE=1 SV=1 - [CNDP2_MOUSE] 0.033988 -0.319591 Thioredoxin reductase 1, cytoplasmic OS=Mus musculus Q9JMH6 GN=Txnrd1 PE=1 SV=3 - [TRXR1_MOUSE] 0.034851 0.491720 Dihydrolipoyllysine-residue acetyltransferase component of pyruvate dehydrogenase complex, mitochondrial OS=Mus Q8BMF4 musculus GN=Dlat PE=1 SV=2 - [ODP2_MOUSE] 0.036660 0.752455 Calponin-3 OS=Mus musculus GN=Cnn3 PE=1 SV=1 - Q9DAW9 [CNN3_MOUSE] 0.037251 -0.315844 Desmin OS=Mus musculus GN=Des PE=1 SV=3 - P31001 [DESM_MOUSE] 0.037309 0.438850 Leucine-rich repeat-containing protein 59 OS=Mus musculus Q922Q8 GN=Lrrc59 PE=2 SV=1 - [LRC59_MOUSE] 0.038064 0.222627 14-3-3 protein gamma OS=Mus musculus GN=Ywhag PE=1 P61982 SV=2 - [1433G_MOUSE] 0.038114 -0.370435 Gasdermin-D OS=Mus musculus GN=Gsdmdc1 PE=2 SV=1 - Q9D8T2 [GSDMD_MOUSE] 0.038862 -0.221215 155

Ras-related protein Rab-8B OS=Mus musculus GN=Rab8b P61028 PE=1 SV=1 - [RAB8B_MOUSE] 0.039400 -0.386915 V-type proton ATPase subunit C 1 OS=Mus musculus Q9Z1G3 GN=Atp6v1c1 PE=1 SV=4 - [VATC1_MOUSE] 0.039686 0.416840 Galectin-3 OS=Mus musculus GN=Lgals3 PE=1 SV=3 - P16110 [LEG3_MOUSE] 0.040796 0.288735 Neurolysin, mitochondrial OS=Mus musculus GN=Nln PE=2 Q91YP2 SV=1 - [NEUL_MOUSE] 0.043903 -0.797659 Tyrosine-protein kinase Fyn OS=Mus musculus GN=Fyn PE=1 P39688 SV=4 - [FYN_MOUSE] 0.049157 1.047266 Protein S100-A11 OS=Mus musculus GN=S100a11 PE=2 P50543 SV=1 - [S10AB_MOUSE] 0.049721 -0.368273 GMP synthase [glutamine-hydrolyzing] OS=Mus musculus Q3THK7 GN=Gmps PE=1 SV=2 - [GUAA_MOUSE] 0.049891 0.518352

156

Table A6: Data of pathway analysis used to generate heat maps of top 30 enriched pathways upon exposure to MWCNTs. The data are separated by increased and suppressed enrichment relative to control.

Increased pathway expression pvalues for exposure versus control PATHWAY AvsC ZvsC UvsC Mitochondrial Dysfunction 8.50539E-11 0.12455625 0.00047011 Oxidative Phosphorylation 4.26434E-10 0.317439266 0.00762093 EIF2 Signaling 0.166472957 4.22482E-11 0.49872569 NRF2-mediated Oxidative Stress Response 1 2.7994E-07 0.00063721 mTOR Signaling 0.18036301 1.46174E-09 1 Regulation of eIF4 and p70S6K Signaling 0.120452094 5.59124E-08 0.43075237 Relaxin Signaling 1 0.014275608 1.0692E-06 Caveolar-mediated Endocytosis Signaling 0.003341355 1 8.8217E-06 Glycolysis I 1 1 3.0537E-08 Corticotropin Releasing Hormone Signaling 1 0.008093559 4.7575E-06 Leukocyte Extravasation Signaling 1 0.168166861 1.0887E-06 Remodeling of Epithelial Adherens Junctions 0.243028864 0.218585124 6.0799E-06 TCA Cycle II (Eukaryotic) 0.00404913 1 9.3543E-05 Ephrin Receptor Signaling 0.525803552 0.14008301 5.9937E-06 IL-1 Signaling 1 0.290527038 1.62E-06 Gαi Signaling 1 0.075509164 8.3747E-06 Signaling by Rho Family GTPases 0.2581817 1 3.7986E-06 Tec Kinase Signaling 0.482373699 1 2.6319E-06 Clathrin-mediated Endocytosis Signaling 0.039313439 0.487643279 7.2728E-05 Integrin Signaling 0.58062164 0.177737273 1.6115E-05 Axonal Guidance Signaling 1 0.222564491 7.9282E-06 Epithelial Adherens Junction Signaling 1 0.100957268 2.2225E-05 Gluconeogenesis I 1 1 2.4992E-06 Inhibition of Angiogenesis by TSP1 0.008974647 1 0.00032095 P2Y Purigenic Receptor Signaling Pathway 1 0.363369102 7.9743E-06 Role of NFAT in Cardiac Hypertrophy 1 0.489629488 6.7192E-06 Folate Polyglutamylation 0.021482936 1 0.00016128 Tetrahydrofolate Salvage from 5,10- methenyltetrahydrofolate 0.021482936 1 0.00016128 GDP-glucose Biosynthesis 0.025724519 1 0.00024128 Adrenergic Signaling 1 0.276681379 2.3115E-05

157

Decreased pathway expression pvalues for exposure versus control PATHWAY AvsC ZvsC UvsC EIF2 Signaling 8.96581E-17 1 2.5552E-24 Regulation of eIF4 and p70S6K Signaling 0.000362621 1 9.0601E-10 mTOR Signaling 0.001374223 1 8.5653E-09 Remodeling of Epithelial Adherens Junctions 4.76263E-06 1 2.5226E-06 phagosome maturation 0.000110903 0.177663948 3.7521E-05 Sertoli Cell-Sertoli Cell Junction Signaling 0.000175587 1 2.5658E-05 14-3-3-mediated Signaling 0.000134807 1 4.4466E-05 Germ Cell-Sertoli Cell Junction Signaling 8.12994E-05 1 0.00016649 Breast Cancer Regulation by Stathmin1 3.75771E-05 1 0.00040295 Epithelial Adherens Junction Signaling 0.000391612 1 0.0001132 Palmitate Biosynthesis I (Animals) 0.013558053 0.003507712 0.00679059 Fatty Acid Biosynthesis Initiation II 0.013558053 0.003507712 0.00679059 Gap Junction Signaling 0.003835862 1 0.00016649 Protein Ubiquitination Pathway 0.49840065 6.30063E-05 0.20267462 Actin Cytoskeleton Signaling 1.21235E-05 1 1 Methylglyoxal Degradation III 0.091160887 0.024305011 0.04659169 Axonal Guidance Signaling 0.008477579 1 0.01453697 Granzyme A Signaling 0.000131655 1 1 RhoGDI Signaling 0.000146324 1 1 RhoA Signaling 0.0001482 1 1 Sucrose Degradation V (Mammalian) 1 0.013959478 0.02689133 AMPK Signaling 0.331949704 0.003458838 0.44877705 Regulation of Cellular Mechanics by Calpain Protease 0.000531923 1 1 Pyrimidine Ribonucleotides Interconversion 0.168407514 1 0.00378177 Pyrimidine Ribonucleotides De Novo Biosynthesis 0.179698956 1 0.00435529 Signaling by Rho Family GTPases 0.000928736 1 1 Aryl Hydrocarbon Receptor Signaling 0.012959718 1 0.07525657 NRF2-mediated Oxidative Stress Response 0.028902027 0.259230456 0.4411416 Asparagine Biosynthesis I 1 1 0.00340099 Neuregulin Signaling 0.116186486 1 0.03465062

158

Table A7: Log2 fold change values for each heat map created for the following pathways:

Nrf-2 mediated oxidative stress response, IL-1 signaling, Inhibition of angiogenesis by TSP1, mTOR signaling, eIF4/p70S6K signaling, and Oxidative phosphorylation.

NRF2-mediated Oxidative Stress Response Gene log2_FC_U log2_FC_A log2_FC_Z GSK3B 0.250281724 -0.048928813 -0.018125294 MAP2K1 0.287675668 0.083860403 0.059456748 GSTO1 -0.112842097 0.013691751 0.052366468 MAP2K2 0.147171323 -0.062030928 -0.317222763 USP14 0.209290744 0.025979293 -0.05488679 PTPN11 0.007592804 0.693524782 -0.922860552 CLPP 0.095473268 0.016822141 -2.084323602 RRAS -0.136670609 0.15443974 -0.67365206 SOD2 -1.731916034 0.13261699 0.007635687 GSTM3 0.137395977 -0.029068923 -0.125564511 STIP1 -0.000219384 -0.095817721 0.027353564 CUL3 -0.424600894 -0.268636113 -0.323242533 VCP 0.200624053 0.022158706 0.022485991 MGST1 0.247070497 0.393458403 -1.987733455 MAP2K4 0.981252828 0.3399336 0.655398318 EPHX1 -0.387571164 0.165405579 0.022524711 MAPK3 -0.279127283 -0.06491787 -0.135089765 p38MAPK 2.160798063 -0.747842147 -0.536422509 DNAJC10 0.339784163 0.66764499 0.325886392 HIP2 -2.12553161 -1.694772415 -0.284599577 CAT -0.995179768 -0.960293821 -1.347895042 TXN 0.22319788 -0.504898802 0.101844642 AKR -0.080488738 -0.006113235 -0.422421173 ACTC1 0.137262668 -0.004162584 0.117321392 GSTM1 -0.016290858 -0.310648341 -0.166371469 PRDX1 -0.140579644 0.007683169 -0.079850972 GSR -0.049668019 -2.049740809 0.245564059 DNAJA2 -0.219778461 -0.443888602 0.136548247 GSTM5 0.137395977 -0.029068923 -0.125564511 FTL 0.782439858 0.87384157 0.373914839 NRAS 1.250304703 -0.687769523 0.93209257 PPIB -0.012637944 -2.629114933 0.211973996

159

MAPK1 -0.499225092 0.305159231 0.247255762 DNAJC8 0.537873025 1.67651931 -3.851188264 ACTG1 0.136609421 0.042284314 0.117186518 SQSTM1 -0.262253948 -0.253893161 1.521643943 SOD1 -0.050453128 -0.172295482 0.140840214 RRAS2 -0.084072682 0.118598552 -0.517753923 CCT7 0.113232771 -0.075538993 0.171370574 GSTP1 0.329754195 0.212707672 0.297727168 DNAJA1 0.376884317 -0.304740129 0.805671895 TXNRD1 -0.284646495 -0.31944531 0.331110718 ERP29 0.179659799 0.074029224 0.290247414 HMOX1 0.460024194 0.727994304 2.844241665

IL-1 Signaling Gene log2_FC_U log2_FC_A log2_FC_Z PRKAR1B 2.50792626 0.06609395 0.081352961 PRKAR1A 0.162896671 0.091969667 0.081352961 GNA11 0.158640386 0.259673 0.115379255 MAP2K4 0.981252828 0.3399336 -0.166225403 GNAI2 0.576655162 0.109206972 0.115379255 RACK1 0.039811404 -0.070159957 0.081352961 MAPK14 2.160798063 -0.747842147 -0.166225403 PRKACA 0.146759628 -0.36502959 0.081352961 PRKACB 0.259565121 -0.773914739 0.081352961 GNG12 0.718464934 0.540707693 0.115379255 GNB2 0.278075051 0.010702264 0.115379255 GNB1 0.295858546 -0.156405682 0.115379255 GNAS 0.823289941 -0.372583439 0.115379255 MAPK1 -0.499225092 0.305159231 -0.166225403 PRKAR2B -0.694571787 -0.811306289 0.081352961 GNAI3 0.646878773 0.410764021 0.115379255

Inhibition of Angiogenesis by TSP1 Gene log2_FC_U log2_FC_A log2_FC_Z MAP2K4 0.981252828 0.3399336 0.655398318 MAPK14 2.160798063 -0.747842147 -0.536422509 Caspase3 1.772052239 1.303382629 0.358182324 MAPK1 -0.499225092 0.305159231 0.247255762 Fyn 1.150297583 0.838043275 0.344153868 HSPG2 -0.142717036 -1.254978965 -2.646607249 CD47 0.709075855 1.244465722 0.973881511 TSP1 0.581266593 2.731460287 0.850508381 160

mTOR Signaling Gene log2_FC_U log2_FC_A log2_FC_Z HMOX1 0.460024194 0.727994304 2.844241665 EIF4G3 -1.258171963 2.049730197 2.240745538 PRKAA1 -0.032953982 0.535433428 1.267549212 EIF3K -0.634701018 0.033942629 1.245008825 RHOQ 1.065645403 0.553054115 1.143482183 NRAS 1.250304703 -0.687769523 0.93209257 EIF3H -0.539702827 -0.447853355 0.503932722 RPS7 -0.525503581 0.207968979 0.489636404 RHEB 0.075087299 1.74039283 0.467100972 FAU -9.347753795 -3.561105046 0.410836353 RPS25 -0.208005686 -0.089216382 0.409351531 RPSA -0.252675811 0.011161586 0.305658955 RPS5 -0.023125517 -0.263946145 0.284758785 EIF3L 0.043326084 0.228797787 0.276315628 RPS28 -0.170145567 0.099446662 0.266380947 RPS17 -0.023913318 -0.253493575 0.2640379 MAPK1 -0.499225092 0.305159231 0.247255762 RPS27A -0.545839619 -0.404176302 0.229536707 RPS19 -0.063693396 0.057909645 0.228577545 RPS6 -0.668958267 -0.609148571 0.219303583 RPS12 -0.095395245 0.073055011 0.209131908 EIF3F -0.079727938 -0.057897002 0.207299328 EIF4A2 0.053501907 0.018975388 0.183972699 RPS27L 0.347249111 0.253005749 0.183820114 eIF4B 0.00663713 -0.04757959 0.183097381 RPS13 -0.484118357 -0.013292541 0.171867015 EIF3A -0.111918926 -0.434900958 0.162789466 RPS18 -0.219916022 0.06222332 0.160080446 EIF3E -0.079974188 0.219035242 0.157783104 RPS24 -0.253252455 -0.578506629 0.154902822 EIF4A1 0.018885494 0.009126021 0.147943241 RPS4Y1 -0.206414774 -0.234129129 0.143706648 EIF4A3 -0.014456509 0.057799719 0.107898281 RPS10 -0.206642652 -0.104752949 0.101617148 PPP2R1A 0.145092484 -0.00955774 0.097878827 RPS9 -0.742552599 -0.388374631 0.092781125 RPS14 -0.144075897 -0.156924436 0.08270853 RPS3 -0.080037198 -0.01677562 0.07469897 161

RPS8 -0.394881315 -0.547214046 0.074249015 EIF3B -0.265431559 0.010485297 0.05429012 RPS26 0.06466578 -0.089506181 0.051558911 EIF3D -0.212870632 -0.06880264 0.047524334 RPS15A -0.330165179 -0.217003764 0.044163754 EIF3C -0.038711936 -3.749815615 0.033152616 RPS23 -0.344471727 -0.643557163 0.032675206 RPS11 -1.062453088 -0.368685443 0.030958292 EIF4G2 -1.265161945 0.19103005 0.012447384 EIF3I 0.141443188 -0.089801984 0.008116796 FKBP1 -0.290373805 -0.538948117 -0.019653254 eIF4E -0.082632034 0.190008783 -0.020457242 RPS2 -0.262139879 -0.066003127 -0.020872489 EIF3M -0.556019731 0.04257994 -0.021945906 EIF4G1 -0.24815628 -0.3413431 -0.030451346 RPS21 -1.055473008 -0.508910856 -0.048386188 PPP2R1B 0.048821168 -0.240326595 -0.057542084 RAC1 0.240903584 -0.2617793 -0.085583848 PPP2CA 0.030785165 -0.110822016 -0.091359735 RPS20 -0.092849941 0.057596983 -0.096586054 RHOC 0.209316718 -0.307995288 -0.109754099 RHOA 0.209316718 -0.307995288 -0.109754099 MAPK3 -0.279127283 -0.06491787 -0.135089765 EIF3G 0.087035768 -0.092883981 -0.141934444 PPP2CB -0.053600551 -0.07965272 -0.198388048 RPS6KA3 0.082571231 -0.272304476 -0.20232041 PPP2R2A -1.195880963 0.399700539 -0.244844951 RHOG 1.217687505 -0.844040032 -0.267611972 RHOB 0.550563873 0.446744292 -0.272055512 RPS16 -0.501949346 -0.006315294 -0.320589744 RRAS2 -0.084072682 0.118598552 -0.517753923 RHOT1 1.243108218 0.834497839 -0.658875108 RRAS -0.136670609 0.15443974 -0.67365206 PTPN11 0.007592804 0.693524782 -0.922860552

Regulation of eIF4 and p70S6K Signaling Genes log2_FC_U log2_FC_A log2_FC_Z RPS26 0.06466578 -0.089506181 0.051558911 MAP2K1 0.287675668 0.083860403 0.059456748 AGO2 -0.26929482 -0.066681744 0.206079807 PABP -0.027096467 -0.085433054 0.005608026 MAP2K2 0.147171323 -0.062030928 -0.317222763 162

RPS20 -0.092849941 0.057596983 -0.096586054 PPP2R1B 0.048821168 -0.240326595 -0.057542084 PTPN11 0.007592804 0.693524782 -0.922860552 EIF4G2 -1.265161945 0.19103005 0.012447384 RRAS -0.136670609 0.15443974 -0.67365206 EIF3I 0.141443188 -0.089801984 0.008116796 EIF3F -0.079727938 -0.057897002 0.207299328 eIF4E -0.082632034 0.190008783 -0.020457242 PPP2CA 0.030785165 -0.110822016 -0.091359735 EIF3D -0.212870632 -0.06880264 0.047524334 RPS27L 0.347249111 0.253005749 0.183820114 RPS13 -0.484118357 -0.013292541 0.171867015 EIF2S1 -0.314231461 0.045192393 0.157240413 EIF3A -0.111918926 -0.434900958 0.162789466 EIF3K -0.634701018 0.033942629 1.245008825 PPP2R2A -1.195880963 0.399700539 -0.244844951 EIF3G 0.087035768 -0.092883981 -0.141934444 EIF3B -0.265431559 0.010485297 0.05429012 EIF3M -0.556019731 0.04257994 -0.021945906 EIF4A3 -0.014456509 0.057799719 0.107898281 MAPK3 -0.279127283 -0.06491787 -0.135089765 RPS3 -0.080037198 -0.01677562 0.07469897 EIF4A1 0.018885494 0.009126021 0.147943241 MAPK14 2.160798063 -0.747842147 -0.536422509 EIF2S3 -0.135362609 0.014996698 0.046619046 EIF3C -0.038711936 -3.749815615 0.033152616 ITGA3 -0.728421669 -0.526486059 -1.731459987 RPS25 -0.208005686 -0.089216382 0.409351531 RPS16 -0.501949346 -0.006315294 -0.320589744 EIF3E -0.079974188 0.219035242 0.157783104 PPP2CB -0.053600551 -0.07965272 -0.198388048 RPS2 -0.262139879 -0.066003127 -0.020872489 RPS17 -0.023913318 -0.253493575 0.2640379 PPP2R1A 0.145092484 -0.00955774 0.097878827 RPS5 -0.023125517 -0.263946145 0.284758785 RPS21 -1.055473008 -0.508910856 -0.048386188 RPS11 -1.062453088 -0.368685443 0.030958292 EIF4G1 -0.24815628 -0.3413431 -0.030451346 EIF4A2 0.053501907 0.018975388 0.183972699 RPS15A -0.330165179 -0.217003764 0.044163754 RPS23 -0.344471727 -0.643557163 0.032675206 RPSA -0.252675811 0.011161586 0.305658955 163

NRAS 1.250304703 -0.687769523 0.93209257 EIF3H -0.539702827 -0.447853355 0.503932722 RPS9 -0.742552599 -0.388374631 0.092781125 MAPK1 -0.499225092 0.305159231 0.247255762 RPS28 -0.170145567 0.099446662 0.266380947 RPS12 -0.095395245 0.073055011 0.209131908 RPS19 -0.063693396 0.057909645 0.228577545 RPS10 -0.206642652 -0.104752949 0.101617148 EIF1AY -0.765360557 -0.488798202 -1.268749956 RRAS2 -0.084072682 0.118598552 -0.517753923 RPS14 -0.144075897 -0.156924436 0.08270853 RPS8 -0.394881315 -0.547214046 0.074249015 EIF2S2 -0.64011478 -0.834080646 0.422351498 EIF1AX -0.650361503 -0.373761105 -1.142375742 ITGB1 -0.183277867 0.270363267 0.423790924 EIF3L 0.043326084 0.228797787 0.276315628 RPS27A -0.545839619 -0.404176302 0.229536707 RPS18 -0.219916022 0.06222332 0.160080446 RPS24 -0.253252455 -0.578506629 0.154902822 EIF4G3 -1.258171963 2.049730197 2.240745538 FAU -9.347753795 -3.561105046 0.410836353 RPS7 -0.525503581 0.207968979 0.489636404 RPS4Y1 -0.206414774 -0.234129129 0.143706648 RPS6 -0.668958267 -0.609148571 0.219303583

Oxidative Phosphorylation Signaling Gene log2_FC_U log2_FC_A log2_FC_Z Cox6c -0.370700409 0.132867772 -0.208027375 NDUFA9 0.282346229 0.048125837 -0.025145948 UQCRFS1 -0.0089546 0.690607963 -0.069194792 ATP5D -0.534884123 2.061496543 -0.28188869 COX4I1 -0.566822565 -0.130668548 -0.247789637 NDUFB10 -0.42643447 -0.281294406 -0.11043006 UQCRB 0.070789195 0.068833714 -0.56193684 COX5A 0.100343296 0.441329473 0.029687487 CYC1 0.105350802 0.456331841 -0.558374168 UQCR11 -0.131137194 2.237284661 -0.775062946 UQCRQ 0.629665407 0.677797441 -0.320290009 NDUFA8 -0.064447826 0.306234363 -0.775581362 ATP5H 0.087984949 0.275814836 0.006173453 NDUFAB1 -2.363503685 -0.673848805 -0.231631104 NDUFB7 0.675634438 0.696342562 0.152320166 164

CYB5A 0.143573767 0.660627565 -0.133889552 NDUFV2 -1.508513087 0.525899904 0.03001422 COX6B1 0.141115608 0.145310874 -0.041972997 NDUFS2 -1.484213371 0.210529766 -1.857882446 SDHA 0.110873439 -0.008879891 -0.41218752 UQCR10 -1.589605255 0.043668357 -0.182612515 NDUFS7 0.670969615 -1.686605722 -1.528917689 SDHC -0.119908328 0.320919766 -2.618607015 NDUFA4 -0.11629972 0.267564899 -0.262161899 COX2 0.169627894 0.5635414 0.447484556 ATP5F1 -0.26899479 -0.188307964 -0.442741997 UQCRC2 -0.083379504 0.273792712 -0.073506067 UQCRC1 0.099258309 0.193388808 -0.123940168 NDUFA2 0.048011649 0.394569608 -0.405587858 NDUFS3 -0.416603623 0.312632282 -1.81973319 ATP5C1 0.109580022 0.124615349 0.06955803 NDUFV1 -0.471361775 0.7878003 -0.536685295 NDUFA7 0.166890069 0.266557628 0.171109979 COX7A2 -1.337016342 -0.247001134 -1.493075834 CYCS -0.04733653 -0.627647141 -0.484503046 ATP5J2 1.42205775 1.826814245 1.612295584 ATP5O 0.356055899 0.498542026 0.349734336 NDUFS1 0.393717616 0.923189472 -0.211829104 ATP5J 0.656503065 0.850303222 0.490236779 NDUFS6 1.056053092 0.594247358 0.606026938 ATP5B 0.12641977 0.140686616 0.162242147 SDHB 0.319867508 0.221058511 -0.476054682 ATP5A1 0.167184889 0.167500461 0.10408527

165

APPENDIX B

Supplemental Figures and Tables: Chapter 3

Figure B1: Bar graphs depicting average log10 complement component 3 protein response of MWCNT 96 hour exposure for: A) Submerged versus ALI basolateral samples, and B)

Submerged versus ALI apical samples.

166

Table B1: LC method gradients. Method A used to process cell samples, and Method B used to process media and wash samples.

Method Time Duration Flow (nL/min) %B A 0 N/A 300 0 A 2 2 300 0 A 180 180 300 40

A 183 1 300 80 A 193 10 300 80 A 194 1 300 0 A 195 1 300 0 A 225 30 300 0

B 0 N/A 300 0 B 2 2 300 0 B 72 70 300 40 B 73 1 300 80 B 78 5 300 80

B 79 1 300 0 B 80 1 300 0 B 90 10 300 0

167

Table B2: Regression model used to fit the media sample data set. Model equation: Protein

Intensity = Intercept + Exposure Method + Exposure + Error

Df Sum Sq Mean Sq F P value

value

Exposure 1 5.466e+18 5.466e+18 19.837 <0.0001

Method

Exposure 1 1.595e+17 1.595e+17 0.579 0.4468

Residuals 2114 5.899e+20

168

Table B3: Number of significant protein (p-value < 0.05) count for MWCNT versus control.

Sample Exposure Location Number of

Type Method Significant Proteins

Wash ALI Apical 16

Media Submerged Apical 46

Media ALI Basal 31

Media Submerged Basal 10

169

Table B4: Upstream mediator z-score for the apical and basolateral 96 hour media proteome.

M-7 MWCNT exposure versus BSA control data was used to general enrichment analysis.

Upstream regulators Apical Basolateral NANOG 2.236067977 -1.889822365 TGFB1 0.781659137 3.199454766 IL5 -1 2.449489743 HDAC6 -1.287452619 1.986798536 NR1H4 1.341640786 -1.632993162 STAT3 -1.453504749 1.133444974 TGFB2 -1.165179711 1.165179711 AKT1 0 2.236067977 SPP1 -0.826084265 1.392746118 EGFR -1.00621124 1.068131932 NFE2L2 1 1.067489992 TNF 0.059628553 1.997325722 let-7 0 -1.997555509 FGF7 0 1.93449426 CCL5 -0.333333333 1.414213562 SMAD3 -0.277350098 1.386750491 COL18A1 0.554700196 -1.066003582 SATB1 -1.431658266 0.118124885 FBN1 0 -1.491374966 IL13 -1.227609323 0.197216242 GATA4 -0.404061018 -0.202030509 TBX5 -0.404061018 -0.202030509 IL1B 0.344315096 0.225313106 MYOCD -0.312228557 0.096673649 CSF1 0 0.15249857

170

Table B5: Upstream mediator z-score for the apical and basolateral 96 hour M-7 MWCNT exposed proteome. ALI versus submerged exposure data was used to general enrichment analysis.

Upstream Apical Basolateral PGR 3.94438329 2.424871131 EGFR 3.765590932 2.774565584 SYVN1 3.615384615 1.726088481 PCGEM1 3.410527668 2.012610844 ERBB2 3.299831646 1.889822365 IL1B 3.261195447 1.228524508 PI3K 3.147573112 2 MYC 3.035227213 1.524001524 EGF 2.889420557 1.765686002 CCL5 2.828427125 2 IL6 2.772808327 1.614592905 Ap1 2.432700719 1.632993162 SMARCA4 2.342606428 0 HIF1A 0.881140208 1.741133465 ERK1/2 0.707106781 1.940118727 MGEA5 -1.290994449 -0.816496581 HSF1 -1.682750674 -0.580797167 JAG2 -2 -2 CST5 -2.19089023 -1.889822365 CUL4B -2.449489743 0 EGLN -2.597626523 0 miR_122_5p -2.629502941 0 CBX5 -3.16227766 -2 mir_122 -4.357224054 0

171

Table B6: Log2 Fold change for proteins associated with upstream mediators listed in Figure

5. M-7 MWCNT exposure / BSA control data was used to calculate fold change.

TNF Network

Genes Apical Basolateral

NID1 0 5.296815005

LTBP2 0 1.88050249

HSPG2 -0.20702839 1.494849381

C3 0.668905608 0.555409629

NQO1 0.759399981 0.423928305

VCL 0.217801898 0.820689631

GCLM 0 0.960403138

KYNU 0.388324639 0.499518721

CXCL5 0.82434384 0

GCLC 0 0.736433959

COL1A2 0.686940784 0.041889056

TP53I3 0 0.688511103

ACTA2 0 0.652966359

IGFBP2 0.308761004 0.303875335

SERPINB1 -0.26384608 0.824926308

MMP2 0 0.54100416

MARCKSL1 -0.49246638 0.930106241

CD44 0.006708984 0.209077794

FST -1.18750448 1.211562024

SERPINE1 -0.10720954 0.12327962

FN1 0.213677659 -0.36758355

HEXB -0.48960021 0

172

CHI3L1 -0.55210864 0

TIMP1 -0.21324067 -0.353799772

MUC5AC -0.2227156 -0.381847074

HEXA -1.12532707 0

PSME2 -2.51329031 1.09749228

LGALS3 -4.48165556 0.927785964

UBQLN2 0 -4.189020044

CTSC -4.34639122 0

SOD2 -4.68001515 0

S100A8 -5.06182441 0

CLEC11A -5.20341626 0

CFD -0.74605171 -5.511066071

CAT -6.43459304 0

SMAD3 Network

Genes Apical Basolateral

COL3A1 0 0.507111947

COL1A1 0.294434862 0.17457655

CDH1 0.331042956 0

RAC1 -1.27981867 1.474930921

SERPINE1 -0.10720954 0.12327962

VIM -0.89963039 0.283502375

ZYX -4.95978057 1.569978193

CCL5 Network

Genes Apical Basolateral

NAMPT 4.984719669 0.484932492

173

CAPN2 1.156008865 0.611403758

CD44 0.006708984 0.209077794

ALCAM 0.461882663 -0.331940693

TUBB4B -0.86747407 0.561915011

VASP -0.38731221 -0.586866459

PLEC -2.09341132 1.117406248

PNP -4.12536506 0

CDC37 -5.54188321 0.764631664

TGFB1 Network

Genes Apical Basolateral

ITGB1 0.854642398 6.302178308

FBN1 0 4.24500575

LTBP2 0 1.88050249

THBS1 1.384042618 0.220320061

IGFBP3 0.610634546 0.38546492

SPARC 0.765344802 0.138491395

COL1A2 0.686940784 0.041889056

ACTA2 0 0.652966359

TGM2 -0.32995746 0.982451263

MMP2 0 0.54100416

COL3A1 0 0.507111947

COL1A1 0.294434862 0.17457655

NRP1 0.0542066 0.298831875

CDH1 0.331042956 0

CD44 0.006708984 0.209077794

TAGLN 0 0.149556902

174

BAX 0 0.020134637

SERPINE1 -0.10720954 0.12327962

FN1 0.213677659 -0.36758355

HEXB -0.48960021 0

CHI3L1 -0.55210864 0

TIMP1 -0.21324067 -0.353799772

VIM -0.89963039 0.283502375

VASP -0.38731221 -0.586866459

HEXA -1.12532707 0

CD59 -4.16756951 0.820846889

ZYX -4.95978057 1.569978193

IL1B Network

Genes Apical Basolateral

SRGN 2.994114475 0

C3 0.668905608 0.555409629

AKR1B1 0.042598554 1.001985681

CXCL5 0.82434384 0

COL1A1 0.294434862 0.17457655

NRP1 0.0542066 0.298831875

SERPINE1 -0.10720954 0.12327962

MIF -1.0594095 1.009038694

LCP1 -1.42729405 1.33216937

HEXB -0.48960021 0

CHI3L1 -0.55210864 0

MUC5AC -0.2227156 -0.381847074

FABP5 -2.70981921 1.950669344

175

PTGDS -0.05104054 -0.877838455

HEXA -1.12532707 0

SOD2 -4.68001515 0

CAT -6.43459304 0

176

Table B7: Log2 Fold change for proteins associated with upstream mediators listed in

Figure 6. ALI / Submerged M-7 exposure data was used to calculate fold change.

PI3K network

Genes Apical Basolateral

FASN 13.54327266 3.818119834

GCLM 9.58309461 0

GCLC 9.316441925 0

MMP9 7.490225024 0

AKR1B10 4.241728359 1.402241197

AKR1C1/AKR1C2 4.786490686 0.774260993

HSPA5 4.709288841 0

PKM 4.375380313 0

NQO1 4.257862397 0

CTSD 2.773510579 0

SERPINE1 0 -2.19433673

IL1B network

Genes Apical Basolateral

SRGN 8.671349178 5.735432011

FGG 9.442539772 4.939949825

CXCL5 2.284699507 11.25134998

C3 2.420111137 9.812669952

HEXB 10.01448151 0

SOD2 9.144226496 0

CHI3L1 0 9.099131394

177

DBI 8.979588971 0

MUC5AC 3.790641203 4.836091419

NRP1 8.324401035 0

PLA2G4A 7.982613632 0

SERPINH1 7.704064036 0

CTSS 7.574870025 0

MMP9 7.490225024 0

LDLR 6.458905458 0

AKR1B1 4.745698596 1.290026651

LCP1 2.166019054 0

APOE -0.7735415 0

COL1A1 0 -0.97360787

SERPINE1 0 -2.19433673

IL6 network

Genes Apical Basolateral

LYZ 9.194415271 8.206265476

FGG 9.442539772 4.939949825

C3 2.420111137 9.812669952

SRC 8.597152041 1.607829309

ENO2 7.836037534 0

CLU 2.680275601 4.273131229

HP 6.861194042 0

LDLR 6.458905458 0

FGB 2.686674822 3.413181538

HSPA5 4.709288841 0

FGA 2.9589437 0

178

TIMP1 0 1.899673763

MAP2K1 0 -0.89754651

CCL5 network

Genes Apical Basolateral

PNP 9.091763597 7.735244422

PLEC 6.359140982 5.813734245

ALCAM 4.009310395 5.420548473

CDC37 8.685029842 0

CAPN2 8.405192364 0

VASP 7.905543202 0

MMP9 7.490225024 0

TUBB4B 4.97592531 0

NAMPT 0 2.334650207

ERK1/2 network

Genes Apical Basolateral

CLEC11A 9.928306404 6.430759076

PCNA 11.21248647 2.657141916

C3 2.420111137 9.812669952

PSMB6 11.61729293 0

DKK1 0 8.870780213

MUC5AC 3.790641203 4.836091419

HSPA5 4.709288841 0

PKM 4.375380313 0

CALR 4.189261008 0

TIMP1 0 1.899673763

179

EZR 0 0.627653108

VNN1 -7.15159249 -0.90079403

180

APPENDIX C

Supplemental Figures and Tables: Chapter 4

Figure C1: Volcano plots for in-vitro experiment of log2 spectral count fold change versus – log10 pvalue calculated by Fisher’s exact test for: (A) U-MWCNT/control, (B) A-

181

MWCNT/control, and (C) U-MWCNT/A-MWCNT. Proteins plotted as a log2 fold change of

3 and -3 represent protein detection in exposed groups, but not control, and vice versa

(respectively). Positive fold change values represent up-regulation in exposed groups. Shaded areas highlight significance of p < 0.05.

182

Table C1: Method for MCX sample clean up.

volume wash step (mL) Reagent 1 1 Methanol

2 1 10% NH4OH in water 3 2 Methanol 4 3 0.1% formic acid in water 5 n Sample 6 1 0.1% formic acid in water 7 1 0.1% formic acid in methanol

8 1 10% NH4OH in methanol

183

Table C2: Gradient method used for LC.

Duration Flow (min) (nl/min) % B 2 300 0 70 300 40 1 300 80 5 300 80 1 300 0 1 300 0 10 300 0

184

t-test control versus uncoated MWCNTs Control Uncoated p-value log peak log peak Expressi Protein Peptide Description 10 10 Control vs area area on level Uncoated average average Protein S100-A8 MVTTECPQFVQNINIEN OS=Mus musculus 7.0050233 up 4.20018E- P27005 1 LFR GN=S100a8 PE=1 SV=3 15 regulated 08 - [S10A8_MOUSE] Myeloperoxidase OS=Mus musculus 7.9626742 up 7.0278E- P11247 IGLDLPALNMQR 1 GN=Mpo PE=2 SV=2 - 44 regulated 08 [PERM_MOUSE] Cathelin-related antimicrobial peptide 8.2810862 up 3.62182E- P51437 AVDDFNQQSLDTNLYR OS=Mus musculus 1 31 regulated 07 GN=Camp PE=2 SV=1 - [CRAMP_MOUSE] Protein S100-A9 OS=Mus musculus 7.7059655 9.1671556 up 4.11537E- P31725 SITTIIDTFHQYSR GN=S100a9 PE=1 SV=3 06 16 regulated 05 - [S10A9_MOUSE] Lactotransferrin OS=Mus 7.8798512 8.3354542 up 0.0011307 P08071 LRPVAAEVYGTK musculus GN=Ltf PE=2 81 71 regulated 27 SV=4 - [TRFL_MOUSE] Inter-alpha-trypsin inhibitor heavy chain H1 7.2877629 7.9054892 up 0.0013673 Q61702 AAVLGESAGLVR OS=Mus musculus 12 27 regulated 23 GN=Itih1 PE=1 SV=2 - [ITIH1_MOUSE] Neutrophil gelatinase- associated lipocalin 9.0569769 9.6250565 up 0.0034095 P11672 WYVVGLAGNAVQK OS=Mus musculus 16 23 regulated 71 GN=Lcn2 PE=1 SV=1 - [NGAL_MOUSE] Fibrinogen beta chain Q8K0E OS=Mus musculus 7.4345695 8.3989491 up 0.0034945 TPCTVSCNIPVVSGK 8 GN=Fgb PE=2 SV=1 - 63 71 regulated 72 [FIBB_MOUSE] Resistin-like alpha Q9EP9 OS=Mus musculus 8.0362870 9.5280989 up 0.0036016 ELLANPANYPSTVTK 5 GN=Retnla PE=1 SV=1 - 94 45 regulated 85 [RETNA_MOUSE] Histone H4 OS=Mus musculus GN=Hist1h4a 8.5094637 up 0.0056745 P62806 VFLENVIR 9.3048319 PE=1 SV=2 - 91 regulated 38 [H4_MOUSE] Vitamin D-binding protein OS=Mus musculus 9.2576021 9.6462679 up 0.0058785 P21614 SLSLILYSR GN=Gc PE=1 SV=2 - 94 93 regulated 89 [VTDB_MOUSE] H-2 class I histocompatibility antigen, Q10 alpha chain OS=Mus 7.2876282 7.6918380 up 0.0061572 P01898 TWTAADVAAIITR musculus GN=H2-Q10 67 75 regulated 08 PE=1 SV=3 - [HA10_MOUSE] 185

Inter alpha-trypsin inhibitor, heavy chain 4 7.5946862 8.1869846 up 0.0065765 A6X935 LGMYELLLK OS=Mus musculus 21 55 regulated 85 GN=Itih4 PE=1 SV=2 - [ITIH4_MOUSE] Alcohol dehydrogenase class 4 mu/sigma chain 7.4119145 7.0566178 down 0.0106695 Q64437 MLTYDPMLLFTGR OS=Mus musculus 33 39 regulated 91 GN=Adh7 PE=2 SV=2 - [ADH7_MOUSE] Apolipoprotein A-IV OS=Mus musculus 8.5832332 9.2414191 up 0.0204679 P06728 LQLTPYIQR GN=Apoa4 PE=2 SV=3 - 63 32 regulated 53 [APOA4_MOUSE] Complement C4-B OS=Mus musculus 9.0768622 up 0.0206227 P01029 LLVSAGSLYPAIAR 8.7376691 GN=C4b PE=1 SV=3 - 63 regulated 15 [CO4B_MOUSE] Pulmonary surfactant- associated protein B 10.077964 9.6267757 down 0.0208053 P50405 FLEQECDILPLK OS=Mus musculus 96 33 regulated 69 GN=Sftpb PE=2 SV=1 - [PSPB_MOUSE] Complement C3 OS=Mus 8.6443264 8.9482554 up 0.0208711 P01027 QIFSAEFEVK musculus GN=C3 PE=1 35 21 regulated 78 SV=3 - [CO3_MOUSE] Uteroglobin OS=Mus musculus GN=Scgb1a1 11.845086 11.258592 down 0.0231338 Q06318 ILTSPLCK PE=1 SV=1 - 07 59 regulated 1 [UTER_MOUSE] Annexin A2 OS=Mus musculus GN=Anxa2 8.6490847 8.3770390 down 0.0243254 P07356 TPAQYDASELK PE=1 SV=2 - 23 98 regulated 68 [ANXA2_MOUSE] WAP four-disulfide core domain protein 2 Q9DAU 9.8637464 9.3953019 down 0.0245583 QGTCPSVDIPK OS=Mus musculus 7 9 2 regulated 12 GN=Wfdc2 PE=2 SV=1 - [WFDC2_MOUSE] Lysozyme C-2 OS=Mus 10.009196 9.5850940 down 0.0286201 P08905 GDQSTDYGIFQINSR musculus GN=Lyz2 PE=1 67 5 regulated 11 SV=2 - [LYZ2_MOUSE] Cadherin-1 OS=Mus musculus GN=Cdh1 8.0005708 7.6877325 down 0.0409016 P09803 VTDDDAPNTPAWK PE=1 SV=1 - 05 79 regulated 49 [CADH1_MOUSE] Murinoglobulin-1 OS=Mus musculus 8.5068473 9.0242913 up 0.0463571 P28665 HVAYAVYSLSK GN=Mug1 PE=1 SV=3 - 99 54 regulated 08 [MUG1_MOUSE] CD166 antigen OS=Mus musculus GN=Alcam 7.2688688 6.3906861 down 0.0464464 Q61490 VLQPVEGEVAILFK PE=1 SV=3 - 5 22 regulated 52 [CD166_MOUSE] Complement component C9 OS=Mus musculus 8.2088143 8.4745837 up 0.0466563 P06683 TSNFNADFALK GN=C9 PE=1 SV=2 - 72 39 regulated 27 [CO9_MOUSE] Lymphocyte-specific protein 1 OS=Mus 7.7197105 7.1347489 down 0.0494928 P19973 TPSCQDIVAGDMSK musculus GN=Lsp1 PE=1 73 57 regulated 02 SV=2 - [LSP1_MOUSE]

186

t-test control versus AO-coated MWCNTs Control AO-coated p-value Protei log peak log peak Expressi Peptide Description 10 10 Control vs n area area on level AOcoated average average Cathelin-related antimicrobial peptide P5143 8.2797501 up 2.98756E- AVDDFNQQSLDTNLYR OS=Mus musculus 1 7 97 regulated 06 GN=Camp PE=2 SV=1 - [CRAMP_MOUSE] Myeloperoxidase OS=Mus P1124 8.1273301 up 3.96412E- IGLDLPALNMQR musculus GN=Mpo PE=2 1 7 11 regulated 05 SV=2 - [PERM_MOUSE] Protein S100-A8 OS=Mus P2700 MVTTECPQFVQNINIEN musculus GN=S100a8 7.2262927 up 5.10925E- 1 5 LFR PE=1 SV=3 - 32 regulated 05 [S10A8_MOUSE] Clusterin OS=Mus Q0689 8.2505590 8.8804523 up 7.97154E- ASGIIDTLFQDR musculus GN=Clu PE=1 0 85 6 regulated 05 SV=1 - [CLUS_MOUSE] Monocyte differentiation P1081 antigen CD14 OS=Mus 7.3180850 8.0466249 up 0.0012033 GLISALCPLK 0 musculus GN=Cd14 PE=1 2 36 regulated 58 SV=1 - [CD14_MOUSE] Complement C3 OS=Mus P0102 8.6443264 8.8405635 up 0.0017902 QIFSAEFEVK musculus GN=C3 PE=1 7 35 12 regulated 36 SV=3 - [CO3_MOUSE] Neutrophil gelatinase- associated lipocalin P1167 9.0569769 9.8634534 up 0.0025016 WYVVGLAGNAVQK OS=Mus musculus 2 16 21 regulated 6 GN=Lcn2 PE=1 SV=1 - [NGAL_MOUSE] Inter alpha-trypsin inhibitor, A6X93 heavy chain 4 OS=Mus 7.5946862 8.1336459 up 0.0032154 LGMYELLLK 5 musculus GN=Itih4 PE=1 21 72 regulated 07 SV=2 - [ITIH4_MOUSE] Protein S100-A9 OS=Mus P3172 musculus GN=S100a9 7.7059655 9.4344530 up 0.0046984 SITTIIDTFHQYSR 5 PE=1 SV=3 - 06 34 regulated 79 [S10A9_MOUSE] Fibrinogen beta chain Q8K0E OS=Mus musculus 7.4345695 8.3094160 up 0.0049506 TPCTVSCNIPVVSGK 8 GN=Fgb PE=2 SV=1 - 63 17 regulated 3 [FIBB_MOUSE] Lactotransferrin OS=Mus P0807 7.8798512 8.5107828 up 0.0100761 LRPVAAEVYGTK musculus GN=Ltf PE=2 1 81 34 regulated 92 SV=4 - [TRFL_MOUSE] Actin-related protein 2 P6116 OS=Mus musculus 6.8935433 7.5909747 up 0.0183229 LCYVGYNIEQEQK 1 GN=Actr2 PE=1 SV=1 - 64 25 regulated 18 [ARP2_MOUSE] Kallikrein 1-related peptidase b16 OS=Mus P0407 8.3633477 8.5851410 up 0.0421137 LDSTCLVSGWGSITPTK musculus GN=Klk1b16 1 05 15 regulated 15 PE=1 SV=2 - [K1B16_MOUSE]

187

t-test uncoated versus AO-coated MWCNTs Uncoated AO-coated p-value Expressio Protein Peptide Description log peak log peak area uncoated vs 10 10 n level area average average AOcoated Tubulin beta-5 chain OS=Mus ALTVPELTQQVFD musculus up P99024 7.131326248 7.679544083 0.001394774 AK GN=Tubb5 PE=1 regulated SV=1 - [TBB5_MOUSE] Monocyte differentiation antigen CD14 up P10810 GLISALCPLK OS=Mus musculus 7.713804903 8.046624936 0.011083482 regulated GN=Cd14 PE=1 SV=1 - [CD14_MOUSE]

188

Table C4: Protein count data with Benjamini corrected p-value for GO enrichment analysis terms.

Biological Process Control vs U-MWCNT Term Count Benjamini p- –Log10 Benjamini p- value value response to stress 9 0.031092671 1.507341972 response to external stimulus 7 0.018232985 1.739142224 immune response 6 0.022547382 1.646903883 immune effector process 4 0.02170684 1.6634034 regulation of response to stimulus 5 0.029521425 1.529862689 response to other organism 4 0.083941091 1.076025388 activation of immune response 3 0.093976607 1.026980241 taxis 3 0.12738425 0.894884266 response to biotic stimulus 4 0.118448593 0.926470094 regulation of immune system process 4 0.114328697 0.941844745 positive regulation of response to stimulus 3 0.237846291 0.623703617 positive regulation of immune system process 3 0.259738232 0.585464119

Biological Process Control vs A-MWCNT Term Count Benjamini –Log10 Benjamini p- value response to other organism 4 0.046324091 1.334193097 response to external stimulus 5 0.036006773 1.443615795 response to stress 6 0.028099547 1.551300678 response to biotic stimulus 4 0.02641378 1.578169439 response to chemical stimulus 5 0.047867662 1.319957786 immune response 3 0.43755505 0.358967299 taxis 2 0.556181235 0.254783668 immune effector process 2 0.560517941 0.251410482

189

Molecular Function Control vs U-MWCNT Term Count Benjamini –Log10 Benjamini p- value enzyme inhibitor activity 8 2.11E-06 5.674985435 endopeptidase inhibitor activity 6 9.09E-05 4.041506387 peptidase inhibitor activity 6 9.38E-05 4.027847196 enzyme regulator activity 8 7.47E-04 3.126465539 serine-type endopeptidase inhibitor activity 4 0.007627558 2.117614469 calcium ion binding 6 0.063491993 1.197281037 phospholipase inhibitor activity 2 0.067073018 1.173452153 lipase inhibitor activity 2 0.067073018 1.173452153 glycosaminoglycan binding 3 0.09274945 1.032688658 pattern binding 3 0.102399222 0.989703344 polysaccharide binding 3 0.102399222 0.989703344 carbohydrate binding 3 0.409063835 0.388208915 lipid binding 3 0.426946957 0.369626078

Molecular Function Control vs A-MWCNT Term Count Benjamini –Log10 Benjamini p- value glycosaminoglycan binding 3 0.130878728 0.883130935 polysaccharide binding 3 0.084291187 1.074217831 pattern binding 3 0.084291187 1.074217831 calcium ion binding 4 0.271163964 0.566768026 carbohydrate binding 3 0.224479367 0.648823572 heparin binding 2 0.441560666 0.35500962

190

APPENDIX D:

Supplemental Table from Chapter 5

Table D1: Top 10 protein PSM for control and corona reported for each experiment.

Exp1 ID Description PSM control PSM Corona ratio (control / corona) P07724 Serum albumin OS=Mus musculus1018 GN=Alb PE=1404 SV=3 - [ALBU_MOUSE]2.51980198 Q921I1 Serotransferrin OS=Mus musculus236 GN=Tf PE=1 SV=156 - [TRFE_MOUSE]4.214285714 P01027 Complement C3 OS=Mus musculus147 GN=C3 PE=1133 SV=3 - [CO3_MOUSE]1.105263158 P06684 Complement C5 OS=Mus musculus109 GN=C5 PE=190 SV=2 - [CO5_MOUSE]1.211111111 Q06318 Uteroglobin OS=Mus musculus GN=Scgb1a1109 PE=179 SV=1 - [UTER_MOUSE]1.379746835 P22599 Alpha-1-antitrypsin 1-2 OS=Mus musculus90 GN=Serpina1b38 PE=1 SV=2 - [A1AT2_MOUSE]2.368421053 Q00896 Alpha-1-antitrypsin 1-3 OS=Mus musculus90 GN=Serpina1c36 PE=1 SV=2 - [A1AT3_MOUSE]2.5 P07759 Serine protease inhibitor A3K OS=Mus87 musculus 40GN=Serpina3k PE=1 SV=2 - [SPA3K_MOUSE]2.175 Q61646 Haptoglobin OS=Mus musculus GN=Hp84 PE=1 SV=126 - [HPT_MOUSE] 3.230769231 P02088 Hemoglobin subunit beta-1 OS=Mus79 musculus GN=Hbb-b181 PE=1 SV=2 -0.975308642 [HBB1_MOUSE]

Exp2 ID Description PSM control PSM Corona ratio (control / corona) P07724 Serum albumin OS=Mus musculus GN=Alb1048 PE=1 SV=3311 - [ALBU_MOUSE] 3.36977492 Q921I1 Serotransferrin OS=Mus musculus GN=Tf205 PE=1 SV=122 - [TRFE_MOUSE] 9.318181818 P01027 Complement C3 OS=Mus musculus GN=C3120 PE=1 SV=393 - [CO3_MOUSE] 1.290322581 P07759 Serine protease inhibitor A3K OS=Mus115 musculus GN=Serpina3k134 PE=1 SV=2 0.858208955- [SPA3K_MOUSE] P60710 Actin, cytoplasmic 1 OS=Mus musculus112 GN=Actb PE=1131 SV=1 - [ACTB_MOUSE]0.854961832 P22599 Alpha-1-antitrypsin 1-2 OS=Mus musculus110 GN=Serpina1b36 PE=1 SV=2 - [A1AT2_MOUSE]3.055555556 P06684 Complement C5 OS=Mus musculus GN=C5107 PE=1 SV=250 - [CO5_MOUSE] 2.14 Q00897 Alpha-1-antitrypsin 1-4 OS=Mus musculus106 GN=Serpina1d33 PE=2 SV=1 - [A1AT4_MOUSE]3.212121212 P50405 Pulmonary surfactant-associated protein100 B OS=Mus musculus37 GN=Sftpb PE=22.702702703 SV=1 - [PSPB_MOUSE] P01942 Hemoglobin subunit alpha OS=Mus musculus95 GN=Hba29 PE=1 SV=2 - [HBA_MOUSE]3.275862069

191

Exp3 ID Description PSM control PSM Corona ratio (control / corona) P07724 Serum albumin OS=Mus musculus GN=Alb574 PE=1 SV=3246 - [ALBU_MOUSE] 2.333333333 Q921I1 Serotransferrin OS=Mus musculus GN=Tf137 PE=1 SV=172 - [TRFE_MOUSE] 1.902777778 P01027 Complement C3 OS=Mus musculus GN=C391 PE=1 SV=3102 - [CO3_MOUSE] 0.892156863 P02088 Hemoglobin subunit beta-1 OS=Mus musculus74 GN=Hbb-b164 PE=1 SV=2 - [HBB1_MOUSE]1.15625 P06684 Complement C5 OS=Mus musculus GN=C566 PE=1 SV=278 - [CO5_MOUSE] 0.846153846 P60710 Actin, cytoplasmic 1 OS=Mus musculus57 GN=Actb PE=11 SV=1 - [ACTB_MOUSE] 57 P07759 Serine protease inhibitor A3K OS=Mus56 musculus GN=Serpina3k84 PE=1 SV=2 0.666666667- [SPA3K_MOUSE] P22599 Alpha-1-antitrypsin 1-2 OS=Mus musculus46 GN=Serpina1b64 PE=1 SV=2 - [A1AT2_MOUSE]0.71875 Q00897 Alpha-1-antitrypsin 1-4 OS=Mus musculus45 GN=Serpina1d45 PE=2 SV=1 - [A1AT4_MOUSE]1 P35242 Pulmonary surfactant-associated protein44 A OS=Mus musculus39 GN=Sftpa1 PE=21.128205128 SV=1 - [SFTPA_MOUSE]

Exp4 ID Description PSM control PSM Corona ratio (control / corona) P07724 Serum albumin OS=Mus musculus GN=Alb396 PE=1 SV=3108 - [ALBU_MOUSE] 3.666666667 Q921I1 Serotransferrin OS=Mus musculus GN=Tf108 PE=1 SV=140 - [TRFE_MOUSE] 2.7 P07759 Serine protease inhibitor A3K OS=Mus93 musculus GN=Serpina3k46 PE=1 SV=2 - 2.02173913[SPA3K_MOUSE] P01027 Complement C3 OS=Mus musculus GN=C386 PE=1 SV=364 - [CO3_MOUSE] 1.34375 Q8VDD5 Myosin-9 OS=Mus musculus GN=Myh973 PE=1 SV=4 -104 [MYH9_MOUSE] 0.701923077 P08074 Carbonyl reductase [NADPH] 2 OS=Mus64 musculus GN=Cbr253 PE=1 SV=1 - [CBR2_MOUSE]1.20754717 P50405 Pulmonary surfactant-associated protein63 B OS=Mus musculus44 GN=Sftpb PE=21.431818182 SV=1 - [PSPB_MOUSE] P28665 Murinoglobulin-1 OS=Mus musculus GN=Mug160 PE=1 51SV=3 - [MUG1_MOUSE]1.176470588 P60710 Actin, cytoplasmic 1 OS=Mus musculus57 GN=Actb PE=172 SV=1 - [ACTB_MOUSE]0.791666667 P17563 Selenium-binding protein 1 OS=Mus musculus56 GN=Selenbp118 PE=1 SV=2 - [SBP1_MOUSE]3.111111111

192

Exp5 ID Description PSM control PSM Corona ratio (control / corona) P07724 Serum albumin OS=Mus musculus GN=Alb469 PE=1 SV=3160 - [ALBU_MOUSE] 2.93125 Q921I1 Serotransferrin OS=Mus musculus GN=Tf171 PE=1 SV=152 - [TRFE_MOUSE] 3.288461538 P07759 Serine protease inhibitor A3K OS=Mus147 musculus GN=Serpina3k54 PE=1 SV=2 2.722222222- [SPA3K_MOUSE] P01027 Complement C3 OS=Mus musculus GN=C3104 PE=1 SV=363 - [CO3_MOUSE] 1.650793651 P06684 Complement C5 OS=Mus musculus GN=C583 PE=1 SV=246 - [CO5_MOUSE] 1.804347826 Q8VDD5 Myosin-9 OS=Mus musculus GN=Myh967 PE=1 SV=4 -79 [MYH9_MOUSE] 0.848101266 P60710 Actin, cytoplasmic 1 OS=Mus musculus66 GN=Actb PE=180 SV=1 - [ACTB_MOUSE] 0.825 Q91X72 Hemopexin OS=Mus musculus GN=Hpx60 PE=1 SV=2 - 9[HEMO_MOUSE] 6.666666667 Q00896 Alpha-1-antitrypsin 1-3 OS=Mus musculus59 GN=Serpina1c51 PE=1 SV=2 - [A1AT3_MOUSE]1.156862745 P28665 Murinoglobulin-1 OS=Mus musculus GN=Mug159 PE=1 28SV=3 - [MUG1_MOUSE]2.107142857

193

APPENDIX E:

Supplemental Table from Chapter 6

Table E1: List of 35 candidate proteins that could be used for the development of a targeted mass spectrometry assay.

Proteins Proteins Actin-related protein 2/3 complex Fibronectin Aflatoxin B1 aldehyde reductase Glutathione-disulfide reducatse CCL5 Heme oxygenase-1 CD166 HSP60 CD44 IL-18 Chaperonin containing TCP1 IL-1beta Col1A1 IL-6 Col1A2 IL-8 Complement 1 Insulin-like growth factor- binding protein Complement 2 Mucin Complement 3 Myeloperoxidase Complement 4 Osteopontin Complement 5 Peroxiredoxin 1 Complement 9 S100 A13 Cullin 1 SOD2 Fibrillin 1 Stress induced phosphoprotein 1 Fibrinogen Thrombospondin 1 Thrombospondin 4

194

REFERENCES

1. Iijima S. Helical Microtubules of Graphitic Carbon. Nature. 1991;354(6348):56-8. doi:Doi 10.1038/354056a0.

2. Li Y, Zhou Z, Shen P, Chen Z. Spin gapless semiconductor-metal-half-metal properties in nitrogen-doped zigzag graphene nanoribbons. ACS nano. 2009;3(7):1952-8. doi:10.1021/nn9003428.

3. Charlier J-C, Blase X, Roche S. Electronic and transport properties of nanotubes. Reviews of Modern Physics. 2007;79(2):677-732. doi:10.1103/RevModPhys.79.677.

4. Choudhary V, Gupt A. Polymer/Carbon Nanotube Nanocomposites. 2011. doi:10.5772/18423.

5. Anantram MP, Léonard F. Physics of carbon nanotube electronic devices. Reports on Progress in Physics. 2006;69(3):507-61. doi:10.1088/0034-4885/69/3/r01.

6. José‐ Yacamán M, Miki‐ Yoshida M, Rendón L, Santiesteban JG. Catalytic growth of carbon microtubules with fullerene structure. Applied Physics Letters. 1993;62(6):657-9. doi:10.1063/1.108857.

7. Kumar M, Ando Y. Chemical Vapor Deposition of Carbon Nanotubes: A Review on Growth Mechanism and Mass Production. Journal of Nanoscience and Nanotechnology. 2010;10(6):3739-58. doi:10.1166/jnn.2010.2939.

8. Chico L, Crespi VH, Benedict LX, Louie SG, Cohen ML. Pure carbon nanoscale devices: Nanotube heterojunctions. Physical review letters. 1996;76(6):971-4. doi:10.1103/PhysRevLett.76.971.

9. Arepalli S. Laser Ablation Process for Single-Walled Carbon Nanotube Production. Journal of Nanoscience and Nanotechnology. 2004;4(4):317-25. doi:10.1166/jnn.2004.072.

10. P M Ajayan TWE. Nanometre-size tubes of carbon. Rep Prog Phys. 1997;60:1025-62.

11. Thess A, Lee R, Nikolaev P, Dai H, Petit P, Robert J et al. Crystalline Ropes of Metallic Carbon Nanotubes. Science. 1996;273(5274):483-7. doi:10.1126/science.273.5274.483.

195

12. Hou PX, Bai S, Yang QH, Liu C, Cheng HM. Multi-step purification of carbon nanotubes. Carbon. 2002;40(1):81-5. doi:10.1016/s0008-6223(01)00075-6.

13. Arora N, Sharma NN. Arc discharge synthesis of carbon nanotubes: Comprehensive review. Diamond and Related Materials. 2014;50:135-50. doi:10.1016/j.diamond.2014.10.001.

14. Eatemadi A, Daraee H, Karimkhanloo H, Kouhi M, Zarghami N, Akbarzadeh A et al. Carbon nanotubes: properties, synthesis, purification, and medical applications. Nanoscale research letters. 2014;9(1):393. doi:10.1186/1556-276X-9-393.

15. Naha S, Puri IK. A model for catalytic growth of carbon nanotubes. Journal of Physics D: Applied Physics. 2008;41(6):065304. doi:10.1088/0022-3727/41/6/065304.

16. Devine CK, Oldham CJ, Jur JS, Gong B, Parsons GN. Fiber containment for improved laboratory handling and uniform nanocoating of milligram quantities of carbon nanotubes by atomic layer deposition. Langmuir : the ACS journal of surfaces and colloids. 2011;27(23):14497-507. doi:10.1021/la202677u.

17. George SM. Atomic layer deposition: an overview. Chemical reviews. 2010;110(1):111- 31. doi:10.1021/cr900056b.

18. Spagnola JC, Gong B, Arvidson SA, Jur JS, Khan SA, Parsons GN. Surface and subsurface reactions during low temperature aluminium oxide atomic layer deposition on fiber- forming polymers. Journal of Materials Chemistry. 2010;20(20):4213. doi:10.1039/c0jm00355g.

19. Dandley EC, Taylor AJ, Duke KS, Ihrie MD, Shipkowski KA, Parsons GN et al. Atomic layer deposition coating of carbon nanotubes with zinc oxide causes acute phase immune responses in human monocytes in vitro and in mice after pulmonary exposure. Particle and fibre toxicology. 2016;13(1):29. doi:10.1186/s12989-016-0141-9.

20. Peng B, Locascio M, Zapol P, Li S, Mielke SL, Schatz GC et al. Measurements of near- ultimate strength for multiwalled carbon nanotubes and irradiation-induced crosslinking improvements. Nature nanotechnology. 2008;3(10):626-31. doi:10.1038/nnano.2008.211.

21. Wei BQ, Vajtai R, Ajayan PM. Reliability and current carrying capacity of carbon nanotubes. Applied Physics Letters. 2001;79(8):1172-4. doi:10.1063/1.1396632.

196

22. Balandin AA. Thermal properties of graphene and nanostructured carbon materials. Nature materials. 2011;10(8):569-81. doi:10.1038/nmat3064.

23. De Volder MF, Tawfick SH, Baughman RH, Hart AJ. Carbon nanotubes: present and future commercial applications. Science. 2013;339(6119):535-9. doi:10.1126/science.1222453.

24. Madani SY, Mandel A, Seifalian AM. A concise review of carbon nanotube's toxicology. Nano reviews. 2013;4. doi:10.3402/nano.v4i0.21521.

25. Bonner JC. Nanoparticles as a potential cause of pleural and interstitial lung disease. Proceedings of the American Thoracic Society. 2010;7(2):138-41. doi:10.1513/pats.200907- 061RM.

26. Tasis D, Tagmatarchis N, Bianco A, Prato M. Chemistry of carbon nanotubes. Chemical reviews. 2006;106(3):1105-36. doi:10.1021/cr050569o.

27. Pumera M. Carbon nanotubes contain residual metal catalyst nanoparticles even after washing with nitric acid at elevated temperature because these metal nanoparticles are sheathed by several graphene sheets. Langmuir : the ACS journal of surfaces and colloids. 2007;23(11):6453-8. doi:10.1021/la070088v.

28. Dahm MM, Evans DE, Schubauer-Berigan MK, Birch ME, Deddens JA. Occupational exposure assessment in carbon nanotube and nanofiber primary and secondary manufacturers: mobile direct-reading sampling. The Annals of occupational hygiene. 2013;57(3):328-44. doi:10.1093/annhyg/mes079.

29. Dahm MM, Schubauer-Berigan MK, Evans DE, Birch ME, Fernback JE, Deddens JA. Carbon Nanotube and Nanofiber Exposure Assessments: An Analysis of 14 Site Visits. The Annals of occupational hygiene. 2015;59(6):705-23. doi:10.1093/annhyg/mev020.

30. Erdely A, Dahm M, Chen BT, Zeidler-Erdely PC, Fernback JE, Birch ME et al. Carbon nanotube dosimetry: from workplace exposure assessment to inhalation toxicology. Particle and fibre toxicology. 2013;10(1):53. doi:10.1186/1743-8977-10-53.

31. WHO. Asbestos. 2017. http://www.who.int/ipcs/assessment/public_health/asbestos/en/.

197

32. Raghu G, Collard HR, Egan JJ, Martinez FJ, Behr J, Brown KK et al. An official ATS/ERS/JRS/ALAT statement: idiopathic pulmonary fibrosis: evidence-based guidelines for diagnosis and management. American journal of respiratory and critical care medicine. 2011;183(6):788-824. doi:10.1164/rccm.2009-040GL.

33. Nalysnyk L, Cid-Ruzafa J, Rotella P, Esser D. Incidence and prevalence of idiopathic pulmonary fibrosis: review of the literature. European respiratory review : an official journal of the European Respiratory Society. 2012;21(126):355-61. doi:10.1183/09059180.00002512.

34. Kendall RT, Feghali-Bostwick CA. Fibroblasts in fibrosis: novel roles and mediators. Frontiers in pharmacology. 2014;5:123. doi:10.3389/fphar.2014.00123.

35. Mercer RR, Hubbs AF, Scabilloni JF, Wang L, Battelli LA, Schwegler-Berry D et al. Distribution and persistence of pleural penetrations by multi-walled carbon nanotubes. Particle and fibre toxicology. 2010;7:28. doi:10.1186/1743-8977-7-28.

36. Tang S, Tang Y, Zhong L, Murat K, Asan G, Yu J et al. Short- and long-term toxicities of multi-walled carbon nanotubes in vivo and in vitro. Journal of applied toxicology : JAT. 2012;32(11):900-12. doi:10.1002/jat.2748.

37. Mercer RR, Hubbs AF, Scabilloni JF, Wang L, Battelli LA, Friend S et al. Pulmonary fibrotic response to aspiration of multi-walled carbon nanotubes. Particle and fibre toxicology. 2011;8:21. doi:10.1186/1743-8977-8-21.

38. Searl A. Biopersistence and durability of nine mineral fibre types in rat lungs over 12 months. The Annals of occupational hygiene. 1999;43(3):143-53. doi:10.1016/s0003- 4878(99)00017-4.

39. Donaldson K, Murphy FA, Duffin R, Poland CA. Asbestos, carbon nanotubes and the pleural mesothelium: a review of the hypothesis regarding the role of long fibre retention in the parietal pleura, inflammation and mesothelioma. Particle and fibre toxicology. 2010;7:5. doi:10.1186/1743-8977-7-5.

40. Muller J, Huaux F, Moreau N, Misson P, Heilier JF, Delos M et al. Respiratory toxicity of multi-wall carbon nanotubes. Toxicology and applied pharmacology. 2005;207(3):221-31. doi:10.1016/j.taap.2005.01.008.

198

41. Ryman-Rasmussen JP, Cesta MF, Brody AR, Shipley-Phillips JK, Everitt JI, Tewksbury EW et al. Inhaled carbon nanotubes reach the subpleural tissue in mice. Nature nanotechnology. 2009;4(11):747-51. doi:10.1038/nnano.2009.305.

42. Shvedova AA, Kisin ER, Mercer R, Murray AR, Johnson VJ, Potapovich AI et al. Unusual inflammatory and fibrogenic pulmonary responses to single-walled carbon nanotubes in mice. American journal of physiology Lung cellular and molecular physiology. 2005;289(5):L698-708. doi:10.1152/ajplung.00084.2005.

43. Shvedova AA, Kisin E, Murray AR, Johnson VJ, Gorelik O, Arepalli S et al. Inhalation vs. aspiration of single-walled carbon nanotubes in C57BL/6 mice: inflammation, fibrosis, oxidative stress, and mutagenesis. American journal of physiology Lung cellular and molecular physiology. 2008;295(4):L552-65. doi:10.1152/ajplung.90287.2008.

44. Ma-Hock L, Treumann S, Strauss V, Brill S, Luizi F, Mertler M et al. Inhalation toxicity of multiwall carbon nanotubes in rats exposed for 3 months. Toxicological sciences : an official journal of the Society of Toxicology. 2009;112(2):468-81. doi:10.1093/toxsci/kfp146.

45. Pauluhn J. Subchronic 13-week inhalation exposure of rats to multiwalled carbon nanotubes: toxic effects are determined by density of agglomerate structures, not fibrillar structures. Toxicological sciences : an official journal of the Society of Toxicology. 2010;113(1):226-42. doi:10.1093/toxsci/kfp247.

46. Porter DW, Hubbs AF, Mercer RR, Wu N, Wolfarth MG, Sriram K et al. Mouse pulmonary dose- and time course-responses induced by exposure to multi-walled carbon nanotubes. Toxicology. 2010;269(2-3):136-47. doi:10.1016/j.tox.2009.10.017.

47. Delorme MP, Muro Y, Arai T, Banas DA, Frame SR, Reed KL et al. Ninety-day inhalation toxicity study with a vapor grown carbon nanofiber in rats. Toxicological sciences : an official journal of the Society of Toxicology. 2012;128(2):449-60. doi:10.1093/toxsci/kfs172.

48. Murray AR, Kisin ER, Tkach AV, Yanamala N, Mercer R, Young SH et al. Factoring-in agglomeration of carbon nanotubes and nanofibers for better prediction of their toxicity versus asbestos. Particle and fibre toxicology. 2012;9:10. doi:10.1186/1743-8977-9-10.

49. Snyder-Talkington BN, Qian Y, Castranova V, Guo NL. New perspectives for in vitro risk assessment of multiwalled carbon nanotubes: application of coculture and

199

bioinformatics. Journal of toxicology and environmental health Part B, Critical reviews. 2012;15(7):468-92. doi:10.1080/10937404.2012.736856.

50. Reduce, refine, replace. Nature immunology. 2010;11(11):971. doi:10.1038/ni1110-971.

51. Sewell F, Edwards J, Prior H, Robinson S. Opportunities to Apply the 3Rs in Safety Assessment Programs. ILAR journal. 2016;57(2):234-45. doi:10.1093/ilar/ilw024.

52. Jud C, Clift MJ, Petri-Fink A, Rothen-Rutishauser B. Nanomaterials and the human lung: what is known and what must be deciphered to realise their potential advantages? Swiss medical weekly. 2013;143:w13758. doi:10.4414/smw.2013.13758.

53. He X, Young SH, Schwegler-Berry D, Chisholm WP, Fernback JE, Ma Q. Multiwalled carbon nanotubes induce a fibrogenic response by stimulating reactive oxygen species production, activating NF-kappaB signaling, and promoting fibroblast-to-myofibroblast transformation. Chemical research in toxicology. 2011;24(12):2237-48. doi:10.1021/tx200351d.

54. Pacurari M, Qian Y, Fu W, Schwegler-Berry D, Ding M, Castranova V et al. Cell permeability, migration, and reactive oxygen species induced by multiwalled carbon nanotubes in human microvascular endothelial cells. Journal of toxicology and environmental health Part A. 2012;75(2):112-28. doi:10.1080/15287394.2011.615110.

55. Ye SF, Wu YH, Hou ZQ, Zhang QQ. ROS and NF-kappaB are involved in upregulation of IL-8 in A549 cells exposed to multi-walled carbon nanotubes. Biochemical and biophysical research communications. 2009;379(2):643-8. doi:10.1016/j.bbrc.2008.12.137.

56. Hilton GM, Taylor AJ, Hussain S, Dandley EC, Griffith EH, Garantziotis S et al. Mapping differential cellular protein response of mouse alveolar epithelial cells to multi- walled carbon nanotubes as a function of atomic layer deposition coating. Nanotoxicology. 2017;11(3):313-26. doi:10.1080/17435390.2017.1299888.

57. Taylor AJ, McClure CD, Shipkowski KA, Thompson EA, Hussain S, Garantziotis S et al. Atomic layer deposition coating of carbon nanotubes with aluminum oxide alters pro- fibrogenic cytokine expression by human mononuclear phagocytes in vitro and reduces lung fibrosis in mice in vivo. PloS one. 2014;9(9):e106870. doi:10.1371/journal.pone.0106870.

200

58. Vietti G, Lison D, van den Brule S. Mechanisms of lung fibrosis induced by carbon nanotubes: towards an Adverse Outcome Pathway (AOP). Particle and fibre toxicology. 2016;13:11. doi:10.1186/s12989-016-0123-y.

59. Azad N, Iyer AK, Wang L, Liu Y, Lu Y, Rojanasakul Y. Reactive oxygen species- mediated p38 MAPK regulates carbon nanotube-induced fibrogenic and angiogenic responses. Nanotoxicology. 2013;7(2):157-68. doi:10.3109/17435390.2011.647929.

60. Wang P, Nie X, Wang Y, Li Y, Ge C, Zhang L et al. Multiwall carbon nanotubes mediate macrophage activation and promote pulmonary fibrosis through TGF-beta/Smad signaling pathway. Small. 2013;9(22):3799-811. doi:10.1002/smll.201300607.

61. Hussain S, Sangtian S, Anderson SM, Snyder RJ, Marshburn JD, Rice AB et al. Inflammasome activation in airway epithelial cells after multi-walled carbon nanotube exposure mediates a profibrotic response in lung fibroblasts. Particle and fibre toxicology. 2014;11:28. doi:10.1186/1743-8977-11-28.

62. Horvath L, Umehara Y, Jud C, Blank F, Petri-Fink A, Rothen-Rutishauser B. Engineering an in vitro air-blood barrier by 3D bioprinting. Scientific reports. 2015;5:7974. doi:10.1038/srep07974.

63. Hermanns MI, Unger RE, Kehe K, Peters K, Kirkpatrick CJ. Lung epithelial cell lines in coculture with human pulmonary microvascular endothelial cells: development of an alveolo- capillary barrier in vitro. Laboratory investigation; a journal of technical methods and pathology. 2004;84(6):736-52. doi:10.1038/labinvest.3700081.

64. Klein SG, Serchi T, Hoffmann L, Blomeke B, Gutleb AC. An improved 3D tetraculture system mimicking the cellular organisation at the alveolar barrier to study the potential toxic effects of particles on the lung. Particle and fibre toxicology. 2013;10:31. doi:10.1186/1743- 8977-10-31.

65. Rothen-Rutishauser B, Blank F, Muhlfeld C, Gehr P. In vitro models of the human epithelial airway barrier to study the toxic potential of particulate matter. Expert opinion on drug metabolism & toxicology. 2008;4(8):1075-89. doi:10.1517/17425255.4.8.1075.

66. Chortarea S, Clift MJ, Vanhecke D, Endes C, Wick P, Petri-Fink A et al. Repeated exposure to carbon nanotube-based aerosols does not affect the functional properties of a 3D human epithelial airway model. Nanotoxicology. 2015;9(8):983-93. doi:10.3109/17435390.2014.993344.

201

67. Hinderliter PM, Minard KR, Orr G, Chrisler WB, Thrall BD, Pounds JG et al. ISDD: A computational model of particle sedimentation, diffusion and target cell dosimetry for in vitro toxicity studies. Particle and fibre toxicology. 2010;7(1):36. doi:10.1186/1743-8977-7- 36.

68. Polk WW, Sharma M, Sayes CM, Hotchkiss JA, Clippinger AJ. Aerosol generation and characterization of multi-walled carbon nanotubes exposed to cells cultured at the air-liquid interface. Particle and fibre toxicology. 2016;13:20. doi:10.1186/s12989-016-0131-y.

69. James P. Protein identification in the post-genome era: the rapid rise of proteomics. Quarterly Reviews of Biophysics. 1997;30(4):279-331.

70. Fleischmann R, Adams M, White O, Clayton R, Kirkness E, Kerlavage A et al. Whole- genome random sequencing and assembly of Haemophilus influenzae Rd. Science. 1995;269(5223):496-512. doi:10.1126/science.7542800.

71. Lander ES, Linton LM, Birren B, Nusbaum C, Zody MC, Baldwin J et al. Initial sequencing and analysis of the human genome. Nature. 2001;409(6822):860-921. doi:10.1038/35057062.

72. Venter JC, Adams MD, Myers EW, Li PW, Mural RJ, Sutton GG et al. The sequence of the human genome. Science. 2001;291(5507):1304-51. doi:10.1126/science.1058040.

73. Schena M, Shalon D, Davis RW, Brown PO. Quantitative Monitoring of Gene Expression Patterns with a Complementary DNA Microarray. Science. 1995;270(5235):467- 70. doi:10.1126/science.270.5235.467.

74. Bowtell DD. Options available--from start to finish--for obtaining expression data by microarray. Nature genetics. 1999;21(1 Suppl):25-32. doi:10.1038/4455.

75. Duggan DJ, Bittner M, Chen Y, Meltzer P, Trent JM. Expression profiling using cDNA microarrays. Nature genetics. 1999;21:10-4. doi:10.1038/4434.

76. Gygi SP, Rochon Y, Franza BR, Aebersold R. Correlation between Protein and mRNA Abundance in Yeast. Mol Cell Biol. 1999;19(3):1720-30. doi:10.1128/mcb.19.3.1720.

77. J L Hargrove FHS. The role of mRNA and protein stability in gene expression. The FASEB Journal. 1989;3:2360-70. 202

78. Schwanhausser B, Busse D, Li N, Dittmar G, Schuchhardt J, Wolf J et al. Global quantification of mammalian gene expression control. Nature. 2011;473(7347):337-42. doi:10.1038/nature10098.

79. Lithwick G, Margalit H. Hierarchy of sequence-dependent features associated with prokaryotic translation. Genome research. 2003;13(12):2665-73. doi:10.1101/gr.1485203.

80. Yarmush ML, Jayaraman A. Advances in proteomic technologies. Annual review of biomedical engineering. 2002;4:349-73. doi:10.1146/annurev.bioeng.4.020702.153443.

81. Yates JR, Ruse CI, Nakorchevsky A. Proteomics by mass spectrometry: approaches, advances, and applications. Annual review of biomedical engineering. 2009;11:49-79. doi:10.1146/annurev-bioeng-061008-124934.

82. Merrick BA, Witzmann FA. The role of toxicoproteomics in assessing organ specific toxicity. 2009;99:367-400. doi:10.1007/978-3-7643-8336-7_13.

83. Schmidt A, Forne I, Imhof A. Bioinformatic analysis of proteomics data. BMC systems biology. 2014;8 Suppl 2:S3. doi:10.1186/1752-0509-8-S2-S3.

84. VerBerkmoes NC, Bundy JL, Hauser L, Asano KG, Razumovskaya J, Larimer F et al. Integrating “Top-Down” and “Bottom-Up” Mass Spectrometric Approaches for Proteomic Analysis ofShewanella oneidensis. Journal of proteome research. 2002;1(3):239-52. doi:10.1021/pr025508a.

85. Wilkins MR, Sanchez J-C, Gooley AA, Appel RD, Humphery-Smith I, Hochstrasser DF et al. Progress with Proteome Projects: Why all Proteins Expressed by a Genome Should be Identified and How To Do It. Biotechnology and Genetic Engineering Reviews. 1996;13(1):19-50. doi:10.1080/02648725.1996.10647923.

86. Klose J, Kobalz U. Two-dimensional electrophoresis of proteins: An updated protocol and implications for a functional analysis of the genome. Electrophoresis. 1995;16(1):1034- 59. doi:10.1002/elps.11501601175.

87. Niessen WMA, Tinke AP. Liquid chromatography-mass spectrometry General principles and instrumentation. Journal of Chromatography A. 1995;703(1-2):37-57. doi:10.1016/0021- 9673(94)01198-n.

203

88. Griffiths J. A brief history of mass spectrometry. Analytical chemistry. 2008;80(15):5678-83. doi:10.1021/ac8013065.

89. Barber M, Bordoli RS, Sedgwick RD, Tyler AN. Fast atom bombardment of solids as an ion source in mass spectrometry. Nature. 1981;293(5830):270-5. doi:10.1038/293270a0.

90. Fenn J, Mann M, Meng C, Wong S, Whitehouse C. Electrospray ionization for mass spectrometry of large biomolecules. Science. 1989;246(4926):64-71. doi:10.1126/science.2675315.

91. Felitsyn N, Peschke M, Kebarle P. Origin and number of charges observed on multiply- protonated native proteins produced by ESI. International Journal of Mass Spectrometry. 2002;219(1):39-62. doi:10.1016/s1387-3806(02)00588-2.

92. Han X, Aslanian A, Yates JR, 3rd. Mass spectrometry for proteomics. Current opinion in chemical biology. 2008;12(5):483-90. doi:10.1016/j.cbpa.2008.07.024.

93. Yates JR, 3rd. Mass spectral analysis in proteomics. Annual review of biophysics and biomolecular structure. 2004;33:297-316. doi:10.1146/annurev.biophys.33.111502.082538.

94. Liu T, Belov ME, Jaitly N, Qian WJ, Smith RD. Accurate mass measurements in proteomics. Chemical reviews. 2007;107(8):3621-53. doi:10.1021/cr068288j.

95. Perry RH, Cooks RG, Noll RJ. Orbitrap mass spectrometry: instrumentation, ion motion and applications. Mass spectrometry reviews. 2008;27(6):661-99. doi:10.1002/mas.20186.

96. Brancia FL. Recent developments in ion-trap mass spectrometry and related technologies. Expert review of proteomics. 2006;3(1):143-51. doi:10.1586/14789450.3.1.143.

97. Domon B, Aebersold R. Mass spectrometry and protein analysis. Science. 2006;312(5771):212-7. doi:10.1126/science.1124619.

98. Makarov A. Electrostatic Axially Harmonic Orbital Trapping: A High-Performance Technique of Mass Analysis. Analytical chemistry. 2000;72(6):1156-62. doi:10.1021/ac991131p.

204

99. Zubarev RA, Makarov A. Orbitrap mass spectrometry. Analytical chemistry. 2013;85(11):5288-96. doi:10.1021/ac4001223.

100. Hu Q, Noll RJ, Li H, Makarov A, Hardman M, Graham Cooks R. The Orbitrap: a new mass spectrometer. Journal of mass spectrometry : JMS. 2005;40(4):430-43. doi:10.1002/jms.856.

101. Senko MW, Canterbury JD, Guan S, Marshall AG. A High-performance Modular Data System for Fourier Transform Ion Cyclotron Resonance Mass Spectrometry. Rapid Communications in Mass Spectrometry. 1996;10(14):1839-44. doi:10.1002/(sici)1097- 0231(199611)10:14<1839::aid-rcm718>3.0.co;2-v.

102. Michalski A, Damoc E, Hauschild JP, Lange O, Wieghaus A, Makarov A et al. Mass spectrometry-based proteomics using Q Exactive, a high-performance benchtop quadrupole Orbitrap mass spectrometer. Molecular & cellular proteomics : MCP. 2011;10(9):M111 011015. doi:10.1074/mcp.M111.011015.

103. Aebersold R, Mann M. Mass spectrometry-based proteomics. Nature. 2003;422(6928):198-207. doi:10.1038/nature01511.

104. Hale JE. Advantageous uses of mass spectrometry for the quantification of proteins. International journal of proteomics. 2013;2013:219452. doi:10.1155/2013/219452.

105. Kumar C, Mann M. Bioinformatics analysis of mass spectrometry-based proteomics data sets. FEBS Lett. 2009;583(11):1703-12. doi:10.1016/j.febslet.2009.03.035.

106. Perkins DN, Pappin DJC, Creasy DM, Cottrell JS. Probability-based protein identification by searching sequence databases using mass spectrometry data. Electrophoresis. 1999;20(18):3551-67. doi:10.1002/(sici)1522- 2683(19991201)20:18<3551::aid-elps3551>3.0.co;2-2.

107. Eng JK, McCormack AL, Yates JR. An approach to correlate tandem mass spectral data of peptides with amino acid sequences in a protein database. Journal of the American Society for Mass Spectrometry. 1994;5(11):976-89. doi:10.1016/1044-0305(94)80016-2.

108. Xu C, Ma B. Software for computational peptide identification from MS-MS data. Drug discovery today. 2006;11(13-14):595-600. doi:10.1016/j.drudis.2006.05.011.

205

109. Craig R, Beavis RC. TANDEM: matching proteins with tandem mass spectra. Bioinformatics. 2004;20(9):1466-7. doi:10.1093/bioinformatics/bth092.

110. Cox J, Neuhauser N, Michalski A, Scheltema RA, Olsen JV, Mann M. Andromeda: a peptide search engine integrated into the MaxQuant environment. Journal of proteome research. 2011;10(4):1794-805. doi:10.1021/pr101065j.

111. Bereman MS, Johnson R, Bollinger J, Boss Y, Shulman N, MacLean B et al. Implementation of statistical process control for proteomic experiments via LC MS/MS. Journal of the American Society for Mass Spectrometry. 2014;25(4):581-7. doi:10.1007/s13361-013-0824-5.

112. Karpievitch YV, Dabney AR, Smith RD. Normalization and missing value imputation for label-free LC-MS analysis. BMC bioinformatics. 2012;13 Suppl 16:S5. doi:10.1186/1471-2105-13-S16-S5.

113. Zhang B, VerBerkmoes NC, Langston MA, Uberbacher E, Hettich RL, Samatova NF. Detecting differential and correlated protein expression in label-free shotgun proteomics. Journal of proteome research. 2006;5(11):2909-18. doi:10.1021/pr0600273.

114. Ting L, Cowley MJ, Hoon SL, Guilhaus M, Raftery MJ, Cavicchioli R. Normalization and statistical analysis of quantitative proteomics data generated by metabolic labeling. Molecular & cellular proteomics : MCP. 2009;8(10):2227-42. doi:10.1074/mcp.M800462- MCP200.

115. Pursiheimo A, Vehmas AP, Afzal S, Suomi T, Chand T, Strauss L et al. Optimization of Statistical Methods Impact on Quantitative Proteomics Data. Journal of proteome research. 2015;14(10):4118-26. doi:10.1021/acs.jproteome.5b00183.

116. Shannon P, Markiel A, Ozier O, Baliga NS, Wang JT, Ramage D et al. Cytoscape: a software environment for integrated models of biomolecular interaction networks. Genome research. 2003;13(11):2498-504. doi:10.1101/gr.1239303.

117. Huang da W, Sherman BT, Lempicki RA. Systematic and integrative analysis of large gene lists using DAVID bioinformatics resources. Nature protocols. 2009;4(1):44-57. doi:10.1038/nprot.2008.211.

206

118. Szklarczyk D, Franceschini A, Wyder S, Forslund K, Heller D, Huerta-Cepas J et al. STRING v10: protein-protein interaction networks, integrated over the tree of life. Nucleic acids research. 2015;43(Database issue):D447-52. doi:10.1093/nar/gku1003.

119. IPA. Ingenutiry Pathway Analysis. QIAGEN. 2016. www.qiagen.com/ingenuity. Accessed 05/05/2016 2016.

120. Wu X, Hasan MA, Chen JY. Pathway and network analysis in proteomics. Journal of theoretical biology. 2014;362:44-52. doi:10.1016/j.jtbi.2014.05.031.

121. Donaldson K, Aitken R, Tran L, Stone V, Duffin R, Forrest G et al. Carbon nanotubes: a review of their properties in relation to pulmonary toxicology and workplace safety. Toxicological sciences : an official journal of the Society of Toxicology. 2006;92(1):5-22. doi:10.1093/toxsci/kfj130.

122. Prato M, Kostarelos K, Bianco A. Functionalized carbon nanotubes in drug design and discovery. Accounts of chemical research. 2008;41(1):60-8. doi:10.1021/ar700089b.

123. Singh S, Kruse P. Carbon nanotube surface science. International Journal of Nanotechnology. 2008;5(9/10/11/12):900. doi:10.1504/ijnt.2008.019826.

124. Hyde GK, McCullen SD, Jeon S, Stewart SM, Jeon H, Loboa EG et al. Atomic layer deposition and biocompatibility of titanium nitride nano-coatings on cellulose fiber substrates. Biomedical materials. 2009;4(2):025001. doi:10.1088/1748-6041/4/2/025001.

125. Peng Q, Sun XY, Spagnola JC, Hyde GK, Spontak RJ, Parsons GN. Atomic layer deposition on electrospun polymer fibers as a direct route to AL2O3 microtubes with precise wall thickness control. Nano letters. 2007;7(3):719-22. doi:10.1021/nl062948i.

126. Parsons GN, George SM, Knez M. Progress and future directions for atomic layer deposition and ALD-based chemistry. MRS Bulletin. 2011;36(11):865-71. doi:10.1557/mrs.2011.238.

127. Poland CA, Duffin R, Kinloch I, Maynard A, Wallace WA, Seaton A et al. Carbon nanotubes introduced into the abdominal cavity of mice show asbestos-like pathogenicity in a pilot study. Nature nanotechnology. 2008;3(7):423-8. doi:10.1038/nnano.2008.111.

207

128. WHO. Asbestos. World Health Organization. 2015. http://www.who.int/ipcs/assessment/public_health/asbestos/en/. Accessed 12/18/2015 2015.

129. NIH. What Is Idiopathic Pulmonary Fibrosis? 2011. http://www.nhlbi.nih.gov/health/health-topics/topics/ipf/. Accessed 11/20/2014.

130. Cheresh P, Kim SJ, Tulasiram S, Kamp DW. Oxidative stress and pulmonary fibrosis. Biochimica et biophysica acta. 2013;1832(7):1028-40. doi:10.1016/j.bbadis.2012.11.021.

131. Dandley EC, Taylor AJ, Duke KS, Ihrie MD, Shipkowski KA, Parsons GN et al. Atomic layer deposition coating of carbon nanotubes with zinc oxide causes acute phase immune responses in human monocytes in vitro and in mice after pulmonary exposure. Particle and fibre toxicology. 2016;13(1). doi:10.1186/s12989-016-0141-9.

132. Costa PM, Fadeel B. Emerging systems biology approaches in nanotoxicology: Towards a mechanism-based understanding of nanomaterial hazard and risk. Toxicology and applied pharmacology. 2015. doi:10.1016/j.taap.2015.12.014.

133. Yamamoto K, Ferrari JD, Cao Y, Ramirez MI, Jones MR, Quinton LJ et al. Type I alveolar epithelial cells mount innate immune responses during pneumococcal pneumonia. Journal of immunology. 2012;189(5):2450-9. doi:10.4049/jimmunol.1200634.

134. Ryman-Rasmussen JP, Tewksbury EW, Moss OR, Cesta MF, Wong BA, Bonner JC. Inhaled multiwalled carbon nanotubes potentiate airway fibrosis in murine allergic asthma. American journal of respiratory cell and molecular biology. 2009;40(3):349-58. doi:10.1165/rcmb.2008-0276OC.

135. Gong B, Peng Q, Jur JS, Devine CK, Lee K, Parsons GN. Sequential Vapor Infiltration of Metal Oxides into Sacrificial Polyester Fibers: Shape Replication and Controlled Porosity of Microporous/Mesoporous Oxide Monoliths. Chemistry of Materials. 2011;23(15):3476- 85. doi:10.1021/cm200694w.

136. Jur JS, Spagnola JC, Lee K, Gong B, Peng Q, Parsons GN. Temperature-dependent subsurface growth during atomic layer deposition on polypropylene and cellulose fibers. Langmuir : the ACS journal of surfaces and colloids. 2010;26(11):8239-44. doi:10.1021/la904604z.

137. NCSU. Center for Electron Microscopy. NCSU. 2005. https://www.ncsu.edu/cem/index.html. Accessed 12/08/2015 2015. 208

138. Gumpertz FGGaML. Planning, Construction, and Statistical Analysis of Comparative Experiments. . John Wiley & Sons; 2004.

139. Waters. Oasis Sample Extraction Products. 2015. http://www.waters.com/waters/en_US/Oasis-Sample-Extraction- Products/nav.htm?cid=513209&locale=en_US. Accessed 12/30/2015 2015.

140. Scientific T. Pierce LDH Cytotoxicity Assay Kit. 2015. https://www.thermofisher.com/order/catalog/product/88953. Accessed 09/30/2015 2015.

141. Bereman MS, Canterbury JD, Egertson JD, Horner J, Remes PM, Schwartz J et al. Evaluation of front-end higher energy collision-induced dissociation on a benchtop dual- pressure linear ion trap mass spectrometer for shotgun proteomics. Analytical chemistry. 2012;84(3):1533-9. doi:10.1021/ac203210a.

142. MacLean B, Tomazela DM, Shulman N, Chambers M, Finney GL, Frewen B et al. Skyline: an open source document editor for creating and analyzing targeted proteomics experiments. Bioinformatics. 2010;26(7):966-8. doi:10.1093/bioinformatics/btq054.

143. Michalski A, Cox J, Mann M. More than 100,000 detectable peptide species elute in single shotgun proteomics runs but the majority is inaccessible to data-dependent LC- MS/MS. Journal of proteome research. 2011;10(4):1785-93. doi:10.1021/pr101060v.

144. Bairoch A. The SWISS-PROT protein sequence data bank and its new supplement TREMBL. Nucleic acids research. 1996;24(1):21-5. doi:10.1093/nar/24.1.21.

145. Kall L, Canterbury JD, Weston J, Noble WS, MacCoss MJ. Semi-supervised learning for peptide identification from shotgun proteomics datasets. Nature methods. 2007;4(11):923-5. doi:10.1038/nmeth1113.

146. Zhang B, Chambers MC, Tabb DL. Proteomic parsimony through bipartite graph analysis improves accuracy and transparency. Journal of proteome research. 2007;6(9):3549- 57. doi:10.1021/pr070230d.

147. Gilbert B, Fakra SC, Xia T, Pokhrel S, Madler L, Nel AE. The fate of ZnO nanoparticles administered to human bronchial epithelial cells. ACS nano. 2012;6(6):4921- 30. doi:10.1021/nn300425a.

209

148. Xia T, Kovochich M, Liong M, Madler L, Gilbert B, Shi H et al. Comparison of the mechanism of toxicity of zinc oxide and cerium oxide nanoparticles based on dissolution and oxidative stress properties. ACS nano. 2008;2(10):2121-34. doi:10.1021/nn800511k.

149. Kensler TW, Wakabayashi N, Biswal S. Cell survival responses to environmental stresses via the Keap1-Nrf2-ARE pathway. Annual review of pharmacology and toxicology. 2007;47:89-116. doi:10.1146/annurev.pharmtox.46.120604.141046.

150. Dong J, Ma Q. Suppression of basal and carbon nanotube-induced oxidative stress, inflammation and fibrosis in mouse lungs by Nrf2. Nanotoxicology. 2015:1-11. doi:10.3109/17435390.2015.1110758.

151. van Berlo D, Wilhelmi V, Boots AW, Hullmann M, Kuhlbusch TA, Bast A et al. Apoptotic, inflammatory, and fibrogenic effects of two different types of multi-walled carbon nanotubes in mouse lung. Archives of toxicology. 2014;88(9):1725-37. doi:10.1007/s00204- 014-1220-z.

152. Dinarello CA. Interleukin-1 in the pathogenesis and treatment of inflammatory diseases. Blood. 2011;117(14):3720-32. doi:10.1182/blood-2010-07-273417.

153. Hirano S, Fujitani Y, Furuyama A, Kanno S. Uptake and cytotoxic effects of multi- walled carbon nanotubes in human bronchial epithelial cells. Toxicology and applied pharmacology. 2010;249(1):8-15. doi:10.1016/j.taap.2010.08.019.

154. Boyles M, Stoehr L, Schlinkert P, Himly M, Duschl A. The Significance and Insignificance of Carbon Nanotube-Induced Inflammation. Fibers. 2014;2(1):45-74. doi:10.3390/fib2010045.

155. Girtsman TA, Beamer CA, Wu N, Buford M, Holian A. IL-1R signalling is critical for regulation of multi-walled carbon nanotubes-induced acute lung inflammation in C57Bl/6 mice. Nanotoxicology. 2014;8(1):17-27. doi:10.3109/17435390.2012.744110.

156. Hamilton RF, Jr., Buford M, Xiang C, Wu N, Holian A. NLRP3 inflammasome activation in murine alveolar macrophages and related lung pathology is associated with MWCNT nickel contamination. Inhalation toxicology. 2012;24(14):995-1008. doi:10.3109/08958378.2012.745633.

210

157. Han J, Lee J, Bibbs L, Ulevitch R. A MAP kinase targeted by endotoxin and hyperosmolarity in mammalian cells. Science. 1994;265(5173):808-11. doi:10.1126/science.7914033.

158. Smith JS, Gorbett D, Mueller J, Perez R, Daniels CJ. Pulmonary hypertension and idiopathic pulmonary fibrosis: a dastardly duo. The American journal of the medical sciences. 2013;346(3):221-5. doi:10.1097/MAJ.0b013e31827871dc.

159. Iozzo RV, San Antonio JD. Heparan sulfate proteoglycans: heavy hitters in the angiogenesis arena. The Journal of clinical investigation. 2001;108(3):349-55. doi:10.1172/JCI13738.

160. Lawler J. Thrombospondin-1 as an endogenous inhibitor of angiogenesis and tumor growth. Journal of Cellular and Molecular Medicine. 2002;6(1):1-12. doi:10.1111/j.1582- 4934.2002.tb00307.x.

161. Tirado-Rodriguez B, Ortega E, Segura-Medina P, Huerta-Yepez S. TGF- beta: an important mediator of allergic disease and a molecule with dual activity in cancer development. Journal of immunology research. 2014;2014:318481. doi:10.1155/2014/318481.

162. Ohta Y, Shridhar V, Kalemkerian GP, Bright RK, Watanabe Y, Pass HI. Thrombospondin-1 expression and clinical implications in malignant pleural mesothelioma. Cancer. 1999;85(12):2570-6. doi:10.1002/(sici)1097-0142(19990615)85:12<2570::aid- cncr12>3.0.co;2-f.

163. Wang AP, Li XH, Yang YM, Li WQ, Zhang W, Hu CP et al. A Critical Role of the mTOR/eIF2alpha Pathway in Hypoxia-Induced Pulmonary Hypertension. PloS one. 2015;10(6):e0130806. doi:10.1371/journal.pone.0130806.

164. He C, Klionsky DJ. Regulation Mechanisms and Signaling Pathways of Autophagy. Annual Review of Genetics. 2009;43(1):67-93. doi:10.1146/annurev-genet-102808-114910.

165. Dunlop EA, Tee AR. mTOR and autophagy: a dynamic relationship governed by nutrients and energy. Seminars in cell & developmental biology. 2014;36:121-9. doi:10.1016/j.semcdb.2014.08.006.

166. Laplante M, Sabatini DM. mTOR signaling in growth control and disease. Cell. 2012;149(2):274-93. doi:10.1016/j.cell.2012.03.017. 211

167. Hatefi Y. The mitochondrial electron transport and oxidative phosphorylation system. Annual review of biochemistry. 1985;54:1015-69. doi:10.1146/annurev.bi.54.070185.005055.

168. Martindale JL, Holbrook NJ. Cellular response to oxidative stress: signaling for suicide and survival. Journal of cellular physiology. 2002;192(1):1-15. doi:10.1002/jcp.10119.

169. Suliman HB, Piantadosi CA. Mitochondrial Quality Control as a Therapeutic Target. Pharmacological reviews. 2016;68(1):20-48. doi:10.1124/pr.115.011502.

170. Ratner V, Starkov A, Matsiukevich D, Polin RA, Ten VS. Mitochondrial dysfunction contributes to alveolar developmental arrest in hyperoxia-exposed mice. American journal of respiratory cell and molecular biology. 2009;40(5):511-8. doi:10.1165/rcmb.2008-0341RC.

171. David W. Kamp VP, Sigmund A. Weitzman, Navdeep Chandel. Asbestos-induced alveolar epithelial cell apoptosis: Role of mitochondrial dysfunction caused by iron-derived free radicals. In: Val Vallyathan XSPD, Vince Castranova, editor. Developments in Molecular and Cellular Biochemistry. vol 37: SpringerLink; 2002. p. 153-60.

172. Nymark P, Wijshoff P, Cavill R, van Herwijnen M, Coonen ML, Claessen S et al. Extensive temporal transcriptome and microRNA analyses identify molecular mechanisms underlying mitochondrial dysfunction induced by multi-walled carbon nanotubes in human lung cells. Nanotoxicology. 2015;9(5):624-35. doi:10.3109/17435390.2015.1017022.

173. Grek CL, Newton DA, Spyropoulos DD, Baatz JE. Hypoxia up-regulates expression of hemoglobin in alveolar epithelial cells. American journal of respiratory cell and molecular biology. 2011;44(4):439-47. doi:10.1165/rcmb.2009-0307OC.

174. Gross SS, Lane P. Physiological reactions of nitric oxide and hemoglobin: a radical rethink. Proceedings of the National Academy of Sciences of the United States of America. 1999;96(18):9967-9.

175. Poynter SE, LeVine AM. Surfactant biology and clinical application. Critical care clinics. 2003;19(3):459-72.

176. Hayashi T, Endo M. Carbon nanotubes as structural material and their application in composites. Composites Part B: Engineering. 2011;42(8):2151-7. doi:http://doi.org/10.1016/j.compositesb.2011.05.011.

212

177. De Volder MF, Tawfick Sh Fau - Baughman RH, Baughman Rh Fau - Hart AJ, Hart AJ. Carbon nanotubes: present and future commercial applications. (1095-9203 (Electronic)).

178. Donaldson K, Murphy FA, Duffin R, Poland CA. Asbestos, carbon nanotubes and the pleural mesothelium: a review of the hypothesis regarding the role of long fibre retention in the parietal pleura, inflammation and mesothelioma. Particle Fibre Toxicol. 2010;7. doi:10.1186/1743-8977-7-5.

179. Pacurari M, Castranova V Fau - Vallyathan V, Vallyathan V. Single- and multi-wall carbon nanotubes versus asbestos: are the carbon nanotubes a new health risk to humans? (1528-7394 (Print)).

180. Maynard AD, Baron PA, Foley M, Shvedova AA, Kisin ER, Castranova V. Exposure to carbon nanotube material: Aerosol release during the handling of unrefined single-walled carbon nanotube material. J Toxicol Env Health Pt A. 2004;67. doi:10.1080/15287390490253688.

181. Shvedova AA, Yanamala N, Kisin ER, Khailullin TO, Birch ME, Fatkhutdinova LM. Integrated Analysis of Dysregulated ncRNA and mRNA Expression Profiles in Humans Exposed to Carbon Nanotubes. PLOS ONE. 2016;11(3):e0150628. doi:10.1371/journal.pone.0150628.

182. Donaldson K, Murphy F Fau - Schinwald A, Schinwald A Fau - Duffin R, Duffin R Fau - Poland CA, Poland CA. Identifying the pulmonary hazard of high aspect ratio nanoparticles to enable their safety-by-design. (1748-6963 (Electronic)).

183. Mercer Rr Fau - Scabilloni JF, Scabilloni Jf Fau - Hubbs AF, Hubbs Af Fau - Wang L, Wang L Fau - Battelli LA, Battelli La Fau - McKinney W, McKinney W Fau - Castranova V et al. Extrapulmonary transport of MWCNT following inhalation exposure. (1743-8977 (Electronic)). doi:D - NLM: PMC3750633 EDAT- 2013/08/10 06:00 MHDA- 2015/04/01 06:00 CRDT- 2013/08/10 06:00 PHST- 2013/04/09 [received] PHST- 2013/08/06 [accepted] AID - 1743-8977-10-38 [pii] AID - 10.1186/1743-8977-10-38 [doi] PST - epublish.

184. Mercer RR, Scabilloni JF, Hubbs AF, Battelli LA, McKinney W, Friend S et al. Distribution and fibrotic response following inhalation exposure to multi-walled carbon nanotubes. Particle and Fibre Toxicology. 2013;10(1):33. doi:10.1186/1743-8977-10-33.

185. Ma-Hock L, Treumann S, Strauss V, Brill S, Luizi F, Mertler M et al. Inhalation toxicity of multiwall carbon nanotubes in rats exposed for 3 months. Toxicol Sci. 2009;112. doi:10.1093/toxsci/kfp146. 213

186. Umeda Y, Kasai T Fau - Saito M, Saito M Fau - Kondo H, Kondo H Fau - Toya T, Toya T Fau - Aiso S, Aiso S Fau - Okuda H et al. Two-week Toxicity of Multi-walled Carbon Nanotubes by Whole-body Inhalation Exposure in Rats. (0914-9198 (Print)). doi:D - NLM: PMC3695335 OTO - NOTNLM.

187. Nichols JE NJ, Vega SP, Argueta LB, Eastaway A, Cortiella J. Modeling the lung: Design and development of tissue engineered macro- and micro-physiologic lung models for research use. . Exp Biol Med (Maywood ) 2014;239:1135-69.

188. Rothen-Rutishauser B BF, Mühlfeld Ch GP. In vitro models of the human epithelial airway barrier to study the toxic potential of particulate matter. . Exp Opinion Drug Metabolism Toxicol 2008;4:1075-89.

189. Krug HF. Nanosafety Research—Are We on the Right Track? Angewandte Chemie International Edition. 2014;53(46):12304-19. doi:10.1002/anie.201403367.

190. Thurnherr T, Brandenberger C Fau - Fischer K, Fischer K Fau - Diener L, Diener L Fau - Manser P, Manser P Fau - Maeder-Althaus X, Maeder-Althaus X Fau - Kaiser J-P et al. A comparison of acute and long-term effects of industrial multiwalled carbon nanotubes on human lung and immune cells in vitro. (1879-3169 (Electronic)).

191. Clift MJD, Endes C, Vanhecke D, Wick P, Gehr P, Schins RPF et al. A Comparative Study of Different In Vitro Lung Cell Culture Systems to Assess the Most Beneficial Tool for Screening the Potential Adverse Effects of Carbon Nanotubes. Toxicological Sciences. 2014;137(1):55-64. doi:10.1093/toxsci/kft216.

192. Wang L, Luanpitpong S, Castranova V, Tse W, Lu Y, Pongrakhananon V et al. Carbon Nanotubes Induce Malignant Transformation and Tumorigenesis of Human Lung Epithelial Cells. Nano Letters. 2011;11(7):2796-803. doi:10.1021/nl2011214.

193. Lenz AG, Karg E, Lentner B, Dittrich V, Brandenberger C, Rothen-Rutishauser B et al. A dose-controlled system for air-liquid interface cell exposure and application to zinc oxide nanoparticles. Particle Fibre Toxicol. 2009;6. doi:10.1186/1743-8977-6-32.

194. Loret T, Peyret E, Dubreuil M, Aguerre-Chariol O, Bressot C, le Bihan O et al. Air- liquid interface exposure to aerosols of poorly soluble nanomaterials induces different biological activation levels compared to exposure to suspensions. Particle and Fibre Toxicology. 2016;13(1):58. doi:10.1186/s12989-016-0171-3.

214

195. Neilson L, Mankus C, Thorne D, Jackson G, DeBay J, Meredith C. Development of an in vitro cytotoxicity model for aerosol exposure using 3D reconstructed human airway tissue; application for assessment of e-cigarette aerosol. Toxicology in Vitro. 2015;29(7):1952-62. doi:http://doi.org/10.1016/j.tiv.2015.05.018.

196. Paur H-R, Cassee FR, Teeguarden J, Fissan H, Diabate S, Aufderheide M et al. In-vitro cell exposure studies for the assessment of nanoparticle toxicity in the lung—A dialog between aerosol science and biology. Journal of Aerosol Science. 2011;42(10):668-92. doi:http://doi.org/10.1016/j.jaerosci.2011.06.005.

197. Hilton GM, Taylor AJ, Hussain S, Dandley EC, Griffith EH, Garantziotis S et al. Mapping differential cellular protein response of mouse alveolar epithelial cells to multi- walled carbon nanotubes as a function of atomic layer deposition coating. Nanotoxicology. 2017:1-14. doi:10.1080/17435390.2017.1299888.

198. Lehmann A BC, Blank F, Gehr P, Rothen-Rutishauser B. . A 3D model of the human epithelial airway barrier. . In: Yarmush ML LR, editor. Alternatives to animal testing. Artech House; 2010. p. 239-60

199. Clift MJ, Gehr P Fau - Rothen-Rutishauser B, Rothen-Rutishauser B. Nanotoxicology: a perspective and discussion of whether or not in vitro testing is a valid alternative. (1432-0738 (Electronic)).

200. Wisniewski JR, Zougman A, Nagaraj N, Mann M. Universal sample preparation method for proteome analysis. Nature methods. 2009;6(5):359-62. doi:10.1038/nmeth.1322.

201. Beri J, Rosenblatt MM, Strauss E, Urh M, Bereman MS. Reagent for Evaluating Liquid Chromatography-Tandem Mass Spectrometry (LC-MS/MS) Performance in Bottom-Up Proteomic Experiments. Analytical chemistry. 2015;87(23):11635-40. doi:10.1021/acs.analchem.5b04121.

202. Cox J, Mann M. MaxQuant enables high peptide identification rates, individualized p.p.b.-range mass accuracies and proteome-wide protein quantification. Nature biotechnology. 2008;26(12):1367-72. doi:10.1038/nbt.1511.

203. Qiagen. Ingenuity Pathway Analysis. 2017. www.qiagen.com/ingenuity. Accessed 04/06/2017 2017.

215

204. Costa PM, Fadeel B. Emerging systems biology approaches in nanotoxicology: Towards a mechanism-based understanding of nanomaterial hazard and risk. Toxicology and applied pharmacology. 2016;299:101-11. doi:10.1016/j.taap.2015.12.014.

205. Lee JL, Streuli CH. Integrins and epithelial cell polarity. Journal of cell science. 2014;127(Pt 15):3217-25. doi:10.1242/jcs.146142.

206. Carmosino M, Valenti G, Caplan M, Svelto M. Polarized traffic towards the cell surface: how to find the route. Biology of the cell. 2009;102(2):75-91. doi:10.1042/BC20090134.

207. Assemat E, Bazellieres E, Pallesi-Pocachard E, Le Bivic A, Massey-Harroche D. Polarity complex proteins. Biochimica et biophysica acta. 2008;1778(3):614-30. doi:10.1016/j.bbamem.2007.08.029.

208. Rodriguez-Boulan E, Kreitzer G, Müsch A. Organization of vesicular trafficking in epithelia. Nature Reviews Molecular Cell Biology. 2005;6(3):233-47. doi:10.1038/nrm1593.

209. Wang AZ, Ojakian GK, Nelson WJ. Steps in the morphogenesis of a polarized epithelium. I. Uncoupling the roles of cell-cell and cell-substratum contact in establishing plasma membrane polarity in multicellular epithelial (MDCK) cysts. Journal of cell science. 1990;95(1):137-51.

210. Ojakian GK, Schwimmer R. Regulation of epithelial cell surface polarity reversal by beta 1 integrins. Journal of cell science. 1994;107(3):561-76.

211. Adamson IY HC, Bowden DH. Epithelial cell-fibroblast interactions in lung injury and repair. The American Journal of Pathology. 1990;137(2):385-92.

212. Fujii T, Hayashi S, Hogg JC, Mukae H, Suwa T, Goto Y et al. Interaction of alveolar macrophages and airway epithelial cells following exposure to particulate matter produces mediators that stimulate the bone marrow. American journal of respiratory cell and molecular biology. 2002;27(1):34-41. doi:10.1165/ajrcmb.27.1.4787.

213. NIH. What is Idiopathic Pulmonary Fibrosis. 2011. http://www.nhlbi.nih.gov/health/health-topics/topics/ipf/. Accessed 04/11/2017 2017.

216

214. Clift MJ, Endes C, Vanhecke D, Wick P, Gehr P, Schins RP et al. A comparative study of different in vitro lung cell culture systems to assess the most beneficial tool for screening the potential adverse effects of carbon nanotubes. Toxicological sciences : an official journal of the Society of Toxicology. 2014;137(1):55-64. doi:10.1093/toxsci/kft216.

215. Snyder-Talkington BN, Schwegler-Berry D, Castranova V, Qian Y, Guo NL. Multi- walled carbon nanotubes induce human microvascular endothelial cellular effects in an alveolar-capillary co-culture with small airway epithelial cells. Particle and fibre toxicology. 2013;10:35. doi:10.1186/1743-8977-10-35.

216. Tabet L, Bussy C, Amara N, Setyan A, Grodet A, Rossi MJ et al. Adverse effects of industrial multiwalled carbon nanotubes on human pulmonary cells. Journal of toxicology and environmental health Part A. 2009;72(2):60-73. doi:10.1080/15287390802476991.

217. Thurnherr T, Brandenberger C, Fischer K, Diener L, Manser P, Maeder-Althaus X et al. A comparison of acute and long-term effects of industrial multiwalled carbon nanotubes on human lung and immune cells in vitro. Toxicology letters. 2011;200(3):176-86. doi:10.1016/j.toxlet.2010.11.012.

218. Hamilton RF, Jr., Wu Z, Mitra S, Shaw PK, Holian A. Effect of MWCNT size, carboxylation, and purification on in vitro and in vivo toxicity, inflammation and lung pathology. Particle and fibre toxicology. 2013;10(1):57. doi:10.1186/1743-8977-10-57.

219. Endes C, Schmid O, Kinnear C, Mueller S, Camarero-Espinosa S, Vanhecke D et al. An in vitro testing strategy towards mimicking the inhalation of high aspect ratio nanoparticles. Particle and fibre toxicology. 2014;11:40. doi:10.1186/s12989-014-0040-x.

220. Lenz AG, Karg E, Lentner B, Dittrich V, Brandenberger C, Rothen-Rutishauser B et al. A dose-controlled system for air-liquid interface cell exposure and application to zinc oxide nanoparticles. Particle and fibre toxicology. 2009;6:32. doi:10.1186/1743-8977-6-32.

221. Press TNA. Toxicity Testing in the 21st Century. 2007.

222. Vairavapandian D, Vichchulada P, Lay MD. Preparation and modification of carbon nanotubes: review of recent advances and applications in catalysis and sensing. Analytica chimica acta. 2008;626(2):119-29. doi:10.1016/j.aca.2008.07.052.

217

223. Thostenson ET, Ren ZF, Chou TW. Advances in the science and technology of carbon nanotubes and their composites: a review. Composites Science and Technology. 2001;61(13):1899-912. doi:Doi 10.1016/S0266-3538(01)00094-X.

224. Endo M, Strano MS, Ajayan PM. Potential applications of carbon nanotubes. Carbon Nanotubes. 2008;111:13-61.

225. Bonner JC. Carbon nanotubes as delivery systems for respiratory disease: do the dangers outweigh the potential benefits? Expert Review of Respiratory Medicine. 2011;5(6):779-87. doi:Doi 10.1586/Ers.11.72.

226. Maynard AD, Baron PA, Foley M, Shvedova AA, Kisin ER, Castranova V. Exposure to carbon nanotube material: aerosol release during the handling of unrefined single-walled carbon nanotube material. Journal of toxicology and environmental health Part A. 2004;67(1):87-107. doi:10.1080/15287390490253688.

227. Lam CW, James JT, McCluskey R, Hunter RL. Pulmonary toxicity of single-wall carbon nanotubes in mice 7 and 90 days after intratracheal instillation. Toxicological sciences : an official journal of the Society of Toxicology. 2004;77(1):126-34. doi:10.1093/toxsci/kfg243.

228. Mitchell LA, Gao J, Wal RV, Gigliotti A, Burchiel SW, McDonald JD. Pulmonary and systemic immune response to inhaled multiwalled carbon nanotubes. Toxicological sciences : an official journal of the Society of Toxicology. 2007;100(1):203-14. doi:10.1093/toxsci/kfm196.

229. Salvador-Morales C, Townsend P, Flahaut E, Venien-Bryan C, Vlandas A, Green MLH et al. Binding of pulmonary surfactant proteins to carbon nanotubes; potential for damage to lung immune defense mechanisms. Carbon. 2007;45(3):607-17. doi:DOI 10.1016/j.carbon.2006.10.011.

230. Klebanoff SJ. Myeloperoxidase: friend and foe. J Leukoc Biol. 2005;77(5):598-625. doi:10.1189/jlb.1204697.

231. Peng Q, Gong B, VanGundy RM, Parsons GN. “Zincone” Zinc Oxide−Organic Hybrid Polymer Thin Films Formed by Molecular Layer Deposition. Chemistry of Materials. 2009;21(5):820-30. doi:10.1021/cm8020403.

218

232. Marichy C, Bechelany M, Pinna N. Atomic layer deposition of nanostructured materials for energy and environmental applications. Advanced materials. 2012;24(8):1017-32. doi:10.1002/adma.201104129.

233. Liao L, McClatchy DB, Yates JR. Shotgun proteomics in neuroscience. Neuron. 2009;63(1):12-26. doi:10.1016/j.neuron.2009.06.011.

234. Asara JM, Christofk HR, Freimark LM, Cantley LC. A label-free quantification method by MS/MS TIC compared to SILAC and spectral counting in a proteomics screen. Proteomics. 2008;8(5):994-9. doi:10.1002/pmic.200700426.

235. Bonner JC, Silva RM, Taylor AJ, Brown JM, Hilderbrand SC, Castranova V et al. Interlaboratory evaluation of rodent pulmonary responses to engineered nanomaterials: the NIEHS Nano GO Consortium. Environmental health perspectives. 2013;121(6):676-82. doi:10.1289/ehp.1205693.

236. UniProt C. The universal protein resource (UniProt). Nucleic acids research. 2008;36(Database issue):D190-5. doi:10.1093/nar/gkm895.

237. Liu H, Sadygov RG, Yates JR. A Model for Random Sampling and Estimation of Relative Protein Abundance in Shotgun Proteomics. Anal Chem. 2004;76(14):4193-201. doi:10.1021/ac0498563.

238. Schilling B, Rardin MJ, MacLean BX, Zawadzka AM, Frewen BE, Cusack MP et al. Platform-independent and label-free quantitation of proteomic data using MS1 extracted ion chromatograms in skyline: application to protein acetylation and phosphorylation. Molecular & cellular proteomics : MCP. 2012;11(5):202-14. doi:10.1074/mcp.M112.017707.

239. Gaut JP, Yeh GC, Tran HD, Byun J, Henderson JP, Richter GM et al. Neutrophils employ the myeloperoxidase system to generate antimicrobial brominating and chlorinating oxidants during sepsis. Proceedings of the National Academy of Sciences of the United States of America. 2001;98(21):11961-6. doi:10.1073/pnas.211190298.

240. Kagan VE, Konduru NV, Feng W, Allen BL, Conroy J, Volkov Y et al. Carbon nanotubes degraded by neutrophil myeloperoxidase induce less pulmonary inflammation. Nature nanotechnology. 2010;5(5):354-9. doi:10.1038/nnano.2010.44.

241. Shvedova AA, Kapralov AA, Feng WH, Kisin ER, Murray AR, Mercer RR et al. Impaired clearance and enhanced pulmonary inflammatory/fibrotic response to carbon 219

nanotubes in myeloperoxidase-deficient mice. PloS one. 2012;7(3):e30923. doi:10.1371/journal.pone.0030923.

242. Actor JK, Hwang SA, Kruzel ML. Lactoferrin as a Natural Immune Modulator. Current Pharmaceutical Design. 2009;15(17):1956-73.

243. Chakraborty S, Kaur S, Guha S, Batra SK. The multifaceted roles of neutrophil gelatinase associated lipocalin (NGAL) in inflammation and cancer. Biochimica et biophysica acta. 2012;1826(1):129-69. doi:10.1016/j.bbcan.2012.03.008.

244. Tambor V, Kacerovsky M, Lenco J, Bhat G, Menon R. Proteomics and bioinformatics analysis reveal underlying pathways of infection associated histologic chorioamnionitis in pPROM. Placenta. 2013;34(2):155-61. doi:10.1016/j.placenta.2012.11.028.

245. Dunkelberger JR, Song WC. Complement and its role in innate and adaptive immune responses. Cell research. 2010;20(1):34-50. doi:10.1038/cr.2009.139.

246. Wright JR. Immunoregulatory functions of surfactant proteins. Nat Rev Immunol. 2005;5(1):58-68. doi:10.1038/nri1528.

247. van Iwaarden JF, Claassen E, Jeurissen SH, Haagsman HP, Kraal G. Alveolar macrophages, surfactant lipids, and surfactant protein B regulate the induction of immune responses via the airways. American journal of respiratory cell and molecular biology. 2001;24(4):452-8. doi:10.1165/ajrcmb.24.4.4239.

248. Liu Z, Chen K, Davis C, Sherlock S, Cao Q, Chen X et al. Drug delivery with carbon nanotubes for in vivo cancer treatment. Cancer research. 2008;68(16):6652-60. doi:10.1158/0008-5472.CAN-08-1468.

249. Wong BS, Yoong SL, Jagusiak A, Panczyk T, Ho HK, Ang WH et al. Carbon nanotubes for delivery of small molecule drugs. Advanced drug delivery reviews. 2013;65(15):1964- 2015. doi:10.1016/j.addr.2013.08.005.

250. Cirillo G, Hampel S, Spizzirri UG, Parisi OI, Picci N, Iemma F. Carbon nanotubes hybrid hydrogels in drug delivery: a perspective review. BioMed research international. 2014;2014:825017. doi:10.1155/2014/825017.

220

251. Rieger C, Kunhardt D, Kaufmann A, Schendel D, Huebner D, Erdmann K et al. Characterization of different carbon nanotubes for the development of a mucoadhesive drug delivery system for intravesical treatment of bladder cancer. International journal of pharmaceutics. 2015;479(2):357-63. doi:10.1016/j.ijpharm.2015.01.017.

252. Zhang W, Zhang Z, Zhang Y. The application of carbon nanotubes in target drug delivery systems for cancer therapies. Nanoscale research letters. 2011;6:555. doi:10.1186/1556-276X-6-555.

253. Wim H De Jong PJB. Drug delivery and nanoparticles: Applications and hazards. Int J Nanomedicine. 2008;3(2):133-49.

254. Strojan K, Leonardi A, Bregar VB, Krizaj I, Svete J, Pavlin M. Dispersion of Nanoparticles in Different Media Importantly Determines the Composition of Their Protein Corona. PloS one. 2017;12(1):e0169552. doi:10.1371/journal.pone.0169552.

255. Whitwell H, Mackay RM, Elgy C, Morgan C, Griffiths M, Clark H et al. Nanoparticles in the lung and their protein corona: the few proteins that count. Nanotoxicology. 2016;10(9):1385-94. doi:10.1080/17435390.2016.1218080.

256. Zanganeh S, Spitler R, Erfanzadeh M, Alkilany AM, Mahmoudi M. Protein corona: Opportunities and challenges. The international journal of biochemistry & cell biology. 2016;75:143-7. doi:10.1016/j.biocel.2016.01.005.

257. Yang ST, Liu Y, Wang YW, Cao A. Biosafety and bioapplication of nanomaterials by designing protein-nanoparticle interactions. Small. 2013;9(9-10):1635-53. doi:10.1002/smll.201201492.

258. Lundqvist M, Stigler J, Elia G, Lynch I, Cedervall T, Dawson KA. Nanoparticle size and surface properties determine the protein corona with possible implications for biological impacts. Proceedings of the National Academy of Sciences of the United States of America. 2008;105(38):14265-70. doi:10.1073/pnas.0805135105.

259. Lynch I, Dawson KA. Protein-nanoparticle interactions. Nano Today. 2008;3(1-2):40-7. doi:10.1016/s1748-0132(08)70014-8.

260. Lundqvist M, Stigler J, Cedervall T, Berggard T, Flanagan MB, Lynch I et al. The evolution of the protein corona around nanoparticles: a test study. ACS nano. 2011;5(9):7503-9. doi:10.1021/nn202458g. 221

261. Mahmoudi M, Lynch I, Ejtehadi MR, Monopoli MP, Bombelli FB, Laurent S. Protein- nanoparticle interactions: opportunities and challenges. Chemical reviews. 2011;111(9):5610-37. doi:10.1021/cr100440g.

262. Sapsford KE, Tyner KM, Dair BJ, Deschamps JR, Medintz IL. Analyzing nanomaterial bioconjugates: a review of current and emerging purification and characterization techniques. Analytical chemistry. 2011;83(12):4453-88. doi:10.1021/ac200853a.

263. Docter D, Distler U, Storck W, Kuharev J, Wunsch D, Hahlbrock A et al. Quantitative profiling of the protein coronas that form around nanoparticles. Nature protocols. 2014;9(9):2030-44. doi:10.1038/nprot.2014.139.

264. Walkey CD, Olsen JB, Song F, Liu R, Guo H, Olsen DW et al. Protein corona fingerprinting predicts the cellular interaction of gold and silver nanoparticles. ACS nano. 2014;8(3):2439-55. doi:10.1021/nn406018q.

265. Eigenheer R, Castellanos ER, Nakamoto MY, Gerner KT, Lampe AM, Wheeler KE. Silver nanoparticle protein corona composition compared across engineered particle properties and environmentally relevant reaction conditions. Environmental Science: Nano. 2014;1(3):238. doi:10.1039/c4en00002a.

266. Fernandez-Iglesias N, Bettmer J. Complementary mass spectrometric techniques for the quantification of the protein corona: a case study on gold nanoparticles and human serum proteins. Nanoscale. 2015;7(34):14324-31. doi:10.1039/c5nr02625c.

267. Zhang H, Wu Ra. Proteomic profiling of protein corona formed on the surface of nanomaterial. Science China Chemistry. 2015;58(5):780-92. doi:10.1007/s11426-015-5395- 9.

268. Tu C, Rudnick PA, Martinez MY, Cheek KL, Stein SE, Slebos RJ et al. Depletion of abundant plasma proteins and limitations of plasma proteomics. Journal of proteome research. 2010;9(10):4982-91. doi:10.1021/pr100646w.

269. Cai X, Ramalingam R, Wong HS, Cheng J, Ajuh P, Cheng SH et al. Characterization of carbon nanotube protein corona by using quantitative proteomics. Nanomedicine : nanotechnology, biology, and medicine. 2013;9(5):583-93. doi:10.1016/j.nano.2012.09.004.

222

270. Shannahan JH, Brown JM, Chen R, Ke PC, Lai X, Mitra S et al. Comparison of nanotube-protein corona composition in cell culture media. Small. 2013;9(12):2171-81. doi:10.1002/smll.201202243.

271. Siepen JA, Keevil EJ, Knight D, Hubbard SJ. Prediction of missed cleavage sites in tryptic peptides aids protein identification in proteomics. Journal of proteome research. 2007;6(1):399-408. doi:10.1021/pr060507u.

272. Bollineni RC, Guldvik IJ, Gronberg H, Wiklund F, Mills IG, Thiede B. A differential protein solubility approach for the depletion of highly abundant proteins in plasma using ammonium sulfate. The Analyst. 2015;140(24):8109-17. doi:10.1039/c5an01560j.

273. Vroman L. Effect of Adsorbed Proteins on the Wettability of Hydrophilic and Hydrophobic Solids. Nature. 1962;196(4853):476-7. doi:10.1038/196476a0.

274. Saptarshi SR, Duschl A, Lopata AL. Interaction of nanoparticles with proteins: relation to bio-reactivity of the nanoparticle. Journal of nanobiotechnology. 2013;11:26. doi:10.1186/1477-3155-11-26.

223