ABSTRACT

FROMEN, CATHERINE ANN. Monodisperse, Uniformly-Shaped Particles for Controlled Respiratory Delivery. (Under the direction of Joseph M. DeSimone).

The majority of the world’s most infectious diseases occur at the air-tissue interface called the mucosa, including HIV/AIDS, , measles, and bacterial or viral gut and respiratory . Despite this, have generally been developed for the systemic immune system and fail to provide protection at the mucosal site. Vaccine delivery directly to the lung mucosa could provide superior lung protection for many infectious diseases, such as TB or influenza, as well as systemic and therapeutic vaccines for diseases such as Dengue fever, asthma, or cancer. Specifically, precision engineered biomaterials are believed to offer tremendous opportunities for a new generation of vaccines. The goal of this approach is to leverage naturally occurring processes of the immune system to produce memory responses capable of rapidly destroy virulent pathogens without harmful exposure. Considerable knowledge of biomaterial properties and their interaction with the immune system of the lung is required for successful translation. The overall goal of this work was to fabricate and characterize nano- and micro- particles using the Particle Replication In Non-wetting Templates (PRINT) fabrication technique and optimize them as pulmonary vaccine carriers. The main objectives of this PhD research included (1) the development of a calibration-quality aerosol system using PRINT, the application of these calibration-quality aerosols to improve understanding of (2) shaped aerosols under flow and (3) their cellular fate in the lung, and (4) the application of this knowledge towards the development of a mucosal vaccine. We hypothesized that the precision particle control afforded by the PRINT technology could advance understanding of the role that particle features, such as size, shape, and surface charge, play on all aspects of pulmonary vaccine formulations. Particles were fabricated and optimized for delivery via the lung, exhibiting the first examples of monodisperse, non-spherical aerosols. The particle control afforded by the PRINT platform was also used to probe the biological function of key lung antigen presenting cells (APCs) in mice. We demonstrate for the first time the role of particle charge on airway APC association, finding that cationic particles were preferentially associated with lung dendritic cells, giving them a distinct advantage over anionic formulations in vaccine platforms. Subsequently, these cationic particles were found to increase production of both systemic and mucosal antibodies, demonstrating the importance of particle characteristics on pulmonary and the importance of surface charge in a T-cell dependent antibody response. This work contributes to the overall understanding of how parameters of precision engineered particles influence both pulmonary delivery and immune programing, towards the ultimate application of a translatable pulmonary vaccine formulation.

© Copyright 2014 Catherine Ann Fromen

All Rights Reserved Monodisperse, Uniformly-Shaped Particles for Controlled Respiratory Vaccine Delivery

by Catherine Ann Fromen

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

Chemical Engineering

Raleigh, North Carolina 2014

APPROVED BY:

______Dr. Joseph M. DeSimone Dr. Michael Dickey Committee Chair

______Dr. Saad Khan Dr. Orlin Velev DEDICATION This dissertation is dedicated to my family for their continual support.

ii BIOGRAPHY The author was born and raised in Medfield, MA to parents Greg and Debby Fromen. She entered the University of Rochester in Rochester, NY in 2005 and graduated with a B.S. in Chemical Engineering in 2009. During this time, she first participated in academic research under the direction of Steve Jacobs, Ken Marshall and Jerry Cox. She entered the PhD graduate program at North Carolina State University in the Department of Chemical and Biomolecular Engineering in the fall of 2009. Despite her original intentions to pursue research in alternative energy, she joined Joseph DeSimone’s lab at the University of North Carolina at Chapel Hill and has performed her PhD research in pulmonary drug delivery. At the time of this dissertation, the author intends to continue academic research and has accepted the University of Michigan President’s Postdoctoral Fellowship to continue in the field of drug delivery with precision engineered biomaterials.

iii ACKNOWLEDGEMENTS Where to start! This PhD process has involved the help and support of so many people who have shaped me into the person I am today. I am beyond grateful to have had such fantastic support throughout these past five years and consider myself quite blessed. First and foremost, I would like to thank my advisor Joe DeSimone, who picked me out of a crowd of wide-eyed first year graduate students and welcomed me into his group. The opportunities that he has catalyzed for me have been immeasurable, exposing me to projects, people, and experiences I never would have believed were possible. Thank you very much for your continued belief in my potential, even when I wasn’t sure of it myself, and for your invaluable support and encouragement to differentiate myself and pursue the unknown. I am truly in your debt and look forward to your continued support in my future endeavors! Next, I would like to thank my NCSU Chemical Engineering family, who, despite my departure to Chapel Hill, has remained steadfast in their support. I regret the many missed conversations and collaborations that I can only envision would have transpired if I had remained local. First, I am incredibly thankful to my committee members: Dr. Orlin Velev, Dr. Saad Khan, and Dr. Michael Dickey. Their continued patience, flexibility, friendship, and scientific advice has helped me maintain my roots in Chemical Engineering and their willingness to engage with my unconventional research direction is greatly appreciated (especially as they embark on reading this dissertation!). I am also hugely thankful to Dr. Dickey for providing me the opportunity to guest lecture in his introductory Chemical Engineering class and the encouragement to better my teaching skills. I would also like to specifically thank a number of CHE faculty, including Dr. Fedkiw, Dr. Genzer, Dr. Haugh, Dr. Ollis, Dr. Reeves and Dr. Hall, who have each in some way positively contributed to my time at NCSU. Finally, I would never have navigated the logistics of graduate school without the incredible support of Ms. Sandra Bailey – thank you! In missing out on more experiences at NCSU, I was lucky to gain incredible connections throughout many departments at UNC. First, I need to thank Dr. Jenny Ting of the Immunology department, for providing me intellectual support and resources of her lab.

iv Some weeks, I spent more time in her lab than my own, and I appreciate her patience as I monopolized her students’ time. I am hugely thankful for the opportunity to work with Reid Roberts, who first introduced me to immunoengineering and is one of the most insightful and inspiring scientists I’ve ever met. I appreciate the coaching of Coy Allen, who taught me all of the lung animal techniques. A thank you also goes to other members of the Ting lab, including Alex, Justin, Yoshi, Aga, Cate, Tim, Les, Leo, Emily and Elizabeth for making me feel at home. I also was known to frequent the lab of David Leith in the Environmental Engineering lab, for which I am extremely grateful. He provided me with the first direction in aerosol characterization and allowed me unrestricted access to his resources. I am incredibly thankful for his expertise. I also received overwhelming assistance from Maryanne Boundy and Joe Pedit and their practical experience and technical support was essential to my success. Next, I would like to acknowledge the professional guidance and mentorship of Ben Maynor and Pete Mack. Through our initial collaborations between the DeSimone lab and the Liquidia Inhalation team, I was fortunate to receive extremely helpful technical input from both Ben and Pete, who exposed me to the industrial size of the pharmaceutical industry. I have very fond memories of conversations at conferences, lunch meetings, or sample exchanges and their mentorship is greatly appreciated. I was also fortunate to receive personal and professional guidance from two special lab managers. I would like to thank Mary for teaching me how to thrive in the DeSimone lab, leading by example, and for encouraging me when I needed it most. I would also like to thank Chris for his willingness to lend an ear, his conviction in my abilities, his much needed life advice, and most importantly, his friendship. Professionally, I would like to thank a number of colleagues at UNC who have enabled my research. Thanks to Amar Kumbhar and Carrie Donley from CHANL for providing technical SEM support. Thanks to Charlene Santos, Alain Valdivia and Mark Ross of the UNC Lineberger Animal core for assistance with injections, housing and care, as well as the entire DLAM staff. Thanks also to Donald, Reggie and Erica, who always managed to brighten my day.

v At this point, I would also like to acknowledge my funding support from the NIH Pioneer Award (1DP1OD006432) and DTRA (HDTRA1-13-1-0045) - without their support, none of this dissertation would have been possible. I have been fortunate to be part of one of the most dynamic academic labs and have been exposed to all sorts of characters! This has made coming to lab an adventure – thank you to all of my current and former lab members for your wonderful conversations, technical skills and general good nature. I would love to acknowledge each of you personally, but for the sake of space, I would like to thank: Stu, John Fain, Abby, Heather, Tojan, Luke, Reid, Tim, Yapei, Roland, Herlihy, Reuter, Jillian, Mark Elsesser, Farrell, John Savage, Chintan, Ying, Cam, Detter, Tiffany, Kai Chen, Charlie, Tilda, Lissett, Merve, Nicole, David, Sarika, and anyone else of the DeSimone lab I may have overlooked. I would also like to thank Vicki for being absolutely incredible and keeping me organized, inspired, and supplied with cake, and Crista for helping me see the best in myself. I would not have survived this experience without the help from my friends and family. I am blessed to have such a supportive group of friends who have encouraged me in this undertaking and kept me grounded. I would like to thank Brian, Dave, Marc, Katie, Chris, and Grace, as well as the many friends I’ve made in NC who helped me through, including Christina, Jonathan, Aaron, Joseph, Phil, Will, Dave, Drew, Dylan, Stephie, Jasmine, Tojan, Justin, Abby, and Stu. Your friendships mean the world to me. I would also like to acknowledge the UNC symphony orchestra for providing a necessary outlet for maintaining my sanity and thank a few fellow orchestra members, including Anya, Will, Margaret, Ash, Emily, Amanda, Lauren, and Tonu. An additional acknowledgement goes to Graham Myhre, who encouraged me to go to graduate school in the first place. I would also like to say a huge thanks to my academic family at the University of Rochester for getting me started in research: Dr. Jacobs, Ken Marshall, and Dr. Cox. I could not imagine graduate school without a few special people who’ve come into my life during this time, and need to say a special thanks to them. Whitney, Melanie, Alessandra, Heather, Stephanie and Katherine – you ladies have provided me more friendship, support, happiness, laughter, and love than anyone ever deserves, giving me a

vi sense of family both in Raleigh and Chapel Hill, and I’m extremely grateful that you have come into my life. Tammy – you have been my partner in crime for the past four years and I’m incredibly indebted to you for picking up my slack, your musical entertainment, and the many guac nights, not to mention the plethora of technical and personal support. You are the best part of the Inhalation team and I will build a snowman with you and Marc anytime! Marc – your companionship has been invaluable through the past five years as we’ve navigated life in both Raleigh and Chapel Hill. Thank you for always encouraging, challenging, and supporting me, for Alias, Archer, and opening-night movie showings, and for sharing your advice, technical prowess and lab citizenship with me. Greg – having you as a colleague has dramatically changed my perspectives on science, research, hard work, and friendship for the better and I am honored to have worked so closely with you over the past two years. All of the treks to-and-from Lineberger in all sorts of weather were certainly worth it - thank you for the unwavering encouragement in all aspects of my life. My final thanks go to my family. I would like to thank Katie Young, my extended family and the true winner of the Best Friend Award – simply put, I could not have done it without you. I am incredibly thankful for the incredible support, praise, and inspiration from my Grandma Jean and Aunt Rita. My sister Jenny also for has provided me with innumerable encouragement, always keeping me grounded, kidnapping me for beach trips, and sharing her love. Finally, I would not be the person I am today without the love support and unwavering support from my parents. They have taught me the meaning of hard work and have been the inspiration behind my achievements and I owe them my eternal thanks and gratitude.

vii TABLE OF CONTENTS LIST OF TABLES ...... xii LIST OF FIGURES ...... xiii LIST OF ABBREVIATIONS ...... xvi LIST OF SYMBOLS ...... xviii

CHAPTER ONE: Introduction to Precision Engineered Biomaterials, Mucosal Immunology, Vaccinology and Pulmonary Drug Delivery...... 1 1.1. Introduction: Precision Engineered Biomaterials for Medical Applications ...... 2 1.2. Immunology Primer ...... 4 1.2.1. Physical Barriers ...... 5 1.2.2. Innate Immune System ...... 6 1.2.3. Adaptive Immune System ...... 10 1.3. Mucosal Immunology ...... 17 1.3.1. Mucosal Structure ...... 18 1.3.2. Components of Mucus Secretions ...... 19 1.3.3. Secretory Antibodies ...... 23 1.3.4. Cells of the Mucosa ...... 25 1.3.5. Secondary Lymphoid Tissues ...... 26 1.3.6. Disease at the Mucosa ...... 28 1.4. Mucosal Vaccines ...... 29 1.4.1 Vaccine Overview ...... 30 1.4.2. Benefits of Vaccination at Mucosal Surfaces ...... 32 1.4.3. Current Mucosal Vaccines ...... 33 1.4.4. Pulmonary Vaccines ...... 34 1.4.5. Adjuvants ...... 36 1.4.6. Existing Needs for Mucosal Vaccines ...... 37 1.5. Pulmonary Therapeutic Delivery ...... 38 1.5.1. Lung Structure ...... 38 1.5.2. Physical Parameters for Aerosol Delivery ...... 40 1.5.3. Types of Inhaler Delivery ...... 43 1.6. Micro and Nanoparticle Fabrication Techniques ...... 44 1.6.1. Bottom-up Fabrication Techniques ...... 44 1.6.2. Top-down Fabrication Techniques ...... 46 1.7. Thesis Overview ...... 49 1.8. References ...... 50

viii CHAPTER TWO: Development of PRINT Particles as Pulmonary Drug Delivery Vehicles ...... 68 2.1. Introduction ...... 69 2.2. Techniques ...... 71 2.2.1. PRINT Particle Fabrication ...... 71 2.2.2. Lyophilization ...... 74 2.2.3. Andersen Cascade Impactor ...... 75 2.3. Materials and Methods ...... 77 2.3.1. Reagents ...... 77 2.3.2. BSA Particle Fabrication ...... 77 2.3.3. HDODA Particle Fabrication ...... 78 2.3.4. Fabrication of Jet-milled Itraconazole ...... 79 2.3.5. Particle Characterization ...... 79 2.3.6. Lyophilization ...... 80 2.3.7. Cascade Impactor Aerosol Studies ...... 81 2.4. Results ...... 82 2.4.1. Fabrication of PRINT Particles for Aerosol Delivery ...... 82 2.4.2. PRINT Aerosols Compared to Jet milled Formulations ...... 83 2.4.3. PRINT BSA Shaped Particle Aerosol Study ...... 84 2.4.4. PRINT HDODA Particle Shaped Aerosol Study ...... 86 2.4.5. Optimization of Dry Powder PRINT Formulations ...... 89 2.5. Discussion ...... 93 2.6. Conclusions ...... 96 2.7. Acknowledgements ...... 97 2.8. References ...... 97

CHAPTER THREE: Synthesis and Characterization of Monodisperse Uniformly Shaped Respirable Aerosols ...... 103 3.1. Introduction ...... 104 3.2. Theory ...... 107 3.2.1. Shaped Particles in Stokes’ Flow ...... 107 3.2.2. Extension of Dynamic Shape Factor in the Non-Stokesian Flow of an Aerodynamic Particle Sizer (APS) ...... 109 3.3. Materials and Methods ...... 110 3.3.1. Particle Fabrication using PRINT ...... 110 3.3.2. Particle Characterization ...... 110 3.3.3. Aerosol Characterization using APS ...... 111

ix 3.3.4. Aerosol Characterization using an ACI ...... 111 3.3.5. APS Model for Shape Factor Calculation ...... 112 3.3.6. Shape Factor Validations ...... 115 3.4. Results and Discussion ...... 117 3.4.1. Particle Fabrication ...... 117 3.4.2. Aerosol Characterization ...... 119 3.4.3. Shape Factor Determination with APS Model ...... 121 3.4.4. Application of Determined Shape Factors ...... 127 3.5. Conclusions ...... 128 3.6. Acknowledgements ...... 129 3.7. References ...... 130

CHAPTER FOUR: Cellular Fate of Nanoparticles in Lung APCs for Pulmonary Vaccines ...... 134 4.1. Introduction ...... 135 4.2. Materials and Methods ...... 136 4.2.1. Animals ...... 136 4.2.2. Reagents ...... 136 4.2.3. Particle Fabrication and Characterization ...... 137 4.2.4. Tissue and Cell Preparation ...... 138 4.2.5. Pulmonary Administration ...... 139 4.2.6. Antibodies: Flow Cytometry and ELISAs ...... 139 4.2.7. Histopathology ...... 139 4.2.8. Statistical Analysis ...... 140 4.3. Results ...... 140 4.3.1. Particle Fabrication and Characterization ...... 140 4.3.2. Evaluation of Particle Tolerability in the Lung ...... 142 4.3.3. Particle Association in Lung APCs ...... 145 4.3.4. Change in Lung APC Population Following Instillation ...... 149 4.3.5. Particle Tracking to Medistinal Lymph Nodes ...... 152 4.4. Discussion ...... 155 4.5. Conclusions ...... 157 4.6. Acknowledgements ...... 158 4.7. References ...... 158

x CHAPTER FIVE: Induction of T-cell Dependent Mucosal and Systemic Antibody Responses Following Pulmonary Delivery of Cationic Hydrogel-based Nanoparticles ...... 163 5.1. Introduction ...... 164 5.2. Materials and Methods ...... 166 5.2.1. Animals ...... 166 5.2.2. Reagents ...... 166 5.2.3. Particle Fabrication and Characterization ...... 166 5.2.4. Tissue and Cell Preparation ...... 167 5.2.5. OT-II Co-culture ...... 167 5.2.6. Quantitative Realtime RT-PCR ...... 168 5.2.7. Pulmonary Delivery and ...... 168 5.2.8. Antibodies: Flow Cytometry & ELISAs ...... 168 5.2.9. Statistical Analysis ...... 169 5.3. Results ...... 169 5.3.1. Particle Fabrication and Functionalization ...... 169 5.3.2. In Vitro BMDC and CD4+ OT-II T Cell Co-Culture ...... 169 5.3.3. In vitro evaluation of upregulation of co-stimulatory receptors, cytokines and chemokines on BMDCs ...... 172 5.3.4. In Vivo Pulmonary ...... 176 5.4. Discussion ...... 183 5.5. Conclusions ...... 186 5.6. Acknowledgements ...... 187 5.7. References ...... 187

CHAPTER SIX: Summary and Outlook ...... 192 6.1. Summary ...... 193 6.2. Impact and Future Directions ...... 194 6.2.1. PRINT as a Platform for Pulmonary Drug Delivery ...... 194 6.2.2. Characterization of Monodisperse PRINT Aerosols ...... 195 6.2.3. Calibration-Quality PRINT Particles for Probing Lung Biology ...... 198 6.2.4. PRINT Nanoparticles for Pulmonary Vaccines ...... 199 6.2.5. Immunoengineering Applications for PRINT Nanoparticles in the Lung ...... 202 6.3. Outlook ...... 205 6.4. Acknowledgements ...... 207 6.5. References ...... 207

xi LIST OF TABLES Table 1.1 Antibody concentrations in human fluids ...... 24 Table 1.2 Representative examples of clinical and preclinical lung vaccines ...... 35 Table 1.3 Re as a function of airway geometry...... 42 Table 2.1 Example of PRINT parameters used for fabrication of BSA particles ...... 78 Table 2.2 Example of PRINT parameters used for fabrication of HDODA particles .....79 Table 2.3 Slow-freeze lyophilization recipe ...... 80 Table 2.4 ACI sizing results for BSA/trehalose PRINT particles ...... 86 Table 2.5 DLS results of lyophilization study...... 92 Table 3.1 Measured and tabulated particle characteristics ...... 119 Table 3.2 ACI results for small and large pollen-mimic powder samples ...... 121 Table 3.3 Comparison of APS method results to other literature methods ...... 123 Table 4.1 DLS results for NPs before and after OVA functionalization ...... 141

xii LIST OF FIGURES Figure 1.1 A timeline of advances in nanomedicine ...... 3 Figure 1.2 Diagram of the Innate vs. Adaptive Immune System ...... 5 Figure 1.3 Diagram of phagocytosis and generalized signaling steps ...... 7 Figure 1.4 Differentiation of white blood cells ...... 8 Figure 1.5 Images and functions of dendritic cells...... 10 Figure 1.6 Two pathways of MHC presentation ...... 12 Figure 1.7 T cell education by DCs...... 13 Figure 1.8 Antibody structure & classes ...... 15 Figure 1.9 Structure of the mucosa ...... 19 Figure 1.10 General structure of mucin glycoproteins ...... 20 Figure 1.11 Changes in human mucus viscosity as a function of shear rate...... 22 Figure 1.12 PCL in the lung ...... 23 Figure 1.13 Structure of IgA ...... 25 Figure 1.14 Structure of Lymph nodes ...... 27 Figure 1.15 Empirical timeline of vaccine development ...... 30 Figure 1.16 Timeline of the modern approach to vaccine development ...... 32 Figure 1.17 Diagram of lung anatomy...... 39 Figure 1.18 Animal models and their relevant lung geometries ...... 40 Figure 1.19 Equivalent volume spheres ...... 41 Figure 1.20 Examples of particles fabricated through bottom-up approaches...... 45 Figure 1.21 Examples of dry powder particles ...... 46 Figure 1.22 Particles fabricated via top-down methods ...... 49 Figure 2.1 The PRINT process ...... 72 Figure 2.2 Scalability of the PRINT Process ...... 73 Figure 2.3 Phase diagram representation of lyophilization process...... 74 Figure 2.4 Diagrams of an Andersen Cascade Impactor ...... 76 Figure 2.5 Aerosol dispersion devices ...... 81 Figure 2.6 BSA PRINT particles of a variety of sizes and shapes ...... 82

xiii Figure 2.7 Log probability plot of particle distributions ...... 84 Figure 2.8 ACI sizing results for BSA/trehalose PRINT particles ...... 85 Figure 2.9 ACI sizing results for donut & pollen series of PRINT HDODA particles .....87 Figure 2.10 ACI sizing results of 10 µm pollen ...... 88 Figure 2.11 Lyophilization solvent study ...... 90 Figure 2.12 Optical images of the lyophilized cakes of 80x320 nm ...... 91 Figure 2.13 ACI sizing results of 3 µm donuts lyophilized in 7 wt% sugar ...... 93 Figure 2.14 Diagram of freezing process of particle suspensions ...... 94 Figure 3.1 Equivalent spheres of a non-spherical particle ...... 106 Figure 3.2 Theoretical velocity profiles of the APS nozzle ...... 113 Figure 3.3 Calculated residence times for particles with varied shape factor ...... 114 Figure 3.4 SEM micrographs of PRINT microparticles...... 118 Figure 3.5 APS number results for fenestrated toroid series...... 120 Figure 3.6 Pollen ACI distributions ...... 121

Figure 3.7 Relationship between experimentally measured DAPS and corrected DAE ...... 122 Figure 4.1 Diagram of mouse LNs, with medistinal LNs ...... 138 Figure 4.2 Ovalbumin functionalization of hydrogel PRINT particles ...... 140 Figure 4.3 Representative SEM micrographs of functionalized NPs...... 141 Figure 4.4 IL-6 cytokine analysis of BALF and serum following NP instillation ...... 142 Figure 4.5 H&E stained lung sections of lungs treated with (ζ+)NP and (ζ-)NP ...... 143 Figure 4.6 IL-6 cytokine analysis of BALF and serum after LPS challenge ...... 144 Figure 4.7 Immunohistochemistry of (ζ+) and (ζ-) NP-treated lungs ...... 145 Figure 4.8 Representative gating to identify NP+ lung APCs...... 146 Figure 4.9 NP association in AMs treated with NP via instillation ...... 147 Figure 4.10 NP association in CD103 DCs treated with NP via instillation ...... 148 Figure 4.11 NP association in CD11b DCs treated with NP via instillation ...... 149 Figure 4.12 Percentage of AMs out of total CD45+ cell from lungs ...... 150 Figure 4.13 Percentage of CD103 DCs out of total CD45+ cell from lungs ...... 151 Figure 4.14 Percentage of CD11b DCs out of total CD45+ cell from lungs ...... 152

xiv Figure 4.15 Representative LN gating...... 153 Figure 4.16 Percentage of CD11c+ DCs from lungs treated with NP via instillation ...... 153 Figure 4.17 Percentage of DC subtypes out of CD11c+ from lungs treated with NP ...... 154 Figure 4.18 NP association in DC subtypes ...... 155 Figure 5.1 Diagram of BMDC, OT-II CD4+ T cell co-culture...... 168 Figure 5.2 Representative gating and purity of BMDCs and OT-II co-culture cells ...... 170 Figure 5.3 CD4+ OT-II T-cell proliferation in response to NP-OVA treated BMDC.....171 Figure 5.4 Average CD4+ OT-II T-cell proliferation ...... 172 Figure 5.5 Internalization of cationic and anionic OVA-conjugated NP by BMDC ...... 173 Figure 5.6 MHCII & T-cell co-receptor expression by BMDC following NP-OVA ...... 174 Figure 5.7 Cytokine and chemokine expression by NP-OVA treated BMDC...... 175 Figure 5.8 Immunization schedule ...... 177 Figure 5.9 GC B cells in the mediastinal LN following 1° and 2° immunization ...... 178 Figure 5.10 CD4+ T-cell activation following pulmonary immunization ...... 179 Figure 5.11 OVA-specific IgG antibody titers following immunization with OVA-NP...180 Figure 5.12 OVA-specific IgA antibody titers in BAL immunization with OVA-NP ...... 181 Figure 5.13 OVA-specific IgG and IgA antibody titers from co-delivery with CpG ...... 182 Figure 5.14 OVA-specific IgG and IgA antibody titers in plasma and BAL of MyD88-/- 183 Figure 6.1 Comparison of PRINT-zanamivir particles with Relenza ...... 194 Figure 6.2 Computer simulated trajectories of equal volume PRINT aerosols ...... 197 Figure 6.3 OVA allergy model...... 203 Figure 6.4 OVA-specific serum IgE levels following instillation challenge ...... 204

xv LIST OF ABBREVIATIONS ACI - Andersen cascade impactor AEM - 2-aminoethylmethacrylate APC - Antigen presenting cell APS - Aerodynamic particle sizer BAL - Broncheoalveolar lavage BALF - Broncheoalveolar lavage fluid BALT - Bronchus-associated lymphoid tissues BCR - B cell receptor BMDCs - Bone marrow derived dendritic cells Bmem - B memory cells CFD - Computational fluid dynamics CJN - Collision jet nebulizer CTL - Cytotoxic T cell DCs - Dendritic cells dLN - Draining Lymph node DPI - Dry powder inhaler ED - Emitted dose Fab - Antigen binding region Fc - Constant region FCA - Freund’s complete adjuvant FDA - Food and Drug Administration FIA - Freund’s incomplete adjuvant FMO - Fluorescence-minus one FPF - Fine particle fraction G - Generation GALT - gut-associated lymphoid tissues GC - Germinal center GSD - Geometric standard deviation H&E - Hematoxylin and eosin stain Hc - Heavy chain HEV - High endothelial venules HIV - Human immunodeficiency virus HP4A - Tetra(ethylene glycol) monoacrylate HPMC - Hydroxypropyl methylcellulose HPV - Human papillomavirus virus i.d. - Intradermal i.m. - Intramuscular i.n. - Intranasal IACUC - Institutional Animal Care and Use Committee Ig - Immunoglobin IgA - Immunoglobin A Lc - Light chain

xvi LN - Lymph node LPP - Large porous particles LPS - Lipopolysaccharide MALT - Mucosa-associated lymphoid tissues MDI - Metered dose inhaler MFI - Median fluorescence intensity MHC - Major histocompatibility complex MMAD - Mass median aerodynamic diameter n.s. - Not significant NMAD - Number median aerodynamic diameter NP - Nanoparticle OVA - Ovalbumin PAMP - Pathogen-associated molecular patterns PCL - Periciliary layer PDI - Polydispersity index PDMS - Polydimethyl siloxane PEG700DA - Poly(ethylene glycol)700 diacrylate PET - Poly(ethylene terephthalate) (PET) PFPE - Perfluoropolyether PRR - Pattern recognition receptors PVOH - Polyvinyl alcohol s.c. - Subcutaneous SC - Secretory component SIgA - Secretory immunoglobin A Spln - Spleen TCR - T cell receptor TH - T helper TH1 - T helper 1 TH2 - T helper 2 TLR - Toll-like receptor Tmem - Memory T cell TPO - Diphenyl(2,4,6-trimethylbenzoyl)phosphine oxide Treg - Regulatory T cell UT - Untreated VLP - Virus-like particles WT - Wild type Zavg - Average hydrodynamic diameter ZP - Zeta potential (symbol)

xvii LIST OF SYMBOLS Q - Flow rate DAE - Aerodynamic diameter Re - Reynolds number χ - Dynamic shape factor FD - Drag force Μ - Fluid viscosity DEV - Diameter of equivalent volume sphere V - Particle velocity CC - Cunningham slip correction factor λ - Mean free path of fluid VTS - Terminal settling velocity ρp - Particle density ρ0 - Standard density G - Acceleration due to gravity U - Fluid velocity φ - Sphericity Mair - Mass of bulk sample in air Mwater - Mass of bulk sample in water ρwater - Density of water ρair - Density of air TexpAPS - Calculated experimental residence time in APS χAPS - Calculated dynamic shape factor from APS model d50 - Midpoint diameter of cumulative fraction distribution T - Residence time DAPS - DAE from experimental APS measurements χSed - Calculated dynamic shape factor from sedimentation measurements χOblate - Calculated dynamic shape factor from oblate approximation χPSA - Calculated dynamic shape factor from projected surface area approach

xviii

CHAPTER ONE

Introduction to Precision Engineered Biomaterials, Mucosal Immunology, Vaccinology and Pulmonary Drug Delivery

Based on: 1. Sections from Petrosko, Fromen, Auyeung, DeSimone, Mirkin, NAE The Bridge, 2013. Reproduced with permission. 2. Preprint of sections of a book chapter from Fromen, Dunn, DeSimone, Handbook of Nanobiomedical Research: Fundamentals, Applications and Recent Developments. © World Scientific Publishing Company, forthcoming 2014. 3. Fromen, DeSimone, in preparation. 1.1. Introduction: Precision Engineered Biomaterials for Medical Applications In 2011, the term “convergence” was used by Phil Sharp and Bob Langer in a perspectives piece to Science Magazine to define a growing effort at the interface of the physical and life sciences. The authors called not just for increased collaboration, but continued and focused consolidation of these seemingly disparate areas in order to hail a “century of biology” and revolutionize medical sciences.1 They identified a number of problems in human health than can only be addressed through application of engineering principles to state-of-the-art biological knowledge. Considerable efforts from engineers, chemists, material scientists, biologists, pharmacologists, physicians, entrepreneurs, and regulators will be required to converge on a similar set of objectives.1-4 Precision-engineered biomaterials are on the order of biological systems and therefore lend themselves to medical breakthroughs. Materials defined on at least one dimension less than 100 nm are the ideal size to efficiently interact with biological structures; these materials can act as scaffolds to attach various biomolecules and organize them into useful architectures. Recent advances in material science engineering have enabled materials to be precisely engineered at this length-scale and thus direct their interactions with biological systems for a specific outcome. In parallel, increased knowledge of cellular mechanisms in response to various stimuli has given insight to cellular functionality throughout the human body. By merging these two parallel research themes, it is increasingly possible to design materials with a diverse range of specific biological responses. This has the potential to transform the way we treat disease and redefine therapies for diseases ranging from cancer to asthma to .5-7 As such, nanotechnology and precision engineered particles have been at the forefront of convergent science. Over the past 20 years, nanotechnology has matured from a field focused on understanding miniaturization and its consequences to one defined by the rational design, synthesis, and manipulation of nanoscale objects. Many advances in nanotechnology have had an extraordinary impact on the medical field, enabling some of the most meaningful developments in diagnostics, imaging, and therapeutics over the past several decades. A timeline highlighting some of these advances is shown in Figure 1.1.

2

Figure 1.1. A timeline of advances in nanomedicine. Reproduced with permission from reference.4

Given the tremendous progress in engineering biomaterials on the nanoscale, one continuing area of application lies at the interface of nanoscience and the human immune system. Immunology is an inherent part of every aspect of medicine and is a field that is continually developing. Nano- and micro-particles interact differently with immune specific cells than small molecules, which offers unique opportunities to mimic biological systems and modulate immune responses. This has a very powerful implication for medical applications; inert particles can be programmed to stimulate the immune system and recognize and eliminate cancer cells or to produce protective antibodies against an infectious disease. Particles capable of immunomodulation could be designed to retrain the immune system as a way to treat autoimmune diseases, such as multiple sclerosis and diabetes. Furthermore, nanoparticles are compatible with many routes of administration (e.g., inhalation, nasal, dermal, and oral), allowing delivery to specific sites in the body to leverage local biological mechanisms.5-7

3 The work presented in this thesis is an example of convergent research. The following content in Chapter 1 will provide the reader with the necessary overview of the immune system, vaccinology, and pulmonary drug delivery, recognizing that the reader is most likely unfamiliar with at least one of these topics. It is important to note these sections are not exhaustive, as each topic could easily comprise entire volumes of work (and have!). This background will provide the reader with the appropriate content to understand the thesis work towards developing a particulate pulmonary vaccine.

1.2. Immunology Primer The human body is constantly exposed to a barrage of bacteria, viruses, toxins, and other environmental stimuli that have lethal intentions. A number of mechanisms have evolved to protect us from continual infection, collectively referred to as the immune system. The human immune system is a complex network of cells and responses which protects us from foreign material. The frontline of defense can be described as the innate portion of the immune system, consisting of a cellular response and the physical barriers of the human body intended to prevent infection. Amazingly, the immune system also can establish a memory response to rapidly respond to fight a previously encounter pathogen. This adaptive immune response is common for all jawed vertebrates and is the basis for prophylactic vaccines.8 While the innate immune system can respond immediately to an invader, the adaptive immune system requires a longer time frame to establish a response, as shown in Figure 1.2.

4

Figure 1.2. Diagram of the Innate vs. Adaptive Immune System. Included are relevant cell types, as well as the approximate timescale for action. Reproduced with permission from reference.9

In order to use precisely designed biomaterials to interact with the immune system, one needs to have an understanding of the immunological mechanisms already in place. To this aim, the following sections will briefly cover these three major areas of an immune response.

1.2.1. Physical Barriers The first line of defense implemented by the immune system is passive prevention.10 Numerous physical barriers prevent pathogens from entering the body by limiting their contact to susceptible tissues. The skin represents one such barrier, which consists of an outer layer of keratinized epithelial cells to prevent physical entry through the epidermis, due to tight junctions between cells and hydrophobic characteristics of keratin fibers.11, 12 Another physical barrier results from the physical geometries of the human anatomy, which also act to

5 limit pathogen exposure. For example, the human lung consists of over 23 bifurcations, resulting in a complex network that restricts pathogen access to the deep alveolar regions in the parenchyma.13-15 Additional geometric pathogen restriction occurs due to the small openings of the gastrointestinal and urogenital tracts.12 To further protect important tissues, a mucosal barrier coats the luminal surfaces of the respiratory, gastrointestinal, and urogenital tracts, as well as the inner ear and conjunctiva of the eye. Mucus is responsible for entrapping, digesting, and clearing foreign pathogens, effectively preventing them from ever reaching the underlying epithelium.12, 16-18 It is a semipermeable membrane, allowing nutrients to diffuse through, but possessing viscoelastic and adhesive properties which physical and chemically restrict pathogen motility.19 Mucus covers over approximately 400 m2 of surface area in an average human body and is a major defender of the immune system.17 The mucosal layer will be discussed in greater depth later in Section 1.3.

1.2.2. Innate Immune System Passive barriers alone are not enough to thwart the intentions of virulent pathogens. Bacteria and viruses have evolved mechanisms to penetrate mucus and proliferate in the human body. The innate immune system is required to actively combat these pathogens. To do so, the innate immune system implements specialized protein and cellular responses which identify a threat and mount an appropriate response; this response must be a balanced show of force to rapidly isolate and destroy the threat, without detrimentally overreacting. One mechanism employed by the innate immune system is the complement cascade, which is a series of programmed protein interactions that work to disrupt membranes of “non-self” organisms. The other portion of the innate immune system involves response from a specialized series of cells which have been specifically purposed to respond to an invader. When a pathogen penetrates the initial physical barrier, it is greeted by cells of the innate immune system. Present throughout the epithelium and surrounding tissues are sentinel cells called macrophages. These cells roam through the tissue clearing away dead

6 cells and foreign pathogens through a process called phagocytosis, which is diagramed in Figure 1.3.

Figure 1.3. Diagram of phagocytosis and generalized signaling steps. Reproduced with permission from reference.20

Macrophages are a type of leukocyte, or white blood cell, that is a professional phagocyte. Other white blood cells that will be discussed later include monocytes, dendritic cells, lymphocytes and granulocytes, which all originate from a multipotential hematopoietic stem cell in the bone marrow. Bone marrow and the thymus are considered primary lymphoid tissues. Monocytes, dendritic cells and granulocytes (neutrophils, eosinophils) differentiate from a common myeloid progenitor, while other white blood cells, such as lymphocytes (B and T cells) differentiate from a common lymphoid progenitor (Figure 1.4).

7

Figure 1.4. Differentiation of white blood cells. Reproduced with permission from reference.21 © T. Winslow.

Macrophages are in prime locations of the lung, skin and gut and are ready to respond to an invasion. When a pathogen is able to penetrate the physical barriers of the skin or mucosa, the macrophage lying in wait responds by engulfing and digesting the pathogen. Depending on the size of the invasion, macrophages are also equipped to call for back-up. They can secrete various protein molecules, called chemokines and cytokines, which create a diffusion gradient, acting as a beacon to other immune cells. These chemical signals secreted by macrophages, and subsequent recruited immune cells such as natural killer (NK) cells and neutrophils, also act to increase blood flow to the area, stimulate nerves and contract blood vessels to allow more fluid flow to the site. This is why a site of infection often swells and turns red, due to the increased fluid and blood flow. An infection site might also ooze pus; this is from a build-up of dead neutrophils, which rapidly proliferate once recruited to infection sites. Neutrophils also phagocytose pathogens and can dump potent enzymes stored in cellular compartments called granules to kill pathogens. One of the most important cells recruited by the chemical signals secreted by macrophages is the dendritic cell (DCs). DCs are considered an integral part of both the innate and adaptive immune system. Like a macrophage and a neutrophil, a dendritic cell is

8 also a phagocyte, capable of internalizing foreign objects. In homeostasis, DCs behave similarly to macrophages, continually sampling the tissue environment and acting as sentinels on the lookout for an invasion.10 This is reflected in their morphology, often appearing with long protrusions, or dendrites, as shown in Figure 1.5. Upon encountering a pathogen, DCs are activated in a few ways. DCs are sensitive to the cytokines secreted by other immune cells, such as macrophages, which can activate the DCs. Additional activation can occur with direct contact with the pathogen. On the surface of all DCs, regardless of maturation state, are specific proteins called pattern-recognition receptors (PRRs), which have evolved to recognize specific pieces of microorganisms. When a special class of these PRRs, called toll-like receptors (TLRs), recognizes a specific portion of the pathogen, called a pathogen-associated molecular pattern (PAMP), a signaling cascade begins to activate the DC.22 An activated DC can readily internalize and digest the invader, saving important peptide segments from pathogen-derived proteins, called antigens. These fragments are efficiently processed and stored by DCs through a process called antigen presentation (discussed in 1.2.3). Activated DCs then leave site of attack and travel to the lymph node, where they play a pivotal role in directing the adaptive immune response.10, 23-29

9

Figure 1.5. Images and functions of dendritic cells. (A) Fluorescence micrograph of dendritic cells of the skin. (B) Function of dendritic cells in various life cycles. (C) Diagram of dendritic cell antigen sampling. © L. Sompayrac. Adapted and reproduced with permissions from references.10, 29, 30

Interestingly, DCs were first discovered only ~40 years ago by Ralph Steinman and Zanvil Cohn in 1973.28 These cells represent a small population in the body and, since their discovery, numerous subsets have been identified, with functionality depending on the tissue location. However, despite their small numbers, it is increasingly clear that these specialized cells play a huge role as the bridge between the innate and adaptive immune system.10, 23-29, 31

1.2.3. Adaptive Immune System The innate immune system is responsible for the long-term, coordinated attack on a pathogen. The combination of the complement system, macrophage sequestration, and neutrophil degranulation attempts to impede the progress of a pathogen invasion. Often, these elements are sufficient to completely prevent a pathogen from taking residence in the host. In the case of more serious pathogens or frequent offenders, the innate immune system requires

10 additional support. While the innate system continues to fend off the attack, the next wave of support comes in the form of the adaptive immune system. The adaptive immune system is an elegant mechanism which can respond to an incredibly diverse range of pathogens, as well as establish a memory response. Unlike the innate immune response, which has a slew of generalized tools intended to kill a pathogen, the adaptive immune response develops a series of weapons which can respond in force to eliminate a very specific pathogen. These weapons include antibodies and effector T cells, which are directed to a specific target through explicit instruction following the coordination of DCs, B cells and T cells, as well as the presence of signaling proteins. This complicated series of interactions is required to keep the adaptive immune system in check and only respond to the attacking pathogen. DCs are central to directing the adaptive immune response. Following encounter with a pathogen, key proteins from the pathogen, or antigens, are digested by the DC and transferred to the surface of the DC in order to display these contents to other immune cells it encounters. This process occurs by two separate pathways, involving two unique proteins: major histocompatibility complex (MHC) I and II. The MHC I protein is expressed on most cells in the body as a way to monitor the events occurring inside of each individual cell. The MHC II protein is found only on antigen presenting cells (APCs), which include DCs, macrophages and B cells. Upon internalization of the antigen, DCs will display the appropriate peptide fragment in either MHC I or MHC II, depending on the method of internalization. This is depicted in Figure 1.6.

11

Figure 1.6. Two pathways of MHC presentation. (A) MHC I pathway for display of internal peptides. (B) MHC II pathway of external peptides. Reproduced and adapted with permission from reference.10 © L. Sompayrac.

Through these surface displays, DCs can help other immune cells, importantly T cells, recognize pieces of the pathogen. This process of antigen presentation to educate T cells is shown in Figure 1.7. In addition to antigen recognition, T cells need further direction; they must be instructed on how to respond to that antigen once it encounters it. DCs also provide this instruction. When the DC first encountered the pathogen, it was activated by a PAMP, which was recognized by one of the TLRs on the cell membrane. In conjunction with other environmental cues, these TLRs initiated a cascade of internal signals via (NF)-kB-dependent signaling, which gave insight to the DC about the best way to respond to the pathogen.26, 32 This information will then be conveyed during T cell activation process through cytokine secretion from the activated DCs. The antigen display on the specific MHC molecule and the secreted cytokine milieu will direct a naïve T cell to a particular function, as diagramed in Figure 1.7.

12

Figure 1.7. T cell education by DCs. DC MHC displays, co-stimulatory molecules and secreted cytokines for (A) CD4+ Th1 T cell education, (B) CD4+ Th2 T cell education, (C) + + CD4 Th17 T cell education, (D) CD4 TH education and corresponding B cell stimulation, (E) CD8+ T cell education against intracellular pathogens, and (F) CD8+ T cell education from antigens of “self”. Adapted and reproduced with permission from reference.33

T cells are lymphocytes that can be defined by the presence of surface receptors called T cell receptors (TCR). Each T cell has a slightly different permutation of TCR, due to genetic rearrangement of the gene segments which code for the TCR, resulting in a huge diversity of TCRs on T cells. An adult human has about 300 billion T cells, which mature in the thymus, enabling the total collection of T cells to recognize almost an infinite amount of protein antigens. It is this variation which makes the adaptive immune system so powerful! Traditional T cells will have TCRs that recognize a specific antigen-MHC complex (either I or II) on the surface of an APC and will also express a specific co-receptor molecule (either CD4+ or CD8+). CD4+ or CD8+ co-receptor proteins aid in the recognition of the proper

13 MHC molecule and are used to further differentiate the type of T cell and its function. Thus, traditional T cells are divided broadly into three major categories: T helper cells (TH), 10 cytotoxic T cells (CTL), and regulatory T cells (Treg). In general, TH and Treg will express CD4+ and recognize peptides in MHC II displays; CTL will express CD8+ and recognize antigens in MHC I displays. TH can be further divided into T helper type I (TH1), T helper type 2 (TH2), T helper type 17 (TH17) or T follicular helper cells (TFH); their functions are shown in Fig. 1.7. Regardless of the type of T cell, once the proper connection between TCR and co- receptor molecules has been made, the T cell will be activated, initiating signaling pathways for rapid proliferation of this specific type of T cell. This process is called clonal selection and expansion. A tremendous increase in an identical population of T cells will result and these T cells can then perform their function at the infection site. CD8+ CTL effector T cells will directly attack infected cells, instructing them to die via apoptosis and thus eliminating 10 the contents of the cells, which is an important containment step for viral infections. TH1,

TH2, and TH17 cells will secrete a specific panel of cytokines that recruit more immune cells to the battle. Although recent discoveries have deemed this an over-simplification, these cytokines historically have been used to generalize the type of a particular immune response, i.e., TH1 responses are considered “pro-inflammatory” to kill intracellular pathogens, while

TH2 responses are considered “anti-inflammatory” and related to allergy. Additionally, TH cells play a critical step in the development of antibody-driven immunity. B cells are another lymphocyte, which are responsible for the production of antibodies. Like T cells, B cells are identified by a surface marker, called the B cell receptor (BCR), and also like T cells, each BCR is selected through genetic scramble of modular design to result in an extremely diverse set of BCRs in the B cell population. While TCRs recognize protein antigens properly displayed in a MHC molecule, BCRs can recognize any organic molecule and do not directly require specific presentation by APCs. Similar to T cells, when a BCR comes in contact with its cognate antigen, it will proliferate by clonal expansion. Following proliferation, each B cell will then mature, proceeding through the process of class switching, somatic hypermutation, and career decision.

14 The ultimate goal of a B cell is to produce antibodies, or immunoglobins. Antibodies are specialized protein complexes which provide an additional weapon to attack the pathogen. Antibodies consist of two protein chains: the heavy chain (Hc) and light chain (Lc). These combine to form the complexes shown in Figure 1.8.

Figure 1.8. Antibody structure & classes. Reproduced with permission from reference.34

Each antibody has an antigen binding region (Fab) and a constant region (Fc). Molecular changes to the Fc region change the class of antibody (Fig. 1.8). Additionally, antibodies and BCRs have the same structure, with slight differences in the Fc region that anchor the BCR to the surface of the B cell. These molecular changes come through rearrangement at the DNA level through modular design and junctional diversity. Further modifications through the genome work to increase the potency of produced antibodies. The term ‘class switching’ refers to the process in B cell maturation which results in changes to the Fc region. All B cells begin by producing IgM, but class switching will allow them to produce one of the other antibody types, which have important differences in application. Also during the maturation process, B cells undergo somatic hypermutation to

15 alter their Fab region, where the rate at which DNA mutations occur is accelerated to change the affinity the antigen binding region has to the cognate antigen. B cells need continual restimulation to continue clonal expansion, naturally selecting mutated BCRs which have an increased binding affinity over those with a lower affinity. The end result of these two changes during B cell maturation is the production of large quantities of high affinity antibodies. Mature plasma B cells can produce as much as 2000 antibodies per second.10 Antibodies secreted by B cells seek out the pathogen and bind to their cognate antigen. Antibody binding is generally a high affinity process, resulting in surface decoration of a pathogen following exposure to an antibody. These antibodies can provide kinetic restrictions, entrapping them in a mucosal layer to encourage clearance. These antibodies can also act as beacons to attract other proteins, such as the complement system. Phagocytes also have surface receptors to recognize the Fc region of an antibody, resulting in increased internalization from macrophages or neutrophils. While B cells are capable of clonal expansion following interaction with their cognate alone, maturation requires T cell help. TH cells provide the necessary co-stimulatory signals which can initiate class switching. As diagramed in Figure 1.7.D, activated TH cells can bind through its TCR to the corresponding MHC II antigen presentation on the B cell and, in combination with additional surface markers on TH cells, such as CD40L, and T cell-secreted cytokines, only B cells which recognize these surface displays will be activated. T help is required for antibody switch from IgM to other antibody classes.31, 35 Additionally, T help is also required for the creation of memory B cells (Bmem), which remain dormant until a secondary exposure to their cognate antigen, resulting in rapid clonal expansion. Having previously undergone somatic hypermutation, these Bmem cells are extra potent at fighting the repeat offender.24, 36-38 The adaptive immune system is incredibly complex! The discussion here merely skims the surface of the various roles and interactions of the adaptive response. The many complicated interactions between APCs, B cells and T cells are really a system of checks and balances, intended to keep control of the tremendous powder adaptive immune system. With proper recognition to an antigen, these cells can do serious damage, but all of the proper

16 recognition steps must occur in order to initiate such a powerful response. Autoimmune diseases, while rare, occur when some piece of this complex puzzle gets out of hand. This incredible ability to develop a response to a specific molecule stems from the diversity of surface molecules. Precision control and organization of the various molecular components of antibodies, BCRs, TCRs, and MHC displays highlights the importance of organization on this length scale; minor adjustments in a single region of any one of these components changes whether a cell is activated or not. The goal of using precision engineered biomaterials is to duplicate pieces of this process in order to produce memory B and T cells that could rapidly destroy virulent pathogens without ever exposing the body to harmful portions of the pathogen. One could imagine duplicating portions of the pathogen, specific regions of the MHC display, or TCR display to produce such results. However, the elegancy and complexity of the adaptive immune response makes this a daunting challenge.

1.3. Mucosal Immunology The previous section provides an overview of the main functions of the systemic immune system, involving primary lymphoid organs of the thymus and bone marrow, as well as the spleen. In addition to the systemic immune system, highly specialized lymphoid organ systems are present throughout the body to associate with mucosal barriers. These mucosa- associated lymphoid tissues (MALT) are compartmentalized systems that balance responses at a particular mucosal interface.17 Unlike the systemic immune system, MALT are continually exposed to foreign material and must recognize a pathogen from inert environmental material. MALT can act independently from the systemic immune system and independently between compartments, resulting in region-specific immune responses. Similar responses can also occur across the entire MALT network, or the “common mucosal immune system”.39 Empirical knowledge of MALT dates to China ca. 900 A.D. to the first documentation of a nasal immunization.40 Smallpox is transmitted through inhalation where it infects the mouth, nose and respiratory tract.41 To prevent the spread of smallpox to

17 uninfected persons, healed pustule scabs were inhaled by an uninfected person directly to the lung and nasal mucosal site. While this ritual would prevent infection for some, in others, the full course of infection would ensue. However, knowledge of this practice would eventually spread to Europe, leading ultimately to the development of the first vaccines.40 Despite the early start, knowledge of the specific anatomical features and functions of the complete mucosal immune system are still being uncovered.40 Possibly the next historical landmark was the identification of gut-associated lymphoid tissue (GALT) in 1673 by Johannes Peyer, who noticed structures surrounding the intestines, now called Peyer’s patches. It wasn’t until the early to mid-1900s that these patches were recognized as an important gut defense mechanism. Identification of other mucosal immune systems has been even slower: bronchus-associated lymphoid tissue (BALT) was first identified in 1973, and the functionality of key lung lymphoid cells is still underway.25, 40, 42-44 Differences exist between the various mucosal compartments, however, many commonalities are present between these subsets. In this section, the specialized aspects of the common mucosal system will be discussed, with a special focus on specific MALT examples where appropriate.

1.3.1. Mucosal Structure The MALT is divided into inductive and effector sites. The inductive sites are secondary lymphoid tissues, such as lymph nodes and Peyer patches, and are regions where education of the adaptive immune response occurs. These will be discussed in Section 1.3.5. They are located in close proximity to the effector sites, which are typically the tissue environment interface, consisting of the eye, nose, throat, gut, lung and urogenital tract. At these effector sites, the mucosa generally consists of a thin layer of epithelial cells, which are covered by a layer of mucus.39 The epithelium of the gut and lung are single layered, while the eye, vagina and esophagus are multilayered.45 In the majority of the respiratory tract, these epithelial cells have protrusions called cilla, which beat regularly, causing the mucus to flow. In contrast, the epithelium in the gut is organized into microvilli.46 Beneath the epithelium is the lamina propria, which anchors the epithelium and

18 allows movement of a dense population of lymphocytes. In multilayered epithelium, the lamina propria also contains glands from which mucins are secreted (mucins in single layered epithelial MALT contain goblet cells which provide this function). This is collectively diagramed in Figure 1.9.

Figure 1.9. Structure of the mucosa. Reproduced with permission from reference.12

1.3.2. Components of Mucus Secretions The mucus blanket covers the interface of the MALT. Mucus is a viscoelastic layer composed of a variety of glycoproteins, lipids and other biomolecules that forms an adhesive layer to physically entrap foreign objects. Mucus is secreted continually by goblet cells and submucosal glands.47 The gut and lung are the most prolific producers of mucus, with the gut producing ~2.5 gal/day. Mucus is comprised mainly of water, ca. 95% by weight.45 The remaining components of mucus are various biomolecules, including glycoproteins, defensins, surfactants, salts, DNA, enzymes, antibodies, dead cells and foreign material. As

19 such, mucus has rather interesting physiochemical properties, often acting as a non- Newtonian fluid.48 The main components of mucus are glycoproteins called mucins. There are over 20 MUC genes identified, producing large molecular weight proteins which generally comprise a common MUC backbone, serine and threonine rich tandem repeating regions, and o-linked carbohydrate chains, as depicted in Figure 1.10.47, 49 These mucins are very diverse, ranging in molecular weight from 0.2-2 x 106 Da, creating challenges in identification, quantification, and determination of functionality.49, 50 However, they are typically polyanionic, obtaining this negative charge from sulfate and sialic acids in the carbohydrate chains.49 Mucins are commonly found as covalently-linked oligomer molecules, and can exist as free-formed macroscopic mucin fibers, ranging 2-10nm in diameter, or surface tethered molecules.48, 49

Figure 1.10. General structure of mucin glycoproteins. Reproduced with permission from reference.47

Another important component in mucus is lipids. Much of the lipophilic properties of mucus come from hydrophobic regions of mucins, encouraging lipid binding to prevent degradation, especially in extreme pH of the intestine.51 In both the lung and the gut, lipids and surfactants are actively secreted by glands and goblet cells for additional functionality. In the lung, airway surfactants are critical components in reducing the surface tension at the air- cell interface and allowing the repeated expansion and contraction of the lung during inspiration. Four lung surfactants, SP-A, SP-B, SP-C, and SP-D, have been identified, and contribute in varying degrees to pathogen binding, opsonization, increased phagocytosis and

20 bacterial death. These same surfactants have been identified in other mucosal surfaces as well.52 Mucus is also composed of bioactive secretions. Within the mucosal layer, a collection of enzymes, proteins, peptides and DNA are continually secreted. These consist of various enzymes, mucinases, sialidases, proteases, antimicrobial peptides called defensins, and collectin proteins, to name a few.52-56 Together, these bioactive components can have tremendous influences on the quality and functionality of mucus, capable of changing bulk viscosity properties. Also present in the mucus is DNA from dead epithelial cells, which can further contribute to dramatic changes in mucus viscosity in diseases such as cystic fibrosis (CF).48 Finally, antibodies are secreted into the mucosal milieu, as will be discussed further in Section 1.3.3. In combination, these biomolecules result in a complex fluid which in a healthy state acts like a shear thinning gel. This allows for a lubricating layer of fluid, with the barrier immediately at the epithelial remaining undisturbed under shear. Glycoproteins are especially known to contribute to the properties classifying mucus as a viscoelastic gel.48 Glycoprotein oligomers are connected through disulfide bonds and noncovalent carbohydrate interactions, which encourages gelation. Additionally, the anionic properties encourage charge stabilization and prevent excessive entanglement between fibers to maintain a decreasing viscosity under shear.49 Salt content also dramatically affects bulk properties, resulting in the swelling or shrinking of the gel.48 Extensive studies on the micro and macrorheology have been performed on human mucus to understand these relations.19, 45, 46, 48, 51, 53, 57-61 An example of viscosity measurements of human mucus is shown in Figure 1.11.

21

Figure 1.11. Plot of changes in human mucus viscosity as a function of shear rate. Reproduced with permission from reference.19

As mentioned previously, minor variations in mucin content due to bioactive components can dramatically change the viscosity. For example, a decrease in the viscosity of cervical mucus has been shown to decrease its protective properties and increase infection rate of sexually transmitted diseases.56 Conversely, an increase in mucus viscosity due to genetic deficiencies in ion pumps in the lung causes the collapse of the mucillary escalator, which in turn leads to stagnant mucus and increased occurrence of infection.61 Unsurprisingly, mucus layers vary between mucosal tissues. Mucus thicknesses depend on the site: thicknesses in the stomach range from 50-450 µm, 110-160 µm in the colon, 1-10 µm in the lung.19 The gut comprises two mucosal layers: a thin inner layer, which is sterile and rarely experiences foreign material, and a thick outer layer, with properties as described above.46 In the lung, the mucus layer is also further divided into two separate phases: a classic mucus layer and a periciliary layer (PCL). The PCL is comprised of membrane-tethered mucins, which lubricate the motile cilia and prevent the mucus gel layer from collapsing. This has been described through a polymer size-exclusion gradient, supported by a gel-on-brush model, as shown in Figure 1.12.61

22

Figure 1.12. PCL in the lung. (A) Histology section of airway. (B) Diagram of gel-on-brush model. Adapted and reprinted with permission from reference.61

The PCL enables the adjacent classical mucus layer to be transported with minimal mixing up the mucociliary escalator, delivering trapped foreign material towards the trachea for digestion. Mucocilary lung transport results in mucus velocity of ~10-100um/s.19 Taken together, the various secretions and resulting complex fluid of mucus actively works to repel invading pathogens and entrap foreign material. Generally, the majority of mucus enters the digestive tract, where these foreign components, as well as their mucus custodians, are digested and excreted.

1.3.3. Secretory Antibodies As shown in Figure 1.8 (in Section 1.2.3), there are five major classes of antibodies, each with their own function. Every B cell begins life secreting IgM, as the large structure and ten antibody binding sites is superior at initiating the compliment cascade for pathogen neutralization and clearance. Following class switch, B cells can secrete different antibody types. These include IgG, the most abundant immunoglobin in the blood, which excels at neutralizing viruses and increasing clearance by phagocytes, and IgE, which is associated with allergy but is productive against parasites. However, the most commonly produced antibody class is IgA, which is specific to the mucosa.10, 62

23 IgA was first discovered by Heremans and colleges in the early 1960s and has been identified as a major defense of the mucosal immune system. Interestingly, IgA is secreted in mother’s milk, coating her child’s gut with protective mucosal antibodies until the child can produce its own. It is estimated that over 3 g of IgA is secreted daily into the mucosa, with concentrations in the mucosa ranging from 2-3 mg/mL.63 A table of the relative antibody concentrations in different fluids is shown in Table 1.1.

Table 1.1. Antibody concentrations in human fluids. Adapted from reference.64 Fluid IgG (mg/mL) IgA (mg/mL) IgM (mg/mL)

Serum 12.0 3.0 1.5 Milk 0.1 1.5 0.4 Saliva 0.004 0.04 0.006 Bile 0.09 0.07 0.02

IgA is generally produced by plasma B cells residing in the lamina propria. As shown in Figure 1.8.E, IgA is a dimeric version form of IgG, which is composed of two Hc and Lc, joined by the J chain. Only IgA and IgM are capable of polymerizing through the J chain. As a result, dimeric forms of IgA have four antigen binding regions and additional molecular bulk, increasing binding avidity and entrapment prowess. The dimeric form also prevents initiation of complement once bound to a pathogen, keeping the mucosal interface from a constant state of inflammation.10 The connection via J chain is required for association of an additional molecule, called a secretory component (SC), which allows IgA to cross the epithelium unto the mucosa. This complex is called secretory IgA (SIgA), as shown in Figure 1.13.

24

Figure 1.13. Structure of IgA. Adapted and reproduced with permissions from reference.65

Transport proceeds through a transmembrane glycoprotein on the surface of mucosal epithelial cells. IgA is extremely hydrophilic and negatively charged. Additionally, humans possess two subclasses of IgA, IgA1 and IgA2; mice have no IgA subclasses.62 IgA has been implicated as a major defense in a number of diseases, as well as neutralization of numerous toxins and enzymes. Selected examples include influenza, rotavirus, enterotoxigenic Escherichia coli, cholera, Chlamydia, and Acanthamoeba.62, 66

1.3.4. Cells of the Mucosa As with other aspects of mucosa discussed thus far, the associated lymphocytes are also specialized. In this section, the important cellular immune players of the mucosa are described. Specialized subsets of APCs exist in all mucosal tissues. For example, the lung has both interstitial and alveolar macrophages, whose main function is phagocytosis and are rarely implied in lymphatic trafficking.44, 67 Conversely, macrophages the lymph nodes are professional antigen presenters and can initiate antibody production.24 Similar divisions are observed for DCs, depending on the mucosal tissue and the required DC function. At each mucosal site, there are DC subsets capable of draining to local lymph nodes. Additionally, there are subsets of DCs that are locally retained, maintaining antigen presentation and T cell priming directly at the effective site.24, 25, 44, 68 DCs can be differentiated through a number of

25 surface receptors, including CD8, CD11c, CX3CR1, CD207, CD11b and CD103. CX3CR1 is a chemokine receptor, and subsets expressing this and CD103 typically represent a migratory population.44, 66 Many of these DC populations in the lung and other mucosal sites are still being identified.23, 25, 27-29, 69 Work presented in Chapter 4 will expand on some of these DC populations specific to the lung. B and T lymphocytes also exhibit unique functions at mucosal sites. During antigen presentation in the secondary lymphoid tissues, T cells are not only activated and instructed by the DCs, but they are also directed to a tissue of action. T cell homing allows them to selectively migrate to the issue under attack; memory T cells are also localized to these 70, 71 tissues. The presence of TH cells in the secondary lymphoid tissues will also encourage the proliferation of IgA-secreting plasma B cells, which can also home to the effector organ. It is this aspect of the adaptive mucosal immune system which highlights why the route of exposure is so important to the resulting response.17, 18, 24, 66, 68 There are a number of other important cells which are localized at the mucosa that have specialized functions. These include M cells, mast cells, eosinophils, and mucosal epithelial cells, which are beyond the scope of this work.12, 16, 18, 24, 32, 39, 66, 68, 72 Additionally, mucosal surfaces are continually exposed to bacteria, such as the flora in the gut. Many of these bacteria are not pathogenic, but actually required for bodily function. This microbial co-dependence places a unique challenge on the mucosal immune system, which must recognize and tolerate these bacteria, while still capable of destroying a virulent invader. This interplay with the microbiota has been extensively reviewed elsewhere.73

1.3.5. Secondary Lymphoid Tissues There are over 300 billion T cells and over 3 billion B cells, with small naïve populations that recognize a single antigen.10 Following an invasion, there is potentially a single APC which is presenting the cognate antigen. How do these cells ever find each other to mount a response? Secondary lymphoid tissues are the inductive sites where the magic of the adaptive immune system of the mucosa takes place. APCs are directed to these tissues immediately

26 after encountering an antigen, where they are exposed to a continual stream of T and B lymphocytes. These two cell types have been migrating from one secondary lymphoid tissue to the next, constantly testing their receptors against those of the APC. By funneling all of the cells to these central locations of the lymph nodes, tonsils, Peyer’s patches, and other regions of the MALT, the odds of a B or T cell meeting its cognate antigen are dramatically increased.74, 75 There are somewhere between 100 and 200 lymph nodes in a human adult. Lymph nodes are specialized lymphoid tissues connecting the lymph system which are intended to localize lymphocytes. T and B cells enter from the blood through high endothelial venules (HEV), where they localized to B and T cell rich areas of the lymph node. These regions are constantly exposed to lymph fluid, which contain APCs and soluble antigens from the mucosal tissues. APCs can also move along on the loose network of fibers which makes up the subcapsular sinus, increasing their exposure to passing B and T cells.74, 75 A diagram of the structure of a lymph node is found in Figure 1.14

Figure 1.14. Lymph nodes. (A) Structure of lymph node, with B cell zones in blue and T cell regions in white. (B) Fluorescence micrograph of lymph nodes stained with sinus stained in green and T cells (CD3) stained with red. Adapted and reprinted with permission from reference.74

27 Following exposure to the cognate antigen, T and B cells will begin clonal expansion. In the lymph node, B cells will proliferate in B-cell follicles that develop into larger complexes called germinal centers (GCs). GCs can be found in all secondary lymphoid tissues and are the regions where B cells will mature, class switch, undergo somatic hypermutation, and develop into plasma or memory cells. They must be continually exposed to APCs, which linger in the local lymph node.

1.3.6. Disease at the Mucosa Most of the world’s most infectious diseases occur at the mucosa. HIV/AIDS, tuberculosis (TB), measles, malaria, diarrhea and respiratory infections combine to cause over 13 million deaths per year, with only malaria having no mucosal component.16 Diarrhea alone is attributed to 2 million deaths annually, particularly in children younger than 5, and another 2 million attributed to tuberculosis.16, 32, 76 The following section details some particular pathogens, with the evolved immune response. Enterotoxigenic Escherichia coli and Vibrio cholera are bacterial gut infections which are common causes for diarrhea. These bacteria colonize the epithelia of the small intestine, secreting toxins that result in watery stool and dehydration. These are combatted by the immune system through increased production of IgA. Escherichia coli is also implicated in urinary tract infections, while pathogens such as chlamydia and Neisseria gonorrhoeae are also major pathogens of the urogenital tract. These and other bacterial infections, such as Streptococcus pneumoniae, causing pneumonia in the lung, are also most effectively cleared + by antibodies, occurring in a TH CD4 dependent pathway. However, the most commonly virulent lung bacteria, Mycobacterium tuberculosis, require a TH1 response to halt disease progression, as antibodies do not provide sufficient protection.76-80 The bacteria responsible for TB targets alveolar macrophages and is capable of maintaining a latent inside these phagosomes; CD4+ TH1 cells require recognition of TB antigens displayed in the phagosome’s MHC surface displays to eliminate the cell.76, 80, 81 Viral infections also occur at the mucosa. The seasonal flu is caused by the influenza virus, which causes more than 16 billion dollars due to health care costs in the US. Following

28 exposure, SP-A and SP-D lung surfactants and defensins work to inhibit viral activity in the mucus, while B cells produce anti-influenza antibodies against the hemagglutinin (HA) portion of the virus. CD8+ T cell CTL responses are also important for influenza protection.32 In the gut, viral infections include noroviruses and rotaviruses, which are major causes for epidemic gastroenteritis and diarrhea. The main immunogenic weapon against noroviruses seems to be secretory IgA, while a major CTL CD8+ T cell response has been observed in rotavirus infections.32 Finally, autoimmune diseases can play a critical role at the mucosa, stemming from the active involvement of the immune system at these sites. Celiac disease results from rampant inflammation of the bowel following exposure to a gluten protein, resulting in chronic constipation and anemia. Other gut autoimmune diseases such as ulcerative colitis and Crohn’s disease are also characterized by inflammation and the infiltration of T cells. The respiratory disease of asthma is often considered an autoimmune disease, as the acute and long-term inflammation originates from an overreaction of the immune system to an otherwise inert environmental molecule. The pathology of these autoimmune diseases is often quite complex, but has dramatic effects on the functionality of the mucosal protective layer. In total, immune responses in MALT are generally sufficient to mount an adaptive immune response to many mucosal specific pathogens. However, these infectious diseases still represent a huge global health burden. Prophylactic prevention would dramatically alter the quality of life, especially in poor regions, and the promise of complete extinction of these pathogens is certainly enticing.

1.4. Mucosal Vaccines Vaccination has been called the greatest success in modern medicine.82 The term is derived from the Latin word for ‘cow’, vacca, originating from the landmark work of in 1798.83 Jenner observed that milkmaids who were exposed to the cowpox disease were immune from the deadly smallpox form of the disease. He then inoculated a small boy

29 with cowpox pustules, finding the patient immune to the virulent form of smallpox, with only mild side effects. This is the first example of an “attenuated” virus, ultimately leading to the complete eradication of smallpox in 1979.41, 83 Since this discovery, a number of vaccines have been developed, which, in general, have targeted the systemic immune system. The advent of mucosal vaccines is relatively recent. As such, a brief overview of vaccines follows, with a focus on current research trends in mucosal vaccinology.

1.4.1. Vaccine Overview Historically, vaccine development falls under two broad approaches. Following the work by Jenner, vaccine development progressed empirically, requiring the identification, isolation, and subsequent inactivation of the pathogen. A timeline of empirical vaccine development is shown in Figure 1.15.

Figure 1.15. Empirical timeline of vaccine development. Adapted and reprinted with permissions from reference.83

30 Vaccines produced by this approach consisted of three different types of vaccines: live attenuated, inactivated, and vaccines. Live attenuated vaccines (LAVs) are live pathogens that have been passaged through another organism to mitigate its ability to replicate in humans. LAVs are often passaged upwards of 200 times in different cell cultures or chicken embryonic eggs, such as live attenuated influenza vaccines, and are generally thought to produce the most robust immune response of vaccine types.83 Examples of LAVs include the measles, mumps and rubella (MMR), chickenpox, TB (the Bacille Calmetter Guerin, BCG vaccine), and nasal form of influenza.84 Inactivated and toxoid vaccines are both forms of deactivated vaccines, originating from either the actual pathogen or bacterial toxins which have been disabled through chemical or thermal exposure.83 An example of an inactivated form is the and examples of toxoid vaccines include diphtheria and tetanus. These forms produce a shorter immune response, often requiring boost doses.84 All of the vaccine types discussed above require isolation from some type of organism, resulting in safety, efficacy and manufacturing challenges. In the 1970s, glycoconjugate vaccines revolutionized vaccinology. Protein antigens derived from pathogens of meningococcus, influenza B, and pneumococcus were conjugated to a polysaccharide backbone, providing the first instance of a vaccine not derived from a microorganism.83 The immune system only responds to specific proteins in a given pathogen, allowing for the separation of these antigens from the virulent pieces of the pathogen. This is called a . Unfortunately, the antigen alone is often not sufficient in producing a robust immune response, requiring an additional stimulatory agent, called an adjuvant. In the case of glycoconjugate vaccines, adjuvant properties are provided by the resultant structure of the self-assembly complex. Other subunit vaccines have been developed through the isolation of particular antigens, recombinant DNA technology, virus-like-particles (VLP), and reverse engineering approaches.83 A timeline of this rational approach of vaccine development is shown in Figure 1.16.

31

Figure 1.16. Timeline of the modern approach to vaccine development. Adapted and reprinted with permissions from reference.83

Currently approved subunit vaccines include hepatitis B, pertussis, injectable influenza and human papillomavirus (HPV).84 The development of corresponding adjuvants required for these vaccines will be discussed in section 1.4.5. Regardless of the type of vaccine, efficacy is measured by a number of parameters. These include antibody titer, T cell quality, memory responses, and survival against a pathogen challenge.12, 17, 18, 38, 66, 82, 85, 86 The overall shift in vaccine design approach came from incorporation of immunology. Further advancement in vaccine development is anticipated from future convergence with chemical and materials engineers.

1.4.2. Benefits of Vaccination at Mucosal Surfaces While most infections involve the mucosa, historically have been developed for the systemic or peripheral immune system. These vaccinations are delivered

32 by intramuscular (i.m.) or subcutaneous (s.c.) injection and often fail to provide protection at the mucosal site, due to the compartmentalization of the mucosal immune system. Clinical and preclinical studies have demonstrated that vaccine delivery directly to the mucosal site can provide equivalent systemic protection and often equivalent or increased mucosal protection. Delivery directly to the mucosa could provide superior protection for many infectious diseases, such as TB or influenza, as well as therapeutic vaccines for diseases such as cancer.6, 71 Mucosal vaccination also has logistical benefits. This approach requires vaccine delivery through routes of administration that are not injectable, including oral, pulmonary or topical for urogenital or ocular delivery. These could be delivered with minimal medical training, eliminating dangerous biohazard waste. These advantages are especially valuable in third world countries, where mass immunization campaigns are most needed. The alternative routes discussed also potentially have manufacturing and storage advantages over injectable vaccines.

1.4.3. Current Mucosal Vaccines As of this thesis, there are currently only a handful of commercially available vaccines in the U.S. which involve the mucosa. The ® subunit vaccine (Merck) is delivered intramuscularly but provides mucosal protection against four types of HPV. Two other vaccines, both live attenuated, are given directly to the mucosa. RotaTeq® (Merck) is delivered orally, providing gut protection against rotavirus, and FluMist® (MedImmune), is given nasally to protect the upper respiratory tract against seasonal influenza. Additionally, the most commonly implemented polio vaccine was the LAV oral dosage form, developed by Sabin and colleagues in 1960, which was hugely successful in almost complete eradication of the disease, due to ease of administration and mucosal protection.66 Despite the relatively few available mucosal vaccines, there is extensive clinical and preclinical research supporting the efficacy of this approach, spanning multiple routes of administration. There are also a number of licensed vaccines outside of the US, especially for oral vaccine administration for cholera and typhoid. Preclinical research has shown

33 promising results from oral vaccines against Shigellosis and enterotoxigenic Escherichia coli. Additional preclinical research has shown that oral administration of HIV vaccines provides responses at distant mucosal sites, as well as systemic protection. However, as HIV depletes important CD4+ T cells, novel approaches to bolstering the immune response are still needed.66 In addition to oral vaccine delivery, other routes of administration have been successfully demonstrated in preclinical studies as well. Preclinical nasal administration to mice of Prevnar® (Wyeth) against pneumonia provided protection to the nose, lung, and gut, while systemic administration did not. Nasal administration of subunit meningococcal vaccines also resulted in superior nasal mucosal IgA when compared to injected forms.66 Finally, vaccination by aerosol administration to the lung has been successfully demonstrated, as reviewed in the following section.

1.4.4. Pulmonary Vaccines Like other mucosal vaccines, delivery of vaccines to the respiratory tract results in superior lung immune responses by using the inherent immune mechanisms of the lung. Limited lung efficacy of traditional i.m. or s.c. vaccines is particular evident in vaccines for influenza, pertussis, and tuberculosis.87 Delivery to the site of infection for these diseases, such as the lung, has been shown to provide superior protection. Additionally, dry powder approaches, such as spray drying, result in manufacturing, storage, and cost benefits, enabling the potential for mass vaccination campaigns, especially to third world areas.87-90 Delivery of inactivated whole pathogen, LAV, toxoid, and subunit vaccines have been successfully demonstrated via the respiratory tract, delivered by nebulized liquid formulations, dry powder microparticles, or nanoparticles. A summary of studies involving pulmonary vaccination is shown in Table 1.2. In general, dry powder formulations have been shown to deliver inactivated and protein antigens with improved lung mucosal involvement, especially in humoral IgA responses, providing some of the leading candidates for translation. Clinical studies have been performed in humans with aerosol delivery of nebulized TB, measles, and influenza antigens, proving safety and efficacy of this

34 approach.91-93 Importantly, one clinical studies has involved a measles vaccination of spray dried microparticles.94

Table 1.2. Representative examples of clinical and preclinical vaccines delivered to the lung. Figure adapted and expanded from refs [87, 89] with permissions. Abbreviations: PPS – Pluronic stabilized polypropylene sulfide nanoparticles, SD – spray dried, NP – nanoparticle, MP – microparticle, HBV – Hepatitus B virus, ICVM – interbilayer-crosslinked multilamellar vesicles , VLP – virus-like particles, i.n. – intranasal, i.d. – intradermal, LAV – live, . Disease Formulation Species Summary Ref Diphtheria SD MP, Guinea pig Comparison of formulation & administration route: pathogen- 95 Toxoid specific production of mucosal & systemic Ig HBV SD MP/NP, subunit Guinea pig Comparison of administration route: pulmonary produced 96 systemic IgG and more IgA than i.m. Emulsion MP, Rats Comparison of particle size: particles DAE ~5 µm by 97 subunit pulmonary produced greater response than DAE ~12 µm Emulsion NP, Rats Comparison of particle formulation: Hydrophobic NPs >500 98 subunit nm elicited more mucosal IgA (lung, vagina, saliva) and splenic cytokines HPV Nebulized VLP, Human Comparison of administration route: pulmonary delivery 99 subunit superior to i.m and i.n. for IgG and IgA titers at cervix

Influenza SD MP, Mouse Comparison of particle formulation: both formulations 100 subunit provided higher Ig following pulmonary administration Liquid & SD MP, Mouse Comparison of formulations: powder provided both humoral & 101 subunit cellular mucosal responses, liquid only provided systemic Nebulized Human Comparison of administration route: pulmonary delivery 92 LAV provided superior protection and lower side effects than traditional s.c. administration Measles SD MP, Cotton rats Comparison of administration route: pulmonary delivery 102 LAV provided equivalent Ig response compared to i.m. SD MP, Macaque Comparison of administration route: pathogen-specific 103 LAV neutralizing Ig formed following pulmonary administration, lower than injection Nebulized Human Comparison of administration route: higher Ig titers following 93 LAV aerosol administration compared to i.m. Nebulized Human Comparison of administration route: equivalent or superior Ig 104 LAV response following aerosol administration compared to traditional i.m. SD MP, Human US Phase I , comparison of administration route: 94 LAV two inhaler types vs. s.c.

Model OVA NP (PPS), subunit, Mouse Comparison of formulation: NPs provided superior CD8 T cell 105 CpG adjuvant responses in lung as compared to antigen alone NP (ICVM), subunit, Mouse Comparison of formulation: pulmonary provided superior CD8 106 poly:I:C & MPLA T cell response at multiple mucosal sites over antigen alone; adjuvants demonstration for both prophylactic & therapeutic applications

Tuberculosis Nebulized Human US Phase I clinical trial, comparison of administration route in 91 LAV boost dosage: pulmonary vs. i.d.

NP (PPS), subunit, Mouse Comparison of route of administration: pulmonary enhanced 107 CpG adjuvant Th1 response and protection upon challenge compared to s.c. SD, MP/NP, subunit Guinea pig Granuloma reduction upon challenge; subunit provided 108 superior protection to live attenuated in boost SD MP/NP, Guinea pig Granuloma reduction upon challenge 77 LAV

35 Design of nanoparticle carriers for pulmonary vaccines has recently gained attention. Nanoparticles provide an additional level of control for aerosol vaccine formulations, including incorporation of adjuvant properties, APC targeting, mucosal penetration, and lymph node accumulation.6, 58, 59, 85, 89, 109-111 Additionally, nanoparticle formulations have been designed to enhance both humoral and cellular immune responses in the lung.105, 106, 112, 113 The induction of a CTL memory CD8+ T cell response could be critical for protection against certain lung pathogens, notably TB.105 Many nanoparticle preclinical studies have involved formulations with ovalbumin (OVA), a model antigen derived from chickens.105, 106, 114-118

1.4.5. Adjuvants Subunit vaccines require co-delivery of both the pathogen-specific antigen and an adjuvant for the body to produce both a humoral and cell-mediated immunity. Indeed, without the addition of an adjuvant, subunit vaccines typically fail to produce T cell responses.119 As with vaccine development, adjuvant development has historically been empirically derived. Aluminum salts, or alum, and oil in water emulsions called Freund’s complete and incomplete adjuvants (FCA and FIA, respectively) were developed in the early 1900s and remain widely in use today.119, 120 Understanding of the immunological mechanism behind alum was only recently realized.121 A gap in adjuvant development occurred and new adjuvant systems did not enter until the 1990s and early 2000s. Oil and water emulsion adjuvants in use today include MF59 (Fluad®, Novartis), AS03 (®, GlaxoSmithKline).119, 122 Chemical modifications of bacterial endotoxin, lipopolysaccharides (LPS), resulted in the production of monophosphoryl lipid A (MPLA), which is now an approved adjuvant in combination with alum, AS04 (®, GlaxoSmithKline).119, 122 Given the very limited options currently for adjuvants, increasing research is underway in this field. MPLA, a known TLR4 ligand, represents the first approved adjuvant with known immunological stimuli. Other TLR ligands are in development as adjuvants, including CpG oligodeoxynucleotides (TLR 9), poly:I:C (TLR 3) and imidazoquinolines

36 (TLR7 and TLR8).119, 123 Nanoparticles offer an additional approach to adjuvant delivery.6, 7, 88, 124, 125 Nanoparticles have been implicated with inherent adjuvant properties. They also offer opportunities to co-deliver antigens with known stimulatory PAMPs, such as TLRs, safely and closely mimicking pathogen surface topographies.6, 7, 85, 88, 124, 125 Importantly, nanoparticles have been shown to penetrate the mucosa and are have potential applications as mucosal adjuvants, which represents a considerable need for mucosal vaccine development.126, 127 As an example of current limitations, the most commonly used adjuvant, alum, is accepted as a poor mucosal adjuvant.128 In all cases, new adjuvants face demanding formulation and regulatory requirements, as they will be expected to provide a nontoxic stimuli for a controlled immunological response.119

1.4.6. Existing Needs for Mucosal Vaccines While the examples discussed in this section demonstrate the promise of vaccination directly to mucosal organs, much work remains to be done. Following the advent of more engineering approaches to vaccine development, the understanding of exact immunological mechanisms at mucosal sites needs to be increased. For example, while the oral polio vaccine is known to be effective, the underlying immunological response still remains a mystery.66 Lymphocyte subset identification and functionality specifically at mucosal sites also needs to be increased in order to further target and direct these populations.85 Indeed, discrete subsets of TLR receptors are known to exist on APCs, making identification key prior to delivery of a corresponding TLR adjuvant.119 Subunit vaccines represent the most effective and safe approaches for new vaccines, which require additional incorporation of an adjuvant for efficacy. Isolation of pathogen- specific antigens is required for increased efficacy against certain disease.122 Specific needs for mucosal vaccines also include the development of safe and effective adjuvants. These adjuvants must be potent enough to stimulate the immune system, without causing irreparable damage to the delicate immunological balance of the mucosal interface. This is particularly true of the lung, which performs critical function of oxygen exchange.119, 120, 124, 126, 129-131 Development of new protein antigens and adjuvants will help realize the “Holy

37 Grail” in mucosal vaccination, which will produce a long-lasting, preventative response following a single administration.66 Finally, mucosal vaccine development must advance to address challenging diseases, such as TB, influenza, and HIV, which originate from clever and/or rapidly mutating pathogens. Novel and innovative approaches towards vaccinating against these diseases are desperately needed.12, 76, 81 Pathogen mimicry, through co-delivery of antigens and adjuvants, as well as other ground-breaking biomolecules, may prove to be elegant solutions to these challenges.

1.5. Pulmonary Therapeutic Delivery Drug delivery directly to the site of action has tremendous pharmacological benefits, including increased bioavailability, limited off-target effects and potent local immune responses. As discussed in section 1.4.2, delivery directly to the mucosa can provide increased vaccine efficacy. In this section, delivery parameters influencing drug delivery to the lung will be discussed. Effective aerosol particle delivery requires knowledge of lung anatomy for accurate flow analysis and avoidance of immune clearance mechanisms.132

1.5.1. Lung Structure The purpose of the lung is to transport oxygen into the blood and exchange with carbon dioxide. This function occurs through the expansion and contraction of the organ, through a muscle called the diaphragm. The lung consists of a complex, branching network, divided into two main regions: the respiratory tract and the alveolar region. The alveolar region, also called the acinus or respiratory zone, is the portion of the lung where oxygen and carbon dioxide are actively exchanged. The respiratory tract, or conducting zone, is the pathway that connects the airway openings at the mouth and nose to the alveolar diffusion region. In the respiratory tract, air flows through the pharynx, the larynx and then trachea, at which point the airway begins to branch. Flow then enters the mainstem bronchi, lobar bronchi, segmental bronchi, and subsegmental bronchi, with the airways dividing as many as

38 15 to 20 times. The many divisions within the respiratory tract end in the terminal bronchioles. Flow then continues to the acinus region, passing through the respiratory bronchioles, the alveolar ducts and the alveoli. An adult human lung contains over 300 million alveoli, with a surface area of approximately 75m2. Oxygen reaching the acinus region diffuses to pulmonary capillaries surrounding the alveolar ducts, which then transports the newly oxygenated blood vessels to the heart and then throughout the body.14, 31, 133 A diagram of the lung is shown in Figure 1.17.

Figure 1.17. Diagram of lung anatomy. Reproduced with permission from reference.134 © T. Winslow.

The human lung is divided in half, consisting of the right and left lung. The right lung contains three lobes while the left has two. Human airways divide through asymmetric bifurcations.135 Pulmonary studies have been performed on a wide variety of animal models, including mice, rats, guinea pigs, rabbits, dogs and sheep. Selection of the appropriate model depends largely on application and availability. In general, smaller animals are used for studies of local acting drugs and mucociliary clearance, while larger animals are used for studies on systemically delivered drugs and pharmacokinetics.136, 137 Mice, for instance, also

39 have five lobes, but only a single left lung and monopodial bifurcations.138 Figure 1.18 diagrams the differences between commonly used animal models for pulmonary delivery.

Figure 1.18. Animal models commonly used for respiratory studies and their relevant lung geometries. Values obtained from references.136, 139, 140 Images reproduced with permissions from references.139, 141, 142

1.5.2. Physical Parameters for Aerosol Delivery For convention, aerosol particles are described for a series of equivalent spheres, where one physical parameter of a nonspherical particle is equated to that of a sphere of a given diameter. Non-spherical aerosols of a geometric diameter, DG, are often defined through an equivalent volume diameter, DEV, and an aerodynamic diameter, DAE. The DEV is the diameter of a sphere which would have the same volume as the non-spherical particle, while DAE is the diameter of a sphere with equivalent settling velocity and a standard density equal to water.14 A diagram of these diameters is shown in Figure 1.19.

40

Figure 1.19. Equivalent volume spheres. Arrows indicate terminal settling velocity (VTS).

Relationships between these spheres will be further discussed in Chapter 3. Inhaled foreign particles frequently enter the respiratory tract, but most will never reach the sterile alveolar region of the lung. As flow enters the mouth or nose, the complex geometry provides the initial physical barrier. Due to flow patterns, airway diameters and increased deposition at branch junctions, foreign objects greater than DAE of 10 μm will deposit in the mouth, pharynx and trachea, while particles 5-10 μm will typically deposit no further than the upper bronchial tubes. Particles smaller than 1 μm rarely deposit on any lung surface and are consequently exhaled. Only particles or bacteria 1-5 μm in diameter deposit in the lower bronchi and alveoli regions.14, 133, 143 Objects deposited in these lower regions can be cleared by coughing, mucociliary transport, or phagocytes.14, 57, 58, 60, 144 Airflow through the lung can be described by the Navier-Stokes equations, while dynamics of individual particles can be described by Newton’s second law.14, 145 However, due to the numerous branching airways, flow cannot be described using bulk properties, as the resulting flows deviate from simple axial. The Reynold’s number (Re) must be described locally to accurately encompass regions of laminar and turbulent flow, as well as the formation of eddies and other flow abnormalities. Complex lung geometries make solving for overall flow or particle trajectories mathematically strenuous.14 Many groups have used computer simulations to model airflow through the lung to numerically solving the Navier-

41 Stokes equation using custom computational fluid dynamics (CFD) software.145-147 Approximate airflow properties as a function of airway bifurcation is shown in Table 1.3.

Table 1.3. Re as a function of airway geometry. Adapted from reference.14 Abbreviations: G – generation number, D. cross-sectional airway diameter.

Aerosol particle deposition is typically described by one of three mechanisms: impaction, sedimentation and diffusion. Impaction is a velocity dependent inertial phenomenon which causes particles to leave the air stream and deposit on lung walls.148, 149 Sedimentation is a gravitational phenomenon and has been shown to be the main mechanism of deposition in breath holding.148, 150 The final deposition mechanism is Brownian diffusion, which is a time dependent mechanism affecting small particles in low flow regions.14 Knowing the dominant mechanisms in each lung regions will be crucial when modeling particle deposition.

42 1.5.3. Types of Inhaler Delivery Following the general understanding of aerosol delivery and deposition into the lung, a number of treatment forms have been developed for drug delivery. Inhaled treatments for conditions such as asthma have been widely employed since the 1960s. The first inhalation treatments used perfume technology; a compressed suspension of drug in a volatile fluid was sprayed directly into the patient’s mouth.150 Aerosol drug delivery has evolved to the use of inhalers and nebulizers, using similar aerosol generation. Nebulizers create an aerosol mist from flow of a high velocity air jet or ultrasonic vibrations directed towards a liquid drug, which forms liquid droplets. Nebulizers are bulky and require a large amount of operational power, limiting their use to hospital and home settings. The pressurized meter-dose inhaler (pMDI) is more convenient and user-friendly. These portable devices release a specific amount of liquid drug at a high pressure, resulting in a droplet spray inhaled by the patient. Use of pMDI requires coordinated breathing with the release of spray to deliver the drug to the lung, resulting in patient-to-patient variability. Dose-delivery efficiencies for asthma inhalers range from 3 to 15% for children and 10 to 30% for adults, indicating that less than a third of the contained drug actually escapes the inhaler to reach the airway opening; the most advanced pMDIs deliver only 60% of the inhaled material to central and intermediate bronchial tubes.145 While the pMDI is limited by the requirement of liquid soluble drugs, use of a propellant and coordination of breathing, advances in dry-powder inhalers (DPI) have recently become a viable alternative. In a DPI, solid particles, often comprised of a lactose base, are inhaled directly through a simple breath-actuated cylinder. Use of solid particles avoids irritant liquids and allows easy dose delivery, and advancement in DPIs is an ongoing and promising area of investigation for pulmonary delivery.143, 145, 150 Particle fabrication techniques for commercially available DPIs, as well as other preclinical research approaches, will be discussed in the following section.

43 1.6. Micro and Nanoparticle Fabrication Techniques There are a huge variety of particle fabrication techniques which have been developed for biomedical applications. Precision defined biomaterials especially have been used to engineer immune responses.5, 6 Here, a few general particle fabrication techniques will be discussed.

1.6.1. Bottom-up Fabrication Techniques Bottom-up fabrication techniques begin with the smallest single base unit, molecular or particulate, which are arranged to create an ordered structure. A wide range of materials can be fabricated through bottom-up approaches, from structured macromolecules to inorganic particulates. The ultimate physical geometries achievable are also diverse. Bottom- up fabrication techniques are often cost effective batch processes and significant advances in recent years have made these approaches appealing for nanoparticle fabrication. Many conventional relevant bottom-up fabrication approaches for biomedicine have been discussed in detail elsewhere, including liposomes,151 micelles,152 inorganic nanoparticles,153 gold nanoparticles,154 silica nanoparticles,155 carbon nanotubes,156 and particle stretching.157-159 Examples of nanoparticles for drug delivery applications by these approaches are shown in Figure 1.20.

44

Figure 1.20. Examples of particles fabricated through bottom-up approaches. (A) TEM image of PEGylated gold nanorods.160 (B) TEM image of 1D chains of gold nanoparticles with a diameter of 16 nm.161 (C) SEM of right bipyramidal silver nanoparticles.162 (D) TEM image of mesoporous silica nanoparticles with an aspect ratio of ca. 5.163 SEM images of (E) UFO and (F) needle shaped polystyrene particles fabricated through stretching.157 (G) TEM image of amphiphilic cylindrical micelles.164 (H) Fluorescence microscopy image of a single filomicelle.165 Reproduced with permissions from references.

Spray-drying is also considered a bottom-up approach. Inhalation delivery was revolutionized by this technique, following a Science publication that demonstrated the application of large porous particles (LPP). These highly porous particles successfully achieved a DAE between 1-3 µm for peripheral lung deposition while maintaining large DG to avoid macrophage uptake.144 This produced a transition in respiratory drug delivery to the development of large, dry powder particles with low densities. Following this discovery, conventional fabrication of these pharmaceutical aerosols for DPIs expanded from micronization (milling) to spray-drying for the production of LPPs.166 Spray-drying results in polydisperse aerosol populations, with large particle size distributions requiring additional separation techniques. Examples of these particles can be seen in Figure 1.21.

45

Figure 1.21. Examples of dry powder particles. (A) Spray dried uncoated Ig particles.167 (B) First example of LPP spray dried particles.144 (C) PulmoSphereTM (Novartis)167 Reproduced with permissions from references.

In these bottom-up fabrication methods, particle chemical composition and aerosol characteristics are inherently coupled. For example, the solubility and drying kinetics of precursor solutions can impact the particle size distribution of spray-dried particles.166 Additional fabrication challenges arise with forming dry, non-agglomerating powders comprised of pure active ingredients, especially biologicals like siRNA, proteins and monoclonal antibodies (mAbs). Indeed, there are currently no marketed dry powder inhaled mAbs or siRNA therapies.168 However, advances in spray drying have enabled the fabrication of LPPs containing highly potent chemotherapeutics and biologics.89, 167 Importantly, dry powder technologies are being actively translated into the clinic.169

1.6.2. Top-down Fabrication Techniques Top-down fabrication techniques are the reverse of bottom-up fabrication; processing steps are taken to selectively remove portions of the starting material to design the end product. Milling is considered a top-down fabrication approach and is commonly used for fabrication of dry powders for pulmonary drug delivery.168 This type of approach is also widely employed through silicon lithographic patterning in the microelectronics industry and has recently found application in particle fabrication techniques. Bottom-up fabrication often yields particles of polydisperse size and shape distributions. Inherent to top-down fabrication is an element of controlled design, enabling an additional level of direct control over the precise details of the product.

46 Photolithography is the most cited example of top-down fabrication and many of the methods described in this section follow from fabrication methods designed for the microelectronics industry. Photolithography is an optical approach that creates well-defined patterns onto a surface by transferring geometric patterns encoded on a mask to a light sensitive photoresist coating. Depending on the nature of the photoresist, exposed (positive resist) or unexposed (negative resist) regions are selectively removed to impose the desired pattern. In the case of the microelectronics industry, this process takes places on a silicon wafer. Critical to many top-down particle fabrication methods discussed here, the pattern feature size is limited by state of the art photolithography techniques and controlled by the system optics and energy sources. Increasing market size and the ever increasing need for faster, more powerful microelectronic devices has driven advances in photolithography and facilitated a steady decrease in obtainable feature sizes since the first transistor in 1947.170, 171 Indeed, at the time of this publication, current transistor sizes on the market are Intel’s 22 nm tri-gate technology.172 Further resolution can be achieved through E-beam, ion beam and dip-pen lithography. The microelectronics industry harnesses the patterning of silicon and deposition of metals, many of which are not ideal for medical applications. Numerous technologies have been inspired by and adapted from traditional photolithography approaches to fabricate particles of relevant materials. These top-down approaches allow for independent control of particle size, shape, and mechanical properties. Soft-lithography techniques evolved from the need to pattern materials outside of traditional lithographic-friendly materials. As the name implies, these techniques employ “soft” materials, often elastomers, and encompass fabrication techniques of molding or imprinting for the creation of micro- or nano-structures. While the materials and specific approaches vary between researchers, most soft lithographic techniques follow the same general process. A master template is made through traditional photolithography onto a hard surface. An inverse to the original pattern is then made using a low surface energy and low modulus elastomer resin, which is able to conform well to the master. The resin is solidified through a thermal or photochemical curing and then removed from the master, resulting in an

47 exact inverse replica of the original master pattern. This inverse replica can then be used for molding or imprinting new materials, or the formation of a complex device based on the pattern, depending on the application and type of mold material used. Soft-lithography offers unique opportunities to pattern sensitive materials for a huge variety of applications extending beyond nanomedicine and particle fabrication. Early examples in the literature of elastomer-based soft lithography have their origins from Whitesides and colleagues, who pioneered initial work concerning polydimethyl siloxane (PDMS) in the early to mid ‘90s.173-178 Stemming from this work this, many examples of PDMS-based techniques have been employed to pattern surfaces and importantly, fabricate particles. Patterns have been transferred through stamping, imprinting, or molding using the features embedded in the PDMS template. Generally, imprinting leaves the molded material on the original substrate following patterning, while molding results in filled template cavities which maintain integrity once the backing substrate has been removed. PDMS is widely used for particle fabrication, being simple, robust, and inexpensive, and has been successfully applied in patterning of a huge range of materials.179- 185 However, common issues with traditional PDMS-based soft lithography approaches include the occurrence of a residual interconnecting flash layer, which results in an embossed film rather than discrete particles, and chemical permeation of PDMS, leading to swelling and loss of geometric fidelity. This has led to research in developing materials other than PDMS for stamping and molding applications and development of additional processing steps to remove flash layers. Successful alternatives to PDMS imprinting have been developed. These include techniques of bilayer nanoimprint lithography (B-NIL),186,187 step- flash imprint lithography (S-FIL),188, 189 gelatin-based mold templates,190 and Particle Replication In Non-wetting Templates (PRINT), which will be discussed in detail in Chapter 2.

48

Figure 1.22. Particles fabricated via top-down methods. (A-C) Examples of particles fabricated through soft-lithography. (A) Fluorescence microscopy of 20x20x240 µm amphiphilic triblock PRINT particles.191 (B) SEM of B-NIL nanopillars from a bilayer substrate of SU-8 on PMMA.186 (C) 400 nm pentagonal PEG diacrylate particles by S-FIL.188 (D) SEM of tipless gold pyramids fabricated using PEEL.192 (E-H) Examples of particles fabricated using microfluidics. (E) Optical image of superparamagnetic STL particles.193 (F) Differential interference contrast image of PEG diacrylate triangles.194 (G) SEM of curved PEG diacrylate particles.194 (H) Optical microscopy image of ellipsoids.195 Reproduced with permissions from references.

Overall, these top-down approaches enable production of particles with superior control over size and shape. Examples of particles fabricated with top-down approaches are shown in Figure 1.22.

1.7. Thesis Overview Given this expansive landscape of mucosal immunity, pulmonary drug delivery, and nano-micro particle fabrication, we now proceed to the main body of work. We hypothesized that the precision particle control afforded by the PRINT technology could advance understanding of the role that particle features, such as size, shape, and surface charge, play on all aspects of pulmonary vaccine formulations. The overall goal

49 of this work was to fabricate and characterize PRINT particles and optimize them as pulmonary vaccine carriers. In Chapter 2, the PRINT particle fabrication platform is investigated and optimized for pulmonary delivery. Optimization of particle fabrication, aerosol dispersion and dry powder formulations are described. Chapter 3 further probes the benefit of using the PRINT platform for aerosol delivery, detailing the first example of monodisperse, non-spherical aerosols and establishing the role that particle shape can play on aerodynamic properties of an aerosol. In Chapter 4, the particle control afforded by the PRINT platform is then used to probe the biological function of key lung APCs in mice, for their ultimate application as a vaccine. Finally, in Chapter 5, PRINT nanoparticles are demonstrated as pulmonary vaccine carriers. The thesis concludes with a summary and recommendations for future work in Chapter 6.

1.8. References 1. Sharp PA, Langer R. Promoting convergence in biomedical science. Science. 2011, 333, 527.

2. Convergence: Facilitating transdisciplinary integration of life sciences, physical sciences, engineering, and beyond. No. National Academy of Sciences, National Academy of Engineering, Institute of Medicine, National Research Council, 2014

3. DeSimone JM, Farrell CL. Driving convergence with human diversity. Science Translational Medicine. 2014, 6, 238ed211.

4. Petrosko SH, Fromen CA, Auyeung E, DeSimone JM, Mirkin CA. Nanotechnology: An enduring bridge between engineering and medicine. NAE The Bridge. 2013, Fall 2013, 7- 15.

5. Hubbell JA, Thomas SN, Swartz MA. Materials engineering for immunomodulation. Nature. 2009, 462, 449-460.

6. Moon JJ, Huang B, Irvine DJ. Engineering nano- and microparticles to tune immunity. Adv Mater. 2012, 24, 3724-3746.

50 7. Smith DM, Simon JK, Baker JR, Jr. Applications of nanotechnology for immunology. Nat Rev Immunol. 2013, 13, 592-605.

8. Flajnik MF, Kasahara M. Origin and evolution of the adaptive immune system: Genetic events and selective pressures. Nat Rev Genet. 2010, 11, 47-59.

9. Mantis NJ. The intestinal epithelium: The interface between host and pathogen. In: Vajdy M. Immunity against mucosal pathogens: Springer, 2008.

10. Sompayrac L. How the immune system works. UK: John Wiley & Sons; 2012.

11. Kupper TS, Fuhlbrigge RC. Immune surveillance in the skin: Mechanisms and clinical consequences. Nat Rev Immunol. 2004, 4, 211-222.

12. Neutra MR, Kozlowski PA. Mucosal vaccines: The promise and the challenge. Nat Rev Immunol. 2006, 6, 148-158.

13. Patton JS, Byron PR. Inhaling medicines: Delivering drugs to the body through the lungs. Nat Rev Drug Discov. 2007, 6, 67-74.

14. Hinds WC. Aerosol technology: Properties, behavior, and measurement of airborne particles. New York: John Wiley & Sons, Inc.; 1999.

15. Lippmann M. Size-selective health hazard sampling. In:Air sampling instruments for evaluation of atmospheric contaminants, Cincinnati: ACGIH, 1995.

16. Mucosal immunology and virology. London: Springer; 2006.

17. Mucosal vaccines: Modern concepts, strategies, and challenges. New York, NY: Springer; 2012.

18. Holmgren J, Czerkinsky C. Mucosal immunity and vaccines. Nat Med. 2005, 11, S45-53.

19. Cone RA. Barrier properties of mucus. Adv Drug Deliv Rev. 2009, 61, 75-85.

20. Underhill DM, Goodridge HS. Information processing during phagocytosis. Nat Rev Immunol. 2012, 12, 492-502.

51 21. Domen J, Wagers A, Weissman IL. Bome marrow (hematopoietic) stem cells. In: Regenerative Medicine: Department of Health and Human Services, 2006.

22. Akira S, Uematsu S, Takeuchi O. Pathogen recognition and innate immunity. Cell. 2006, 124, 783-801.

23. Banchereau J, Steinman RM. Dendritic cells and the control of immunity. Nature. 1998, 392, 245-252.

24. Hu W, Pasare C. Location, location, location: Tissue-specific regulation of immune responses. J Leukoc Biol. 2013, 94, 409-421.

25. Iwasaki A. Mucosal dendritic cells. Annu Rev Immunol. 2007, 25, 381-418.

26. Iwasaki A, Medzhitov R. Toll-like receptor control of the adaptive immune responses. Nat Immunol. 2004, 5, 987-995.

27. Shortman K, Liu YJ. Mouse and human dendritic cell subtypes. Nat Rev Immunol. 2002, 2, 151-161.

28. Steinman RM. Dendritic cells: Versatile controllers of the immune system. Nat Med. 2007, 13, 1155-1159.

29. Steinman RM, Banchereau J. Taking dendritic cells into medicine. Nature. 2007, 449, 419-426.

30. Gallorini S, O'Hagan DT, Baudner BC. Concepts in mucosal immunity and mucosal vaccines. In: Das Neves J, Sarmento B. Mucosal delivery of biopharmaceuticals: Biology, challenges and strategies, New York: Springer, 2014.

31. Inaba K, Steinman RM, Van Voorhis WC, Muramatsu S. Dendritic cells are critical accessory cells for thymus-dependent antibody responses in mouse and in man. Proc Natl Acad Sci U S A. 1983, 80, 6041-6045.

32. Immunity against mucosal pathogens. New York, NY: Springer; 2008.

33. Dendritic cells: Controllers of adaptive immunity. Available at: nature.com/nri/posters/dendriticcells.

52 34. Different types of immunoglobulins. Igg, iga, igd, ige, and igm. 2014.

35. Jacobson EB, Caporale LH, Thorbecke GJ. Effect of thymus cell injections on germinal center formation in lyphoid tissues of nude (thymusless) mice. Cell Immunol. 1974, 13, 416-430.

36. Heesters BA, Myers RC, Carroll MC. Follicular dendritic cells: Dynamic antigen libraries. Nat Rev Immunol. 2014, 14, 495-504.

37. Kasturi SP, Skountzou I, Albrecht RA, Koutsonanos D, Hua T, Nakaya HI, Ravindran R, Stewart S, Alam M, Kwissa M, Villinger F, Murthy N, Steel J, Jacob J, Hogan RJ, Garcia-Sastre A, Compans R, Pulendran B. Programming the magnitude and persistence of antibody responses with innate immunity. Nature. 2011, 470, 543-547.

38. Seder RA, Darrah PA, Roederer M. T-cell quality in memory and protection implications for vaccine design. Nat Rev Immunol. 2008, 8, 247-258.

39. Cesta MF. Normal structure, function, and histology of mucosa-associated lymphoid tissue. Toxicol Pathol. 2006, 34, 599-608.

40. Mestecky J, Bienenstock J, McGhee JR, Lamm ME, Strober W, Cebra JJ, Mayer L, Ogra PL. Historical aspects of mucosal immunology. In: Mestecky J, Lamm ME, Ogra PL, Stober W, Bienenstock J, McGhee JR, Mayer L. Mucosal immunology: Academic Press, 2005.

41. Henderson DA, Inglesby TV, Bartlett JG, Ascher MS, Eitzen E, Jahrling PB, Hauer J, Layton M, McDade J, Osterholm MT, O'Toole T, Parker G, Perl T, Russell PK, Tonat K. Smallpox as a biological weapon. JAMA. 1999, 281, 2127-2137.

42. Furuhashi K, Suda T, Hasegawa H, Suzuki Y, Hashimoto D, Enomoto N, Fujisawa T, Nakamura Y, Inui N, Shibata K, Nakamura H, Chida K. Mouse lung cd103+ and cd11bhigh dendritic cells preferentially induce distinct cd4+ t-cell responses. Am J Respir Cell Mol Biol. 2012, 46, 165-172.

43. Suzuki Y, Suda T, Furuhashi K, Shibata K, Hashimoto D, Enomto N, Fujisawa T, Nakamura Y, Inui N, Nakamura H, Chida K. Mouse cd11bhigh lung dendritic cells have more potent capability to induce iga than cd103+ lung dendritic cells in vitro. Am J Respir Cell Mol Biol. 2012, 46, 773-780.

53 44. Guilliams M, Lambrecht BN, Hammad H. Division of labor between lung dendritic cells and macrophages in the defense against pulmonary infections. Mucosal Immunol. 2013, 6, 464-473.

45. Smart JD. The basics and underlying mechanisms of mucoadhesion. Adv Drug Deliv Rev. 2005, 57, 1556-1568.

46. McGuckin MA, Linden SK, Sutton P, Florin TH. Mucin dynamics and enteric pathogens. Nat Rev Microbiol. 2011, 9, 265-278.

47. Rose MC, Voynow JA. Respiratory tract mucin genes and mucin glycoproteins in health and disease. Physiol Rev. 2006, 86, 245-278.

48. Lai SK, Wang YY, Wirtz D, Hanes J. Micro- and macrorheology of mucus. Adv Drug Deliv Rev. 2009, 61, 86-100.

49. Strous GJ, Dekker J. Mucin-type glycoproteins. Crit Rev Biochem Mol Biol. 1992, 27, 57-92.

50. Mucins: Methods and protocols; 2012.

51. Lamont JT. Mucus: The front line of intestinal mucosal defense. Ann NY Acad Sci. 1992, 664, 190-201.

52. Wright JR. Immunoregulatory functions of surfactant proteins. Nat Rev Immunol. 2005, 5, 58-68.

53. Garcia-Verdugo I, Descamps D, Chignard M, Touqui L, Sallenave JM. Lung protease/anti-protease network and modulation of mucus production and surfactant activity. Biochimie. 2010, 92, 1608-1617.

54. Lehrer RI. Primate defensins. Nat Rev Microbiol. 2004, 2, 727-738.

55. Sallenave JM, Shapiro S. Proteases and antiproteases in development, homeostasis and disease: The old, the new, and the unknown. Int J Biochem Cell Biol. 2008, 40, 1066- 1067.

56. Wiggins R, Hicks SJ, Soothill PW, Millar MR, Corfield AP. Mucinases and sialidases: Their role in the pathogenesis of sexually transmitted infections in the female genital tract. Sex Transm Inf. 2001, 77, 402-408.

54 57. Kaliner M, Marom Z, Patow C, Shelhamer J. Human respiratory mucus. J Allergy Clin Immunol. 1984, 73, 318-323.

58. Lai SK, O'Hanlon DE, Harrold S, Man ST, Wang YY, Cone R, Hanes J. Rapid transport of large polymeric nanoparticles in fresh undiluted human mucus. Proc Natl Acad Sci U S A. 2007, 104, 1482-1487.

59. Wang YY, Lai SK, Suk JS, Pace A, Cone R, Hanes J. Addressing the peg mucoadhesivity paradox to engineer nanoparticles that "slip" through the human mucus barrier. Angew Chem Int Ed Engl. 2008, 47, 9726-9729.

60. Evans CM, Koo JS. Airway mucus: The good, the bad, the sticky. Pharmacol Ther. 2009, 121, 332-348.

61. Button B, Cai LH, Ehre C, Kesimer M, Hill DB, Sheehan JK, Boucher RC, Rubinstein M. A periciliary brush promotes the lung health by separating the mucus layer from airway epithelia. Science. 2012, 337, 937-941.

62. Mucosal immune defense: Immunoglobin a. New York, NY: Springer; 2007.

63. Metzger DW. Iga and respiratory immunity. In: Kaetzel CS. Mucosal immune defense: Immunoglobin a, New York, NY: Springer, 2007;269-290.

64. Underdown BJ, Mestecky J. Mucosal immunoglobulins. In: Ogra PL, Strober W, Mestecky J, McGhee JR, Lamm ME, Bienenstock J. Handbook of mucosal immunology, New York: Academic Press, Inc, 1994.

65. Woof JM. The structure of iga. In: Kaetzel CS. Mucosal immune defense: Immunoglobin a, New York, NY: Springer, 2007;1-24.

66. Ravindran R, Pulendran B. Mucosal vaccines. In: Rappuoli R, Bagnoli F. Vaccine design : Innovative approaches and novel strategies, Norfolk, UK: Caister Academic, 2011;171- 209.

67. Balhara J, Gounni AS. The alveolar macrophages in asthma: A double-edged sword. Mucosal Immunol. 2012, 5, 605-609.

68. Woodrow KA, Bennett KM, Lo DD. Mucosal vaccine design and delivery. Annu Rev Biomed Eng. 2012, 14, 17-46.

55 69. Banchereau J, Briere F, Caux C, Davoust J, Laebecque S, Liu YJ, Pulendran B, Palucka K. Immunobiology of dendritic cells. Annu Rev Immunol. 2000, 18, 767-811.

70. Sheridan BS, Lefrançois L. Regional and mucosal memory t cells. Nature Immunology. 2011, 131, 485-491.

71. Sandoval F, Terme M, Nizard M, Badoual C, Bureau MF, Freyburger L, Clement O, Marcheteau E, Gey A, Fraisse G, Bouguin C, Merillon N, Dransart E, Tran T, Quintin- Colonna F, Autret G, Thiebaud M, Suleman M, Riffault S, Wu TC, Launay O, Danel C, Taieb J, Richardson J, Zitvogel L, Fridman WH, Johannes L, Tartour E. Mucosal imprinting of vaccine-induced cd8(+) t cells is crucial to inhibit the growth of mucosal tumors. Science Translational Medicine. 2013, 5, 172ra120.

72. Borges O, Lebre F, Bento D, Borchard G, Junginger HE. Mucosal vaccines: Recent progress in understanding the natural barriers. Pharm Res. 2010, 27, 211-223.

73. Backhed F, Ley RE, Sonnenburg JL, Peterson DA, Gordon JI. Host-bacterial mutualism in the human intestine. Science. 2005, 307, 1915-1920.

74. Girard JP, Moussion C, Forster R. Hevs, lymphatics and homeostatic immune cell trafficking in lymph nodes. Nat Rev Immunol. 2012, 12, 762-773.

75. von Andrian UH, Mempel TR. Homing and cellular traffic in lymph nodes. Nat Rev Immunol. 2003, 3, 867-878.

76. Rook GAW, Dheda K, Zumla A. Immune responses to tuberculosis in developing countries: Implications for new vaccines. Nat Rev Immunol. 2005, 5, 661-667.

77. Garcia-Contreras L, Wong YL, Muttil P, Padilla D, Sadoff J, Derousse J, Germishuizen WA, Goonesekera S, Elbert K, Bloom BR, Miller R, Fourie PB, Hickey A, Edwards D. Immunization by a bacterial aerosol. Proc Natl Acad Sci U S A. 2008, 105, 4656-4660.

78. Jeyanathan M, Heriazon A, Xing Z. Airway luminal t cells: A newcomer on the stage of tb vaccination strategies. Trends Immunol. 2010, 31, 247-252.

79. Yu F, Wang J, Dou J, Yang H, He X, Xu W, Zhang Y, Hu K, Gu N. Nanoparticle-based adjuvant for enhanced protective efficacy of DNA vaccine ag85a-esat-6-il-21 against mycobacterium tuberculosis infection. Nanomedicine. 2012, 8, 1337-1344.

56 80. Tubo NJ, Jenkins MK. Cd4+ t cells: Guardians of the phagosome. Clin Microbiol Rev. 2014, 27, 200-213.

81. Delogu G, Manganelli R, Brennan MJ. Critical research concepts in tuberculosis vaccine development. Clin Microbiol Infect. 2014, 20, 59-65.

82. Ogra PL, Faden H, Welliver RC. Vaccination strategies for mucosal immune responses. Clin Microbiol Rev. 2001, 14, 430-445.

83. De Gregorio E, Rappuoli R. From empiricism to rational design: A personal perspective of the evolution of vaccine development. Nat Rev Immunol. 2014, 14, 505-514.

84. Different types of vaccines. Available at: historyofvaccines.org/content/articles/different- types-vaccines. Accessed: July 7, 2014.

85. Fahmy TM, Demento SL, Caplan MJ, Mellman I, Saltzmann WM. Design opportunities for actively targeted nanoparticle vaccines. Nanomedicine. 2008, 3, 343-355.

86. Platt A, Wetzler L. Innate immunity and vaccines. Curr Top Med Chem. 2013, 13, 2597- 2608.

87. Tonnis WF, Kersten GF, Frijlink HW, Hinrichs WLJ, de Boer AH, Amorij J. Pulmonary vaccine delivery: A realistic approach? J Aerosol Med Pulm Drug Deliv. 2012, 25, 249- 260.

88. Pulliam B, Sung JC, Edwards DA. Design of nanoparticle-based dry powder pulmonary vaccines. Expert Opin Drug Del. 2007, 4, 651 - 663.

89. Sou T, Meeusen EN, de Veer M, Morton DA, Kaminskas LM, McIntosh MP. New developments in dry powder pulmonary vaccine delivery. Trends Biotechnol. 2011, 29, 191-198.

90. Blank F, Stumbles P, von Garnier C. Opportunities and challenges of the pulmonary route for vaccination. Expert Opin Drug Del. 2011, 8, 547-563.

91. Safety of tuberculosis vaccine, mva85a, administered by the aerosol route and the intradermal route. Available at: clinicaltrials.gov/ct2/show/NCT01497769?term=aerosol+vaccine&rank=1. Accessed: July 7, 2014.

57 92. Waldman RH, Mann JJ, Small PA. Immunization against influenza: Prevention of illness in man by aerosolized . JAMA. 1969, 207, 520-524.

93. Castro JF, Bennett JV, Rincon HG, Munoz MT, Sanchez LA, Santos JI. Evaluation of immunogenicity and side effects of triple viral vaccine (mmr) in adults, given by two routes: Subcutaneous and respiratory (aerosol). Vaccine. 2005, 23, 1079-1084.

94. A clinical trial to assess the safety of a (dry powder) administered by two different devices (pmv-001). Available at: clinicaltrials.gov/ct2/show/NCT01557699?term=measles+vaccine&rank=4. Accessed: July 7, 2014.

95. Amidi M, Pellikaan HC, Hirschberg H, de Boer AH, Crommelin DJ, Hennink WE, Kersten G, Jiskoot W. Diphtheria toxoid-containing microparticulate powder formulations for pulmonary vaccination: Preparation, characterization and evaluation in guinea pigs. Vaccine. 2007, 25, 6818-6829.

96. Muttil P, Prego C, Garcia-Contreras L, Pulliam B, Fallon JK, Wang C, Hickey AJ, Edwards D. Immunization of guinea pigs with novel hepatitis b antigen as nanoparticle aggregate powders administered by the pulmonary route. AAPS J. 2010, 12, 330-337.

97. Thomas C, Gupta V, Ahsan F. Particle size influences the immune response produced by formulated in inhalable particles. Pharm Res. 2010, 27, 905-919.

98. Thomas C, Rawat A, Hope-Weeks L, Ahsan F. Aerosolized pla and plga nanoparticles enhance humoral, mucosal and cytokine responses to hepatitis b vaccine. Mol Pharm. 2011, 8, 405-415.

99. Nardelli-Haefliger D, Lurati F, Wirthner D, Spertini F, Schiller JT, Lowy DR, Ponci F, De Grandi P. Immune responses induced by lower airway mucosal immunisation with a human papillomavirus type 16 virus-like particle vaccine. Vaccine. 2005, 23, 3634-3641.

100. Saluja V, Amorij JP, Kapteyn JC, de Boer AH, Frijlink HW, Hinrichs WL. A comparison between spray drying and spray freeze drying to produce an influenza subunit vaccine powder for inhalation. J Control Release. 2010, 144, 127-133.

101. Amorij JP, Saluja V, Petersen AH, Hinrichs WL, Huckriede A, Frijlink HW. Pulmonary delivery of an inulin-stabilized influenza subunit vaccine prepared by spray- freeze drying induces systemic, mucosal humoral as well as cell-mediated immune responses in balb/c mice. Vaccine. 2007, 25, 8707-8717.

58 102. Kisich KO, Higgins MP, Park I, Cape SP, Lindsay L, Bennett DJ, Winston S, Searles J, Sievers RE. Dry powder measles vaccine: Particle deposition, virus replication, and immune response in cotton rats following inhalation. Vaccine. 2011, 29, 905-912.

103. de Swart RL, LiCalsi C, Quirk AV, van Amerongen G, Nodelman V, Alcock R, Yuksel S, Ward GH, Hardy JG, Vos H, Witham CL, Grainger CI, Kuiken T, Greenspan BJ, Gard TG, Osterhaus AD. Measles vaccination of macaques by dry powder inhalation. Vaccine. 2007, 25, 1183-1190.

104. Dilray A, Cutts FT, de Castro JF, Wheeler JG, Brown D, Roth C, Coovadia HM, Bennett JV. Response to different measles vaccine strains given by aerosol and subcutaneous routes to schoolshilder: A randomised trial. Lancet. 2000, 255, 798-803.

105. Nembrini C, Stano A, Dane KY, Ballester M, van der Vlies AJ, Marsland BJ, Swartz MA, Hubbell JA. Nanoparticle conjugation of antigen enhances cytotoxic t-cell responses in pulmonary vaccination. Proc Natl Acad Sci U S A. 2011, 108, E989-997.

106. Li AV, Moon JJ, Abraham W, Suh H, Elkhader J, Seidman MA, Yen M, Im E, Foley MH, Barouch DH, Irvine DJ. Generation of effector memory t cell-based mucosal and systemic immunity with pulmonary nanoparticle vaccination. Science Translational Medicine. 2013, 5, 204-130.

107. Ballester M, Nembrini C, Dhar N, de Titta A, de Piano C, Pasquier M, Simeoni E, van der Vlies AJ, McKinney JD, Hubbell JA, Swartz MA. Nanoparticle conjugation and pulmonary delivery enhance the protective efficacy of ag85b and cpg against tuberculosis. Vaccine. 2011, 29, 6959-6966.

108. Lu D, Garcia-Contreras L, Muttil P, Padilla D, Xu D, Liu J, Braunstein M, McMurray DN, Hickey AJ. Pulmonary immunization using antigen 85-b polymeric microparticles to boost tuberculosis immunity. AAPS J. 2010, 12, 338-347.

109. Kunda NK, Somavarapu S, Gordon SB, Hutcheon GA, Saleem IY. Nanocarriers targeting dendritic cells for pulmonary vaccine delivery. Pharm Res. 2013, 30, 325-341.

110. Hardy CL, Lemasurier JS, Mohamud R, Yao J, Xiang SD, Rolland JM, O'Hehir RE, Plebanski M. Differential uptake of nanoparticles and microparticles by pulmonary apc subsets induces discrete immunological imprints. J Immunol. 2013, 191, 5278-5290.

111. Choi HS, Ashitate Y, Lee JH, Kim SH, Matsui A, Insin N, Bawendi MG, Semmler- Behnke M, Frangioni JV, Tsuda A. Rapid translocation of nanoparticles from the lung airspaces to the body. Nat Biotechnol. 2010, 28, 1300-1303.

59 112. Stano A, Nembrini C, Swartz MA, Hubbell JA, Simeoni E. Nanoparticle size influences the magnitude and quality of mucosal immune responses after intranasal immunization. Vaccine. 2012, 30, 7541-7546.

113. Moon JJ, Suh H, Li AV, Ockenhouse CF, Yadava A, Irvine DJ. Enhancing humoral responses to a malaria antigen with nanoparticle vaccines that expand tfh cells and promose germinal center induction. Proc Natl Acad Sci U S A. 2012, 109, 1080-1085.

114. Korsholm KS, Agger EM, Foged C, Christensen D, Dietrich J, Andersen CS, Geisler C, Andersen P. The adjuvant mechanism of cationic dimethyldioctadecylammonium liposomes. Immunology. 2007, 121, 216-226.

115. Kwon YJ, Standley SM, Goh SL, Frechet JM. Enhanced antigen presentation and immunostimulation of dendritic cells using acid-degradable cationic nanoparticles. J Control Release. 2005, 105, 199-212.

116. Pesce I, Monaci E, Muzzi A, Tritto E, Tavarini S, Nuti S, De Gregorio E, Wack A. Intranasal administration of cpg induces a rapid and transient cytokine response followed by dendritic and natural killer cell activation and recruitment in the mouse lung. J Innate Immun. 2010, 2, 144-159.

117. Slutter B, Bal SM, Ding Z, Jiskoot W, Bouwstra JA. Adjuvant effect of cationic liposomes and cpg depends on administration route. J Control Release. 2011, 154, 123- 130.

118. Slutter B, Jiskoot W. Dual role of cpg as immune modulator and physical crosslinker in ovalbumin loaded n-trimethyl chitosan (tmc) nanoparticles for nasal vaccination. J Control Release. 2010, 148, 117-121.

119. Reed SG, Orr MT, Fox CB. Key roles of adjuvants in modern vaccines. Nat Med. 2013, 19, 1597-1608.

120. Petrovsky N, Aguilar JC. Vaccine adjuvants: Current state and future trends. Immunol Cell Biol. 2004, 82, 488-496.

121. Lambrecht BN, Kool M, Willart MA, Hammad H. Mechanism of action of clinically approved adjuvants. Curr Opin Immunol. 2009, 21, 23-29.

122. Rappuoli R, Mandl CW, Black S, De Gregorio E. Vaccines for the twenty-first century society. Nat Rev Immunol. 2011, 11, 865-872.

60 123. De Gregorio E, D'Oro U, Wack A. Immunology of tlr-independent vaccine adjuvants. Curr Opin Immunol. 2009, 21, 339-345.

124. De Temmerman ML, Rejman J, Demeester J, Irvine DJ, Gander B, De Smedt SC. Particulate vaccines: On the quest for optimal delivery and immune response. Drug Discov Today. 2011, 16, 569-582.

125. Zhao L, Seth A, Wibowo N, Zhao CX, Mitter N, Yu C, Middelberg AP. Nanoparticle vaccines. Vaccine. 2014, 32, 327-337.

126. Vujanic A, Wee JL, Snibson KJ, Edwards S, Pearse M, Quinn C, Moloney M, Taylor S, Scheerlinck JP, Sutton P. Combined mucosal and systemic immunity following pulmonary delivery of iscomatrix adjuvanted recombinant antigens. Vaccine. 2010, 28, 2593-2597.

127. Lai SK, Wang YY, Hanes J. Mucus-penetrating nanoparticles for drug and gene delivery to mucosal tissues. Adv Drug Deliv Rev. 2009, 61, 158-171.

128. Lawson LB, Norton EB, Clements JD. Defending the mucosa: Adjuvant and carrier formulations for mucosal immunity. Curr Opin Immunol. 2011, 23, 414-420.

129. Fujihashi K, Koga T, van Ginkel FW, Hagiwara Y, McGhee JR. A dilemma for mucosal vaccination: Efficacy versus toxicity using enterotoxin-based adjuvants. Vaccine. 2002, 20, 2431 - 2438.

130. Jones KS. Biomaterials as vaccine adjuvants. Biotechnol Prog. 2008, 24, 807-814.

131. Kaufmann SH, McElrath MJ, Lewis DJ, Del Giudice G. Challenges and responses in human vaccine development. Curr Opin Immunol. 2014, 28, 18-26.

132. Controlled pulmonary drug delivery. New York: Springer; 2011.

133. Weinberger SE, Cockrill BA, Mandel J. Principles of pulmonary medicine. Philadelphia, PA: Elsevier; 2008.

134. General information about small cell lung cancer. Available at: cancer.gov/cancertopics/pdq/treatment/small-cell-lung/Patient/page1. Accessed: July 8, 2014.

61 135. Lee D, Park SS, Ban-Weiss GA, Fanucchi MV, Plopper CG, Wexler AS. Bifurcation model for characterization of pulmonary architecture. Anat Rec (Hoboken). 2008, 291, 379-389.

136. Fernandes CA, Vanbever R. Preclinical models for pulmonary drug delivery. Expert Opin Drug Del. 2009, 6, 1231-1245.

137. Cryan SA, Sivadas N, Garcia-Contreras L. In vivo animal models for drug delivery across the lung mucosal barrier. Adv Drug Deliv Rev. 2007, 59, 1133-1151.

138. Irvin CG, Bates JHT. Measuring the lung function in the mouse: The challenge of size. Respir Res. 2003, 4.

139. Schreider JP, Hutchens JO. Morphology of the guinea pig respiratory tract. Anat Rec. 1980, 196, 313-321.

140. Chaturvedi A, Lee Z. Three-dimensional segmentation and skeletonization to build an airway tree data structure for small animals. Phys Med Biol. 2005, 50, 1405-1419.

141. Oldham MJ, Phalen RF. Dosimetry implications of upper tracheobronchial airway anatomy in two mouse varieties. Anat Rec. 2002, 268, 59-65.

142. Fractals in us. Available at: classes.yale.edu/fractals/WorldOfFractals/Us/Us.html. Accessed: July 8, 2014.

143. Edwards DA. Delivery of biological agents by aerosols. AIChE Journal. 2002, 48, 2- 6.

144. Edwards DA, Hanes J, Caponetti G, Hrkach J, Ben-Jebria A, Eskew ML, Mintzes J, Deaver D, Lotan N, Langer R. Large porous particles for pulmonary drug delivery. Science. 1997, 276, 1868-1971.

145. Kleinstreuer C, Zhang Z, Donohue JF. Targeted drug-aerosol delivery in the human respiratory system. Annu Rev Biomed Eng. 2008, 10, 195-220.

146. Kleinstreuer C, Zhang Z, Li Z, Roberts WL, Rojas C. A new methodology for targeting drug-aerosols in the human respiratory system. International Journal of Heat and Mass Transfer. 2008, 51, 5578-5589.

62 147. Kleinstreuer C, Seelecke S. Inhaler system for targeted maximum drug-aerosol delivery. In: USPTO, ed.: North Carolina State University, 2011.

148. Darquenne C, Paiva M, Prisk GK. Effect of gravity on aerosol dispersion and deposition in the human lung after periods of breath holding. J Appl Physiol. 2000, 89, 1787-1792.

149. Darquenne C, Prisk GK. Aerosol deposition in the human respiratory tract breathing air and 8020 heliox. J Aerosol Med. 2004, 17, 278-285.

150. Crowder TM, Rosati JA, Schroeter JD, Hickey AJ, Martonen TB. Fundamental effects of particle morphology on lung delivery. Pharm Res. 2002, 19, 239-245.

151. Malam Y, Loizidou M, Seifalian AM. Liposomes and nanoparticles: Nanosized vehicles for drug delivery in cancer. Trends Pharmacol Sci. 2009, 30, 592-599.

152. Torchilin VP. Micellar nanocarriers: Pharmaceutical perspectives. Pharm Res. 2007, 24, 1-16.

153. Sun C, Lee JS, Zhang M. Magnetic nanoparticles in mr imaging and drug delivery. Adv Drug Deliv Rev. 2008, 60, 1252-1265.

154. Huang X, Neretina S, El-Sayed MA. Gold nanorods: From synthesis and properties to biological and biomedical applications. Advanced Materials. 2009, 21, 4880-4910.

155. Tang F, Li L, Chen D. Mesoporous silica nanoparticles: Synthesis, biocompatibility and drug delivery. Adv Mater. 2012, 24, 1504-1534.

156. Lacerda L, Bianco A, Prato M, Kostarelos K. Carbon nanotubes as nanomedicines: From toxicology to pharmacology. Adv Drug Deliv Rev. 2006, 58, 1460-1470.

157. Champion JA, Katare YK, Mitragotri S. Particle shape: A new design parameter for micro- and nanoscale drug delivery carriers. J Control Release. 2007, 121, 3-9.

158. Champion JA, Mitragotri S. Shape induced inhibition of phagocytosis of polymer particles. Pharm Res. 2009, 26, 244-249.

159. Yoo JW, Doshi N, Mitragotri S. Adaptive micro and nanoparticles: Temporal control over carrier properties to facilitate drug delivery. Adv Drug Deliv Rev. 2011, 63, 1247- 1256.

63 160. Gormley AJ, Greish K, Ray A, Robinson R, Gustafson JA, Ghandehari H. Gold nanorod mediated plasmonic photothermal therapy: A tool to enhance macromolecular delivery. Int J Pharm. 2011, 415, 315-318.

161. Sardar R, Shumaker-Parry JS. Asymmetrically functionalized gold nanoparticles organized in one-dimensional chains. Nano Lett. 2008, 8, 731-736.

162. Cobley CM, Skrabalak SE, Campbell DJ, Xia Y. Shape-controlled synthesis of silver nanoparticles for plasmonic and sensing applications. Plasmonics. 2009, 4, 171-179.

163. Huang X, Li L, Liu T, Hao N, Liu H, Chen D, Tang F. The shape effect of mesoporous silica nanoparticles on biodistribution, clearance, and biocompatibility in vivo. ACS Nano. 2011, 5, 5390-5399.

164. Zhang K, Fang H, Chen Z, Taylor J-SA, Wooley KL. Shape effects of nanoparticles conjugated with cell-penetrating peptides (hiv tat ptd) on cho cell uptake. Bioconjugate Chem. 2008, 19, 1880-1887.

165. Geng Y, Dalhaimer P, Cai S, Tsai R, Tewari M, Minko T, Discher DE. Shape effects of filaments versus spherical particles in flow and drug delivery. Nat Nanotechnol. 2007, 2, 249-255.

166. Chow AH, Tong HH, Chattopadhyay P, Shekunov BY. Particle engineering for pulmonary drug delivery. Pharm Res. 2007, 24, 411-437.

167. Vehring R. Pharmaceutical particle engineering via spray drying. Pharm Res. 2008, 25, 999-1022.

168. Garcia A, Mack P, Williams S, Fromen C, Shen T, Tully J, Pillai J, Kuehl P, Napier M, Desimone JM, Maynor BW. Microfabricated engineered particle systems for respiratory drug delivery and other pharmaceutical applications. J Drug Deliv. 2012, 2012, 941243.

169. Daniher DI, Zhu J. Dry powder platform for pulmonary drug delivery. Particuology. 2008, 6, 225-238.

170. Moore GE. Cramming more components onto integrated circuits. Electronics. 1965, 38.

171. Ito T, Okazaki S. Pushing the limits of lithography. Nature. 2000, 406, 1027-1031.

64 172. Intel(r) 22nm technoogy. Available at: http://www.intel.com/content/www/us/en/silicon-innovations/intel-22nm- technology.html. Accessed: October 20, 2012.

173. Hidber PC, Helbig W, Kim E, Whitesides GM. Microcontact printing of palladium colloids: Micron-scale patterning by electroless deposition of copper. Langmuir. 1996, 12, 1375-1380.

174. Kim E, Xia Y, Whitesides GM. Polymer microsctructures formed by moulding in capillaries. Nature. 1995, 376, 581-584.

175. Kim E, Xia Y, Zhao X-M, Whitesides GM. Solvent-assisted microcontact molding: A convenient method for fabricating three-dimensional structures on surfaces of polymers. Adv Mater. 1997, 9, 651-654.

176. Kumar A, Whitesides GM. Features of gold having micrometer to centimeter dimensions can be formed through a combination of stamping with an elastomeric stamp and an alkanethiol ‘‘ink’’ followed by chemical etching. Appl Phys Lett. 1993, 63, 2002.

177. Xia Y, Whitehead GS. Soft lithography. Angew Chem Int Ed Engl. 1998, 37, 550- 575.

178. Xia Y, Whitehead GS. Soft lithography. Annu Rev Mater Sci. 1998, 28, 153-184.

179. Lu Y, Chen SC. Micro and nano-fabrication of biodegradable polymers for drug delivery. Adv Drug Deliv Rev. 2004, 56, 1621-1633.

180. Truskett VN, Watts MP. Trends in imprint lithography for biological applications. Trends Biotechnol. 2006, 24, 312-317.

181. Pregibon DC, Doyle PS. Optimization of encoded hydrogel particles for nucleic acid quantification. Anal Chem. 2009, 81, 4873-4861.

182. Lewis CL, Choi C-H, Lin Y, Lee C-S, Yi H. Fabrication of uniform DNA-conjugated hydrogel microparticles via replica molding for facile nuceic acid hydridization assays. Anal Chem. 2010, 82, 5851-5858.

183. Cox GP, Marshall KL, Lambropoulos JC, Leitch M, Fromen C, Jacobs SD. Modeling the effects of microencapsulation on the electro-optic behavior of polymer cholesteric liquid crystal flakes. J Appl Phys. 2009, 106, 124911.

65 184. Torres CMS, Zankovych S, Seekamp J, Kam AP, Cedeno CC, Hoffmann T, Ahopelto J, Reuther F, Pfeiffer K, Bleidiessel G, Gruetzner G, Maximov MV, Heidari B. Nanoimprint lithography: An alternative nanofabrication approach. Mater Sci Eng C. 2003, 23, 23-31.

185. Guo LJ. Nanoimprint lithography: Methods and material requirements. Adv Mater. 2007, 19, 495-513.

186. Buyukserin F, Aryal M, Gao J, Hu W. Fabrication of polymeric nanorods using bilayer nanoimprint lithography. Small. 2009, 5, 1632-1636.

187. Tao L, Zhao XM, Gao JM, Hu W. Lithographically defined uniform worm-shaped polymeric nanoparticles. Nanotechnology. 2010, 21, 095301.

188. Glangchai LC, Caldorera-Moore M, Shi L, Roy K. Nanoimprint lithography based fabrication of shape-specific, enzymatically-triggered smart nanoparticles. J Control Release. 2008, 125, 263-272.

189. Le NV, Dauksher WJ, Gehoski KA, Nordquist KJ, Ainley E, Mangat P. Direct pattern transfer for sub-45nm features using nanoimprint lithography. Microelectronic Eng. 2006, 83, 839-842.

190. Acharya G, Shin CS, McDermott M, Mishra H, Park H, Kwon IC, Park K. The hydrogel template method for fabrication of homogeneous nano/microparticles. J Control Release. 2010, 141, 314-319.

191. Wang JY, Wang Y, Sheiko SS, Betts DE, DeSimone JM. Tuning multiphase amphiphilic rods to direct self-assembly. J Am Chem Soc. 2012, 134, 5801-5806.

192. Lee J, Hasan W, Stender CL, Odom TW. Pyramids: A platform for designing multifunctional plasmonic particles. Accounts of Chemical Reserach. 2008, 41, 1762- 1771.

193. Suh SK, Yuet K, Hwang DK, Bong KW, Doyle PS, Hatton TA. Synthesis of nonspherical superparamagnetic particles: In situ coprecipitation of magnetic nanoparticles in microgels prepared by stop-flow lithography. J Am Chem Soc. 2012, 134, 7337-7343.

194. Dendukuri D, Pregibon DC, Collins J, Hatton TA, Doyle PS. Continuous-flow lithography for high-throughput microparticle synthesis. Nat Mater. 2006, 5, 365-369.

66 195. Xu S, Nie Z, Seo M, Lewis P, Kumacheva E, Stone HA, Garstecki P, Weibel DB, Gitlin I, Whitesides GM. Generation of monodisperse particles by using microfluidics: Control over size, shape, and composition. Angew Chem Int Ed Engl. 2005, 44, 724-728.

67

CHAPTER TWO

Development of PRINT Particles as Pulmonary Drug Delivery Vehicles

Based on: 1. Sections from Garcia, Mack, Williams, Fromen, Shen, Pillai, Kuehl, Napier, DeSimone, Maynor, Journal of Drug Delivery, 2011. Reproduced with permission. 2. Shen, Fromen, Kai, Roberts, Luft, DeSimone, Distribution and Clearance of PEGylated Particles in the Lung, in preparation. 3. Preprint of sections of a book chapter from Fromen, Dunn, DeSimone, Handbook of Nanobiomedical Research: Fundamentals, Applications and Recent Developments. © World Scientific Publishing Company, forthcoming 2014.

68 2.1. Introduction Drug delivery via the pulmonary route offers fast and noninvasive drug delivery to a variety of targets, both local and throughout the body. The lung is comprised of a large surface area and a small epithelial barrier, allowing for high areas of local particle absorption and relative ease of diffusion to the circulatory system for systemic delivery. Local pulmonary delivery is considered more efficient than oral delivery for many lung conditions; this route targets conditions within the lung by delivering high concentrations of drug directly to the site and limiting systemic toxicity. Systemic pulmonary delivery benefits by particle avoidance of the liver and first pass metabolism. Slow release times of drugs deposited in the lung can also result in prolonged drug residence times within the body.1-8 These attributes make pulmonary delivery a highly desirable delivery route for treatment of numerous disease, including asthma, lung cancer, cystic fibrosis (CF), chronic obstructive pulmonary disease (COPD), tuberculosis, emphysema, and diabetes.7 Respiratory diseases specifically are a significant cause of morbidity and mortality worldwide, with an estimated 10 million lung disease related deaths in 2004 globally and with health care costs in the US alone of a projected $173 billion in 2010.9, 10 Self-administered inhalation therapy using pressurized metered dose inhalers (pMDI), dry powder inhalers (DPI), and nebulizers is an attractive route for treatment of respiratory diseases, allowing for local delivery of high concentrations of therapeutics in the lung and avoidance of systemic toxicities associated with oral or injectable therapies.2, 6, 7 However, these current therapeutic aerosol delivery vehicles are surprisingly inefficient. Dose-delivery efficiencies for asthma inhalers range from 3 to 15% for children and 10 to 30% for adults, indicating that less than a third of the contained drug actually escapes the inhaler to reach the airway opening; the most advanced pMDIs deliver only 60% of the inhaled material to central and intermediate bronchial tubes.2 The use of DPIs can mitigate some of the associated issues, such as drug solubility and the difficulties of coordinated breathing, but despite significant advances in the field of microparticle fabrication, preparation of respirable aerosol particles with reproducible and tunable aerodynamic properties remains a challenge.2, 7, 8, 11, 12 Conventional fabrication of these pharmaceutical aerosols for DPIs is accomplished

69 by techniques such as micronization (milling) or spray-drying.13, 14 Use of highly porous particles is an additional approach that has successfully achieved small aerodynamic diameter (DAE) particles for peripheral lung deposition and systemic drug delivery by decreasing particle density.5 However, these fabrication techniques result in polydisperse aerosol populations, with large particle size distributions and limited control over particle shape. Additional fabrication challenges arise with forming dry, non-agglomerating powders comprised of pure active ingredients, especially biologicals like siRNA, proteins and monoclonal antibodies (mAbs). Indeed, there are currently no marketed dry powder inhaled mAbs or siRNA therapies. Precise control of particle geometry is potentially a tunable parameter capable in improving drug delivery to the lung and represents an unexplored opportunity in aerosol pharmaceutics. Given the limitations of current fabrication methods and the numerous barriers of respiratory lung delivery, there is a great demand to increase efficacy of existing therapeutics and a growing need to deliver new drugs and biologics. In this work, we demonstrated the use of a top-down, roll-to-roll particle micro-molding technology, (PRINT®, Particle Replication in Non-wetting Templates) to fabricate monodispersed, non- spherical particles with unprecedented control over size, shape and composition and optimize controlled drug delivery to the lungs.15-21 PRINT inherently creates monodisperse particles in solution, but it is important to ensure this monodispersity can be maintained when particles are aerosolized, in order to fully investigate individual particle properties in aerosol form. Dry powder particle samples were created by lyophilization of the particle solution and dispersed aerosols characterized via Andersen Cascade Impactor (ACI).22 Optimization of lyophilization parameters, such as solvent and freezing time, as well as an investigation into the role of particle shape and method of dispersion, was performed to maximize PRINT aerosol efficiency.

70 2.2. Techniques 2.2.1. PRINT Particle Fabrication Particle Replication in Non-Wetting Templates (PRINT) is a platform particle drug delivery technology that co-opts the precision afforded by lithographic techniques in the microelectronics industry to produce precisely engineered particles. PRINT is an intrinsically dry process with mild processing conditions, making it ideal for molding sensitive biologics and pharmaceutical agents.15-21 The PRINT process described here was used throughout this work, as well as the remainder of this thesis. PRINT is a micro- and nano- molding platform that utilizes molds from fluorinated elastomers. Photocurable perfluoropolyether (PFPE) molds are fabricated through replication of a silicon master template which has been patterned using traditional lithographic techniques (Figure 2.1 top). The photocurable, perfluoropolyether (PFPE) resin has low surface energy, which enables excellent conformation to the master, clean surface release upon photocure, and improved master-to-template fidelity when compared to PDMS replicas. Combined with the ever advancing field of traditional lithography, these properties endow PRINT molds with superior control over particle geometry. Using PRINT, particles of a variety of geometries can be fabricated, with feature sizes ranging from 20 nm to 10 μm, with the variety of achievable shapes further adapted through a process of stretching intermediate master replicas.15-19, 21-25

71

Figure 2.1. The PRINT process. PFPE Fluorocur (green) molds are made from a patterned silicon wafer. Molds are laminated to a pre-particle thin film (red) and passed through the nip of a heated laminator to fill the mold with no interconnecting layer. Filled molds are then laminated to an adhesive transfer sheet (yellow), which is dissolved to disperse particles.

Unlike PDMS, PRINT templates are non-wetting and non-swelling to organic and inorganic materials, which facilitates the molding of a variety of compositions. PRINT is a versatile and gentle fabrication method, and has been shown to fabricate particles of small molecules, protein biologics, siRNA, and hydrophilic or acid-labile polymers containing therapeutic or diagnostic cargos.18, 22, 24, 26-30 Particle fabrication occurs through spontaneous filling of the cavities by capillary or convective forces and removal of excess material through contact with a higher energy surface (such as poly(ethylene terephthalate), PET) prevents formation of an inter-connecting “flash” layer between mold cavities (Figure 2.1 middle). Depending on the pre-particle composition, materials are solidified in the mold cavities through vitrification, crystallization, or gelation before contacting the mold with a sacrificial adhesive harvesting layer to liberate particles (Figure 2.1 bottom). At this point, free flowing powders or stable dispersions can be obtained by dissolving away the adhesive

72 layer from the particles with the option to then be further purified, chemically modified, or analyzed. Particles can be used as suspensions or dried using evaporation or lyophilization to produce dry powders.22 Uniquely, the PRINT process is readily scalable via a roll-to-roll system, by curing molds with a PET backing film.19 As shown in Figure 2.2, the various steps outlined in this section have been adapted to continuous roll-to-roll process through a custom apparatus available in the DeSimone laboratory, courtesy of Liquidia Technologies.

Figure 2.2. Scalability of the PRINT Process. Patterns on a silicon wafer are transferred to linear feet of PRINT mold, which can be used on a continuous roll-to-roll line. DeSimone roll-to-roll imagine courtesy of Marc Kai

Additionally, the PRINT manufacturing process has been demonstrated at scales relevant to support preclinical and clinical studies. At the time of this thesis, Liquidia Technologies has completed a first-in-man, Phase I clinical study of a PRINT vaccine candidate and demonstrated the production of Good Manufacturing Practice (GMP) compliant pharmaceutical materials using this novel nanofabrication process, at scale relevant for clinical development.31

73 2.2.2. Lyophilization Lyophilization, or freeze-drying, is a gentle process by which solvents are removed by sublimation to generate a dry sample and is widely employed for preservation purposes in pharmaceutical and food industries. This process was implemented to generate dry powder PRINT samples suitable for inhalation powders. Conventional lyophilization occurs through four steps, as shown in the phase diagram in Figure 2.3. These steps consist of freezing, sublimation, primary (and sometimes secondary) drying and storage. An aqueous suspension of solvent and solute are frozen and the pressure is lowered past the triple point to sublime the solvent. Following sublimation, the sample is heated above room temperature, often in repeated cycles, to evaporate any residual solvent and completely dry the sample. The dry sample is then returned to room temperature and atmospheric pressure for immediate storage.32-34

Figure 2.3. Phase diagram representation of lyophilization process in pure water. 1) Aqueous suspensions of particles at room temperature and pressure are frozen to obtain a solid. 2) The pressure is lowered past the solid-gas equilibrium line to sublime the solvent. 3) The temperature is raised over room temperature. 4) The pressure is then raised to atmospheric pressure for an extended drying period to evaporate any residual solvent before returning to room temperature.

74 The lyophilization procedure must be optimized for each new system, as both cooling and heating rates and durations can dramatically affect the stability of the dry sample. Alterations in the freezing step are specifically important. A controlled, slow freeze (<1° C/min) minimizes nucleation sites, increases nucleation temperature, allows solvent molecules kinetic freedom to crystalize, and provides larger regions of frozen solvent molecules; this in turn results in faster sublimation, larger pore sizes in the lyophilized powder and overall more repeatable cake formation. Conversely, a rapid freeze (>2° C/min) results in uncontrolled nucleation and kinetically traps solvent molecules, which yields heterogeneous cake morphology and potentially incomplete lyophilization. Similarly, duration of the sublimation step must be long enough to ensure complete drying without unnecessarily increasing the duration of the process. The weight percent of solids will also alter the cake formation; too low and the cake will collapse, too high and the cake may not completely dry. Many pharmaceutical formulations require the addition of stabilizers, such as sugars or surfactants, to adequately protect the cargo and ensure proper cake formation during the freeze-drying process.34

2.2.3. Andersen Cascade Impactor An Andersen cascade impactor (ACI, MSP Corporation) was used to determine mass median aerodynamic diameters (MMAD) of aerosols.35-40 Use of an ACI is the gold standard of aerosol sizing of the pharmaceutical community, as deposition on the various stages has known correlations to lung deposition.41 The ACI is an eight-stage impactor which sizes particles based on inertial momentum. Dispersed particles are pulled through a vacuum and flowed through a series of stages containing nozzles of decreasing diameter. Particles of a certain cut-off diameter will impact on the subsequent collection plate, based on their acceleration through each nozzle as demonstrated in Figure 2.4.A. Particles smaller than the cut-off diameter will circumvent the collection plate, follow the streamlines, and enter the next series of nozzles.37, 40

75

Figure 2.4. Diagrams of an Andersen Cascade Impactor. A) Cross-section of ACI illustrating particle trajectories through the device. B) Diagram of ACI assembly with stage designations and corresponding cut-off diameters. Adapted and reprinted from ref. [40].

A diagram of an ACI with stage cut-offs is shown in Figure 2.4.B, with stage designations and corresponding cut-off diameters for two relevant flow rates. An ACI was initially designed to size MDIs and operates at 28.3 L/min, which corresponds to an arbitrary flow rate (Q) of 1 ft3/min. However, ACI sizing for dry powder inhalers requires additional considerations, as the patient inspiration actuates the particle dispersion; thus, the duration of inspiration and pressure drop across the inhaler defines particle dispersion and aerosol generation. To take this into consideration, Q through the ACI must be adjusted to obtain a realistic pressure drop across the inhaler device. This can typically be achieved using a flow rate of 60 L/min to provide the 4 kPa drop across a low resistance DPI, but will change the stage cut-offs designed for 28.3 L/min. These can be calculated with empirical relations, but by using a different set of stages, as shown in red in Figure 2.4B., similar cut-offs to the 28.3 L/min operating conditions can be generated for Q equal to 60 L/min. Additionally, the inspiration duration must be altered between the two operating conditions to correspond to the total air volume of a healthy adult.38-40, 42 To adhere to USP Pharmacopea standards, an ACI operating at 28.3 L/min will sample aerosol for 8 seconds and an ACI operating at 60

76 L/min will sample aerosol for 4 seconds, resulting in a total sample volume equivalent to 4 L.39, 42

2.3. Materials and Methods 2.3.1. Reagents Solvents and buffers of reagent grade were obtained by Fisher Scientific. PRINT molds were provided by Liquidia Technologies. Pre-particle reagents of poly(ethylene glycol)700 diacrylate (PEG700DA), 1,6-hexanediol diacrylate (HDODA), diphenyl(2,4,6- trimethylbenzoyl)phosphine oxide (TPO), bovine serum albumin (BSA), trehalose and leucine were obtained from Sigma. Dyes incorporated into particle matrices included methacryloxyethyl thiocarbamoyl rhodamine B, obtained from PolySciences, and fluorescein-o-acrylate, obtained from Fisher. Rolls of poly(ethylene terephthalate) (PET) were obtained from KRS Plastics. Thin films of poly(vinyl alcohol) (PVOH) on PET sheets were prepared in house. Tert-butanol and sucrose was obtained from Sigma. 1 µm polystyrene latex spheres were purchased from PolySciences.

2.3.2. BSA Particle Fabrication Fabrication of bovine serum albumin (BSA) protein, used as a model protein, into PRINT particles was performed through a melt-solidification strategy, similar to those described previously.18, 30 Leucine and trehalose were used as plasticizers; trehalose was also chosen for advantageous dispersion properties for dry powder particles, as it is commonly used in the literature as an excipient. The three pre-particle components of BSA, leucine and trehalose were dissolved in water at ~10 wt% at a typical ratio of 1.5 BSA: 1.5 leucine: 1 trehalose by weight. A thin film was cast on a sheet of PET using a Meyer rod and dried using a heat gun to evaporate the solvent. The PET delivery sheet containing the pre-particle thin film was laminated to a PRINT mold using a small hand roller. This sandwich was passed through a heated laminator to allow the thin film to flow and the sandwich was split immediately upon exiting the nip. The brief nip residence time allowed for minimal heat

77 exposure, preventing the protein from being denatured, and the subsequent rapid cooling of the mold resulted in solidification of the pre-particle components into the PRINT cavities. Filled molds were then laminated to a thin film of PVOH-coated PET sheets, passed through a heated nip, cooled briefly on the counter-top and delaminated to transfer the particles. Transfer sheets were then harvested using a bead harvester using ~800uL of isopropanol (IPA) per foot of particles. Particles were washed three times with IPA through a centrifugation step and then transferred to the lyophilization solvent, tert-butanol. Through an initial iterative step, the absolute weight percent solids, Meyer rod number, amount of pre-particle solution deposited, fill temperature and pressure was adjusted to each mold for every batch to generate appropriate films which maximize mold filling and minimize scum. An example of these parameters is shown in Table 2.1.

Table 2.1. Example of PRINT parameters used for fabrication of BSA particles. Pre-particle Settings Laminator Settings Mold 3um donuts Fill Transfer Wt% solids 10% Roller speed 1.5 Roller speed 3 Mayer Rod #5 Pressure 45psi Pressure 80 psi Pre-particle volume 120 uL Temperature 130°C Temperature 130°C Ratio BSA:leucine:trehalose 1.5:1.5:1

2.3.3. HDODA Particle Fabrication Fabrication of non-degradable, highly cross-linked polymer PRINT particles were fabricated using photosensitive pre-particle components.23, 26, 29 A pre-particle solution composed of 97 wt% 1,6-hexanediol diacrylate (HDODA), 2 wt % methacryloxyethyl thiocarbamoyl rhodamine B (PolySciences) and 1 wt% diphenyl(2,4,6- trimethylbenzoyl)phosphine oxide (dissolved in minimal dimethylformamide, DMF) were applied drop-wise to a PET sheet. This was then sandwiched to a patterned PRINT mold and passed through a heated laminator, spreading the liquid monomer solution and filling the

78 PRINT cavities. The sandwich was split immediately upon exiting the nip to remove residual pre-particle solution against the PET sheet. Open-faced, filled molds were photocured under ultraviolet light in a N2-purged, UV-LED oven for 5 minutes. Solid particles were then harvested onto a thin film of PVOH coated on PET and collected by dissolving the film in water. Additional washes were performed to remove excess PVOH. Laminator settings were varied with each batch to optimize filling; an example of settings is shown in Table 2.2.

Table 2.2. Example of PRINT parameters used for fabrication of HDODA particles. Pre-particle Settings Laminator Settings Mold 3um donuts Fill Transfer Wt% solids 97% Roller speed 1.0 Roller speed 1.0 Mayer Rod - Pressure 50 psi Pressure 80 psi Pre-particle volume 50 uL Temperature 150°C Temperature 150°C

2.3.4. Fabrication of Jet-milled Itraconazole Itraconazole powder (Spectrum Chemical) was micronized for aerodynamic particle size comparison testing with PRINT particles. Micronization was performed using one pass through the Glen Mills Laboratory Jet Mill, which utilize compressed air to pulverize powders.22

2.3.5. Particle Characterization Particle uniformity and morphology was confirmed using scanning electron microscopy (SEM). Samples were sputter-coated with 1-5 nm of Au/Pd (Cressington Scientific Instruments) and imaged (Hitachi model S-4700). Nanometer-sized particles were measured by dynamic light scattering (Zetasizer Nano ZS, Malvern Instruments, Ltd.) to determine hydrodynamic diameter (Zavg) and poly dispersity index (PDI). Thermogravimetric analysis (TGA) was used to determine particle concentrations. Protein content was

79 determined by Bradford assay (Thermo Fisher, SpectraMax M5 plate-reader), following manufacturer’s instructions, or HPLC courtesy of Liquidia Technologies.

2.3.6. Lyophilization Many lyophilization variables were explored and optimized for PRINT particle formulations. Solvents used included both tert-butanol and water; sucrose was investigated as a stabilizer and the impact of varying initial particle mass was explored. A rapid freeze rate was achieved by immersing samples in liquid nitrogen (N2), while a controlled freeze rate was achieved by programming the stage temperature of the lyophilizer. Two lyophilizers were used for these studies: FreeZone Plus Cascade freeze-dryer (Labconco) and the AdVantage Plus freeze-dryer (SP Scientific). Samples which underwent a rapid freeze were lyophilized using the Labconco tree FreeZone freeze-dryer. Parafilm (Fisher) was used to cover the opening of the vial and a small hole was created to allow the sample to be exposed to the chamber pressure. Vials were immersed in liquid N2 and immediately placed on the freeze-dryer scaffold already under maximum vacuum (~0.03mBar = 3 Pa). Samples remained on the lyophilizer overnight and were immediately stoppered and stored under desiccant upon removal from the vacuum. Samples which underwent a controlled freeze were lyophilized using the SP Scientific AdVantage Plus freeze-dryer. Samples were parafilmed as before and placed on a shelf in the freeze-dryer. The recipe in Table 2.3 was performed; samples were immediately stoppered and stored under desiccant upon completion.

Table 2.3 Slow-freeze lyophilization recipe. Freezing Primary Drying Secondary Drying Ramp to 5°C Hold -45°C, 20 Pa for 1 hr Post heat treatment at 27°C, 88 kPa for 1 hr Hold 5°C for 5 min Ramp to -10°C, 20 Pa Ramp to -45°C Hold -10°C, 20 Pa for 2 hr Hold -45°C for 3 hr Ramp to 25°C, 20 Pa Hold 25°C, 20 Pa for 4 hr Ramp to 5°C, 20 Pa

80 2.3.7. Cascade Impactor Aerosol Studies For ACI analysis, particles were dispersed from one of three device types. Dry powder samples were dispersed from hydroxypropyl methylcellulose (HPMC) capsules (Plastiape SpA) in a Monodose Inhaler (Mod. 8, Plastiape SpA) with fill weights ~10 mg of dry powder. Dry powder samples were also dispersed from a PennCentury insufflator (PennCentury) with fill weights ~2-5 mg. Finally, particle samples were also dispersed from a suspension of methanol using a collision jet nebulizer (CJN, MRE 1-jet, BGI Inc.) into tubing which was connected to the APS through four feet of tubing which acted as a drying column. All three devices (Figure 2.5) were physically connected to the throat of the ACI for sizing.

Figure 2.5. Aerosol dispersion devices. (A) Monodose inhaler, (B) PennCentury, (C) CJN. Reproduced from refs. [43, 44].

Two flow rates were used in these studies: 28.3 L/min, and 60 L/min, which were confirmed prior to aerosol sampling using a flow meter placed at the ACI throat (MSP Corporation).

Stages were coated with PEG300 (Sigma Aldrich) prior to deposition to prevent particle bounce and particles were collected with water and the solutions analyzed. Typical analysis is performed gravimetrically, however, in this project, protein content was measured via Bradford assay and fluorescence imaging was used to determine distributions of

81 fluorescently labeled particles. From the relative mass distributions, the MMAD was calculated by determining the median value, or midpoint diameter (d50), from the cumulative fraction distribution.37 The geometric standard deviation (GSD) was calculated as a ratio of the mass value at the 84% percentile over the d50. Emitted dose (ED) was calculated as the total recovered minus the amount remaining in the device. Fine particle fraction (FPF) was defined as the percentage of ED which was sized less than 5 µm.37-40

2.4. Results 2.4.1. Fabrication of PRINT Particles for Aerosol Delivery Micron-sized BSA particles were successfully fabricated using PRINT, including a range of different particle sizes and shapes. Particle geometries included cylinders, donuts and pollen-mimic, as shown in Figure 2.6. Pollen-mimic shapes were chosen for their biomimicry of known aerosols and donut shapes were designed to contain a “macropore” intended to decrease porosity. Each shape was fabricated with a maximum dimension of ~1 μm or 3 μm to investigate differences between particle volume, while maintaining constant shape.

Figure 2.6. BSA PRINT particles of a variety of sizes and shapes.

82 Following fabrication, the absolute content of BSA particles was determined using HPLC, which was performed in collaboration with the inhalation group at Liquidia Technologies. Trehalose content varied from 35%-55% of the total particle mass, BSA content ranged from 40%-65% and leucine ranged from 10%-25%. Highly cross-linked HDODA particles were also successfully fabricated using PRINT. Particle geometries of micron-sized particles included donuts and pollen-mimics. Additionally, 80x320 nm rods were successfully fabricated to rapidly screen lyophilization parameters. SEMs will be shown with corresponding aerosol sizing in the following.

2.4.2. PRINT Aerosols Compared to Jet milled Formulations To compare the size distributions of PRINT aerosols to conventional fabrication techniques, itraconazole, a small molecule antifungal agent, was jet-milled to fabricate dry powder aerosols and compared to the particle size distribution of 3 μm PRINT HDODA donuts. PRINT HDODA donuts were lyophilized from tert-butanol after flash freezing to obtain a dry powder; both samples were dispersed from a Monodose inhaler and sized via ACI at 28 L/min. From ACI sizing, jet-milled aerosols yielded a typical log-normal population, as indicated by the linear trend (Figure 2.7). However, PRINT aerosols did not form log- normally distributed aerosols, but are rather closer to populations that are normally distributed. Figure 2.7 also shows the population distribution of a theoretical monodisperse aerosol; the sample of 3 μm PRINT HDODA particles fell somewhere between these two extremes.

83

Figure 2.7. Log probability plot of particle distributions. 3μm donut PRINT particles (blue) are compared to jet-milled particles (purple), with SEMS of each type to the right.

Overall, the PRINT aerosol exhibited a narrower distribution and a higher fraction of drug in the respirable range (less than 5 μm) than the jet-milled sample, indicating that the PRINT formulation would yield aerodynamic properties for improved lung deposition.

2.4.3. PRINT BSA Shaped Particle Aerosol Study Aerosol distributions were further modified by changing the initial particle geometries. The same series of BSA geometries shown in Figure 2.6 was converted to a dry powder by lyophilizing a snap-frozen suspension in tert-butanol. Dry powders were filled into HPMC capsules and dispersed using a Monodose inhaler to be sized via ACI. Both 28.3 L and 60 L/min flow rates were evaluated, although 60 L/min sizing results are considered most relevant considering the dispersion technique. Average sizing results for the six BSA geometries is shown in Figure 2.8, with calculated average MMAD, GSD, ED, and FPF shown in Table 2.4.

84

Figure 2.8. ACI sizing results for BSA/trehalose PRINT particles of a variety of sizes and shapes. Sizing was performed under two flow rate conditions and particles were aerosolized from a hydrocellulose capsule in a Monodose inhaler. Mass deposited on each stage was determined by Bradford assay, n=3.

From these results, a number of observations emerge. Comparing between flow rates, the aerosols sized at 28.3 L/min resulted in considerably larger MMADs in all samples, although the larger set of ~3 μm samples were less affected than the ~1 μm geometries. ED and FPF were also dramatically different in the two sizing conditions; 3 μm particles exhibited a slight increase in ED but minimal change in FPF at the higher Q, while ~1 μm particles exhibited a minimal increase in ED but a dramatic increase in FPF. These trends are readily observed in Figure 2.8, where 60 L/min samples clearly deposited on later ACI stages

85 in greater quantities than the slower flow rate. These trends observed for both particle sets indicate that a Q of 60 L/min resulted in improved actuation and dispersion of most dry powder samples, especially so for the smaller three particles. Interestingly, 3 μm cylinders were the only geometry unaffected by ACI Q; these particles showed minimal difference in dispersion, ED, FPF and MMAD between the two flow rates. Additionally, no increase in ED was observed for 10 μm pollen samples at the higher Q.

Table 2.4. ACI sizing results for BSA/trehalose PRINT particles of a variety of sizes and shapes. Sizing was performed under two flow rate conditions and particles were aerosolized from a hydrocellulose capsule in a Monodose inhaler. Mass deposited on each stage was determined by Bradford assay, n=3. 28.3 L/min 60 L/min MMAD GSD ED FPF (%ED) MMAD GSD ED FPF (%ED) 1.5 µm donuts 3.32 1.78 69% 45% 2.53 1.53 81% 79% 1 µm cylinders 3.23 1.84 79% 43% 2.34 1.48 79% 73% 1 µm pollen 3.67 2.33 75% 34% 2.04 1.58 76% 71% 3 µm donuts 3.95 1.57 55% 33% 3.56 1.48 74% 58% 3 µm cylinders 5.03 1.33 71% 22% 5.32 1.52 70% 20% 10 µm pollen 4.52 1.29 61% 34% 4.24 1.55 56% 42%

As 60 L/min sizing produced the desired 4 kPa drop across the DPI and thus more accurately represented the aerosol formation as experienced by the patient, this set of sizing results was further probed for the influence of shape on particle distribution. At this Q, 1 μm pollen particles had the smallest MMAD and 1.5 μm donuts had the largest ED and FPF of the samples tested (Table 2.4). 3 μm cylinders had the largest MMAD and smallest FPF of all samples, although the ED of the 10 μm pollen sample was considerably lower than the other five samples. These results indicate the complex role of particle geometry, device dispersion and inspiration rate on the ultimate efficacy of an inhaled formulation.

2.4.4. PRINT HDODA Particle Shaped Aerosol Study The effects of particle shape on aerodynamic properties were further explored using HDODA particles, which were produced with higher fidelity than the micron-sized BSA

86 particles. Two series of donut and pollen particles were fabricated, with each series containing particles of three different volumes but constant shape. These particles were lyophilized from a flash-frozen suspension of tert-butanol and sized with the ACI at 28.3 L/min from a Penn Century device; SEMs and ACI results are shown in Figure 2.9.

Figure 2.9. ACI sizing results for donut and pollen series of PRINT HDODA particles. (A) Donut Series. Top, SEMs from left to right of 1.5 µm donuts, 3 µm donuts, and 6 µm donuts. Bottom, ACI particle distributions. (B) Pollen Series. Top, SEMs from left to right of 1.µm pollen, 3 µm pollen, and 10 µm pollen. Bottom, ACI particle distributions. ACI sizing was performed at 28.3 L/min and particles were aerosolized from a PennCentury device. Mass deposited on each stage was determined by fluorescence, n=4 with standard deviation shown.

From both of these series of particles, overall physical volume was observed to be the major indicator of particle deposition. Increasing the outer diameter of the donut particles from 1.5 µm, to 3 µm, to 6 µm resulted in a corresponding increase in MMAD from 1.8 µm, to 2.96 µm, to 4.65 µm. A similar trend of increasing MMAD was observed in the pollen series of particles; the MMADs for the 1 µm, 3 µm, and 10 µm pollen were 2.07 µm, 3.39 µm, and 3.57 µm respectively. Interestingly, the 10 µm pollen had an MMAD very similar to the 3

87 µm pollen, despite the 6 µm difference in geometric length between the two particle geometries. To further explore the interesting aerodynamic characteristics of the 10 µm pollen particle, pollen particle samples were dispersed from a suspension of methanol in a CJN to ensure the creation of a truly monodisperse aerosol. These aerosols were sized with the ACI at 28.3 L/min; ACI distributions for 10 µm pollen particle and control 1 µm polystyrene latex spheres are shown in Figure 2.10

Figure 2.10. ACI sizing results of 10 µm pollen, shown with fluorescence micrographs of ACI stages 3 (A1, A2) and 4 (B). A1 and A2 correspond to the same field of view, but different focal point. ACI sizing was performed at 28.3 L/min and particles were aerosolized from a CJN device, with control 1 µm polystyrene latex spheres (PSL) shown in gray. Mass deposited on each stage was determined by fluorescence.

Size distributions of the aerosols formed from the CJN both demonstrated GSD less than 1.22, which is considered a monodisperse aerosol by definition. While control PSL spheres deposited mainly on stage 5, the 10 µm pollen samples were distributed evenly across stages

88 3 and 4. Looking at the optical images of the particles on each of these stages, the orientation of the particles deposited on the stage appears to be different. On stage 3, there are a surprising number of pollen particles standing vertically on the plate, whereas on stage 4, the majority of the particles are lying flat.

2.4.5. Optimization of Dry Powder PRINT Formulations Results from both the BSA and HDODA shaped particle studies demonstrated the challenge in identifying the role of shape on the various aspects of aerosol delivery, especially from powder samples. Dry powder PRINT aerosols dispersed from the Monodose DPI routinely produced normally distributed aerosols, rather than truly monodisperse aerosols (Figure 2.7), making it difficult to draw strong conclusions as to the exact impact of particle shape due to the aggregated state, and thus variable geometries, of the aerosols. Particle distributions from powders dispersed from a Penn Century yielded similar population distributions, and only dispersions from a CJN were truly monodisperse. Additionally, it was qualitatively observed in both the DPI and Penn Century dispersion techniques that the quality of the dry powder dramatically impacted the ACI sizing results; MMAD increased and ED suffered in various conditions, including extreme high or low ambient humidity. It was also observed that slight differences in lyophilization conditions also dramatically changed ACI sizing results. This was especially true for the smaller sized particles, typically 3 μm and smaller, as observed from aggregated aerosol sizing results. As a result, an in depth study of lyophilization optimization was initiated. Lyophilization solvent and weight percent solids were first investigated, using a single batch of 3 μm donuts comprised of HDODA lyophilized under different conditions. The theoretical MMAD for these particles was approximated as 2.19 μm based on calculations from a model using particle projected surface area (more in Chapter 3).45 Aerosol samples were created, dispersed from a Penn Century insufflator and measured with an ACI for MMAD determination (shown n≥3); results are shown in Figure 2.11A.

89

Figure 2.11. Lyophilization solvent study. A) MMADs of 3 μm donut particles lyophilized from different solvents and weight percent solids at a rapid freeze rate. B) Same results plotted as a function of ambient relative humidity. n≥3 with standard deviation shown. Linear regression on average MMADs shown for B.

Dispersion was seen to vary both with changes in percent solids prior to lyophilization and the lyophilization solvent. HDODA particles lyophilized from water, rather than tert-butanol, resulted in MMADs more closely resembling the expected MMAD of 2.19. No clear trend emerged from changing the initial weight percent of particles in the range tested here, although differences were observed. MMADs were also plotted against the relative ambient humidity, yielding a weak positive correlation which suggests that increases in humidity at the time of the ACI sizing increased MMAD sizing results (Figure 2.11B). While initial lyophilization conditions showed tremendous variability in MMAD between samples, no sample achieved an MMAD approaching the theoretical value for 3 μm donuts by varying solid content or lyophilization solvent. Turning to the limited literature of lyophilization of particles for dry powder samples, changes in freezing rate prior to sublimation were next studied, as well as the addition of a sugar stabilizer for cake formation.32-34 To rapidly screen these effects, 1 mg of 80x320 nm crosslinked particles were lyophilized from water with varied sucrose contents and initial freezing rates, resuspended in

90 water, and immediately sized using the DLS. This method could not be implemented using 3 μm donuts due to sizing limitations of the Zetasizer. Example cakes are shown in Figure 2.12 and sizing results for the particle hydrodynamic diameter (Zavg) and polydispersity index (PDI) are shown in Table 2.5.

Figure 2.12. Optical images of the resulting cakes of 80x320 nm particles after lyophilization with sucrose stabilizer following a slow freeze step (~0.2°/min from a ramp to -45°C over 6 hr). The ratio of NP : sucrose is A) 1:1, B) 1:10, C) 1:30, and D) 1:92.5, n=3.

From both the optical images of the cakes, as well as corresponding DLS results, a controlled slow freeze resulted in considerable particle aggregation, regardless of the addition of sucrose stabilizer. Cakes in Fig. 2.12 were not well formed; all cakes were uneven, were slightly tacky and showed evidence of both collapse and phase separation during the freezing process, which were pronounced in 2.#B and 2.#C. This was further supported by the DLS results, as all samples resulted in considerably larger Zavg and PDI than the initial particle sample. The addition of the sucrose stabilizer had a slight increased effect on resulting particle stability, as increased sugar content resulted in decreased Zavg and PDI, but did not completely prevent particle aggregation during the lyophilization process. Particles without the sugar stabilizer actually yielded samples with the smallest Zavg, suggesting that the sucrose contributed to particle aggregation in the slow freezing process.

91 Table 2.5. DLS results of lyophilization study. 80x320 nm particles were lyophilization with sucrose stabilizer at various sucrose concentrations and underwent either a fast (~5 min in liquid nitrogen) or slow freeze (~0.2°/min from a ramp to -45°C over 6 hr). n=3 with standard deviations shown. RATIO Z NP : sucrose % sucrose avg PDI

initial 0 249 ± 2 0.059 ± .007 Slow Freeze 1:0 0 277± 2 0.136± 0.02 1:1 0.1 511± 26 0.363 ± 0.07 1:10 1 318.6± 2 0.184 ± 0.01 1:30 3 298.6 ± 2 0.153 ± 0.01 1: 92.5 9.25 292.3± 3 0.114± 0.01 Fast Freeze 1:0 0 251.8 ± 2 0.061 ± .007 1:1 0.1 492 ± 25 0.345 ± 0.01 1:10 1 265.9 ± 2 0.090± 0.02 1:30 3 265.1 ± 1 0.035± 0.02 1:46 4.6 255.9 ± 1 0.053± 0.01 1:70 7 258.7 ± 3 0.060 ± 0.2 1:92.5 9.25 258.7± 4 0.059 ± 0.05

Improvement in nanoparticle stability was observed for nanoparticle samples rapidly frozen prior to lyophilization. Unmodified samples did not suffer from increases in aggregation, while the addition of sucrose was observed to have a positive effect and restored sample stability at sucrose amounts greater than 4.6%. While the addition of sucrose was not required to restore 80x320 nm particle stability in suspension, we sought to extend these results to aerosol samples to investigate if the addition of sucrose would result in better dispersed aerosols. 3 µm donuts were lyophilized in 7 wt% sugar and sized with ACI out of Penn Century device. The resulting size distribution can be seen in Figure 2.13.

92

Figure 2.13. ACI sizing results of 3 µm donuts lyophilized in 7 wt% sugar and sized with ACI at 28.3 L/min out of Penn Century.

The resulting powder had an MMAD of 3.04 µm and a GSD of 1.5. Compared to previous ACI sizing results, the addition of sugar did not alter the aerosol dispersion.

2.5. Discussion The work presented here represents the first efforts of applying precision designed particles for respiratory drug delivery. Using the PRINT fabrication approach, particle geometries were directly controlled, which is a differentiating attribute as compared to traditional particle generation approaches. In the PRINT process, the particle geometry is directly derived from the semiconductor wafer, bringing inherent nanoscale precision to the particle geometry and offering the capability to generate unique, non-spherical shapes, as shown in Figure 2.6. Due to the implicit connection to the original silicon wafer, every batch of particles can show high uniformity and batch-to-batch consistency, regardless of batch size; this a feature which makes the PRINT technology attractive from the perspective of compliance with Quality-by-Design directives from the FDA.46 In traditional fabrication methods, particle chemical composition and physical characteristics such as geometric or aerodynamic size are inherently coupled; for example, the molecular properties of a small molecule pharmaceutical ingredient are known to impact the particle size distribution of

93 micronized particles, whereas the solubility and drying kinetics of precursor solutions can impact the particle size distribution of spray-dried particles.13, 14 In contrast, micro-molded particle engineering has the ability to define the particle size and shape independent of the input material properties. Micro-molding particles specifically for respiratory delivery resulted in high performance aerosols that possessed tunable aerodynamic diameters and narrow aerodynamic size distributions. This tunable control over aerosol characteristics was demonstrated across a wide range of aerodynamic diameters within the respirable range (Fig 2.9). In addition, compared to a conventional method for particle fabrication, PRINT particles demonstrated improved properties better suited for inhalation therapies. PRINT aerosols achieved an increased respirable dose and decreased MMAD compared to aerosols generated by traditional micronization processes (Fig. 2.7). Furthermore, improvement of the dry powder formation through optimization of freeze-drying methods increased emitted doses and produced well-dispersed aerosols, specifically by controlling freeze rates and solvent composition. Importantly, a rapid freeze rate was required to prevent irreversible particle aggregation; cake formation from a slow freeze yielded dense, moist cakes (Figure 2.12 A-C) which is hypothesized have stemmed from phase separation during the freezing process (Figure 2.14).

Figure 2.14. Diagram of freezing process of particle suspensions following A) rapid or B) slow freezing rates. A rapid freeze rate provides swift nucleation throughout the solvent, effectively trapping the particles in their suspended orientation. A slow freeze rate with limited nucleation sites allows for a solvent front form during the freezing process, which exclude the particles from the solid phase.

94 These attributes of PRINT which enable tunable geometries, particle compositions and dry powder formation are expected to translate into more efficient respiratory drug delivery for a wide range of therapeutics that are intended to deposit in the lung periphery. Importantly, the dispersion of PRINT particle dry powders did not require the use of bulking excipients, such as lactose or sucrose, for particle dispersion, as is often the case for dry powder products. Indeed, the addition of sucrose offered no improvement to aerosol dispersions. Elimination of bulking agents potentially simplifies the chemistry, manufacturing, and control processes required to develop dry powder products, as well as mitigating the potential for excipient- induced user side effects. This work also represents one of the few efforts to explore the role of particle shape on aerosol characteristics. Though particle shape is known to be a critical factor of aerosol properties, thorough exploration of its effect has been limited by current fabrication methods of aerosol particles.47 Controlling particle shape through PRINT thus provides an opportunity to systematically optimize the effect of shape on these stages of drug delivery. Work presented here demonstrated potential advantages of introducing particle shape to DPI formulations (Fig. 2.8, Table 2.4). ACI sizing studies with shaped BSA particles showed that the 1 μm pollen geometry yielded the smallest MMAD; this shape was the most complex geometry studied and points to the ability to design shaped particles to decrease DAE and thus increase lung penetration. Unexpected advantages were also observed from sizing results of BSA 1.5 μm donuts, which were found to have the largest ED and FPF of the samples tested, suggesting that the addition of a macropore aids dispersion, possibly through limiting contact area between particles. In contrast to the non-spherical particles, 3 μm cylinders, which represent the simplest geometry tested, were sized with the largest MMAD and smallest FPF of all samples. Additionally, this powder was unaffected by changes in ACI flow rate, suggesting that dispersion of particles of this geometry 3 μm and larger requires a larger pressure drop across the DPI. These initial studies point to the potential role of particle geometry on tailored airway deposition, but were confounded by the difficult in generating truly monodisperse aerosols, despite having monodisperse particles in solution. Prior to optimizing dry powder

95 formulations, we utilized a CJN with a rapidly evaporating solvent to ensure that monodisperse aerosols could in fact be generated using PRINT. This approach resulted in the successful generation of the first monodisperse non-spherical aerosol; more of this work will be presented in Chapter 3. However, initial work presented here yielded interesting observations of the aerodynamic properties of 10 μm pollen particles (Fig. 2.10). Size distributions of the 10 μm pollen aerosols were distributed evenly across ACI stages 3 and 4, with optical images of the stages showing differences in the orientation of deposited particles. Using a model to predict DAE based on the projected surface area of a particle, we could predict how differences in orientation would alter the DAE of the 10 μm pollen; particles oriented with the 10 µm feature parallel to the flow would project a smaller surface area and have a predicted MMAD of 3.33 μm, while particles oriented perpendicular to the flow would have a predicted MMAD of 2.89 μm.45 This correlates surprisingly well with the deposition observed on the ACI stages (Figure 2.10), as the first prediction should yield vertical particles on stage 3 and horizontal particles on stage 4. While particles could have been reoriented after initial impact and deposition on the ACI plates, this observation suggests that pollen particles sample multiple orientations throughout the complex airflow of the ACI, as opposed to maintaining an orientation which minimizes drag (such as particles oriented parallel to the air stream). This observation is important because it is the first example of deposition variance due to particle geometric orientations. Combined with novel smart inhaler devices capable of precise timing and orientation release of an individual particle at the mouth, region-specific lung targeting could be realized, dramatically improving the delivered dose to an exact region of the lung.48, 49

2.6. Conclusions Microfabrication techniques such as PRINT offer a promising strategy to control particle geometry, making it well-suited for the production of aerosol particles for tailored respiratory drug delivery. Precise control over size and shape allowed for defined aerodynamic properties, which in turn, could lead to improved aerosol performance and

96 differential tunable lung deposition in vivo. Optimization of process parameters affecting the formation of dry powder PRINT powders and aerosol formation through various dispersion techniques were explored here. Overall, PRINT micro-molding is a viable particle design strategy that may address challenges existing in respiratory drug delivery, thereby constituting a promising opportunity for the development of next generation therapeutics.

2.7. Acknowledgements The author would like to thank Abby Larus, Kevin Herlihy, Tojan Rahhal, Joe Pedit, and Maryanne Boundy for their assistance with the aerosol characteristics; a special thanks goes also to David Leith for his helpful discussions, aerosol class notes (UNC course ENVR 416 Aerosol Technology) and unlimited access to the resources in his lab. Thanks to Jillian Perry, Stu Dunn, Abby Larus, Nicole Forman, and Tammy Shen for PRINT fabrication and lyophilization, as well as the inhalation team at Liquidia, especially Pete Mack, Ben Maynor, Katie Horvath, Andres Garcia, and Janet Tully for fabrication, analysis, and characterization of BSA particles. We acknowledge Liquidia Technologies for providing PRINT molds, and the core facilities at UNC, including the CHANL imaging facility. This work was funded in part by the NIH Pioneer Award to J.M.D. (1DP1OD006432) and a sponsored research agreement with Liquidia Technologies.

2.8. References 1. Weers JG, Bell J, Chan HK, Cipolla D, Dunbar C, Hickey AJ, Smith IJ. Pulmonary formulations: What remains to be done? J Aerosol Med Pulm Drug Deliv. 2010, 23 Suppl 2, S5-23.

2. Kleinstreuer C, Zhang Z, Donohue JF. Targeted drug-aerosol delivery in the human respiratory system. Annu Rev Biomed Eng. 2008, 10, 195-220.

3. Azarmi S, Roa WH, Lobenberg R. Targeted delivery of nanoparticles for the treatment of lung diseases. Adv Drug Deliv Rev. 2008, 60, 863-875.

97 4. Sakagami M, Byron PR. Respirable microspheres for inhalation. Clin Pharmacokinet. 2005, 44, 263-277.

5. Edwards DA, Hanes J, Caponetti G, Hrkach J, Ben-Jebria A, Eskew ML, Mintzes J, Deaver D, Lotan N, Langer R. Large porous particles for pulmonary drug delivery. Science. 1997, 276, 1868-1971.

6. Patton JS, Byron PR. Inhaling medicines: Delivering drugs to the body through the lungs. Nat Rev Drug Discov. 2007, 6, 67-74.

7. Mansour HM, Rhee Y, Wu X. Nanomedicine in pulmonary delivery. Int J Nanomedicine. 2009, 4, 299-319.

8. Weers JG, Tarara TE, Clark AR. Design of fine particles for pulmoanry drug delivery. Expert Opin Drug Del. 2007, 4, 297-313.

9. The World Health Organization. The global burden of disease 2004 update. Available at: who.int/healthinfo/global_burden_disease/GBD_report_2004update_full.pdf.

10. National Heart, Lung, and Blood Institute. Morbidity and mortality: 2009 chart book on cardiovascular, lung, and blood diseases. Available at: nhlbi.nih.gov/resources/docs/2009_ChartBook.pdf.

11. Xie Y, Zeng P, Wiedmann TS. Disease guided optimization of the respiratory delivery of microparticulate formulations. Expert Opin Drug Del. 2008, 5, 269-289.

12. Son YJ, McConville JT. Advancements in dry powder delivery to the lung. Drug Dev Ind Pharm. 2008, 34, 948-959.

13. Chow AH, Tong HH, Chattopadhyay P, Shekunov BY. Particle engineering for pulmonary drug delivery. Pharm Res. 2007, 24, 411-437.

14. Vehring R. Pharmaceutical particle engineering via spray drying. Pharm Res. 2008, 25, 999-1022.

15. Canelas DA, Herlihy KP, DeSimone JM. Top-down particle fabrication: Control of size and shape for diagnostic imaging and drug delivery. WIREs Nanomed and Nanobiotechnol. 2009, 1, 391-404.

98 16. Euliss LE, DuPont JA, Gratton S, DeSimone J. Imparting size, shape, and composition control of materials for nanomedicine. Chem Soc Rev. 2006, 35, 1095-1104.

17. Gratton SE, Williams SS, Napier ME, Pohlhaus PD, Zhou Z, Wiles KB, Maynor BW, Chen C, Olafsen T, Samulski ET, DeSimone JM. The pursuit of a scalable nanofabrication platform for use in material and life science appliactions. Acc Chem Res. 2008, 41, 1685-1695.

18. Kelly JY, DeSimone JM. Shape-specific, monodisperse nano-molding of protein particles. J Am Chem Soc. 2008, 130, 5438-5439.

19. Merkel TJ, Herlihy KP, Nunes J, Orgel RM, Rolland JP, DeSimone JM. Scalable, shape- specific, top-down fabrication methods for the synthesis of engineered colloidal particles. Langmuir. 2010, 26, 13086-13096.

20. Petrosko SH, Fromen CA, Auyeung E, DeSimone JM, Mirkin CA. Nanotechnology: An enduring bridge between engineering and medicine. NAE The Bridge. 2013, Fall 2013, 7- 15.

21. Rolland JP, Maynor BW, Euliss LE, Exner AE, Denison GM, DeSimone JM. Direct fabrication and harvesting of monodisperse shape-specific nanobiomaterials. J Am Chem Soc. 2005, 127, 10069-10100.

22. Garcia A, Mack P, Williams S, Fromen C, Shen T, Tully J, Pillai J, Kuehl P, Napier M, Desimone JM, Maynor BW. Microfabricated engineered particle systems for respiratory drug delivery and other pharmaceutical applications. J Drug Deliv. 2012, 2012, 941243.

23. Fromen CA, Shen TW, Larus AE, Mack P, Maynor BW, Luft JC, DeSimone JM. Synthesis and characterization of monodisperse uniformly shaped respirable aerosols. AIChE Journal. 2013, 59, 3184-3194.

24. Merkel TJ, Jones SW, Herlihy KP, Kersey FR, Shields AR, Napier M, Luft JC, Wu H, Zamboni WC, Wang AZ, Bear JE, DeSimone JM. Using mechanobiological mimicry of red blood cells to extend circulation times of hydrogel microparticles. Proc Natl Acad Sci U S A. 2011, 108, 586-591.

25. Wang Y, Merkel TJ, Chen K, Fromen CA, Betts DE, DeSimone JM. Generation of a library of particles having controlled sizes and shapes via the mechanical elongation of master templates. Langmuir. 2011, 27, 524-528.

99 26. Dunn SS, Tian S, Blake S, Wang J, Galloway AL, Murphy A, Pohlhaus PD, Rolland JP, Napier ME, DeSimone JM. Reductively responsive sirna-conjugated hydrogel nanoparticles for gene silencing. J Am Chem Soc. 2012, 134, 7423-7430.

27. Enlow EM, Luft JC, Napier ME, DeSimone JM. Potent engineered plga nanoparticles by virtue of exceptionally high chemotherapeutic loadings. Nano Lett. 2011, 11, 808-813.

28. Hasan W, Chu K, Gullapalli A, Dunn SS, Enlow EM, Luft JC, Tian S, Napier ME, Pohlhaus PD, Rolland JP, DeSimone JM. Delivery of multiple sirnas using lipid-coated plga nanoparticles for treatment of prostate cancer. Nano Lett. 2012, 12, 287-292.

29. Perry JL, Reuter KG, Kai MP, Herlihy KP, Jones SW, Luft JC, Napier M, Bear JE, DeSimone JM. Pegylated print nanoparticles: The impact of peg density on protein binding, macrophage association, biodistribution, and pharmacokinetics. Nano Lett. 2012, 12, 5304-5310.

30. Xu J, Wang J, Luft JC, Tian S, Owens G, Jr., Pandya AA, Berglund P, Pohlhaus P, Maynor BW, Smith J, Hubby B, Napier ME, DeSimone JM. Rendering protein-based particles transiently insoluble for therapeutic applications. J Am Chem Soc. 2012, 134, 8774-8777.

31. Clinical Trial NCT01224262. A study evaluating the safety and tolerability of a seasonal containing liq001 (lift). Available at: clinicaltrials.gov/ct2/show/NCT01224262. Accessed: July 1, 2014.

32. Abdelwahed W, Degobert G, Stainmesse S, Fessi H. Freeze-drying of nanoparticles: Formulation, process and storage considerations. Adv Drug Deliv Rev. 2006, 58, 1688- 1713.

33. Chen G, Wang W. Role of freeze drying in nanotechnology. Drying Technology. 2007, 25, 29-35.

34. Gieseler H, Gieseler M. Freeze drying of pharmaceuticals and biologicals. Lecture material from lyolearn educational series by sp scientific. Durham, nc, november 14, 2011. Available at: Spscientific.Com/webinars/archives/.

35. Mitchell JP, Nagel MW. Cascade impactors for the size characterization of aerosols from medical inhalers. J Aerosol Med. 2003, 16, 341-377.

100 36. Shekunov BY, Chattopadhyay P, Tong HH, Chow AH. Particle size analysis in pharmaceutics: Principles, methods and applications. Pharm Res. 2007, 24, 203-227.

37. Hinds WC. Aerosol technology: Properties, behavior, and measurement of airborne particles. New York: John Wiley & Sons, Inc.; 1999.

38. Good cascade impactor practices, aim and eda for orally inhaled products. New York: Springer; 2013.

39. US Pharmacopeia A. <601> aerosols, nasal sprays, metered-dose inhalers, and dry powder inhalers. In: Pharmacopeial Forum. vol. 30:1342.

40. Quality solutions for inhaler testing. In: Scientific C, ed., 2010:1-88.

41. Thiel CG. Can in vitro particle size measurements be used to predict pulmonary deposition of aerosol from inhalers? J Aerosol Med. 1998, 11, S43-S52.

42. van Oort M, Downey B, Roberts W. Verification of operating the andersen cascade impactor at different flow rates. Pharmacopeial Forum. 1996, 22, 2211-2219.

43. Plastiape. Dry powder inhaler rs01: How to use. Available at: plastiape.com/en/content/1635/dry-powder-inhaler-rs01-how-use. Accessed: July 8, 2014.

44. BGI Incorporated. Collision nebulizers. Available at: bgiusa.com/agc/collison.htm. Accessed: July 8. 2014.

45. Leith D. Drag on nonspherical objects. Aerosol Science and Technology. 1987, 6, 153- 161.

46. Rathore AS, Winkle H. Quality by design for biopharmaceuticals. Nat Biotechnol. 2009, 27, 26-34.

47. Hassan MS, Lau RW. Effect of particle shape on dry particle inhalation: Study of flowability, aerosolization, and deposition properties. AAPS PharmSciTech. 2009, 10, 1252-1262.

48. Kleinstreuer C, Seelecke S. Inhaler system for targeted maximum drug-aerosol delivery. In: USPTO, ed.: North Carolina State University, 2011.

101 49. Kleinstreuer C, Zhang Z, Li Z, Roberts WL, Rojas C. A new methodology for targeting drug-aerosols in the human respiratory system. International Journal of Heat and Mass Transfer. 2008, 51, 5578-5589.

102

CHAPTER THREE

Synthesis and Characterization of Monodisperse Uniformly Shaped Respirable Aerosols

Fromen, Shen, Larus, Mack, Luft, Maynor, DeSimone, AIChE Journal, 2013. Reproduced with permission.

103 3.1. Introduction Due to the drastic improvements in aerosol device and particle engineering technologies in the past decade, the delivery of pulmonary therapeutics from a dry powder has become an area of great interest. Dry powder devices offer many benefits including patient actuation, a capacity to deliver a wide range of therapeutics, and the avoidance of cold chain storage issues. Despite the encouraging prospects, inhalation powders currently have limited application due to the complexities of these systems. Inhalation powders are typically multicomponent systems, which must be entrained and deagglomerated in a desired manner prior to flowing through complex airway geometries and depositing in a target area of the lung.1-4 Unsurprisingly, performance of these powders can vary dramatically. A key cause of this variability stems from the limited understanding of the interplay between an individual particle and its dynamic environment. Few studies have been performed on powders of controlled particle geometry, or even drug-only powder formulations, to address these concerns. Current technologies used to generate powders for inhalation products typically involve bottom-up approaches such as spray drying or precipitation reactions; these approaches often have restricted flexibility in particle composition and size distribution.5, 6 The ability to engineer aerosols in a predetermined fashion with complete control over particle size, shape, and composition, and thus aerosolization, is lacking. Owing to fabrication difficulties in creating monodisperse particles, the creation of truly monodisperse aerosols for inhalation studies has only been achieved in limited cases.7, 8 A perfectly monodisperse aerosol would be comprised of a uniform population of identical, unaggregated particles, indicated by a geometric standard deviation (GSD) equal to 1. In practice, a GSD less than 1.22 is typically referred to as ‘monodisperse’.7-9 The formation of monodisperse aerosols would lend insight into the many physical principles at play in inhalation powders, contributing to optimized performance and improved dose uniformity to the patient. Of the many physical parameters at play in aerosol formulations, particle shape is known to be a critical factor of behavior, and yet, in practice, little has been experimentally investigated.5 Non-spherical aerosols in flow are not easily described through analytic approaches and are typically characterized in the aerosol community by a dynamic shape

104 factor (χ).9-11 A shape factor is used to correct for any differences in particle motion resulting from a non-spherical geometry by relating the drag of the non-spherical particle to that of an equivalent volume sphere. This term is defined in a Stokes’ flow regime (where Reynolds number, Re, is less than one), which is relevant to low-velocity airstreams encountered in the broncheoalveolar regions of the lung (branched generations 18 and higher).9, 12 A particle’s dynamic shape factor contributes to the definition of the particle’s aerodynamic diameter 5, 9, 11 (DAE), the main predictor of lung deposition. Determining an appropriate χ for irregularly shaped particles is not trivial and much work has gone into the determination of the χ for specific geometries. A limited data set for χ determined through macroscale models exists for simple geometric shapes and dusts, with empirical relations extending for symmetric objects, such as spheroids, oblates and prisms.9, 13-17 A broader approach to define χ using projected area and surface area has been established which successfully predicts the current set of literature values.18 Finally, shape factors have been indirectly measured through use of elutriation devices, or simultaneous measurements of both an aerodynamic particle sizer (APS) spectrometer and an Andersen cascade impactor (ACI).18, 19 However, a rapid method for experimentally determining shape factor is lacking. It is important to emphasize that χ is an approximation; any contribution to drag experienced by the particle will be lumped into this single correction term. As with any such correction factor, it is worth briefly addressing the strengths and weaknesses of this approach. The dynamic shape factor approximation is commonly used in the aerosol literature to reduce the otherwise complicated physical characteristics of a non-spherical particle to a much simpler spherical one. By discussing particles in terms of equivalent spheres, as illustrated in Figure 3.1, established properties of spherical aerosols can be rapidly translated to non-spherical ones.

105

Figure 3.1. Diagram illustrating the relationship between a non-spherical particle and its equivalent spheres. Parameters shared between the equivalent sphere and the non-spherical particle are highlighted.

In terms of respiratory drug delivery, this enables predictions of particle deposition in different lung regions based solely on DAE. In most situations of aerosol work, this approximation is extremely useful; long length and time scales in most flow applications result in a reasonable simplification to spherical properties. However, in cases where these length or time scales are reduced, use of χ will not appropriately account for individual particle dynamics. The precise drag of a non-spherical particle depends on the geometry and the orientation of the particle, which are not differentiated using a shape factor simplification.13 Additionally, the influence of torque and lift, known to contribute to aerodynamics of non-spherical particles, are not taken into account.10 While understanding of individual particle dynamics based on geometry will be increasingly influential to the

106 literature, χ characterization will continue to be the standard in the aerosol community as it is an extremely useful simplification. By quantifying χ for non-spherical particles, the role of particle geometry can be investigated for entrainment, deagglomeration, flow and deposition in airways. For respiratory delivery, particle geometry is also known to influence mucocilliary clearance, macrophage uptake and drug release kinetics.4, 20-23 Investigating particle behavior and shape effects in each of these areas could lead to novel strategies for the design of efficient inhaled drug delivery systems. As such, there is an unmet need to create monodisperse shaped particles to perform these studies. Knowledge gained on the behavior of precisely shaped particles as powders and aerosols could be used to optimize performance in the various stages of pulmonary drug delivery. PRINT, a top-down, roll-to-roll nano- and micro-molding technique has been used to fabricate particles with unprecedented control over size, shape and composition; this platform has been applied to the creation of dry power formulations, with the application for inhaled drug delivery.22, 24-26 While previous work has shown that the PRINT micromolding technique is capable of fabricating monodisperse particles as dispersions in liquid, their quality as dry powder aerosols has yet to be explored in depth. The object of this work was to fabricate monodisperse particles relevant for dry powder therapeutics and explore the influence of particle shape on flow and dispersion. Enabled by these inherently monodisperse particles, we also explored a novel method for characterizing aerosol shape factors through APS measurements in attempt to further characterize aerosol behavior of non-spherical particles.

3.2. Theory 3.2.1. Shaped Particles in Stokes’ Flow

Aerosols of respirable size (DAE ~1-5 μm) that settle in air are typically considered to fall under the assumption of Stokes’ flow.9 Under these conditions, an aerosol’s dynamic shape factor is implemented to correct the drag force on non-spherical particles from that of a sphere. In Stokes’ flow, the drag force of a non-spherical particle becomes:

107

(1)

where FD is the drag force, μ is the fluid viscosity, DEV is the diameter of a sphere with a volume equivalent to the shaped particle, V is the particle velocity and CC is the Cunningham slip correction factor, defined as:

[ ( )] (2) where λ is the mean free path of the fluid, typically air.9, 11 These expressions are valid for Re less than 1 and assumes a no slip boundary condition at the surface of the particle. As the particle size drops below 10 μm, the no slip condition is no longer valid; the CC term is needed to correct for slip in this regime and its contribution becomes increasingly significant as particle size decreases.9 Eq. 2 is considered valid for particle sizes 10 μm and below and, as it is a function only of DEV, it can be reasonably applied to correct slip for non-spherical particles.9, 11 Dynamic shape factor is experimentally defined by measuring the terminal settling velocity (VTS) of a particle. Equating the Stokes’ drag force in Eq. 1 with the gravitational force allows for the expression of VTS for an aerosol:

(3)

where ρp is the particle density, g is the acceleration due to gravity. Eq. 3 is valid over all 9 particle sizes, assuming Stokes flow. From Eq.3, measurement of VTS for a non-spherical particle in a situation of known fluid properties allows for the determination of the corresponding dynamic shape factor.27, 28 Importantly, χ can be used to determine the particle’s DAE through the following relation:

( ) (4)

3 9 where ρ0 is a standard particle density (1.0 g/cm ).

108 3.2.2. Extension of Dynamic Shape Factor in the Non-Stokesian Flow of an Aerodynamic Particle Sizer (APS) The previous relation for particle drag can be applied only for Re less than one. However, the APS used to size aerosols in the following experiments operates at a flow rate which corresponds to a Re between 0.01 and 160.29 Due to this operation outside Stokes’ regime, corrections to the APS for errors measuring coincidence events, particle density and shape have been investigated in the literature. 9, 19, 29-33 As shown previously, the APS technique is known to undersized non-spherical particles.9, 19, 32, 34, 35 To account for this and correct APS measurements for non-spherical PRINT particles, an expression for shaped particles outside Stokes’ flow must be considered. As Re increases, the particle drag in Eq. 1 becomes dependent on flow conditions. In the non-Stokesian regime, an additional factor is included in the drag constant. For Re less than ~100, which is within the conditions of the APS for particles less than 10 μm, there are a few commonly used ultra- or near-Stokesian correction terms.29 In this work, the following expression was used to describe the particle drag:

⁄ (5) where U is the fluid velocity.19 The sphericity term φ is related to particle shape factor by:

(6)

In the literature, the sphericity term in Eq. 6 has been validated for χ ranging from 1 to 1.5. In this work, will be considered over all shape factors, as other ultra-Stokesian corrections assume empirical values that are equally untested at χ greater than 1.5. In tandem with the flow profile present in an APS spectrometer, these expressions can be used to correct for the aerodynamic effects of particles with known volume.34

109 3.3. Materials and Methods 3.3.1. Particle Fabrication using PRINT Particles were fabricated using the PRINT micro-molding technique, described in detail previously.25, 26 Highly cross-linked polymer networks were chosen as a model particle composition to explore the macroscale physical properties of shaped aerosols. PRINT molds (Liquidia Technologies) were filled by a lamination technique. A pre-particle solution composed of 97 wt% 1,6-hexanediol diacrylate (HDODA), 2 wt % methacryloxyethyl thiocarbamoyl rhodamine B (PolySciences) and 1 wt% diphenyl(2,4,6- trimethylbenzoyl)phosphine oxide as a photo-initiator was cast onto a sheet of poly(ethylene terephthalate) (PET) and applied to the patterned PRINT molds. Open-faced, filled molds were photocured under ultraviolet light in a N2-purged, UV-LED oven for 30 seconds. Solid particles were then harvested onto a thin film of polyvinyl alcohol (PVA) coated on PET and collected by dissolving the film in water. Additional washes were performed to remove excess PVA. Finally, particles were lyophilized from tert-butanol to obtain dry powder samples. This process was repeated for each of the fourteen geometries investigated in these studies. All chemicals and reagents were obtained from Sigma Aldrich unless noted.

3.3.2. Particle Characterization Particle uniformity and morphology was confirmed using scanning electron microscopy (SEM). Samples were sputter-coated with 1-5 nm of Au/Pd (Cressington Scientific Instruments) and imaged (Hitachi model S-4700). SEM micrographs were used to measure feature sizes of each shape using ImageJ.36 From these measurements, geometric properties, such as particle volume, DEV, and total surface area, were calculated. The density of the particle matrix was assumed to be equivalent to the density of a bulk sample of HDODA. Bulk samples were prepared by polymerizing 1 mL of pre-particle solution on a glass slide in a N2-purged UV-LED oven for 5 minutes. Samples were then weighed in air and water using a density conversion kit for an analytical balance (Mettler Toledo). Bulk density was then calculated:

110

(7)

where Mair is the mass of the bulk sample in air, Mwater is the mass of the sample in water, 37 ρwater is the density of water, and ρair is the density of air.

3.3.3. Aerosol Characterization using APS Aerosol sizing was performed using an APS spectrometer (Model # 3321, TSI Inc.). An APS sizes aerosol particles by directing flow through a nozzle past two parallel lasers, where scattering events lead to measurements of a residence time; this residence time was calibrated to accurately size spherical aerosols of unit density. Particles were dispersed from methanol using a collision jet-nebulizer (CJN, MRE 1-jet, BGI Inc.) which was connected to the APS through four feet of tubing which acted as a drying column. Due to the propensity for plastic tubing to charge aerosols through triboelectrification, glass tubing was chosen to minimize electrostatic charging of the aerosols and care was taken to ground the tubing.9 A controlled leak was placed into the tubing to allow additional air into the column, ensuring conditions for the methanol to evaporate and resulting in methanol-free, monodisperse particles entering the APS. The APS sampled the nebulized aerosol for 5 minutes and readings of the number median aerodynamic diameter (NMAD) and GSD were recorded.

3.3.4. Aerosol Characterization using an Andersen Cascade Impactor (ACI) An Andersen cascade impactor (ACI, ThermoScientific) was used to size powder aerosol samples. Samples were dispersed by three different devices to evaluate the effectiveness in creating aerosols. A PennCentury insufflator device and a volume-calibrated hand pump (PennCentury Inc.) were used to disperse dry powder samples, with typical fill weights of 2-5 mg. A Monodose inhaler device (Mod. 8, Plastiape SpA) was also used during ACI sizing, with 10 mg of dry powder samples loaded into hydroxypropyl methylcellulose (HPMC) capsules (Plastiape SpA). Finally, a CJN was employed as a control method for ensuring the creation of monodisperse aerosols.

111 ACI measurements were performed following standard protocol.38 Prior to testing, collection plates were coated with polyethylene glycol 300 and, following assembly, the flow rate was set and ambient conditions recorded. To avoid charging, the ACI was carefully grounded. The dispersion device was then attached to the ACI throat and tested at 28.3 L/min for 8 seconds. Deposited particles were collected from the device, throat, collection stages, and filter using a known volume of water. Polymer particles were analyzed through fluorescence spectrometry (SpectraMax M5 plate-reader) to obtain fluorescence intensities from the rhodamine B dye in the particle matrix. This corresponded directly to the relative mass deposited on each plate; from these, mass median aerodynamic diameters (MMADs) were calculated as the midpoint diameter (d50) from the cumulative fraction distribution. In a lognormal population, the geometric standard deviation (GSD) can be calculated from ratios of various diameters of the cumulative fraction distribution:

( ) (8)

These expressions are equivalent for an ideal aerosol with a lognormal distribution, which is not true of aerosols with different distributions. As the aerosols studied here are not expected to be lognormally distributed, reported GSDs were consistently calculated using the first 9 expression: σg = d84 / d50.

3.3.5 APS Model for Shape Factor Calculation As mentioned previously, the APS operates under non-Stokesian flow conditions. The correction for shaped particles, as derived in Eq. 5, is not taken into account during APS measurements and, as such, the APS incorrectly sizes non-spherical particles. 9, 14, 24-26 For a particle of both unknown size and unknown shape, it is difficult to correct for this underestimation. However, with a uniformly shaped particle of known dimensions, such as the particles made with PRINT, particle volume can be established. Measured APS results were corrected and the degree of underestimation from these results was used to determine particle shape factors. Following similar adjustments determined from previous work, we

112 extended an appropriate model to account solely for the deviation of particle shape and were able to establish shape factors for PRINT geometries. The drag experienced by a shaped particle was shown in Eq. 5. Applying Newton’s second law, the following expression for particle velocity in the nozzle was established.19, 34

⁄ (9)

This again assumes a Re less than 100. Assuming inviscid, isothermal, compressible flow and using the geometry of the APS nozzle, an expression for the axial profile for air velocity through the nozzle, U, was applied.34 This background profile is shown in Figure 3.2.

Figure 3.2. Theoretical velocity profiles are shown as a function of position through the APS nozzle. The grey curve represents the background air profile. Curves for particle velocity were calculated using E8 for a sphere (solid black) and particles with a shape factor equal to two with an equivalent volume (short dash) and equivalent aerodynamic diameter (long dash) to the spherical particle.33 Particle diameters are measured in μm.

Eq. 9 was solved numerically to generate a particle velocity profile through the APS nozzle as a function of physical parameters.39 The calculated velocity within the APS nozzle for three different particle types is also shown in Figure 3.2: a sphere of unit density and two

113 non-spherical particles, each with a χ of 2 but with different volumes. This figure highlights the error inherent to the APS; not only does a particle of different χ but equivalent DEV have a markedly different velocity profile in the nozzle, but a particle of equivalent DAE also experiences a different velocity profile due to the non-Stokesian flow regime. This deviation accounts for the underestimation attributed to the APS sizing of non-spherical particles. In the numerical approach, two timing lasers were defined at position 2.2 and 2.323 mm based on the APS nozzle geometry. The residence time between the two lasers was determined by integrating the particle velocity over the laser positions.34

∫ (10)

The residence time, T, for spheres of unity density was then calculated to establish the base calibration used by the APS; these residence time measurements are used by the APS to define particle size. This calibration is shown by the solid curve (χ = 1) in Figure 3.3.

Figure 3.3. Calculated residence times for particles of different sizes with varied shape factor. Curves for particle residence time were calculated using E8 and E9.33 Particles of greater shape factor traveled with greater velocity through the timing region and this had a smaller residence time.

114 However, varying the particle physical parameters, such as χ or ρ, has a direct influence on

APS residence time. Eq. 9 can be used to predict T as a function of χ, DEV and ρ. Figure 3.3 illustrates the residence time of non-spherical particles as a function of increasing χ over a range of DEV. Particles of a larger χ will travel through the timing region at a higher velocity (as in Fig. 3.2) and will obtain a smaller residence time (as in Fig. 3.3). A few important points highlighting APS sizing errors emerge from Figure 3.3. The deviation in residence time between a spherical and non-spherical particle becomes increasingly pronounced for larger particle volumes (DEV). Figure 3.3 also emphasizes that a given residence time can correspond to numerous particle geometries, contrary to the APS analytics. In a typical system of unknown particle size, shape and density, this oversight cannot be easily corrected. With particles of uniform size, shape and density, however, the only unknown is particle shape factor. In tandem with experimental measurements, particle shape factors were determined by correcting APS measurements. From the values of residence time of unit density spheres, a calibration curve predicting DAPS was generated, as shown in Figure 3.3. Accordingly, for

DAPS values measured experimentally from monodisperse aerosols, a corresponding residence time (TexpAPS) was determined. Assuming a known DEV from SEM measurements and known particle density from bulk sample measurements, shape factor was the remaining unknown in Eq. 9 and Eq. 10. Solving numerically and iterating over χ values to fit calculated residence 39 time to TexpAPS, shape factors (χAPS) were determined for each geometry.

3.3.6. Shape Factor Validations

To validate the χAPS approach, shape factors for the non-spherical PRINT geometries presented here were compared to existing approaches in the literature. Three commonly used approaches were selected to validate the APS shape factor determinations: 1) use of a sedimentation tank, 2) calculation of shape factor using projected and surface area relations, and 3) calculation of shape factor by approximating geometries of PRINT aerosols to an oblate.

115 Experimental determination of shape factor in a sedimentation tank (χSed) was performed following procedures similar to those in the literature.15, 27, 28 A clear polycarbonate tank (diameter 20 cm height 30 cm) was filled with high viscosity synthetic air compressor oil ISO grade 220, SAE grade 50 “Chemlube822” (UltraChem Inc) so that Re was maintained less than 0.1. Macroscale models of PRINT aerosols were created using a rapid prototyping method and were scaled two thousand times the original dimensions (volumes ~0.25 cm3). Fluid viscosity was calibrated as a function of temperature throughout the experiment using polystyrene spheres (diameters ranging from 0.1 to 0.9 cm), which were timed over a drop of 15 cm.27 Spheres and macroscale models were released ~5 cm beneath surface of the fluid and in the center of the tank to neglect edge effects. Model parts were released from a controlled orientation and timed over a 15 cm drop. Each trial was replicated ten times for each starting orientation, and then replicated over five copies of each model part. Initial release included a horizontal position (Exp-H) to maximize the surface area facing the bottom of the tank and a vertical position (Exp-V) to minimize the surface area facing the bottom of the tank. Using Eq. 3, VTS was determined by dividing the known fall distance over time. Assuming a CC of one, shape factor (χSed) values were computed. An additional approach existing in the literature for shape factor determination involves a relationship between the projected area and surface area.18 This approach makes use of two additional equivalent spheres; for each non-spherical particle, a sphere of equivalent surface area and equivalent projected area is defined in addition to those shown in Figure 3.1. Empirical relationships have been established to relate these four equivalent spheres to calculate shape factor, which fit to previously determined shape factors in the literature. Calculations for shape factor (χPSA) of PRINT geometries were evaluated by assuming the maximum projected area of a given particle, the equivalent to orientation Exp- H in the settling tank experiment. Using SEM measurements of particle geometries, equivalent spheres were established and χPSA calculated using the expressions derived by Leith.18 The final comparison of shape factor determination was performed by approximating 13 each shape as an oblate (χOblate). Empirical expressions exist for the drag on oblate

116 geometries; a common approach in the literature is to assume oblate dimensions which can encase non-spherical particles in order to apply these expressions. This approach is somewhat similar to the use of equivalent spheres, but is instead relating to an equivalent oblate. Calculations for χOblate for PRINT particles were performed by assuming dimensions of an oblate which completely encapsulated the PRINT particle dimensions. PRINT particle dimensions were determined through SEM measurements and χOblate were evaluated for two orientations of the encapsulating oblate using expressions by Loth.13

3.4. Results and Discussion 3.4.1. Particle Fabrication Particles were fabricated using the PRINT molding platform, yielding high fidelity particles. A total of fourteen shaped particles were investigated in these studies; representative SEM micrographs of these shapes are shown in Figure 3.4. The most basic of these shapes was a cylinder and a two dimensional ellipsoid. Increasing in shape complexity, two overlaid ellipsoids comprised the Lorenz particle design. A series of pollen-mimic particles was inspired to replicate the highly dispersible properties of pollen spores. Figure 3.4A shows the cylinder, ellipsoid, Lorenz particles and the pollen-mimic series. To investigate the effect of fenestrations or macrosized pores, a series of three toroids of increasing diameter, as well as a hexagon with a circular cut-out dubbed “hexnut”, were fabricated. Finally, a “ball-and-stick” family of four shapes, referred to as a lollipop, v- boomerang, l-dumbbell and helicopter, was fabricated to explore effects of asymmetry.

117

Figure 3.4. SEM micrographs of PRINT microparticles. (A) Curvature series and Pollen- mimic series: A1) Cylinder, A2) Ellipsoid, A3) Lorenz, A4) Small Pollen, A5) Medium Pollen, A6) Large Pollen. (B) Fenestrated Series: B1) Small Toroid, B2) Medium Toroid, B3) Large Toroid, B4) Hexnut. (C) Ball and Stick family: C1) V-Boomerang, C2) Lollipop, C3) Helicopter, C4) L-Dumbbell.

Figure 3.4B shows the fenestrated series and Figure 3.4C shows the ball-and-stick family. Measurements from these and similar micrographs produced particle dimensions. These are shown in Table 3.1, along with calculated values for DEV and surface area. Length and width measurements listed represent the maximum dimension needed to enclose the particle in a solid oblate.

118 Table 3.1. Measured and tabulated particle characteristics. Physical dimensions (length, width and height) for each geometry obtained through SEM measurements are shown, with calculated values for surface area and DEV. Aerosol properties are also shown, with APS sizing results for particles dispersed from a CJN and DAE corrected using the APS model. GEOMETRIC DIMENSIONS AEROSOL PROPERTIES APS Sizing Length Width Height Surface DEV Corrected 2 Name (μm) (μm) (μm) Area (μm ) (μm) NMAD (μm) GSD DAE (μm) Cylinder 1.21 1.21 0.78 5.29 1.20 1.262 1.079 1.26 Ellipsoid 6.49 3.62 0.97 52.53 3.24 2.679 1.070 2.88 Lorenz 5.98 5.98 0.97 63.54 3.37 2.452 1.075 2.69 Small Pollen 1.45 1.45 0.60 4.69 1.02 1.120 1.121 1.11 Medium Pollen 3.39 3.39 0.35 17.51 1.63 1.215 1.201 1.27 Large Pollen 10.90 10.90 0.43 117.2 3.48 2.163 1.109 2.22 Small Toroid 1.26 1.26 0.74 6.25 1.13 1.111 1.100 1.12 Medium Toroid 3.36 0.36 0.87 28.09 2.41 2.088 1.066 2.19 Large Toroid 6.36 6.37 1.93 108.07 4.73 3.529 1.071 3.98 Hexnut 2.79 2.79 0.95 15.01 1.88 1.636 1.089 1.70 Lollipop 7.16 3.91 0.95 48.91 3.04 2.590 1.076 2.75 V-Boomerang 10.05 3.62 0.95 48.25 2.85 2.073 1.103 2.24 L-Dumbbell 6.19 4.05 0.88 41.16 2.71 2.184 1.073 2.33 Helicopter 6.11 6.11 0.88 47.57 2.74 1.978 1.103 2.14

Bulk density measurements established the sample density to be 1.17 g/cm3. This value was assumed for the density of PRINT particles.

3.4.2. Aerosol Characterization Shaped PRINT particles were aerosolized with a CJN and sized with an APS. The CJN nebulizes a solution through a single jet, which collides with the chamber wall. This decreases the resultant liquid aerosol droplet size to 2.5 μm with a GSD of 1.8, as reported by the manufacturer.40 For our purposes, the liquid droplet features did not control final aerosol properties, as the methanol droplet completely evaporated to liberate particles. To confirm this, a simple calculation was performed to determine the rate of evaporation of a pure methanol droplet in dry conditions; the calculated lifetime of the droplet was ~3 milliseconds.9 The lengthy drying column and placement of controlled leaks in the tubing ensured favorable conditions for evaporation. By adding a dilute suspension of particles in

119 the liquid reservoir of the CJN, single particles were encapsulated in the emitted liquid droplets, which resulted in the creation of a monodisperse aerosol upon methanol evaporation. A complete listing of APS results is shown in Table 3.1, with representative results for three toroid particles plotted in Figure 3.5. For all shaped particles tested, aerosol samples were highly monodisperse, yielding narrow distributions. As demonstrated in Fig. 3.5, slight variations in geometry resulted in finely tuned NMADs. Reported GSDs for all shapes were less than 1.20, indicating that all samples displayed monodisperse characteristics in aerosol form. Even highly non-spherical particles, such as the large pollen-mimic, resulted in a GSD of less than 1.22. This characterization demonstrated that under proper dispersal conditions, PRINT particles can result in truly monodisperse aerosols.

Figure 3.5. APS number results for fenestrated toroid series.

Additionally, two PRINT particle types were aerosolized using three aerosol devices and sized with an ACI. ACI measurements are the gold standard for obtaining aerodynamic diameters, especially in the pharmaceutical industry, as results can directly correlate to lung deposition.41 Figure 3.6A and B show ACI mass distributions of small and large pollen- mimic aerosol samples, with MMAD and GSD reported in Table 3.2. Differences in

120 dispersion device greatly impacted the quality of the aerosol sample; these devices will be discussed further following shape factor results.

Figure 3.6. ACI distributions plotted as % deposition on each stage for (A) small pollen- mimic and (B) large pollen-mimic powder samples, n=4.

Table 3.2. ACI results for small and large pollen-mimic powder samples, dispersed from three different device types. CJN PennCentury Monodose DAE Shape DEV Predicted MMAD GSD MMAD GSD MMAD GSD Small Pollen 1.92 1.11 1.73 1.18 2.45 1.79 2.29 1.49 Large Pollen 3.48 2.22 4.35 1.23 4.72 1.32 4.34 1.32

3.4.3. Shape Factor Determination with APS Model

The APS numerical approach was utilized to determine χAPS and correct DAE from

APS measurements. Corrected DAE are shown in Table 3.1 for each geometry. In Figure 3.7, measured values of DAPS are plotted against corrected DAE, emphasizing the importance of correcting APS values for non-spherical aerosols. Also included in Figure 3.7 are two reference curves predicted by this model: the relation of DAPS with DAE for 1) a sphere, which, by APS calibration has a slope of 1 and 2) a non-spherical particle with χ of 2. These curves were generated using the APS model over a range of particle sizes. The deviation

121 between DAE and DAPS increased as the particle volume increased, agreeing with previous results.19, 32, 34, 35 Non-spherical PRINT particles fall within a range of the two reference curves, illustrating the difference between the APS measurement and the corrected DAE.

Figure 3.7. Relationship between experimentally measured DAPS values plotted against corrected DAE.

This approach uniquely utilizes APS measurements to determine shape factors for a particle of known volume. Values for χAPS ranged from slightly over one, to a maximum of

2.88 for the large pollen-mimic particles; χAPS for all shapes are reported in Table 3.3. The smallest deviation between DAPS and DAE was observed for the cylinder particle, while the smallest reported shape factor was predicted for the small pollen-mimic. The size series of toroids and pollen-mimic particles exhibited increasing χAPS values with increasing particle diameter. Here, increase in overall volume emphasizes drag differences due to shape. The fenestrated design feature, such as those of the toroid series, resulted in increased particle

χAPS as compared to a cylinder. However, intentional mass asymmetries from the placement of unequal arm lengths, such as the lollipop and l-dumbbell, did not result in increased χAPS.

122 More symmetric analogs, such as the helicopter and v-boomerang, were determined to have a larger χAPS.

Table 3.3. Shape factor results determined from APS method, compared to results determined from other literature methods: the PSA method, oblate approximations, and tank sedimentation. χ Sedimentation χ Sedimentation Shape χ APS χ PSA χ Oblate -H χ Oblate -V Exp-H Exp-V

Orientation Averaged

Cylinder 1.09 - - 1.06 1.08 1.00 Ellipsoid 1.48 1.31 1.00 1.34 1.61 1.20 Lorenz 1.82 1.31 1.16 1.40 1.57 1.18 Small Pollen 1.03 - - 1.15 1.20 1.01 Medium Pollen 1.91 - - 1.55 1.82 1.31 Large Pollen 2.88 1.60 1.21 1.94 2.49 1.71 Small Toroid 1.22 - - 1.17 1.10 1.00 Medium Toroid 1.44 1.05 1.00 1.28 1.36 1.08 Large Toroid 1.69 1.44 1.21 1.25 1.30 1.05 Hexnut 1.46 - - 1.16 1.26 1.04 Lollipop 1.43 - - 1.35 1.68 1.23 V-Boomerang 1.87 1.70 1.52 1.39 1.87 1.34 L-Dumbbell 1.57 1.52 1.23 1.37 1.64 1.21 Helicopter 1.91 2.01 1.69 1.43 1.63 1.21

An increased shape factor offers unique opportunities specifically for respiratory drug delivery. A large shape factor will reduce the DAE while maintaining a certain volume. Larger volume particles can deliver an increased therapeutic payload, leading to increased efficacy and dose sparing, while particles of a smaller DAE can penetrate deeper into the lung parenchyma at higher efficiencies.11 This has been established using spray dried, large porous 42 particles where the particle density modulates the DAE. Here, particle shape is instead implemented to modulate DAE through enhancement of particle shape factor. Controlled geometry has other potential implications for respiratory drug delivery, as the interaction between the particle and the local lung environment will be heavily dependent on particle

123 geometry. Particles with large shape factors will be capable of delivering a higher therapeutic deeper into the lung as compared to a spherical particle. From these results, flat, symmetric designs and fenestrated features contributed to increased shape factor; future designs combining these two qualities may yield even more sizeable shape factor predictions. To validate these findings for shape factor on this set of particle geometries, comparisons can be made to the literature. There is a limited set of accepted shape factors for specific particles; these shape factors have generally been established for regular geometries and dust particles using sedimentation experiments.9, 14-17 From this limited collection, some correlations can be made to APS numerical model results. In the literature, the dynamic shape factor for a cylinder with an aspect ratio of 2 is reported to range from 1.01 (vertical orientation) to 1.14 (horizontal), with the averaged orientation 1.09. This averaged value matches exactly with the shape factor reported by the APS model. Further comparisons to shape factors reported in the literature can be made between talc powder and the symmetric, plate-like particles, such as the large pollen-mimic and helicopter geometries. Talc particles are thin symmetric plates with a high aspect ratio and large surface area, resulting in reported experimental shape factors as high as 2.3. Additionally, a plate-like geometry has been theorized to yield shape factors approaching 3 for certain orientations.14 These values correlate well with the large shape factor value of 2.88 for the large pollen-mimic geometry, which closely resembles a flat plate. In many cases, direct comparisons to existing shape factor values cannot be performed due to the lack of existing data; alternative approximations must be taken to determine shape factors of a given geometry. To further validate the χAPS approach, shape factors for the novel shapes presented here were compared to three existing approaches: 1) use of a sedimentation tank (χSed), 2) calculation of shape factor using projected and surface area relations (χPSA), and 3) calculation of shape factor by approximating geometries of

PRINT aerosols to an oblate (χOblate). Results from the three approaches are compared to those of the APS model in Table 3.3. As is readily apparent from Table 3.3, these four approaches yield variable results for shape factors of non-spherical PRINT particles. Shape factors determined through the sedimentation tank and oblate approximation should reflect

124 the maximum (Exp-H, Oblate-H) and minimum (Exp-V, Oblate-V) values for shape factors. Generally, the sedimentation tank is considered the gold standard as actual geometries are considered. APS model shape factors from a dynamic flow were anticipated to fall somewhere between these two extremes. As anticipated, χAPS were larger than the values for all χSed and χOblate in the vertical drag Exp-V/Oblate-V orientations; however, χAPS was also generally larger than those from horizontal Exp-H/Oblate-H orientations. Similarly, χAPS values were also generally larger in magnitude than χPSA. The largest disagreement between

χAPS and the three other approaches was always observed for the large pollen-mimic geometry, with the fenestrated particles (toroids and hexnut) and the Lorenz particles also consistently larger than the three literature predictions. However, this model did agree very well with simpler particle geometries. This agreement was especially true for cylindrical shaped particles, the most spherical shape tested, in which χPSA and χOblate fell within 3% of

χAPS. The slight overestimation of the APS model might stem from a few sources. It is worthwhile to consider limitations of the mathematical expressions used for the model; the expression for sphericity in Eq. 6 has only been extended for shape factors ranging from 1 to 1.5, a range which is often exceeded with these highly non-spherical particles. Adding further complications, the particles tested here likely exhibited tumbling and rotational properties by design, which imply behaviors not represented in a single shape factor. Rotation of particles, especially non-spherical ones, tremendously influences particle aerodynamic properties and their behavior under flow.10 Furthermore, dynamic shape factors do not take into account particle orientation in the flow stream. These are two major considerations which are not included in any shape factor classification. Indeed, the standard approaches for determining particle shape factor used as validation here also fail to capture dynamic behavior. Values of shape factor determined by each of the three commonly implemented methodologies also do not account for rotation of particles in flow. For example, the simplification of particles to an oblate approximation based on the maximum length, width, and height of the particle removes any possible effect of mass asymmetry and will fail to capture realistic particle motion of the true geometry. Additionally, these three literature approaches all relate to

125 macroscale models, rather than true aerosols; χPSA and χOblate were derived to fit χSed literature results. Poor agreement for shapes which are increasingly non-spherical, such as the pollen- mimic and toriods that have geometrical design features such as fenestrations and high aspect ratio, contributed to the hypothesis that these asymmetric PRINT particles were exhibiting characteristics which cannot be well represented with current approaches of shape factor. In spite of these discrepancies, characterization of shape factor still holds value as a characterization parameter for non-spherical PRINT aerosols. The rapid conversion to aerodynamic diameter and the ability to harness knowledge of particle deposition, especially for respiratory delivery, makes shape factor a powerful tool. Indeed, knowledge of particle geometries which are no longer well described by shape factors will aid in additional research focus and design parameters. Future studies into particle dynamics of unique geometries will be required to adequately model the individual features of each design. Computational approaches using a Lagrangian framework and solid body dynamics can provide insight to the complicated relations between particle geometry and surrounding flow; an accurate model of such a system would consider fluid-particle coupling and the role of torque and lift on a single particle.10, 43 This more complex approach would account for these potentially considerable influences and support the hypothesis that non-spherical particles presented here are exhibiting interesting dynamic behavior. Such a model would also provide guidelines of particle geometries where a reduction to a single shape factor value would be no longer useful. Despite the limitations of using a dynamic shape factor, the rapid screening approach presented here yielded reasonable and descriptive shape factors for these novel shapes and likely captured more of the dynamic behavior of a true aerosol under flow than standard approaches used for comparison. Fitting to experimental APS results is also preferable to the tedious experiments involved with a sedimentation tank. Overall, the model presented here for determining shape factors by correcting APS results agreed reasonably with existing approaches, with generally only slight overestimation of dynamic shape factors for most geometries. The agreement between χAPS and literature values, χPSA, and χOblate for the cylinder validate the APS numerical approach and allowed for classification of these non-spherical PRINT particles.

126 3.4.4. Application of Determined Shape Factors Determining shape factors for non-spherical particles allowed for the prediction of

DAE for a given particle geometry. This is valuable when evaluating more realistic powder samples, such as the small and large pollen-mimic samples aerosolized in Figure 3.6 and Table 3.2. Dispersing dry powders and overcoming aggregation for particles of this size range remains a challenge for formulation scientists; electrostatics and van der Waals forces play a significant role in powder agglomeration. To assess which device more adequately dispersed dry powder samples, sizing results of these true dry powder aerosols were compared to the theoretical prediction of the same shaped monodisperse particle. In combination with ACI sizing, the various aerosol dispersion devices resulted in aerosol populations with larger MMADs than the predicted monodisperse aerosol. This can in part be attributed to the low resolution of the ACI stage cut-off values, highlighting discrepancies between aerosol sizing techniques. From previous APS sizing, the aerosol resulting from the CJN was assumed to be monodisperse; indeed, the large and small pollen-mimic aerosols dispersed from the CJN had the smallest MMADs of the three devices and GSDs of about 1.2. Interestingly, the monodisperse large pollen-mimic aerosol deposited evenly on stages 3 and 4, which contributes to the increased MMAD. Compared to these monodisperse aerosols, samples dispersed from either the Monodose inhaler or the PennCentury exhibited increased signs of aggregation. Both devices resulted in larger MMADs than predicted for the two particle geometries; of the three, the MMAD increase for aerosol samples created by the PennCentury was the most substantial. However, deposition on earlier stages and the ACI throat was indicative of larger agglomerates present in aerosols from both devices. These trends were apparent in the distribution presented in Figure 3.6 and supported by the elevated MMAD, but were not necessarily apparent through comparison of each sample’s GSD, the typical indicator of aggregation. Comparison to the predicted monodisperse DAE for each particle geometry allowed for more detailed evaluation of these devices by providing a theoretical reference. Defining a deviation from the theoretical monodisperse particle lends qualitative insight as to the nature of the aerosols produced. This will be especially useful in comparing

127 dispersion of non-spherical aerosols and predicting deposition. Future compositions of the same physical particle geometry can be evaluated against the theoretical monodisperse value. Establishing baseline features of individual particle dynamics, including properties such as

DAE, is critical to probing more complicated processes of realistic samples. The ACI results of dry powder PRINT samples illustrated that agglomerates were still present, but also that a large fraction of both samples behaved as individual particles, indicated by deposition on the same stage as the monodispersed aerosol. Establishing the theoretical monodisperse DAE allows for differentiation of the dose fraction comprised of individual particles. Deagglomeration under dynamic conditions is representative of typical respiratory doses; larger agglomerates will impact in the throat and liberated individual particles will deposit in the lung as predicted by DAE. Typical ACI sizing which does not compare to a monodisperse sample artificially inflates the MMAD and fails to identify the fraction of the dose which deposits in the region of interest. While initial dispersion from the inhaler device is important, deposition in the target lung region will be achieved by individual particles. Following deposition, the rate of mucocilliary clearance, cellular interactions, macrophage uptake, and drug release kinetics will also be a function of individual particle properties. As such, the physical, chemical, and aerodynamic characterization of individual particles holds tremendous value. In this context, our method for determining shape factors of non-spherical novel geometries provides a rapid approach to better characterize well-defined aerosol particles. This methodology further can be applied to any aerosol of controlled particle geometry, as only DEV of a uniform population is required.

3.5. Conclusions The work presented here introduces a class of novel shaped aerosols; using PRINT technology, truly monodisperse shaped aerosols were generated. From this complete control over particle uniformity resulting in a known particle size, a correction to APS sizing techniques was employed to characterize particle shape factors for this series of particles. This approach yielded shape factors in reasonable agreement with the limited approaches

128 present in the literature. Limitations of using particle shape factor as a descriptive parameter were also discussed, which will be critical in further design approaches to non-spherical aerosols. As more complex geometries of aerosol particles are explored, additional characterization parameters will be required to accurately represent dynamic behavior. However, modifying the APS use as described enables a rapid classification technique for determining useful and translatable features of shaped particles, which can be implemented for respiratory delivery indications. Classifying shaped particles with a dynamic shape factor allows for a quick calculation of a theoretical DAE, deviation from which lends insight to sources of particle aggregation when aerosolized from standard inhalers. Implementing designed non-spherical particles for respiratory drug delivery offers unique aerodynamic properties; increasing shape factor will lead to decreased DAE and deeper lung deposition. Improving our analysis of this set of non-spherical PRINT aerosols enhances their applicability to address increasingly complicated questions in aerosol and respiratory topics. Controlling particle shape of a dry powder formulation allows for the opportunity to create monodisperse shaped aerosols with potential to systematically optimize the effect of shape on the many stages of drug delivery. Ideally, a priori knowledge of particle behaviors of a given formulation and dispersion method could be used to design a controlled therapeutic response. Future directions for our work include addressing how particle shape can mitigate powder agglomeration and aid in appropriate cellular delivery. We also intend to explore advanced modeling approaches to better describe the role of geometry in particle aerodynamics in the lung. Overall, this approach for predicting and characterizing particle shape factor on calibration quality particles extends our knowledge of particle interactions as aerosols and may ultimately be used to better guide the design and engineering of particles for pulmonary delivery.

3.6. Acknowledgements We thank Joe Pedit, Maryanne Boundy and David Leith for sharing their APS and for many helpful discussions of aerosol sizing, Steve Emanuel for fabrication of macroscale

129 model parts, the CHANL facility at UNC for use of the SEM, Stuart Dunn and John Fain, Sorin Mitran, and Katherine Phillips for useful discussions regarding PRINT particle fabrication, rigid body dynamics, and MATLAB respectively, and for and Liquidia Technologies for providing molds. This work was supported by Liquidia Technologies and the National Institutes of Health Director’s Pioneer Award (1DP1OD006432).

3.7. References 1. Weers JG, Bell J, Chan HK, Cipolla D, Dunbar C, Hickey AJ, Smith IJ. Pulmonary formulations: What remains to be done? J Aerosol Med Pulm Drug Deliv. 2010, 23 Suppl 2, S5-23.

2. Daniher DI, Zhu J. Dry powder platform for pulmonary drug delivery. Particuology. 2008, 6, 225-238.

3. Calvert G, Ghadiri M, Tweedie R. Aerodynamic dispersion of cohesive powders: A review of understanding and technology. Advanced Powder Technology. 2009, 20, 4-16.

4. Kleinstreuer C, Zhang Z, Donohue JF. Targeted drug-aerosol delivery in the human respiratory system. Annu Rev Biomed Eng. 2008, 10, 195-220.

5. Hassan MS, Lau RW. Effect of particle shape on dry particle inhalation: Study of flowability, aerosolization, and deposition properties. AAPS PharmSciTech. 2009, 10, 1252-1262.

6. Garcia-Verdugo I, Descamps D, Chignard M, Touqui L, Sallenave JM. Lung protease/anti-protease network and modulation of mucus production and surfactant activity. Biochimie. 2010, 92, 1608-1617.

7. Usmani OS, Biddiscombe MF, Barnes PJ. Regional lung deposition and bronchodilator response as a function of beta2-agonist particle size. Am J Respir Crit Care Med. 2005, 172, 1497-1504.

8. Brand P, Meyer T, Sommerer K, Weber N, Scheuch G. Alveolar deposition of monodisperse aerosol particles in the lung of patients with chronic obstructive pulmonary disease. Experimental Lung Research. 2002, 28, 39-54.

130 9. Hinds WC. Aerosol technology: Properties, behavoir and measurement of airborne particles: Wiley-Interscience; 1999.

10. Zastawny M, Mallouppas G, Zhao F, van Wachem B. Derivation of drag and lift force and torque coefficients for non-spherical particles in flows. International Journal of Multiphase Flow. 2012, 39, 227-239.

11. Crowder TM, Rosati JA, Schroeter JD, Hickey AJ, Martonen TB. Fundamental effects of particle morphology on lung delivery: Predictions of stokes' law and the particlular relevance to dry powder inhaler formulation and development. Pharm Res. 2002, 19, 239-245.

12. Lippmann M. Size-selective health hazard sampling. In:Air sampling instruments for evaluation of atmospheric contaminants, Cincinnati: ACGIH, 1995.

13. Loth E. Drag of non-spherical solid particles of regular and irregular shape. Powder Technology. 2008, 182, 342-353.

14. Cheng Y-S, Yeh H-C, Allen MD. Dynamic shape factor of a plate-like particle. Aerosol Science and Technology. 1988, 8, 109-123.

15. Johnson DL, Leith D, Reist PC. Drag on non-spherical, orthotropic, aerosol particles. J Aerosol Sci. 1987, 18, 87-97.

16. Cheng Y-S. Drag forces on nonspherical aerosol particles. Chemical Engineering Communications. 1991, 108, 201-223.

17. Davies CN. Particle-fluid interaction. J Aerosol Sci. 1979, 10, 477-513.

18. Leith D. Drag on nonspherical objects. Aerosol Science and Technology. 1987, 6, 153- 161.

19. Brockmann JE, Rader DJ. Aps response to nonspherical particles and experimental determination of dynamic shape factor. Aerosol Science and Technology. 1990, 13, 162- 172.

20. Geiser M, Schurch S, Gehr P. Influence of surface chemistry and topography of particles on their immersion into the lung's surface-lining layer. J Appl Physiol. 2003, 94, 1793- 1801.

131 21. Champion JA, Mitragotri S. Shape induced inhibition of phagocytosis of polymer particles. Pharm Res. 2009, 26, 244-249.

22. Gratton SE, Ropp PA, Pohlhaus PD, Luft JC, Madden VJ, Napier ME, DeSimone JM. The effect of particle design on cellular internalization pathways. Proc Natl Acad Sci U S A. 2008, 105, 11613-11618.

23. Patton JS, Byron PR. Inhaling medicines: Delivering drugs to the body through the lungs. Nat Rev Drug Discov. 2007, 6, 67-74.

24. Rolland JP, Maynor BW, Euliss LE, Exner AE, Denison GM, DeSimone JM. Direct fabrication and harvesting of monodisperse, shape-specific nanobiomaterials. J Am Chem Soc. 2005, 127, 10096-10100.

25. Kelly JY, DeSimone JM. Shape-specific, monodisperse nano-molding of protein particles. J Am Chem Soc. 2008, 130, 5438-5439.

26. Garcia A, Mack P, Williams S, Fromen C, Shen T, Tully J, Pillai J, Kuehl P, Napier M, Desimone JM, Maynor BW. Microfabricated engineered particle systems for respiratory drug delivery and other pharmaceutical applications. J Drug Deliv. 2012, 2012, 941243.

27. MacLennan ICM. Germinal centers. Annu Rev Immunol. 1994, 12, 117-139.

28. Kasper G, Niida T, Yang M. Measurements of viscous drag on cylinders and chains of spheres with aspect ratios between 2 and 50. Journal of Aerosol Science. 1985, 16, 535- 556.

29. Baron PA. Calibration and use of the aerodynamic particle sizer (aps 3300). Aerosol Science and Technology. 1986, 5, 55-67.

30. Wang H-C, John W. A simple iteration procedure to correct for the density effect in the aerodynamic particle sizer. Aerosol Science and Technology. 1989, 10, 501-505.

31. Tsai C-J, Chen S-C, Huang C-H, Chen D-R. A universal calibration curve for the tsi aerodynamic particle sizer. Aerosol Science and Technology. 2004, 38, 467-474.

32. Marshall IA, Mitchell JP, Griffiths WD. The behaviour of regular-shaped non-spherical particles in a tsa aerodynamic particle sizer. Journal of Aerosol Science. 1991, 22, 73-89.

132 33. Cheng YS, Chen BT, Yeh HC. Behaviour of isometric nonsperical aerosol particles in the aerodynamic particle sizer. Journal of Aerosol Science. 1990, 21, 701-710.

34. Cone RA. Barrier properties of mucus. Adv Drug Deliv Rev. 2009, 61, 75-85.

35. Cheng YS, Chen BT, Yeh HC, Marshall IA, Mitchell JP, Griffiths WD. Behavior of compact nonspherical particles in the tsi aerodynamic particle sizer model aps33b: Ultra- stokesian drag forces. Aerosol Science and Technology. 1993, 19, 255-267.

36. Rasband WS. Imagej. In: Health USNIo, ed. Bethesda, MD, 1997-2012.

37. Density determination of solids. In:Instruction manual for mettler-toledo density kit: Mettler-Toledo, 2002.

38. Chapter 601-physical tests and determinations: Aerosols usp 26-nf 21. 2003, 2105-2123.

39. Matlab. In. Natick, MA: The MathWorks Inc, 2003.

40. BGI Incorporated. Collision nebulizer droplet number output. Available at: http://www.bgiusa.com/agc/droplet_number_output.htm. Accessed: July 1 2012, 2012.

41. Thiel CG. Can in vitro particle size measurements be used to predict pulmonary deposition of aerosol from inhalers? Journal of Aerosol Medicine. 1998, 11, S-43 - S-52.

42. Edwards DA, Hanes J, Caponetti G, Hrkach J, Ben-Jebria A, Eskew ML, Mintzes J, Deaver D, Lotan N, Langer R. Large porous particles for pulmonary drug delivery. Science. 1997, 276, 1868-1971.

43. Tu J, Kiao I, Ahmadi G. Computational fluid and particle dynamics in the human respiratory system: Springer; 2013.

133

CHAPTER FOUR

Cellular Fate of Nanoparticles in Lung APCs for Pulmonary Vaccines

Fromen, Robbins, Shen, Kai, Rahhal, Luft, Ting, DeSimone, in preparation

134 4.1. Introduction The next generation of vaccines can be achieved by pulmonary delivery of precision- engineered particles.1-11 Engineered micro- and nanoparticles (NP) offer elegant solutions for pathogen mimicry, while providing increased safety and efficacy over current vaccine strategies.3, 4, 9 Additionally, engineered particles can be designed such that aerosol properties, lung deposition and cellular interactions are independently taken into consideration.2, 3 While there is extensive literature describing how physical particle properties can influence aerodynamic diameter and thus pulmonary deposition, including work presented in Chapter Two, there is minimal understanding of how these same particle properties influence interactions with lung cells and their subsequent immune responses.1, 2, 12, 13 In the lung, there are numerous cell types; of particular interest to vaccine design is a specific subset called antigen presenting cells (APCs), which comprise B cells, dendritic cells (DCs) and alveolar macrophages (AM).4, 14-20 Alveolar macrophages are the main phagocytic cell in the lung; they roam the airway epithelium, where they are responsible for internalizing, sequestering, and digesting any foreign material.16, 21-25 While they are generally considered APCs, their main function in the lung is more maintenance and clearance, rather than initiating adaptive responses.16, 23, 25 In contrast, DCs are considered “professional” APCs and are a specialized cell type which act as a sentinel in the lung, monitoring and sampling foreign material to mount adaptive immune responses.4, 19, 20 Lung DCs are responsible for internalizing foreign particulate, digesting and presenting antigen by major histocompatibility complex (MHC) II, migrating to LNs and educating T cells.15-17, 19, 20, 26-28 In the lung, there are two conventional myeloid-derived DC subsets, CD11b and CD103 DCs, which have distinct functions.16, 19, 29 CD103 DCs are found protruding through the epithelial of the lung, are considered the main migratory population, and have been implicated in skewing the lung towards Th1, Th2 and Th17 responses given different stimuli.16, 19, 22, 27-32 CD11b DCs are also known to migrate to the LNs, but also have also been shown to prime IgA production in the lung and are the major producers of soluble protein mediators, chemokines and cytokines.15, 19, 22, 26, 30, 32, 33 These cells are found

135 predominately under the basement membrane in the conducting airways.16 The goal of NP vaccination is to preferentially target these DCs cells, while avoiding the inevitable AM uptake. Nanoparticles have been explored as pulmonary vaccine carriers, due to their potential to diffuse through mucosa, their avoidance of AMs, and their ability to co-deliver both adjuvants and antigens.4, 6, 7, 9, 10, 14, 22, 34, 35 To date, the role of NP charge on lung APC association in the lung remains poorly understood, as the majority of studies have focused on anionic NP vaccine carriers.5, 6, 36 To directly test the effects of NP charge on vaccine responses, we covalently attached the model antigen ovalbumin (OVA) to hydrogel-based NP that varied only in surface charge and otherwise had identical size, shape and antigen loading. We hypothesized that particle charge and protein conjugation would play a critical role in the uptake of lung APCs and subsequent trafficking of nanoparticles to the medistinal LNs. Our results show that cationic NP formulations offer considerable benefits for lung DC association, which suggests that cationic NPs are a desirable candidate platform for pulmonary based vaccine delivery.

4.2. Materials and Methods 4.2.1. Animals All studies were conducted in accordance with National Institutes of Health guidelines for the care and use of laboratory animals and approved by the Institutional Animal Care and Use Committee (IACUC) at UNC. All animals were maintained in pathogen-free facilities at UNC and were between 8 and 15 weeks of age. C57BL/6 were obtained from Jackson Laboratories.

4.2.2. Reagents Solvents and buffers of reagent grade were obtained by Fisher Scientific. PRINT 80 x 320nm molds were provided by Liquidia Technologies. Pre-particle reagents of 2- aminoethylmethacrylate (AEM), poly(ethylene glycol)700 diacrylate (PEG700DA) and

136 diphenyl(2,4,6-trimethylbenzoyl)phosphine oxide (TPO) were obtained from Sigma; tetra(ethylene glycol) monoacrylate (HP4A) was synthesized in house via previously described methods. Particle dye maleimide-Dylight 650 was obtained from Fisher. Coupling reagents for the carboiimide conjugation of 1-ethyl-3-[3-dimethylaminopropyl]carbodiimide hydrochloride (EDC) and N-hydroxysulfosuccinimide (s-NHS) were obtained from Thermo Fisher. CpG-B 1826 oligonucleotide (5′-TCCATGACGTTCCTGACGTT-3′), OVA grade V, 10% FBS, 1xPen/Step, and lipopolysaccharide (LPS) were obtained from Sigma Aldrich.

4.2.3. Particle Fabrication and Characterization Amine-containing 80x320nm hydrogel rod-shaped NPs were fabricated on a continuous roll-to-roll PRINT method as described previously.37 Preparticle solutions contained 1% TPO (photoinitiator), 20% AEM (functional groups), 10% PEG700DA

(crosslinker), 0-1% functional fluorescent dye, 68-9% HP4A (monomer) by weight. OVA functionalization was achieved using water soluble carboiimide chemistry. For (ζ+)NP-OVA, OVA was activated with five molar excess of EDC and s-NHS for 30 mins in 0.1M MES buffer at pH 6, reacted with (ζ+)NP in a 1:100 molar ratio (OVA:NP) for 2hrs at 0.1M sodium phosphate buffer pH 7.5, then washed through centrifugation three times in sterile water. For (ζ-)NP-OVA, (ζ+)NPs in DMF were incubated with 100M excess succinic anhydride for 30 minutes, washed first with borate buffer pH 9.5 and then three additional washes in water to achieve (ζ-)NP. (ζ-)NP were then activated with five molar excess of EDC and s-NHS for 30 mins in 0.1M MES buffer at pH 6, reacted with OVA a 1:85 molar ratio (OVA:NP) for 2hrs at 0.1M sodium phosphate buffer pH 7.5, then washed through centrifugation three times in sterile water. Particle morphology monitored using SEM; samples were sputter-coated with 1-5 nm of Au/Pd (Cressington Scientific Instruments) and imaged (Hitachi model S-4700). Particle size and zeta potential were measured by dynamic light scattering (Zetasizer Nano ZS, Malvern Instruments, Ltd.) and thermogravimetric analysis (TGA) was used to determine particle concentrations. Protein concentration on the NPs was determined by BCA assay (Thermo Fisher, SpectraMax M5 plate-reader), following manufacturer’s instructions.

137 Particle samples were assayed at 1 mg/mL and the amount of bound OVA to NPs was determined using a standard curve of soluble OVA and unreacted particles of equivalent concentration as the absorbance blank. Endotoxin levels were determined using a Limulus Amebocyte Lysate (LAL) Quantification kit (Pierce) following manufacturer’s instructions.

4.2.4. Tissue and Cell Preparation Whole blood was obtained through cardiac puncture and collected in eppendorfs; from these, serum was obtained by centrifugation. Following perfusion of 10mL PBS, BAL was performed, then lung and draining medistinal LNs were collected (Fig. 4.1).

Figure 4.1. Diagram of mouse LNs, with medistinal LNs highlighted in yellow. Reproduced with permission from reference.38

LNs were physically agitated between crushed glass slides and passed through a 70 µm sieve to obtain a single cell suspension38. BAL was performed by inserting a cannula in an incision in the trachea and flushing the lungs with 1 mL HBSS. BALF was obtained by centrifugation, separating BALF cells from supernatant. For single cell lung suspensions, lungs were resected, physically agitated and digested in 5 mg/mL Collagenase D, 20 units/mL DNase in HBSS with 2% FBS for 1 hr at 37°C. These cells were then passed through a 70µm sieve and overlaid on a lymphoprep gradient to isolate lung lymphocytes. For histology sectioning, lungs were inflated through the cannulae with a mixture of 1:1

138 PBS:OCT using a gravity inflation; inflated lungs were embedded in OCT and flash frozen using an isopentane, dry ice slurry.

4.2.5. Pulmonary Administration NP and control formulations were delivered to the lungs of anesthetized mice through an orotracheal instillation in a 50 µL volume in PBS. NP doses were 100 µg NP/mouse, corresponding to 10 µg OVA/mouse which was used as the control soluble OVA dose; in studies with adjuvant, 2.5 µg CpG/mouse was also delivered. For LPS challenge study, 20 µg of LPS at 0.4 mg/mL was administered via orotracheal instillation 1 hr following initial NP administration.

4.2.6. Antibodies: Flow Cytometry and ELISAs Single cell suspensions from either tissue or cell culture were kept on ice and blocked with anti-CD16/32 (Fc block, eBioscience) and stained with the following antibodies to mouse cell surface molecules; IA/IE-PE-Cy7, CD11b-APC-Cy7, CD11c-PB, F4/80-PE-Cy5, CD45-BV510, CD103-PE, CD3- BV510, CD19-PE-Cy7, were from BioLegend. Cells were fixed using 2% PFA in PBS. All data were collected using LSRII (BD Biosciences) flow cytometer and analyzed using FlowJo software (Tree Star). Enzyme-linked immunosorbent assay (ELISA) kits for IL-6, TNF-α, and IL-1β were purchased from BD Biosciences and used following manufacturer’s suggestions.

4.2.7. Histopathology Lung histology was performed by K. McNaughton in the UNC Histology Facility of the Department of Cell and Molecular Physiology. 7 µm frozen serial sections were obtained using a Leica 1950 cryostat, mounted to a glass slide and underwent staining. Hematoxylin and eosin (H&E) staining was performed by the core facility. Additional staining of CD11c+ sections was performed by first fixing the slides in chilled acetone for 2 minutes, then blocked and stained with Alexa Fluor 555 Phalloidin, anti-CD11c (Invitrogen), Alexa Fluor 488 Rabbit Anti-Rat IgG (H+L) (Invitrogen), and 14 mM DAPI (Invitrogen).

139 4.2.8. Statistical Analysis Statistical analyses were performed with GraphPad Prism version 6. Analysis of groups was performed as indicated in figures, where asterisks indicating p values of *<0.05 and **<0.01 and n.s indicating not significant. All data points were included in the analyses and no outliers were excluded in calculations of means or statistical significance. Particle batches, cell assays, and immunization studies were repeated in at least two independent experiments, with the number of replicates (particles, cells, mice) indicated in figures.

4.3. Results 4.3.1. Particle Fabrication and Characterization To determine the role of surface charge on NP based vaccines, we utilized a conjugation scheme to functionalize PRINT NPs that allowed reproducible control over both protein loading and zeta potential. Following NP fabrication, the model antigen ovalbumin (OVA) was covalently bound through a carboiimide bond to positively charged, amine- functional PRINT nanoparticles (referred to as (ζ+)NPs) through the reaction in Figure 4.2A. In parallel, the same starting particles were first succinylated, converting the amine groups within the NPs to terminal carboxylate groups and yielding a net negative charge on the NPs (referred to as (ζ-)NP), followed by OVA conjugation (Fig. 4.2B).

Figure 4.2. Ovalbumin functionalization and characterization of hydrogel PRINT particles. (A) Model antigen OVA is covalently linked to cationic (ζ+) PRINT particles using EDC/s- NHS carbodiimide chemistry yielding (ζ+)NP-OVA. (B) Amine groups in (ζ+)NP are converted to carboxylic groups using succinic anhydride to yield anionic (ζ-)NP that are covalently linked to OVA using the same chemistry as in (A).

140 By tuning the reaction conditions in these schemes, OVA was controllably loaded at 100 ± 10 µg OVA / mg NP (n≥5) on both cationic and anionic NP (abbreviated (ζ+)NP-OVA and (ζ-)NP-OVA, respectively). Unreacted amine or carboxylic groups contributed to the remaining overall net charge. Stock OVA concentration was determined to have endotoxin levels <0.5 EU/µg OVA. Despite considerable protein loading, representative scanning electron microscopy (Fig. 4.3), shows unaggregated uniform particles following OVA functionalization.

Figure 4.3. Representative SEM micrographs of functionalized NPs. (ζ+) NP-OVA shown.

This was also confirmed by dynamic light scattering (DLS) results (Table 4.1), which yielded consistent hydrodynamic diameters (Zavg) and a low polydispersity index (PDI) for all four particle formulations.

Table 4.1. DLS results for NPs before and after functionalization (average +/- SD), n≥5 of repeated particle batches. Z (nm) PDI ζ (mV) AVG + (ζ ) NP 273 ± 10 0.06 ± 0.02 45 ± 3 + (ζ ) NP-OVA 285 ± 4 0.06 ± 0.01 37 ± 3 - (ζ ) NP 248 ± 8 0.05 ± 0.02 -38 ± 2 - (ζ ) NP-OVA 264 ± 7 0.05 ± 0.01 -38 ± 3

141 From these measurements, we noticed a slight increase in the Zavg upon functionalization likely due to OVA conjugated to NP surface. Additionally, slight variation in the Zavg was observed overall between cationic and anionic formulations, amounting to slight differences in hydrogel swelling in the particular aqueous conditions of the DLS measurements and the general error associated with assigning a spherical diameter to a non-spherical NP. However, extensive SEM measurements ensure fidelity of both (ζ+)NP-OVA and (ζ-)NP-OVA in all three dimensions. Taken together, our results indicate that an equivalent amount of OVA was conjugated to both cationic and anionic PRINT NPs and did not induce particle aggregation.

4.3.2. Evaluation of Particle Tolerability in the Lung To establish that both cationic and anionic nanoparticles are viable approaches for lung-specific vaccine carriers, we treated mouse lungs with both particle types and tested for tolerability. We first tested whether NP-OVA treatment to the lung caused any adverse inflammatory response directly in the lung, with PBS and CpG as negative and positive controls. We found no significant increase in IL-6 (Fig. 4.4A), TNF-α, or IL-1β (not shown) levels in the BALF over saline treatment at 24 and 72 hrs.

Figure 4.4. IL-6 cytokine analysis of (A) BALF and (B) serum following NP and CpG instillation at 24 and 72 hours in C57B/6 mice, n=3. Statistics performed by 2-way ANOVA with Tukey multiple comparisons test.

142 This is in stark contrast to the effect of CpG administration to the lung, which resulted in ~1000 fold increase in inflammatory cytokine production at both time points. Additionally, minimal increase, if any, in systemic production of these same inflammatory markers was observed for all treatment types (Fig. 4.4B). This indicates that both formulations of NP- OVA do not activate the inflammasome and have no increase in local or systemic inflammation over three days. This time point is anticipated to be sufficient in capturing the onset of acute inflammation. The lack of inflammatory responses was corroborated by immunohistochemistry staining of NP-treated lungs. H&E staining of representative lung pathology as shown in Fig. 4.5 shows no evidence of leukocyte recruitment to the lung vasculature, as would be indicated by the increased presence of blue-purple nuclei. This is in comparison to the massive amounts of cellular recruitment and presence of fluid observed for lungs treated with a bacterial lipopolysaccharide (LPS), a TLR4 ligand.39

Figure 4.5. H&E stained whole lung sections resected 24 hrs following NP instillation. Images A (2x) and C (10x) are from lungs treated with (ζ+)NP, images B (2x) and D (10x) are from lungs treated with (ζ-)NP.

143 Additionally, we wanted to see if cationic NP administration would be detrimental to lung function, as previous studies have shown that anionic NPs are well tolerated. To this aim, we pre-treated mouse lungs with cationic NPs and followed 1 hr later with a challenge of LPS to confirm that lung cells maintained the ability to mount an inflammatory response. No change in IL-6 (Fig. 4.6), TNF-α, or IL-1β (not shown) production in either the BALF or the serum was observed, indicating that cationic NP administration did not hamper lung functionality.

Figure 4.6. IL-6 cytokine analysis of (A) BALF and (B) serum in C57B/6 mice which were treated with (ζ+)NP then challenged 1 hr later with LPS, both via instillation. BALF and serum were collected 24 hrs after LPS challenge, n=3. Statistics performed by 2-way ANOVA with Tukey multiple comparisons test with no statistical significance observed.

In total, these data support that conclusion that both cationic and anionic NP formulations studied are well tolerated in mice lungs and do not cause adverse acute inflammation. This is critical in establishing the viability of these particles as vaccine platforms.

144 4.3.3. Particle Association in Lung APCs To analyze charge-depended particle association in lung APCs and this correlation to potential LN trafficking, we sampled lung tissue at 72 hrs following instillation of fluorescently-labeled NPs. We sought to explore differences in lung association in the presence of antigen, and compared association between NP and NP-OVA formulations of both charges. Additionally, we sought to explore NP fate during a mimic a pulmonary immunization, and explored the changes in particle association following NP co-delivery with the TLR adjuvant, CpG. Lung pathology of anionic and cationic administration was performed to observe the localization of fluorescent NPs in the airways (Fig. 4.7).

Figure 4.7. Histology of NP-treated lungs. Sections stained for fluorescent NPs (red), phalloidin (gray), DAPI (blue), and CD11c+ (green). (A-C) (ζ+) NPs. (D-F) (ζ-) NPs.

145 Both particle types were observed throughout the entire lung and most often appear as punctate fluorescent regions, presumably internalized into phagocytic cells. We stained these sections for CD11c+ and find that the majority of the punctate fluorescence corresponded with this surface marker (Fig 4.6 B-C). To further elucidate the function of the CD11c+ cells visually associated with NPs, we utilized flow cytometry to quantify NP uptake in three critical APC lung populations: AM, CD11b DCs and CD103 DCs. Representative gating used to analyze these populations is shown in Fig. 4.8.

Figure 4.8. Representative gating to identify NP+ lung APCs.

In the lung, macrophages and DCs were identified as CD45+, CD11c+ populations and were separated by MHC II expression. MHC IIlo macrophages were confirmed to be F4/80+ and CD11b-. MHC IIhi DCs were further separated into a CD103+, CD11b- population, identifying the CD103 DCs, and a CD103- CD11b+ population of CD11b DCs. Following

146 population identification, each cell population was analyzed for NP fluorescence, gated by the PBS-treated, NP fluorescence-minus one (FMO) control (Fig. 4.8 bottom).40 Particle association was quantified as the percentage of each cell type which had a detectable NP signal, as well as the median fluorescence intensity (MFI) of NP+ cells. The percentage of NP+ cells indicates how likely a given cell type is to associate with any amount of NPs, while the MFI lends insight to how many particles each individual cell internalizes. Alveolar macrophages readily internalized both anionic and cationic NPs, regardless of the presence of the OVA antigen (Fig. 4.9). At the 100 µg NP dose given, between 60- 100% of AMs were NP+, with no statistical difference observed between the NP groups.

Figure 4.9. NP association in AMs of C57B/6 mice treated with NP via instillation, n≥3. (A) Percentage of NP+ AMs. (B) MFI of NP+ AMs. Statistics performed by 2-way ANOVA with Tukey multiple comparisons test, where * correspond to p values of *<0.05 and **<0.01.

This corresponds well with the known function of AM, which are the primary phagocytic cell in the lung. However, while the percentage of NP+ cells was no different between groups, the MFI of (ζ-)NP-OVA and (ζ-)NP were a factor of 10 higher than the corresponding cationic NPs (Fig. 4.9B). This indicates that a single AM internalized a significantly greater number of anionic NPs, suggesting that either AM have a preference towards internalization of

147 anionic NPs, or the total availability of these NPs was greater than cationic NPs, presumably due to other clearance methods. Interestingly, the introduction of the CpG adjuvant yielded a decrease in the percentage of AMs which associated with (ζ+)NP-OVA and (ζ+)NP. While the percentage of AM which were NP+ following anionic NP treatment remained consistent with CpG co-delivery, significant decreases in the MFI of these cells were observed, signifying that CpG also effected the amount of internalization for anionic NPs. While the AMs seem to prefer anionic NPs, both DC subtypes show increased association with cationic NPs. The percentage of NP+ CD103 DCs is consistent regardless of the presence of OVA, however slight increases in NP+ populations are observed following (ζ+)NP-OVA and (ζ+)NP (Fig. 4.10A). Unlike the AM, the MFI of NP+ CD103 DCs was consistent across all groups, indicating an equivalent amount of NP internalization (Fig. 4.10 B). Interestingly, both the percentage of CD103 DCs and the corresponding MFI were not influenced by CpG administration, which is consistent with the literature.30

Figure 4.10. NP association in CD103 DCs of C57B/6 mice treated with NP via instillation, n≥3. (A) Percentage of NP+ CD103 DCs. (B) MFI of NP+ CD103 DCs. Statistics performed by 2-way ANOVA with Tukey multiple comparisons test, n.s.

CD11b DCs were also found to have an increased percentage of NP+ cells following (ζ+)NP-OVA and (ζ+)NP (Fig. 4.11A) and this did not correspond to any increases in MFI

148 (Fig. 4.11B). However, NP uptake in CD11b DCs was affected by co-administration of CpG. The adjuvant resulted in an overall increase in NP+ cells, significantly increasing the amount associated with anionic NPs. This potentially accounts for some of the MFI loss observed for AM NP-association following CpG administration with (ζ-)NP-OVA and (ζ-)NP. This increase also corresponded to an increase in MFI of CD11b NP+ DCs, indicating the increased frequency of NP+ DCs were also actively internalizing more particles.

Figure 4.11. NP association in CD11b DCs of C57B/6 mice treated with NP via instillation, n≥3. (A) Percentage of NP+ CD11b DCs. (B) MFI of NP+ CD11b DCs. Statistics performed by 2-way ANOVA with Tukey multiple comparisons test, where ** correspond to p values of **<0.01.

4.3.4. Change in Lung APC Population Following Instillation In addition to NP-association, we also quantified changes in APC lung populations following NP treatment. We hypothesized that changes in the balance of APC populations could indicate increased trafficking to the draining lymph node or the influx of other populations. From the same experiment as above, we quantified the total percentage of each APC population out of total CD45+ cells.

149 Treatment of soluble OVA had no change on the overall AM population, as compared to PBS, but both cationic and anionic NP formulations resulted in a decrease in total AM percentage, with cationic NP-treatment having a slightly larger impact (Fig. 4.12).

Figure 4.12. Percentage of AMs out of total CD45+ cell from lungs of C57B/6 mice treated with NP via instillation, n≥3. Statistics performed by 2-way ANOVA with Tukey multiple comparisons test, where ** correspond to p values of **<0.01. PBS served as a negative control for the assay and was not included in the analysis.

This decrease was not as pronounced as the change in AM population observed with co- delivery of CpG, which resulted in dramatic decrease of AM populations. Clearly, the addition of the CpG adjuvant changes the dynamic of local lung populations, implying either a detrimental effect to AMs, a change in their phenotype which would impact the gating scheme, or an influx of other CD45+ cells, potentially DCs or granulocytes. AMs are not thought to migrate from the lung to the lymph node. A similar trend was observed in the population changes of CD103 DCs (Fig. 4.13). While soluble OVA and anionic NPs had no impact on the CD103 population, cationic treatment resulted in a slight decrease in this cell type, and CpG treatment resulted in the complete decrease of these cells. Unlike the AMs, CD103 DCs are known to migrate to the

150 local lymph node, which could contribute to their decreased presence in the lung, in addition to the recruitment of other cell types.

Figure 4.13. Percentage of CD103 DCs out of total CD45+ cell from lungs of C57B/6 mice treated with NP via instillation, n≥3. Statistics performed by 2-way ANOVA with Tukey multiple comparisons test, where * correspond to p values of *<0.05, **<0.01. PBS served as a negative control for the assay and was not included in the analysis.

Interestingly, CD11b DC populations exhibit the inverse effect from AMs and CD103 DCs (Fig. 4.14). Particle treatment results in the increased occurrence of CD11b DCs over soluble OVA and PBS treatment, with no preference based on NP change. CpG treatment results in an increase of CD11b DCs following soluble OVA treatment, but minimal change for any other treatment type. Given the backdrop of the disappearance of both AM and CD103 DC populations following CpG administration, which hints at the infiltration of another cell population, the lack of change to the overall frequency of CD11b DCs suggests this population has actually increased.

151

Figure 4.14. Percentage of CD11b DCs out of total CD45+ cell from lungs of C57B/6 mice treated with NP via instillation, n≥3. Statistics performed by 2-way ANOVA with Tukey multiple comparisons test, where * correspond to p values of *<0.05. PBS served as a negative control for the assay and was not included in the analysis.

4.3.5. Particle Trafficking to Medistinal Lymph Nodes We hypothesized that the observed changes in lung populations would translate to differences in the cellularity and nanoparticle frequency in the lung draining LN. 72 hrs after instillation, we resected the medistinal LNs and used flow cytometry to quantify cell and nanoparticle frequencies as in the lung. Representative LN gating is shown in Fig. 4.14; from this scheme, CD3+ T cells, CD19+ B cells, CD11c+ DCs, and CD11b and CD103 DCs were identified. Nanoparticle association was determined as before in each cell population by gating by the PBS-treated, NP FMO control, with percentage of NP+ populations and NP+ MFI determined.

152

Figure 4.15. Representative LN gating.

Given the dramatic changes in DC populations as a function of instilled treatment, we anticipated these changes to also be observed in the LN. As anticipated, no treatment type influenced B or T cell populations. However, looking at the overall CD11c+, presumably DC, population in the LN, we found that all treatment types resulted in an increased in DC frequency over PBS treatment (Figure 4.16).

Figure 4.16. Percentage of CD11c+ DCs out of non B/T cell (CD3-, CD19-) from lungs of C57B/6 mice treated with NP via instillation, n≥3. Statistics performed by 2-way ANOVA with Tukey multiple comparisons test, n.s. PBS served as a negative control for the assay and was not included in the analysis.

153 While not statically significant, slight increases were observed in the DC frequency following cationic NP treatment and an overall increase in total DCs after CpG administration. However, no differences were observed in the two specific lung DC phenotypes present in the LN. Both CD103 DCs (Fig. 4.17A) and CD11b DCs (Fig. 4.17B) were unchanged by CpG treatments, which corresponded well with the population data for CD11b DCs, but not CD103 DCs.

Figure 4.17. Percentage of DC subtypes out of CD11c+ non B/T cells from lungs of C57B/6 mice treated with NP via instillation, n≥3. (A) CD103 DCs, (B) CD11b DCs. Statistics performed by 2-way ANOVA with Tukey multiple comparisons test, n.s. PBS served as a negative control for the assay and was not included in the analysis.

Particle association in LN cells was similarly unaffected by lung treatment type at this 72 hr time point. Both B and T cells were negative for NP+ following all treatment types (data not shown). However, between 10-20% of CD103 DCs (Fig 4.18A-B) and 5-10% of CD11b DCs (4.18C-D) in the LN were found to be particle positive, indicating that these cells actively transported NPs from the lung. However, no difference in percent associated or MFI for either DC subtype was observed following CpG treatment, indicating that the inclusion of an adjuvant does not increase lung-LN trafficking.

154

Figure 4.18. NP association in DC subtypes of C57B/6 mice treated with NP via instillation, n≥3. (A) % NP+ CD103 DCs, (B) MFI of NP+ CD103 DCs, (C) % NP+ CD11b DCs, (D) MFI of NP+ CD11b DCs. Statistics performed by 2-way ANOVA with Tukey multiple comparisons test, n.s.

4.4. Discussion As pulmonary delivery of nanoparticles continues to be studied as an approach for the next generation of vaccines, it will be increasingly important to understand how NP properties influence and modify pulmonary immune responses.4, 36 Use of the PRINT platform allowed us to use the same initial particle hydrogel polymerization, NP structure, crosslinking network, amount of functional groups and composition to isolate NP surface charge as the sole variable in these studies. From this, we systematically studied the influence NP surface charge and antigen attachment had on lung APC populations, detailing the fate of

155 the NPs, tolerability in the lung, changes in lung APC populations, and the propensity to traffic from the lung to the medistinal LNs. Our results show that cationic NPs are preferentially associated with two important lung DC populations and stimulate the APC lung populations in a similar manner to a TLR agonist, indicating that cationic NPs may have a slight adjuvant property in the lung. A caveat worth mentioning when using flow cytometry is that markers used to identify certain cell populations may change over time, especially in response to a stimulus. The gating method used in this study to distinguish AMs from DCs relied solely on variations in MHC II expression; AMs are typically identified as MHC IIlo while DCs are MHC IIhi. However, some studies have shown that AMs can upregulate both CD11b and MHC II during inflammation. While the pro-inflamatory cytokine was absent in the BALF following particle administration, the co-delivery of CpG may have caused inflammation and the upregulation of various surface markers, including MHC II. This may account for the supposed disappearance of the AM population in the flow cytometry results and future studies will utilize additional markers unaffected by inflammation, such as Siglec-F.41 Cationic NPs preferentially associated with lung DCs, while anionic NPs were more readily internalized by AMs. While AMs are considered APCs, their main function in the lung is to clear and sequester foreign material and maintain lung homeostasis.16, 22, 23 As a result, AMs are not the primary target of NP vaccines and it is more important to target DCs, the “professional” APCs. From our studies, anionic NPs sequestered in AMs would be less successful at eliciting an adaptive immune response. The two lung DC populations studied here have both been implicated in critical immune capabilities, including stimulation of IgA production, CD4+ and CD8+ T cell priming, and migration to LNs.15, 16, 22, 27-30, 32 As both CD103 and CD11b DCs showed a trend of increased association with cationic NP, it is hypothesized that cationic NPs will result in superior immune stimulation, stemming directly from their increased association. Additionally, increased association in CD11b DCs following CpG administration may imply an opportunity to skew the overall lung immune status, as CD11b DCs have been implicated as a main mediator of Th2 responses.30, 42, 43

156 In addition to increased DC association, cationic NPs also offer the potential to act as an adjuvant, similar to CpG. CpG oligonucleotides are a TLR adjuvant, which have been explored as a mucosal adjuvant due to the potent Th1 response generated.44, 45 In the lung, CpG has been shown to cause the production of numerous pro-inflammatory cytokines and the recruitment of a number of cells, including plasmacytoid DCs, natural killer (NK) cells, and neutrophils.46 In our work, co-administration of CpG with NP formulations resulted in the decrease of AM and CD103 DC lung populations, an increase in the overall CD11c+ DC in the LN, an increase in CD11b NP-uptake and corresponding decrease AM NP-uptake. Interesting to note, these same trends existed following the delivery of cationic NPs, to a less extreme degree. These findings suggest that cationic NPs themselves may act as an adjuvant, with potential benefits over the use of aggressive TLR agonists which might impair lung function. From both histology and cytokine analysis of BALF, cationic NPs were well tolerated in the lung and did not impair function. Despite considerable variability in the lung populations and NP cellular association, we do not find that changes in the lung environment have affected trafficking to the LN for the time point tested. Previous work in the literature has shown the role of NP surface charge on passive diffusion from the lung to the LNs, finding that NPs larger than 34 nm do not rapidly translocate from the lung47 Thus, we hypothesize that the NPs detected in the LN were actively transported there by lung APCs, which can be confirmed by using CFSE to selectively stain airway cells prior to migration.15, 36, 48 However, it is currently unclear how much antigen or how many activated DCs must be trafficked to the LN to mount a sufficient immune response, which is the ultimate purpose of these NP carriers. We anticipate that the amount of NP detected in the LNs will be sufficient to engender an antigen-specific response, which will be tested in future studies.

4.5. Conclusions Overall, these studies are critical in establishing the role on NP properties on pulmonary immune cells. Our findings suggest that cationic NP may offer increase

157 preference towards DCs, making surface charge an important parameter for pulmonary NP vaccine design. To our knowledge, this is the first study where NP surface charge has been directly investigated in pulmonary vaccines. Work presented in Chapter 4 will further evaluate these NPs following pulmonary immunization.

4.6. Acknowledgements We thank R. Roberts, K. Reuter, J. Perry, S. Tian, A. Pandya, K. McNaughton, N. Fisher, S. Coquery, for useful discussions and technical assistance. We acknowledge Liquidia Technologies for providing PRINT molds, and the core facilities at UNC, including the Nucleic Acids Core Facility, CHANL imaging facility, the Histology Facility of the Department of Cell and Molecular Physiology, and DLAM facility. This work was funded in part by the NIH Pioneer Award to J.M.D. (1DP1OD006432) and DTRA award (HDTRA1- 13-1-0045).

4.7. References 1. Fromen CA, Shen TW, Larus AE, Mack P, Maynor BW, Luft JC, DeSimone JM. Synthesis and characterization of monodisperse uniformly shaped respirable aerosols. AIChE Journal. 2013, 59, 3184-3194.

2. Garcia A, Mack P, Williams S, Fromen C, Shen T, Tully J, Pillai J, Kuehl P, Napier M, Desimone JM, Maynor BW. Microfabricated engineered particle systems for respiratory drug delivery and other pharmaceutical applications. J Drug Deliv. 2012, 2012, 941243.

3. Blank F, Stumbles P, von Garnier C. Opportunities and challenges of the pulmonary route for vaccination. Expert Opin Drug Del. 2011, 8, 547-563.

4. Moon JJ, Huang B, Irvine DJ. Engineering nano- and microparticles to tune immunity. Adv Mater. 2012, 24, 3724-3746.

5. Li AV, Moon JJ, Abraham W, Suh H, Elkhader J, Seidman MA, Yen M, Im E, Foley MH, Barouch DH, Irvine DJ. Generation of effector memory t cell-based mucosal and

158 systemic immunity with pulmonary nanoparticle vaccination. Science Translational Medicine. 2013, 5, 204-130.

6. Nembrini C, Stano A, Dane KY, Ballester M, van der Vlies AJ, Marsland BJ, Swartz MA, Hubbell JA. Nanoparticle conjugation of antigen enhances cytotoxic t-cell responses in pulmonary vaccination. Proc Natl Acad Sci U S A. 2011, 108, E989-997.

7. De Temmerman ML, Rejman J, Demeester J, Irvine DJ, Gander B, De Smedt SC. Particulate vaccines: On the quest for optimal delivery and immune response. Drug Discov Today. 2011, 16, 569-582.

8. Bachmann MF, Jennings GT. Vaccine delivery: A matter of size, geometry, kinetics and molecular patterns. Nat Rev Immunol. 2010, 10, 787-796.

9. Smith DM, Simon JK, Baker JR, Jr. Applications of nanotechnology for immunology. Nat Rev Immunol. 2013, 13, 592-605.

10. Pulliam B, Sung JC, Edwards DA. Design of nanoparticle-based dry powder pulmonary vaccines. Expert Opin Drug Del. 2007, 4, 651 - 663.

11. Gillies RJ. In vivo molecular imaging. J Cell Biochem Suppl. 2002, 39, 231-238.

12. Patton JS, Byron PR. Inhaling medicines: Delivering drugs to the body through the lungs. Nat Rev Drug Discov. 2007, 6, 67-74.

13. Johnson DL, Leith D, Reist PC. Drag on non-spherical, orthotropic aerosol particles. J Aerosol Sci. 1987, 18, 87-97.

14. Kunda NK, Somavarapu S, Gordon SB, Hutcheon GA, Saleem IY. Nanocarriers targeting dendritic cells for pulmonary vaccine delivery. Pharm Res. 2013, 30, 325-341.

15. Sakagami M. In vivo, in vitro and ex vivo models to assess pulmonary absorption and disposition of inhaled therapeutics for systemic delivery. Adv Drug Deliv Rev. 2006, 58, 1030-1060.

16. Guilliams M, Lambrecht BN, Hammad H. Division of labor between lung dendritic cells and macrophages in the defense against pulmonary infections. Mucosal Immunol. 2013, 6, 464-473.

159 17. Azarmi S, Roa WH, Lobenberg R. Targeted delivery of nanoparticles for the treatment of lung diseases. Adv Drug Deliv Rev. 2008, 60, 863-875.

18. Hu W, Pasare C. Location, location, location: Tissue-specific regulation of immune responses. J Leukoc Biol. 2013, 94, 409-421.

19. Hassan MS, Lau RW. Effect of particle shape on dry particle inhalation: Study of flowability, aerosolization, and deposition properties. AAPS PharmSciTech. 2009, 10, 1252-1262.

20. Kleinstreuer C, Zhang Z, Li Z, Roberts WL, Rojas C. A new methodology for targeting drug-aerosols in the human respiratory system. International Journal of Heat and Mass Transfer. 2008, 51, 5578-5589.

21. Kleinstreuer C, Zhang Z, Donohue JF. Targeted drug-aerosol delivery in the human respiratory system. Annu Rev Biomed Eng. 2008, 10, 195-220.

22. Hardy CL, Lemasurier JS, Mohamud R, Yao J, Xiang SD, Rolland JM, O'Hehir RE, Plebanski M. Differential uptake of nanoparticles and microparticles by pulmonary apc subsets induces discrete immunological imprints. J Immunol. 2013, 191, 5278-5290.

23. Fernandes CA, Vanbever R. Preclinical models for pulmonary drug delivery. Expert Opin Drug Del. 2009, 6, 1231-1245.

24. Sakagami M, Byron PR. Respirable microspheres for inhalation. Clin Pharmacokinet. 2005, 44, 263-277.

25. Thiel CG. Can in vitro particle size measurements be used to predict pulmonary deposition of aerosol from inhalers? J Aerosol Med. 1998, 11, S43-S52.

26. Inaba K, Steinman RM, Van Voorhis WC, Muramatsu S. Dendritic cells are critical accessory cells for thymus-dependent antibody responses in mouse and in man. Proc Natl Acad Sci U S A. 1983, 80, 6041-6045.

27. Naito T, Suda T, Suzuki K, Nakamura Y, Inui N, Sato J, Chida K, Nakamura H. Lung dendritic cells have a potent capability to induce production of immunoglobulin a. Am J Respir Cell Mol Biol. 2008, 38, 161-167.

28. Xie Y, Zeng P, Wiedmann TS. Disease guided optimization of the respiratory delivery of microparticulate formulations. Expert Opin Drug Del. 2008, 5, 269-289.

160 29. Abdelwahed W, Degobert G, Stainmesse S, Fessi H. Freeze-drying of nanoparticles: Formulation, process and storage considerations. Adv Drug Deliv Rev. 2006, 58, 1688- 1713.

30. Furuhashi K, Suda T, Hasegawa H, Suzuki Y, Hashimoto D, Enomoto N, Fujisawa T, Nakamura Y, Inui N, Shibata K, Nakamura H, Chida K. Mouse lung cd103+ and cd11bhigh dendritic cells preferentially induce distinct cd4+ t-cell responses. Am J Respir Cell Mol Biol. 2012, 46, 165-172.

31. Chen G, Wang W. Role of freeze drying in nanotechnology. Drying Technology. 2007, 25, 29-35.

32. Suzuki Y, Suda T, Furuhashi K, Shibata K, Hashimoto D, Enomto N, Fujisawa T, Nakamura Y, Inui N, Nakamura H, Chida K. Mouse cd11bhigh lung dendritic cells have more potent capability to induce iga than cd103+ lung dendritic cells in vitro. Am J Respir Cell Mol Biol. 2012, 46, 773-780.

33. Chan HK. What is the role of particle morphology in pharmaceutical powder aerosols? Expert Opin Drug Del. 2008, 5, 909-914.

34. Wang YY, Lai SK, Suk JS, Pace A, Cone R, Hanes J. Addressing the peg mucoadhesivity paradox to engineer nanoparticles that "slip" through the human mucus barrier. Angew Chem Int Ed Engl. 2008, 47, 9726-9729.

35. Zhao L, Seth A, Wibowo N, Zhao CX, Mitter N, Yu C, Middelberg AP. Nanoparticle vaccines. Vaccine. 2014, 32, 327-337.

36. Kasper G, Niida T, Yang M. Measurements of viscous drag on cylinders and chains of spheres with aspect ratios between 2 and 50. J Aerosol Sci. 1985, 16, 535-556.

37. Perry JL, Reuter KG, Kai MP, Herlihy KP, Jones SW, Luft JC, Napier M, Bear JE, DeSimone JM. Pegylated print nanoparticles: The impact of peg density on protein binding, macrophage association, biodistribution, and pharmacokinetics. Nano Lett. 2012, 12, 5304-5310.

38. Van den Broeck W, Derore A, Simoens P. Anatomy and nomenclature of murine lymph nodes: Descriptive study and nomenclatory standardization in balb/canncrl mice. J Immunol Methods. 2006, 312, 12-19.

161 39. Roberts RA, Shen T, Allen IC, Hasan W, DeSimone JM, Ting JP. Analysis of the murine immune response to pulmonary delivery of precisely fabricated nano- and microscale particles. PLoS One. 2013, 8, e62115.

40. Mitchell JP, Nagel MW. Cascade impactors for the size characterization of aerosols from medical inhalers. J Aerosol Med. 2003, 16, 341-377.

41. van Oort M, Downey B. Cascade impaction of mdis and dpis: Induction port, inlect cone, and preseparator lid designs recommended for inclusion in the general test chapter aerosols <601>. Pharmacopeial Forum. 1996, 22, 2204-2210.

42. Larhrib H, Martin GP, Marriott C, Prime D. The influence of carrier and drug morphology on drug delivery from dry powder formulations. International Journal of Pharmaceutics. 2003, 257, 283-296.

43. Mansour HM, Rhee Y, Wu X. Nanomedicine in pulmonary delivery. Int J Nanomedicine. 2009, 4, 299-319.

44. Kline JN. Eat dirt: Cpg DNA and immunomodulation of asthma. Proc Am Thorac Soc. 2007, 4, 283-288.

45. Moldoveanu Z, Love-Homan L, Huang WQ, Krieg AM. Cpg DNA, a novel immune enhancer for systemic and mucosal immunization with influenza virus. Vaccine. 1998, 16, 1216-1224.

46. Pesce I, Monaci E, Muzzi A, Tritto E, Tavarini S, Nuti S, De Gregorio E, Wack A. Intranasal administration of cpg induces a rapid and transient cytokine response followed by dendritic and natural killer cell activation and recruitment in the mouse lung. J Innate Immun. 2010, 2, 144-159.

47. Choi HS, Ashitate Y, Lee JH, Kim SH, Matsui A, Insin N, Bawendi MG, Semmler- Behnke M, Frangioni JV, Tsuda A. Rapid translocation of nanoparticles from the lung airspaces to the body. Nat Biotechnol. 2010, 28, 1300-1303.

48. Pal I, Ramsey JD. The role of the lymphatic system in vaccine trafficking and immune response. Adv Drug Deliv Rev. 2011, 63, 909-922.

162

CHAPTER FIVE

Induction of T-cell Dependent Mucosal and Systemic Antibody Responses Following Pulmonary Delivery of Cationic Hydrogel-based Nanoparticles

Fromen, Robbins, Shen, Kai, Ting, DeSimone, in preparation

163 5.1. Introduction The lung and gastrointestinal tract are primary sites of pathogen entry and therefore are critical targets for mucosal vaccines. Conventionally administered vaccines (e.g. subcutaneous or intramuscular injection) tend to provide strong humoral protection, but often fail to generate mucosal immunity, especially in the form of IgA. Furthermore, delivering vaccines directly to mucosal sites not only provides local protection, but also protects at distal mucosal sites and confers systemic immunity.1-3 Recent work with pulmonary vaccines has demonstrated the benefit of delivering spray-dried protein antigens for measles, TB and influenza, leading to enhanced local and systemic protection.4-6 In contrast, other studies generated mixed results following pulmonary vaccination that may be attributed to the complex nature of immune responses in the lung, as well as antigen formulation approaches. For example, pulmonary delivery of protein subunit vaccines in the absence of adjuvant does not provide strong humoral or mucosal responses and in some cases may induce immunological tolerance, which could be detrimental for prophylactic vaccines.7, 8 A common approach for augmenting lung based immunization is to co-deliver adjuvants with antigen; however, there is concern in the pulmonary field that use of aggressive adjuvants could induce acute and/or chronic inflammation resulting in impaired lung function.9 Protective mucosal and systemic antibody responses involve complex cross talk between innate and adaptive immune cells. Antigen is first encountered by professional antigen presenting cells (APCs), such as dendritic cells (DC), at the site of infection or injury, resulting in DC maturation and migration to the draining lymph node (dLN). DCs process the antigen into peptides for display on MHCII that allows for activation of antigen specific CD4+ T-cells. Activated T-cells instruct cognate antigen specific B-cells to form germinal centers (GC) in the LN, which are sites for B-cell expansion, affinity maturation and Ig class switch recombination. The final result is the production of highly specific antibodies with specialized functions that help clear pathogens. This multistep process of requires all three cell types (DC, CD4+ T-cells, and B-cells), as T-cell deficient mice fail to form GCs and DC depleted mice exhibit reduced antibody production.10, 11 Therefore, vaccines designed to

164 evoke strong mucosal and systemic antibody responses should induce DC and T-cell activation, as well as robust germinal center formation. The emergence of engineered nanoparticle (NP) formulations to improve subunit vaccine approaches has gained considerable attention in recent years. Using NP as a scaffold for antigen/adjuvant delivery can induce pathogen mimicry while still maintaining the safety of a subunit vaccine.12-16 In the lung, NP formulations offer potential solutions to overcome biological barriers, inherently targeting APCs and penetrating the mucosal layer.1, 9, 17-19 Cationic NP formulations have been shown to increase mucosal antibody production following pulmonary or intranasal administration20, 21, while other groups have utilized cationic NP for CD8+ T-cell responses to harness endosomal escape mechanisms.22, 23 Independent studies using, anionic NP approaches have been shown to specifically enhance cellular based protection in the lung by inducing CD8+ T-cell responses, but these same NP formulations show minimal improvement in antibody production over immunization with soluble protein alone.24, 25 To date, no direct evaluation of the role of NP surface charge on these responses has been performed, likely due to limitations with NP formulation that prevent such a comparison without dramatically changing NP composition. As such, there are conflicting directions in the literature as to the exact role of NP charge in designing a prophylactic mucosal vaccine. To directly test the effects of NP charge on vaccine responses we covalently attached the model antigen ovalbumin (OVA) to hydrogel-based NP that varied only in surface charge and otherwise had identical in size, shape and antigen loading. In this work, we show that antigen-conjugated cationic NPs induced strong DC/T-cell responses in vitro and stimulated mucosal and systemic antibody production at levels consistent with the known TLR agonist CpG following pulmonary immunization, whereas anionic NP did not. Further, co-delivery of antigen-coated cationic NP with soluble CpG enhanced mucosal IgG and IgA production and suggests that positively charged NPs possess inherent adjuvant activity that synergizes with TLR signaling. This work suggests that cationic NPs are a desirable candidate platform for pulmonary based vaccine delivery.

165 5.2 Materials and Methods 5.2.1. Animals All studies were conducted in accordance with National Institutes of Health guidelines for the care and use of laboratory animals and approved by the Institutional Animal Care and Use Committee (IACUC) at UNC. All animals were maintained in pathogen-free facilities at UNC and were between 8 and 15 weeks of age. C57BL/6 were obtained from Jackson Laboratories. OT-II transgenic mice were bred in-house.

5.2.2. Reagents Solvents and buffers of reagent grade were obtained by Fisher Scientific. PRINT 80 x 320nm molds were provided by Liquidia Technologies. Pre-particle reagents of 2- aminoethylmethacrylate (AEM), poly(ethylene glycol)700 diacrylate (PEG700DA) and diphenyl(2,4,6-trimethylbenzoyl)phosphine oxide (TPO) were obtained from Sigma; tetra(ethylene glycol) monoacrylate (HP4A) was synthesized in house via previously described methods 26. Dyes of cell trace violet and pHrodo red succinimidyl ester were obtained from Invitrogen, and maleimide-Dylight 650 was obtained from Fisher. Coupling reagents for the carboiimide conjugation of 1-ethyl-3-[3-dimethylaminopropyl]carbodiimide hydrochloride (EDC) and N-hydroxysulfosuccinimide (s-NHS) were obtained from Thermo Fisher. Both CpG-B 1826 oligonucleotide (5′-TCCATGACGTTCCTGACGTT-3′) and OVA grade V were obtained from Sigma Aldrich. RPMI media, NEAA, Na-Pyruvate, L- glutamine, 2-ME were from Invitrogen and 10% FCS, 1xPen/Step from Sigma.

5.2.3. Particle Fabrication and Characterization Amine-containing 80x320nm hydrogel rod-shaped NPs were fabricated on a continuous roll-to-roll PRINT method as described previously.27 Preparticle solutions contained 1% TPO (photoinitiator), 20% AEM (functional groups), 10% PEG700DA

(crosslinker), 0-1% functional fluorescent dye, 68-9% HP4A (monomer) by weight. OVA functionalization was achieved using water soluble zero length, carboiimide chemistry and characterized as described in Section 4.2.3.

166 5.2.4. Tissue and Cell Preparation BMDCs used in co-culture and NP uptake studies were isolated as described previously. Bone marrow was obtained from C57B/6 mice, red blood cells lysed with ACK buffer, washed and seeded in media with GM-CSF. After 6 days, cells were harvested by cell scraping and enriched using CDllc+ MACs beads (Miltenyl Biotec). Splenocytes were obtained by resecting spleens of OT-II mice, physically agitating between crushed glass slides and passed through a 70µm sieve. RBCs were lysed with ACK buffer. CD4+ T-cells were purified using through a CD25+ depletion and CD4+ enrichment using MACs beads (Miltenyl Biotec). For immunization studies, whole blood was obtained through submandibular bleed or cardiac puncture and collected in heparin-coated tubes; from these, plasma was obtained by centrifugation. Following perfusion of 10mL PBS, draining medistinal LNs were collected, physically agitated between crushed glass slides and passed through a 70 µm sieve to obtain a single cell suspension28. BAL was performed by inserting a cannula in an incision in the trachea and flushing the lungs with 1 mL HBSS. BALF was obtained by centrifugation, separating BALF cells from supernatant.

5.2.5. OT-II Co-culture On day 0, BMDCs were seeded in RMPI (gibco) and 10% FBS in a round-bottom 96 well plate. NPs or soluble OVA were added on day 1 based on mass of OVA per well. On day 2, splenocytes from OT II mice were harvested and purified as above. CD4+ OT-II T- cells were then labeled using Cell Trace Violet following manufacturer’s protocol (Life Technologies™). T-cells were added to NP-treated BMDCs. Cells were harvested 72hrs after T-cell addition. A diagram of this assay is shown in Figure 5.1.

167

Figure 5.1. Diagram of BMDC, OT-II CD4+ T cell co-culture.

5.2.6. Quantitative Realtime RT-PCR Total RNA was harvested using TRIzol® (Invitrogen) followed by oligo(dT) primed reverse transcription using M-MLV (Invitrogen) according to manufacturer’s instructions. Quantitative PCR was for murine Cd80, Cd86, H2aa, Tnfa, Il6, Il12, Il18, Cxcl10, Il10, Ifnb, Il1b,Ccl2, Tnfa, Tgfb1 and Actb (used for normalization) was performed using TaqMan® primer/probe sets and master mix (Applied Biosystems). Courtesy Greg Robbins.

5.2.7. Pulmonary Delivery and Immunizations NP and control formulations were delivered to the lungs of anesthetized mice through an orotracheal instillation in a 50 µL volume. NP doses were 100 µg NP/mouse, corresponding to 10 µg OVA/mouse which was used as the control soluble OVA dose; in studies with adjuvant, 2.5 µg CpG/mouse was also delivered. For single dose immunization studies, a dose was give on day 0 and euthanasia performed on day 11. For prime and boost immunization studies, doses were given on day 0 and 12, with submandibular mandibular bleeds performed on day 10 and euthanasia on day 20.

5.2.8. Antibodies: Flow Cytometry & ELISAs Single cell suspensions from either tissue or cell culture were kept on ice and blocked with anti-CD16/32 (Fc block, eBioscience) and stained with the following antibodies to mouse cell surface molecules; CD80-APC, CD86-PE, GL7-eFluor®-450, CD62L-APC-

168 eFluor®-780, CD62L-PE-Cy7 CD19-PE-Cy7, CD4-Fitc, CD4-PE-Cy7, CD3 Fitc, were from eBioscience; and IA/IE-PacificBlue, CD11b-PE-Cy7, CD11c-APC-Cy7, CD4-BV510, CD44-APC, CD25-PE, Vβ5.1.-APC were form BioLegend. Cells were fixed using 2% PFA in PBS. All data were collected using LSRII (BD Biosciences) flow cytometer and analyzed using FlowJo software (Tree Star). ELISA kits for IL-6 and IL-12 were purchased from BD Biosciences. For OVA- specific indirect ELISAs, plates were coated with 10ng/mL OVA, incubated with samples and followed by OVA-specific secondary antibodies of either goat anti-mouse IgG-HRP or goat anti-mouse IgA-HRP (SouthernBiotech).

5.2.9. Statistical Analysis Statistical analyses were performed with GraphPad Prism version 6. Analysis of groups was performed as indicated in figures, where asterisks indicating p values of *<0.05 and **<0.01 and n.s indicating not significant. All data points were included in the analyses and no outliers were excluded in calculations of means or statistical significance. Particle batches, cell assays, and immunization studies were repeated in at least three independent experiments, with the number of replicates (particles, cells, mice) indicated in figures.

5.3. Results 5.3.1. Particle Fabrication and Functionalization Please refer to section 4.3.1. 5.3.2. In Vitro BMDC and CD4+ OT-II T Cell Co-Culture In order to determine how NP charge affects uptake, antigen processing, and presentation by MHCII, we treated bone marrow derived dendritic cells (BMDC) (gating and purity in Fig. 5.2 A) with (ζ+)NP-OVA or (ζ-)NP-OVA and tested their ability to induce antigen specific T-cell proliferation using cell trace labeled CD4+ T-cells from transgenic OT-II mice (Purity in Fig. 5.2 B).

169

Figure 5.2. Representative gating and purity of co-culture cells. (A) BMDC purification from total WT bone marrow cultured in GM-CSF, with representative flow analysis shown for both pre-CD11c and post-enrichment with magnetic MACs beads. (B) Representative flow cytometric analysis of total OVA-specific OT-II transgenic splenocytes pre- and post- depletion of CD25+ cells (Regulatory and activated T-lineage). The CD25 depleted fraction was then enriched for CD4+ cells and the frequency of OVA-specific T-cells was determined using TCR-Vβ5.1 staining.

BMDCs treated with soluble (soluble) OVA induced T-cell proliferation in a dose dependent manner with a 10ug/ml dose resulting in 77% of OT-II T-cells undergoing >3 divisions (Fig. 5.3). Both cationic and anionic NP-OVA induced the same response using 10 fold lower OVA concentrations. T-cell proliferation in response to (ζ+)NP-OVA treated DC was even

170 stronger at 0.1ug/ml OVA dose, whereas the response to (ζ-)NP-OVA treated DC declined precipitously and was undetectable with sol OVA. The trend continued with a subsequent 10 fold dilution of (ζ+)NP-OVA (0.01ug/ml) inducing 28% of T-cells to undergo >3 divisions with undetectable proliferation in response to (ζ-)NP-OVA and soluble OVA treated DC.

Figure 5.3. CD4+ OT-II T-cell proliferation in response to NP-OVA treated BMDC. (A) Representative gating for Cell trace proliferation, analyzed only in the OVA specific TCR- Vβ5.1+CD4+ T-cell population. Number above each peak in far right plot represents the number of divisions. (B) Representative Cell Trace dilution plots of CD4+VB5.1+ OT-II T- cell division after 72 hour co-culture with BMDC treated with equivalent does of OVA protein. Soluble OVA (upper), (ζ+) NP-OVA (middle), and (ζ-) NP-OVA (lower). Number represents frequency OVA specific CD4+ OT-II T-cells that underwent >3 divisions. N.D., not done.

171 The ability of DCs treated with (ζ+)NP-OVA to induce robust T-cell proliferation was highly reproducible, as similar results were observed with multiple batches independently synthesized of NP-OVA (Fig. 5.4). From these data, we conclude that OVA conjugated NPs lower the amount of antigen needed to induce T-cell proliferation, with cationic NP requiring ~100 fold less antigen, which suggests that (ζ+)NP induce more potent antigen presenting cells.

Figure 5.4. CD4+ OT-II T-cell proliferation in response to NP-OVA treated BMDC. Combined data from 3 experiments using independently synthesized NP-OVA batches. **p<0.001 2-way ANOVA with Sidak’s multiple comparisons test.

5.3.3. In vitro evaluation of upregulation of co-stimulatory receptors, cytokines and chemokines on BMDCs The ability of BMDC treated with cationic NP to induce strong OTII T-cell proliferation at ~100 fold lower concentrations than soluble OVA could be explained by multiple mechanisms. The simplest explanation is that cationic NPs bind more readily to the negatively charged BMDC surface, resulting in increased antigen uptake and presentation compared to anionic NP or soluble OVA. We tested this possibility by incorporating a pH

172 sensitive dye (pHrodo) into the NP during fabrication that fluoresces upon NP internalization in the endosome (Fig. 5.5).

Figure 5.5. Internalization of cationic and anionic OVA-conjugated NP by BMDC using the pH-sensitive dye pHrodo. Cells were incubated with 10ug/ml of (ζ-) NP-OVA (upper) or (ζ+) NP-OVA (lower) at 4°C (no internalization) and at 37°C for the indicated times. Number indicates the frequency of BMDCs in the pHrodo+ gate that have internalized NP (gate set based on 4°C control).

The largest difference we observed in NP uptake occurred at the 24 hour time point with 20% fewer pHrodo+ cells in (ζ-)NP-OVA treated BMDC cultures compared to (ζ+)NP-OVA. This difference began tapering at 48 hours and by 72 hours, ≥94% of DC were pHrodo+ regardless of NP charge, suggesting that internalization at longer time points is not grossly different between (ζ+)NP-OVA and (ζ-)NP-OVA. We also hypothesized that differences in T-cell proliferation in response to NP treated DC could be due to increased surface MHCII and T-cell co-activating receptor expression on DC treated with cationic NP. Flow cytometric analysis indicated that surface MHCII, CD80 and CD86 were upregulated at 72 hours post (ζ+)NP-OVA treatment, which were comparable to the expression levels observed on LPS treated DC after 24 hours (Fig. 5.6A). In contrast, anionic NP only induced a modest increase MHCII and co-receptor expression compared to

173 untreated controls. Interestingly, the maximal level of MHCII and co-receptor expression induced by (ζ+)NP-OVA was dependent on direct OVA conjugation to the NP surface, as blank (ζ+)NP, or (ζ+)NP administered with soluble OVA, were unable to induce strong surface expression of these molecules (Fig. 5.6A). The increase in co-receptor expression was not due to general upregulation of cell surface molecules, as CD11b expression remained constant amongst all treatment groups. Similar results were observed at the mRNA level using qRT-PCR for CD80 and CD86 at 24 hours post NP treatment; however, the mRNA levels of H2aa (MHCII encoding) remained unchanged, suggesting that upregulation of surface MHCII occurs post-transcriptionally (Fig 4.6B).

Figure 5.6. MHCII and T-cell co-receptor expression by BMDC following NP-OVA treatment. (A) Flow cytometric analysis of surface MHCII (IA/IE), CD80, CD86 and CD11b (control) expression on CD11b+CD11c+ gated BMDC 72 hours after NP treatment. Dashed line indicates average expression level on untreated cells (UT), number indicates MFI. (B) qRT-PCR for mRNA expression of Cd80 Cd86 and H2aa (MHCII) by cells treated for 48 hours as in A. Data are normalized to β-actin (Actb) mRNA and graphed as fold change over UT. Equivalent OVA dose [1 µg/ml] corresponds to [10µg/mL] NP dose; LPS treatment for 24 hours at [10 ng/ml]. *p<0.05, **p<0.001; 1-way ANOVA with Tukey’s multiple comparisons test. LPS serves as a positive control for the assay and was excluded from statistical analysis.

174 We further tested whether BMDC cytokine profiles changed following NP treatment and if NP charge was a contributing factor. We found significant increases in IL-6 and IL-12 mRNA expression and protein secretion by DC treated with (ζ+)NP-OVA compared untreated and (ζ-)NP-OVA treatment (Fig. 5.7).

Figure 5.7. Cytokine and chemokine expression by NP-OVA treated BMDC. (A) qRT-PCR for mRNA expression of indicated cytokines by NP treated BMDC at 24 hours (Ifnb, 48 hours). Data are normalized to β-actin (Actb) mRNA and graphed as fold change over untreated control. Equivalent OVA dose [1ug/ml]; LPS dose [10ng/ml]. *p<0.05, **p<0.001; 1-way ANOVA with Tukey’s multiple comparisons test. LPS serves as a positive control for the assay and was excluded from statistical analysis. Representative of 2 independent experiments using independently synthesized NP and NP-OVA batches.

Increases in other key cytokines/chemokines, including Il1b, Il18, Cxcl10, l10 and Ifnb, were significantly upregulated after DC treatment with (ζ+)NP-OVA, as compared to untreated or (ζ-)NP-OVA treated DC (Fig 5.7A). Significant increases in Il18 and Cxcl10 were also observed DC treated with (ζ-)NP-OVA compared to untreated controls; however at a lower level than (ζ+)NP-OVA (Fig. 5.7A). Il1b mRNA expression was elevated in (ζ+)NP-OVA

175 treated DC compared to untreated controls (Fig. 5.7A); however, no IL-1β protein was detected in the supernatant (data not shown), which is consistent with previous findings that this formulation of NPs does not induce inflammasome activation 29. We also assessed the mRNA expression of several other cytokines, including IL-4, and found that they were either not expressed or were no different than untreated cells (not shown). Similar to T-cell co- receptor expression, the increased cytokine expression/secretion required direct OVA conjugation to the NP, as blank NP and blank NP + soluble OVA did not induce strong cytokine responses (Fig. 5.7). This observation suggests that the cytokine responses and increased co-receptor expression are not due to TLR-ligand contaminants (e.g. endotoxin) within the OVA stock, and rather suggests that cationic NPs induce a unique activation profile within DCs that also requires direct conjugation of protein antigen to the NP surface.

5.3.4. In Vivo Pulmonary Immunization It is well established that effective antibody responses to protein antigens require CD4+ T-cell help and in the absence of T-cells, affinity maturation, a process that enhances antibody specificity, and Ig class switch recombination (CSR) from IgM to IgG, IgE or IgA are severely hindered 10. Affinity maturation and CSR occur within specified structures in the spleen and LNs called germinal centers (GC) that are comprised mostly of proliferating B cells that can be identified by upregulation of the surface marker GL7. We used a model of orotracheal NP lung instillation to assess primary and secondary CD4+ T-cell dependent immune responses, including GC formation and antibody CSR using the schedule described in Figure 5.8.

176

Figure 5.8. Immunization schedule: Primary (1°) orotracheal instillation of 10 µg OVA conjugated to NP (100 µg NP) and soluble controls was performed on day 0 with secondary (2°) immunizations occurring on day 10. Mice were bled or euthanized on day 9 or day 20 to obtain plasma and BALF.

We tested whether particulate OVA can induce GC formation in the medistinal LN following primary and secondary lung instillations and found a strong induction of CD19+GL7+ cells GC B-cells in (ζ+)NP-OVA treated mice that was significantly higher than (ζ-)NP-OVA or soluble OVA alone (Fig. 5.9). These changes in GC B-cell populations in (ζ+)NP-OVA treated mice were similar to those treated with soluble CpG/OVA and were confined to the draining mediastinal LN and not observed in the spleen. This suggests that the induction of GC B-cells was due to localized adjuvant activity in the lung and not due to systemic increases in GC B-cell populations in these mice.

177

Figure 5.9. (A) Gating scheme used to define the GC B cell population (CD19+GL7+) in spleen and LN of immunized mice. Data for 1 and 2 immunization are representative of 2 independent experiments. (B) Representative flow plots for frequency of CD19+GL7+ GC B cells in the mediastinal LN following 1° and 2° immunization. (C) Combined data from B; 1° immunization, (n≥5) per group; 2° immunization, n≥10 per group. Data in 1° are representative of 2 independent experiments. Data in 2° are combined from 2 independent experiments. Each experiment used independently synthesized NP and NP-OVA batches.

178 We also investigated the activation status of CD4+ T-cells in the mediastinal lymph node after the secondary instillation. We found a significant increase in the antigen experienced CD4+CD44hiCD62lo T-cell population in the mediastinal LN of (ζ+)NP-OVA and soluble OVA/CpG instilled mice compared to mice treated soluble OVA alone (Fig. 5.10). In contrast, the antigen experienced CD4+ T-cell population was not increased in (ζ-)NP-OVA treated mice compared to soluble OVA alone and we found no difference in the CD4+CD44hiCD62lo T-cell populations in spleen, suggesting that the increase in T effector memory cells is specific to the dLN.

Figure 5.10. CD4+ T-cell activation following pulmonary immunization. (A) Gating scheme used to define the antigen experienced CD4+CD62LloCD44hi T-cells in spleen and LN of immunized mice. (B) Combined data for frequency of CD4+CD62LloCD44hi T-cells in spleen and LN following 2° (n≥10) lung immunizations. Data are combined from 2 independent experiments.

179 Consistent with increased GC formation and antigen experienced CD4+ T-cell populations, we readily detected OVA-specific IgG in the plasma and BALF of (ζ+)NP-OVA treated mice following primary and secondary immunization with titers comparable to those treated with soluble OVA/CpG (Fig. 5.11).

Figure 5.11. OVA-specific IgG antibody titers in plasma and BAL of mice following pulmonary immunization with OVA-NP. (A) Plasma IgG (n≥15), BALF IgG (n≥5) following 1° immunization. (B) Plasma IgG, BALF IgG (n≥10) following 2° immunization. *p<0.05, **p<0.001; 1-way ANOVA with Tukey’s multiple comparisons test. Data in 1° are representative of 2 independent experiments. Data in 2° are combined from 2 independent experiments.

Additionally, three of six mice treated with (ζ+)NP-OVA had detectable levels of OVA specific IgA in BALF, whereas no antigen-specific IgA was detected in any other group following primary immunization, including soluble OVA/CpG (Fig. 5.12).

180

Figure 5.12. OVA-specific IgA antibody titers in BAL of mice following pulmonary immunization with OVA-NP. *p<0.05, **p<0.001; 1-way ANOVA with Tukey’s multiple comparisons test. Data in 1° are representative of 2 independent experiments (n≥5). Data in 2° are combined from 2 independent experiments (n≥10).

Mice treated with (ζ-)NP-OVA had very low OVA-specific plasma and BALF IgG antibody titers that were not statistically different from mice treated with soluble OVA alone (Fig. 5.11). OVA-specific BALF IgA was only detectable in (ζ+)NP-OVA and soluble OVA/CpG treated mice following secondary immunization (Fig. 5.12), indicating that (ζ+)NP-OVA has an adjuvant affect capable of inducing similar systemic and mucosal antibody responses to the TLR ligand CpG that is not observed with (ζ-)NP-OVA. In order to determine whether the adjuvant effect(s) of (ζ+)NP-OVA work through a similar mechanism to CpG, we co- delivered soluble CpG with (ζ+)NP-OVA and investigated the presence of additive effects on OVA-specific antibody titers. We observed that co-delivery of soluble CpG with (ζ+)NP- OVA significantly increased levels of OVA-specific BALF IgG and IgA compared to soluble OVA/CpG (Fig. 5.13), suggesting that (ζ+)NP-OVA and CpG work in an additive fashion to induce more robust antibody responses, especially for mucosal IgA.

181

Figure 5.13. OVA-specific IgG and IgA antibody titers in BAL of mice following pulmonary immunization with OVA-NP and soluble CpG. *p<0.05, **p<0.001; 1-way ANOVA with Tukey’s multiple comparisons test. Data are combined from 2 independent experiments (n≥10).

We also observed that co-delivery of soluble CpG rescued the ability of (ζ-)NP-OVA to induce BALF IgG and IgA responses (Fig. 5.13), suggesting that anionic NPs can still serve as a platform for antigen delivery, but do not contain the inherent adjuvant activity displayed by cationic NPs. Finally, we performed pulmonary immunizations in MyD88 knockouts (MyD88-/-) to evaluate if the observed adjuvant-like property of (ζ+)NP-OVA is MyD88 dependent. This would indicate adjuvant properties independent from most TLR signaling.30, 31 As shown in Figure 5.14, OVA-specific responses were observed in the BALF and serum only for (ζ+)NP- OVA, although at smaller titers than previously observed in WT mice. No signal was observed for co-delivery of soluble OVA and CpG, as expected.

182

Figure 5.14. OVA-specific IgG and IgA antibody titers in plasma and BAL of MyD88-/- mice following pulmonary immunization with NP-OVA. (n≥4).

From these data, we conclude that cationic NPs contain an inherent adjuvant activity in the lung that is equivalent to, and can further synergize with, known TLR-stimulating adjuvants such as CpG.

5.4. Discussion Our studies characterize the effects of NP surface charge on the induction of T-cell dependent mucosal and systemic antibody responses. By using PRINT and a controlled OVA-conjugation scheme to include the same number of reactive sites, we ensured identical antigen loading between cationic and anionic NPs, allowing equivalent mass and numbers of particles for both (ζ+)NP-OVA and (ζ-)NP-OVA formulations to be delivered at equivalent doses. Our results show that cationic hydrogel NPs have an adjuvant-like effect that yields potent local and systemic antibody responses following pulmonary delivery. These responses appear T-cell dependent, as they increased GC formation and induced Ig class switch recombination, which are known to require T-cell help. In vivo, we found a significant increase in the presence of antigen experienced (CD44hiCD62Llo) CD4+ T-cells in the dLN + following (ζ )NP-OVA instillation, suggesting that a T-helper (TH) cell response was

183 induced; however, more studies are needed to characterize type of TH response induced. This may include T-follicular helper cells, since there were increased GC B-cell populations and we hypothesize that there may be a TH1 component supported by our in vitro studies showing that (ζ+)NP-OVA treated DCs induced an array of cytokine expression, including some known for TH1 skewing (IL-12, IL-18), but not TH2 skewing (IL-4). Additional ex vivo studies demonstrated enhanced antigen specific CD4+ T-cell responses to (ζ+)NP-OVA treated DCs, suggesting that improvement of humoral and mucosal responses in vivo may emanate from enhanced DC maturation. Consistent with this, we observed increased expression of T-cell activating co-receptors and MHCII on (ζ+)NP-OVA treated DC in vitro. While these and other cationic nanostructures have been used as potential adjuvants and/or vaccine carriers several studies, 20-23, 32-37 to our knowledge, this is the first report of particle surface charge being the primary variable in driving T-cell dependent antibody responses in vivo. Previous vaccine related studies using cationic liposomes showed increased antibody responses over soluble controls; however, this response was solely attributed to enhanced binding to APC surface and no differences in co-stimulatory molecule or cytokine expression were reported.34, 36, 38 We observed that cationic NPs bound more readily to DC and macrophage cell surfaces than anionic NPs (data not shown); however, the degree of NP internalization was not grossly different between these groups. Our finding that T-cell activating co-receptors and cytokines are induced with antigen-bound cationic NPs, but not with anionic NPs, suggests a clear role of NP charge in DC maturation and effector function. Similar studies using positively charged chitosan based-NPs have also been shown to enhance immune responses over whole protein controls 21, 33, 39 and recent cationic NP studies using the macrophage cell line RAW 264.7 have also demonstrated upregulation of T-cell co- stimulatory molecules and cytokines of CD80, CD86, IL-6, and IL-12 among others.32 Interestingly, our studies suggest that the positive charge of the NP alone is not sufficient for upregulation of co-receptor and cytokine production from DC and that maximal expression requires protein conjugation to the cationic NP surface. This suggests that protein on the surface of cationic NPs is sensed by the DC and the combined effects of NP size, shape,

184 charge and protein content may provide microbial mimicry that triggers DC maturation. Additionally, residual endotoxin contamination from purchased OVA may contribute to enhanced stimulation with the added protein. Importantly, particles of this size (~200nm) have been established as potent simulators of CD4+ T-cell responses in the lung,40 but require active transport form the lung to the dLN which is dependent on lung resident DCs.41, 42 Other studies have used cationic NP formulations as proton sponges to induce endosomal escape, allowing NP cargo to access the cytosol for siRNA delivery or antigen cross- presentation via the MHC I pathway to elicit cytotoxic CD8+ T-cell responses for cancer vaccines.22, 23 The role of NP-surface charge on CD8+ T-cell responses following pulmonary vaccination requires a complimentary set of experiments that are currently underway. Other groups have shown enhancement of CD8+ T-cell responses using elegant NP systems and pulmonary administration, but it is unclear if cationic charge on PRINT based NPs will have an additive effect on harnessing endosomal escape.24, 25 The response of DCs to (ζ+)NP-OVA overlapped with those induced by the TLR4- ligand LPS; however, the kinetics and magnitude of these responses differed. While surface MHC-II and T-cell co-receptor expression by (ζ+)NP-OVA treated DC ultimately reached that of LPS, it took the NP-treated DC an additional 48 hours in vitro. Also, the magnitude of cytokine expression by LPS treated DC was considerably higher than (ζ+)NP-OVA treated DC at the doses tested. However, concerning levels of LPS endotoxin was detected in the purchased OVA stock from Sigma, as has been reported elsewhere, which may be concentrated on NP during particle functionalization.43 However, both cationic and anion particles were conjugated with the same levels of protein, amounting to equivalent levels of potential LPS contamination. OVA-specific antibody generation in MyD88-/- further suggests that the inherent adjuvant activity of these particles is independent of canonical TLR signaling, the main source of stimulation from LPS impurities. Finally, the addition of soluble CpG to cationic NPs induced stronger IgG and IgA responses in the lung, suggesting that cationic particles and CpG signal through distinct, but complimentary pathways to enhance antibody based immune responses. Co-delivery of soluble CpG with anionic NPs was sufficient to rescue antigen specific BALF IgG and IgA responses in vivo, suggesting

185 that antigen on anionic NPs can be processed and presented, but that these particles lack the inherent adjuvant properties found with cationic NP. NP design features are well established for long circulating intravenous delivery; however, there is little known regarding NP design parameters for other applications and routes of administration, especially pulmonary delivery. Mucosal IgA prevents infection by sequestering pathogens to the mucous layer and preventing attachment to the mucosal epithelium. Mucosal vaccines tend to provide antibody based protection at distal mucosal sites in addition to the site of primary immunization, which means that vaccines targeted against intestinal and urogenital pathogens could be delivered via the pulmonary route.2, 3 This may greatly expand the use of pulmonary vaccines beyond pathogens with lung etiology. Our work demonstrates the role of NP charge on CD4+ T-cell mediated responses to a pulmonary vaccine, but the immunostimulatory effects of cationic hydrogel NPs may have applications in promoting clearance of endosomal pathogens by activating a subset of effector CD4+ T-cells critical for managing phagosomal infections (e.g. tuberculosis),44 or for ameliorating asthma by skewing away from allergy promoting TH2 responses to TH1. Soluble

CpG has been delivered to the lung to shift the balance TH2 to TH1 with some therapeutic success, but NP delivery may offer a more controlled and sustained approach.45, 46

5.5. Conclusions Well-defined nanoparticles offer tremendous opportunities for a new generation of vaccination. The degree of control and NP scalability afforded by PRINT allows us to design more complicated vaccine carriers in the future, with the capacity to alter NP dimensions, utilize cleavable linkers for antigen and adjuvant conjugation and formulate aerosols to provide a portable pulmonary route of administration.29, 47, 48 Ultimately, understanding how these particle parameters affect immune programing in the context of vaccination route will be important for eliciting intended immune responses. Our finding that cationic NPs have an adjuvant effect when delivered to the lung will hopefully contribute to future pulmonary based vaccine development.

186 5.6. Acknowledgements We thank R. Roberts, C. Luft, T. Rahhal, K. Reuter, J. Perry, S. Tian, A. Pandya, N. Fisher, S. Coquery, for useful discussions and technical assistance. We acknowledge Liquidia Technologies for providing PRINT molds, and the core facilities at UNC, including the Nucleic Acids Core Facility, CHANL imaging facility and DLAM facility. This work was funded in part by the NIH Pioneer Award to J.M.D. (1DP1OD006432).

5.7. References 1. Blank F, Stumbles P, von Garnier C. Opportunities and challenges of the pulmonary route for vaccination. Expert Opin Drug Del. 2011, 8, 547-563.

2. Borges O, Lebre F, Bento D, Borchard G, Junginger HE. Mucosal vaccines: Recent progress in understanding the natural barriers. Pharm Res. 2010, 27, 211-223.

3. Neutra MR, Kozlowski PA. Mucosal vaccines: The promise and the challenge. Nat Rev Immunol. 2006, 6, 148-158.

4. Sou T, Meeusen EN, de Veer M, Morton DA, Kaminskas LM, McIntosh MP. New developments in dry powder pulmonary vaccine delivery. Trends Biotechnol. 2011, 29, 191-198.

5. LiCalsi C, Christensen T, Bennett JV, Phillips E, Witham CL. Dry powder inhalation as a potential delivery method for vaccines. Vaccine. 1999, 17, 1796 - 1803.

6. Garcia-Contreras L, Wong YL, Muttil P, Padilla D, Sadoff J, Derousse J, Germishuizen WA, Goonesekera S, Elbert K, Bloom BR, Miller R, Fourie PB, Hickey A, Edwards D. Immunization by a bacterial aerosol. Proc Natl Acad Sci U S A. 2008, 105, 4656-4660.

7. Fujihashi K, Koga T, van Ginkel FW, Hagiwara Y, McGhee JR. A dilemma for mucosal vaccination: Efficacy versus toxicity using enterotoxin-based adjuvants. Vaccine. 2002, 20, 2431 - 2438.

8. de Swart RL, LiCalsi C, Quirk AV, van Amerongen G, Nodelman V, Alcock R, Yuksel S, Ward GH, Hardy JG, Vos H, Witham CL, Grainger CI, Kuiken T, Greenspan BJ, Gard TG, Osterhaus AD. Measles vaccination of macaques by dry powder inhalation. Vaccine. 2007, 25, 1183-1190.

187 9. Pulliam B, Sung JC, Edwards DA. Design of nanoparticle-based dry powder pulmonary vaccines. Expert Opin Drug Del. 2007, 4, 651 - 663.

10. Jacobson EB, Caporale LH, Thorbecke GJ. Effect of thymus cell injections on germinal center formation in lyphoid tissues of nude (thymusless) mice. Cell Immunol. 1974, 13, 416-430.

11. Inaba K, Steinman RM, Van Voorhis WC, Muramatsu S. Dendritic cells are critical accessory cells for thymus-dependent antibody responses in mouse and in man. Proc Natl Acad Sci U S A. 1983, 80, 6041-6045.

12. Bachmann MF, Jennings GT. Vaccine delivery: A matter of size, geometry, kinetics and molecular patterns. Nat Rev Immunol. 2010, 10, 787-796.

13. De Temmerman ML, Rejman J, Demeester J, Irvine DJ, Gander B, De Smedt SC. Particulate vaccines: On the quest for optimal delivery and immune response. Drug Discov Today. 2011, 16, 569-582.

14. Moon JJ, Huang B, Irvine DJ. Engineering nano- and microparticles to tune immunity. Adv Mater. 2012, 24, 3724-3746.

15. Smith DM, Simon JK, Baker JR, Jr. Applications of nanotechnology for immunology. Nat Rev Immunol. 2013, 13, 592-605.

16. Zhao L, Seth A, Wibowo N, Zhao CX, Mitter N, Yu C, Middelberg AP. Nanoparticle vaccines. Vaccine. 2014, 32, 327-337.

17. Hardy CL, Lemasurier JS, Mohamud R, Yao J, Xiang SD, Rolland JM, O'Hehir RE, Plebanski M. Differential uptake of nanoparticles and microparticles by pulmonary apc subsets induces discrete immunological imprints. J Immunol. 2013, 191, 5278-5290.

18. Kunda NK, Somavarapu S, Gordon SB, Hutcheon GA, Saleem IY. Nanocarriers targeting dendritic cells for pulmonary vaccine delivery. Pharm Res. 2013, 30, 325-341.

19. Wang YY, Lai SK, Suk JS, Pace A, Cone R, Hanes J. Addressing the peg mucoadhesivity paradox to engineer nanoparticles that "slip" through the human mucus barrier. Angew Chem Int Ed Engl. 2008, 47, 9726-9729.

20. Debin A, Kravtzoff R, Santiago JV, Cazales L, Sperandio S, Melber K, Janowicz Z, Betbeder D, Moynier M. Intranasal immunization with recombinant antigens associated

188 with new cationic particles induces strong mucosal as well as systemic antibody and ctl responses. Vaccine. 2002, 20, 2752-2763.

21. Gupta NK, Tomar P, Sharma V, Dixit VK. Development and characterization of chitosan coated poly-(varepsilon-caprolactone) nanoparticulate system for effective immunization against influenza. Vaccine. 2011, 29, 9026-9037.

22. Kwon YJ, Standley SM, Goh SL, Frechet JM. Enhanced antigen presentation and immunostimulation of dendritic cells using acid-degradable cationic nanoparticles. J Control Release. 2005, 105, 199-212.

23. Wilson JT, Keller S, Manganiello MJ, Cheng C, Lee C, Opara C, Convertine A, Stayton PS. Ph-responsive nanoparticle vaccines for dual-delivery of antigens and immunostimulatory oligonucelotides. ACS Nano. 2013, 7, 3912-3925.

24. Nembrini C, Stano A, Dane KY, Ballester M, van der Vlies AJ, Marsland BJ, Swartz MA, Hubbell JA. Nanoparticle conjugation of antigen enhances cytotoxic t-cell responses in pulmonary vaccination. Proc Natl Acad Sci U S A. 2011, 108, E989-997.

25. Li AV, Moon JJ, Abraham W, Suh H, Elkhader J, Seidman MA, Yen M, Im E, Foley MH, Barouch DH, Irvine DJ. Generation of effector memory t cell-based mucosal and systemic immunity with pulmonary nanoparticle vaccination. Science Translational Medicine. 2013, 5, 204-130.

26. Guzman J, Iglesias MT, Riande E. Synthesis and polymerization of acrylic monomers with hydrophilic long side groups. Polymer. 1997, 38, 5227-5232.

27. Perry JL, Reuter KG, Kai MP, Herlihy KP, Jones SW, Luft JC, Napier M, Bear JE, DeSimone JM. Pegylated print nanoparticles: The impact of peg density on protein binding, macrophage association, biodistribution, and pharmacokinetics. Nano Lett. 2012, 12, 5304-5310.

28. Van den Broeck W, Derore A, Simoens P. Anatomy and nomenclature of murine lymph nodes: Descriptive study and nomenclatory standardization in balb/canncrl mice. J Immunol Methods. 2006, 312, 12-19.

29. Roberts RA, Shen T, Allen IC, Hasan W, DeSimone JM, Ting JP. Analysis of the murine immune response to pulmonary delivery of precisely fabricated nano- and microscale particles. PLoS One. 2013, 8, e62115.

189 30. De Gregorio E, D'Oro U, Wack A. Immunology of tlr-independent vaccine adjuvants. Curr Opin Immunol. 2009, 21, 339-345.

31. Iwasaki A, Medzhitov R. Toll-like receptor control of the adaptive immune responses. Nat Immunol. 2004, 5, 987-995.

32. Koppolu B, Zaharoff DA. The effect of antigen encapsulation in chitosan particles on uptake, activation and presentation by antigen presenting cells. Biomaterials. 2013, 34, 2359-2369.

33. Slutter B, Jiskoot W. Dual role of cpg as immune modulator and physical crosslinker in ovalbumin loaded n-trimethyl chitosan (tmc) nanoparticles for nasal vaccination. J Control Release. 2010, 148, 117-121.

34. Christensen D, Agger EM, Andreasen LV, Kirby D, Andersen P, Perrie Y. Liposome- based cationic adjuvant formulations (caf): Past, present, and future. J Liposome Res. 2009, 19, 2-11.

35. Henderson A, Propst K, Kedl R, Dow S. Mucosal immunization with liposome-nucleic acid adjuvants generates effective humoral and cellular immunity. Vaccine. 2011, 29, 5304-5312.

36. Korsholm KS, Agger EM, Foged C, Christensen D, Dietrich J, Andersen CS, Geisler C, Andersen P. The adjuvant mechanism of cationic dimethyldioctadecylammonium liposomes. Immunology. 2007, 121, 216-226.

37. Yu F, Wang J, Dou J, Yang H, He X, Xu W, Zhang Y, Hu K, Gu N. Nanoparticle-based adjuvant for enhanced protective efficacy of DNA vaccine ag85a-esat-6-il-21 against mycobacterium tuberculosis infection. Nanomedicine. 2012, 8, 1337-1344.

38. Ma Y, Zhuang Y, Xie X, Wang C, Wang F, Zhou D, Zeng J, Cai L. The role of surface charge density in cationic liposome-promoted dendritic cell maturation and vaccine- induced immune responses. Nanoscale. 2011, 3, 2307-2314.

39. van der Lubben IM, Verhoef JC, Borchard G, Junginger HE. Chitosan and its derivatives in mucosal drug and vaccine delivery. Eur J Pharm Sci. 2001, 14, 201-207.

40. Stano A, Nembrini C, Swartz MA, Hubbell JA, Simeoni E. Nanoparticle size influences the magnitude and quality of mucosal immune responses after intranasal immunization. Vaccine. 2012, 30, 7541-7546.

190 41. Choi HS, Ashitate Y, Lee JH, Kim SH, Matsui A, Insin N, Bawendi MG, Semmler- Behnke M, Frangioni JV, Tsuda A. Rapid translocation of nanoparticles from the lung airspaces to the body. Nat Biotechnol. 2010, 28, 1300-1303.

42. Furuhashi K, Suda T, Hasegawa H, Suzuki Y, Hashimoto D, Enomoto N, Fujisawa T, Nakamura Y, Inui N, Shibata K, Nakamura H, Chida K. Mouse lung cd103+ and cd11bhigh dendritic cells preferentially induce distinct cd4+ t-cell responses. Am J Respir Cell Mol Biol. 2012, 46, 165-172.

43. Watanabe J, Miyazaki Y, Zimmerman GA, Albertine KH, McIntyre TM. Endotoxin contamination of ovalbumin suppresses murine immunologic responses and development of airway hyper-reactivity. J Biol Chem. 2003, 278, 42361-42368.

44. Tubo NJ, Jenkins MK. Cd4+ t cells: Guardians of the phagosome. Clin Microbiol Rev. 2014, 27, 200-213.

45. Pesce I, Monaci E, Muzzi A, Tritto E, Tavarini S, Nuti S, De Gregorio E, Wack A. Intranasal administration of cpg induces a rapid and transient cytokine response followed by dendritic and natural killer cell activation and recruitment in the mouse lung. J Innate Immun. 2010, 2, 144-159.

46. Constabel H, Stankov MV, Hartwig C, Tschernig T, Behrens GM. Impaired lung dendritic cell migration and t cell stimulation induced by immunostimulatory oligonucleotides contribute to reduced allergic airway inflammation. J Immunol. 2009, 183, 3443-3453.

47. Fromen CA, Shen TW, Larus AE, Mack P, Maynor BW, Luft JC, DeSimone JM. Synthesis and characterization of monodisperse uniformly shaped respirable aerosols. AIChE Journal. 2013, 59, 3184-3194.

48. Garcia A, Mack P, Williams S, Fromen C, Shen T, Tully J, Pillai J, Kuehl P, Napier M, Desimone JM, Maynor BW. Microfabricated engineered particle systems for respiratory drug delivery and other pharmaceutical applications. J Drug Deliv. 2012, 2012, 941243.

191

CHAPTER SIX

Summary and Outlook

192 6.1.Summary The main objectives of this PhD research included (1) the development of a calibration-quality aerosol system using PRINT, the application of these calibration-quality aerosols to improved understanding of (2) shaped aerosols under flow and (3) their cellular fate in the lung, and applying this knowledge towards the (4) development of a mucosal vaccine. Utilizing the PRINT platform for pulmonary drug delivery allowed us to generate precision-controlled aerosol particles in a scalable manner. We successfully optimized various PRINT formulations and developed methods for generating aerosols, including an approach that led to truly monodisperse shaped aerosols (Chapter 2). From this experimental approach, we developed a model to rapidly characterize the particle’s dynamic shape factor by adjusting for changes in particle drag experienced in the ultra-Stokesian flow regime of an APS (Chapter 3). Prior to this work, monodispersed, non-spherical aerosols had not been demonstrated in the literature, and characterization of dynamic shape factors remained an empirical process. The method demonstrated here is a translatable approach to predict the aerodynamic diameter of a non-spherical particle and thus area of deposition in a human lung. Feeling confident that we understood the role physical particle properties on airway deposition, we next set out to characterize the role of particle properties in the cellular association in the lung, focusing specifically on airway APC (Chapter 4). For the first time, we demonstrated the role of particle charge on airway APC association, finding that cationic particles were preferentially associated with lung DCs over MC, giving them a distinct advantage over anionic particles for vaccine platforms. Additionally, these particles were shown to traffic to the lung draining LNs, which is a critical step for a vaccine platform. From this fundamental knowledge of the fate of particles in the lung, we developed a model nanoparticle vaccine and demonstrated that the surface charge was critical in magnitude of the resulting immune response (Chapter 5). These cationic NP resulted in increased production of both systemic and mucosal antibodies, demonstrating the benefit of pulmonary vaccination and the importance of surface charge in this T-cell dependent response.

193 6.2.Impact and Future Directions 6.2.1. PRINT as a Platform for Pulmonary Drug Delivery The PRINT fabrication approach predictably controls particle geometric and aerodynamic features, a differentiating attribute as compared to traditional particle generation approaches.1, 2 In particular, micromolding strategies such as PRINT represent one of the only methods to precisely control particle shape and size. For PRINT, the particle geometry is directly derived from the semiconductor wafer, bringing inherent nanoscale precision to the particle geometry and offering the capability to generate unique, non-spherical shapes of a wide range of materials with minimal batch-to-batch variability.1 This is absolutely essential in preparation of a pulmonary therapeutic, offering tremendous advantage over current pulmonary therapeutics; simply reformulating a commercial jet-milled therapeutic zanamivir (Relenza, GlaxoSmithKline) with PRINT resulted in ~3-4 times the respirable dose.

Figure 6.1. Favorable properties of PRINT aerosols for dry powder pharmaceutical use. Comparison of 1.5µm donut PRINT-zanamivir particles against the marketed product Relenza (active pharmaceutical ingredient zanamivir) sized using a Next Generation Impactor (NGI). PS = pre-separator, n=3.1

194 From these results and those presented in Chapter 2, it is apparent that the presence of monodisperse particles does not immediately translate into the formation of monodisperse aerosols. The work presented in Chapter 2 offers an important start to the understanding of the various steps of formulating a consistent, well-controlled dry powder aerosol, including the role of particle materials, particle geometry, method of aerosol-dispersion and even basic lyophilization approaches. Given the calibration-quality formation of PRINT particles, future studies should delve into many important fundamental details of how particle composition and geometry affects particle packing; how this influences lyophilization and powder formation; how macroscale powder structure affects deagglomeration and aerosolization. Specifically, atomic force microscopy (AFM) experiments varying particle shape and composition should be performed using controlled particle orientations on specialized AFM tips to quantify particle-particle interactions.3 Moreover, these fundamental studies could lead to the design of particle compositions and geometries which minimize interactions, leading to predictable aerosol formation and thus quality-by-design pharmaceutics. Without these, current product development at the commercial level will require optimization of these steps for each formulation.

6.2.2. Characterization of Monodisperse PRINT Aerosols To the knowledge of the authors, we are the first to demonstrate the generation of a monodisperse population of non-spherical aerosols, demonstrated in Chapter 2 and 3. Additionally, we are the first to produce a systematic model to determine the shape factor of a particle of known physical geometry (Chapter 3). This is a critical step in employing particle geometry to tailor airway deposition. Similar to the modulation of particle density using porous particles, variation of particle shape offers an additional parameter to control both airway deposition and cellular interaction through an increased element of particle design. Moreover, the shape factors determined in this work can be applied to a particle of any volume with the same shape, by maintaining the exact ratios of dimensions of the geometries studied here. The new particle would have a defined volume, shape factor, and density, allowing for a priori determination of the particle’s aerodynamic diameter. Thus,

195 airway deposition can be independently modified while maintaining the features of a given particle shape. Continuing from the work presented here, a new line of PRINT molds could be developed which scale the same shapes studied to span pre-determined DAE while truly maintaining the same physical geometries. The fluid dynamics modelling used to determine shape factors presented in Chapter 3 is rather simple and makes a number of assumptions which reduces the power of such an approach. Indeed, individual dynamics of non-spherical particles are not considered, rather assuming bulk drag properties. While these assumptions were sufficient to correct for the drag in an ultra-Stokesian APS nozzle, they are widely insufficient for predictive lung deposition models. Computational fluid dynamics approaches should be performed on the various shapes presented here to first validate APS findings and model non-spherical particle trajectories in simple flow conditions. From these predictive dynamics, non-spherical particle trajectories could be allowed to interact with independently modeled airflow in a representative lung geometry. An example of preliminary work using this approach is shown in Figure 6.2, which suggests the ability of shape to preferentially choose airway bifurcations (6.2.L). This would enable region-specific deposition as a function of shape.

196

Figure 6.2. Computer simulated trajectories of equal volume PRINT aerosols (A-I) over 100mm free fall in air under gravity and Stokes flow conditions and (J-L) in airway bifurcation. A) Disk, B) Donut, C).Ellipsoid, D) Fenestrated Ellipsoid, E) Lorenz, F) Fenestrated Lorenz, G) Lollipop, H) Helicopter, I) Pollen, J) Sphere, K) Donut and L) Lorenz. (A-I) Trajectories illustrate the variability of particle motion resulting from changes in aerosol shape. (J-L) Particle trajectories in a realistic bifurcation are shown with blue path lines and particle positions shown over the same time steps between figures from a random initial orientation. J) Spheres distribute 50:50 to right:left branches, K) Donuts are delayed from spheres but distribute ~50:50. L) Lorenz distributes 58:42 between right:left branch. J-L Courtesy of Sorin Mitran.

From these still simplistic approaches, non-spherical particles should be studied in realistic lung computational models for predictive lung deposition. These could then be applied to smart inhaler technology for translational changes in lung deposition. In combination with patient imaging data, predictive particle aerodynamics and deposition could deliver expensive and potent reagents directly to disease site. Immediate collaborations

197 should be initiated with groups with existing lung computational models and smart inhaler technologies to establish proof of concept experiments of region-specific deposition with 4-6 non-spherical particles.

6.2.3. Calibration-Quality PRINT Particles for Probing Lung Biology Pulmonary immunology has come a long way in the past two decades. While the function of AMs has been well established since before the 1930s, the conventional DC populations were only identified in 1970s and their exact function in the lung is still under investigation. As such, the majority of studies involving these cell populations attempt to uncover biological function and their immediate role in immune responses. Despite the increasing evidence that these are essential populations to target for pulmonary vaccines, there have been relatively few studies to address how particle design can preferentially target these DC populations, and the few attempts lack an appropriate antigen-loaded vaccine carrier, utilizing instead polystyrene spheres.7 Our work represents the first attempt to characterize the role of particle charge on internalization and trafficking specifically in lung APCs and uses a model antigen-loaded nanoparticle carrier, with appropriate unmodified controls. In the future, it would be beneficial to study the changes in surface receptors, cytokine and chemokine expression in vivo following particle treatment to the lung. This could be achieved by utilizing flow cytometer to sort populations of interest, extracting mRNA, and performing qt-PCR, as in the in vitro work of Chapter 5. DC receptors of interest include: MHC II, CD86 and CD80. It would also be important to monitor cytokine upregulation of IL-6, IL-4, IL-12, TNF-α, IFN-γ, IL-10, IL-8, and TGF-β. In combination, the overall immune response following particle exposure could be characterized. Additional work is also required to understand the mechanism of migration of particles and cells to the lung draining lymph nodes. Currently, it is hypothesized that particles are actively carried by lung DCs but this needs to be confirmed, as it is possible that particles are able to passively diffuse. This experiment should be performed by instilling a fluorescent membrane marker, prior to particle instillation. Upon resection of the lymph

198 nodes, cells which have migrated from the lungs would be identified through the co- localization of particle and lung cell fluorescence.8 Finally, iterations of the studies performed in Chapter 4 and suggested here should be performed on particles of varied size, shape and composition over a range of time points to gain an understanding of the kinetic ramifications. Unpublished work in the DeSimone laboratory has demonstrated the persistence of NPs in lymph nodes following a s.c. footpad injection for over 50 days. Such studies could be performed in the lung to draw parallels between administration routes. As knowledge of lung biology continues to evolve, these studies should also be expanded to include the role of other subtypes of macrophages and DCs, such as interstitial macrophages, plasmacytoid DCs, and monocyte-derived DCs, which are still being understood from a fundamental immunology standpoint. In sum, such rigorous evaluation of the interplay between lung APCs and particle vehicles would lead to particle designs capable of directing or avoiding a specific immune response.

6.2.4. PRINT Nanoparticles for Pulmonary Vaccines Both cationic and anionic nanoparticles have been studied for pulmonary vaccines, with no consensus in the literature as to the role of charge on the resulting immune response. Due to fabrication limitations, there has never been a direct comparison of the role of particle charge in vaccine carriers. Potentially stemming from a hold-over of the design rules of intravenous delivery (and thus the avoidance of cationic materials), elegant anionic NP systems have been developed which elicit a robust CD8+ T cell response in the lung, but fail to generate an improved antibody response, which is consistent to our findings.9-11 In contrast, cationic materials have often implied in pulmonary vaccine-related studies, utilizing formulations that contain specific materials empirically found to enhance immune responses. Vaccination with cationic liposomes have been shown to elicit enhanced antibody production over soluble controls; however, they do not result in increased surface marker expression or relevant cytokines, and thus their enhanced immunogenicity has been attributed to increased binding and uptake into relevant APC populations.12-15 In contrast to liposome formulations,

199 but similar to the cationic particles described here, chitosan based-NPs have also been shown to enhance immune responses over whole protein controls and recent NP studies in RAW cells have demonstrated upregulation of similar co-stimulatory markers to those studied here.16-19 Related, cationic NP formulations have been studied extensively to harness endosomal escape pathways, such as the proton sponge, to access the cytosol for siRNA delivery or antigen cross-presentation via the MHC I pathway to illicit cytotoxic CD8 T cell responses for cancer vaccines.20-22 While these and other cationic nanostructures have been used as potential adjuvants and/or vaccine carriers in numerous studies, this is the first direct isolation of particle surface charge as the main adjuvant-like variable for CD4+ T cell-driven responses. We anticipate that this well-controlled study will inform the field and help to establish firm guidelines on vaccine design characteristics. From this initial study, there are many follow-up experiments one could envision. First and foremost, endotoxin-free OVA must be used in NP formulations to corroborate the presence of adjuvant-like properties of cationic NPs. These studies are currently underway. Importantly, improved characterization of T cell responses following NP administration is needed. While CD4+ T helper cell functions were implied, these could be further classified using antigen-specific tetramers to look at improved activation and education of this small, but important subset. These studies could determine whether all antigen specific CD4+ T cells are helper cells, or if a particular particle formulation can promote expansion of cytotoxic CD4+ T cells, which is the natural cell type required to combat phagosomal pathogens. Implication of particle parameters which stimulate cytotoxic CD4+ T cells would be a paramount advance to vaccine development against such pathogens, such as tuberculosis.23, 24 These studies would require better evaluation of cytokine upregulation following pulmonary immunization, such as INF-γ isolation of OVA-specific CD4+ T cells using tetramer staining in combination with additional cell markers such as CD28.23, 25 Given promising results of the generation of cytotoxic CD4+ T cells, the OVA antigen should be replaced with novel TB-derived antigens, such as Ag85B or MVA85A.26-28 Following immunization with these cationic NPs, guinea pigs could then be challenged with TB, measuring frequency of granuloma formation and survival.27-29

200 The few NP vaccines developed for pulmonary delivery have shown increases with CD8+ T cells responses following anionic NP administration with certain adjuvants. It will be critical to investigate if these responses can be achieved with unadjuvanted cationic particles presented here. These types of responses have implications for therapeutic cancer vaccines, working to skew the local tumor suppressive environment to an active Th1 response which reinvigorates T cells to fight the tumor.11, 30 Lung cancer could benefit tremendously from such novel therapeutics, as there are currently poor prognosis for patients and over 150,000 deaths in the U.S. per year.31 Following the work in this thesis, the literature also suggests that the route of administration of cancer vaccines dramatically affects efficacy.32 First, the cationic NP-OVA formulations here should be evaluated for CD8 T cell responses following pulmonary immunizations, requiring identification of CD8 T cell subsets, through tetramer staining, quantification of TNF-α and IFN-γ producing CD8 T cells, through isolation and ex vivo OVA restimulation, and generation of CD8 memory T cell populations, through flow cytometry. Following these studies, the potency of cationic NPs could be tested in an efficacy model against a xenograft of OVA-expressing cells, acting as a model tumor antigen.11 Ultimate therapeutic translation would be demonstrated by repeating these studies, using NPs functionalized with a tumor-specific peptide. OVA provides an excellent, inexpensive model antigen which can be very informative for the development of optimal particle characteristics for vaccine delivery. However, the ultimate goal of our work is to provide a translatable therapeutic. This could be readily translated to incorporation of pathogen-derived antigens, such as Dengue. Currently, no vaccine exists for Dengue, which is responsible for over 50 million infections globally per year. One major challenge in the development of Dengue vaccines is the importance of vaccination against multiple serotypes of the virus.33 From the work presented here, cationic NPs provide superior mucosal and systemic humoral immune stimulation following pulmonary delivery. Four independent particle batches could be functionalized with different Dengue serotypes and co-delivered through pulmonary instillation to initiate the systemic immune response. Through a series of proof-of-concept experiments varying ratios of NPs

201 between serotypes in the instillation, the simultaneous and balanced immunization against all four serotypes could be realized by measuring antibody titers.

6.2.5. Immonoengineering Applications for PRINT Nanoparticles in the Lung Modification of NP surfaces with specific antigens such as OVA has been used for immunostimulatory applications, such as the generation of a prophylactic vaccine in Chapter 5. However, the potential to further direct the immune system does not end there. For example, NPs could be developed which actually suppress an overactive immune response. Preliminary work in this area has yielded very interesting results. Anionic NP-OVAs were studied in a mouse model of asthma. This model requires sensitization to OVA through two i.p. injections of OVA and alum. Following this exposure, mice are challenged via pulmonary instillation. This timeline is shown in Figure 6.3A. In this model, soluble delivery of OVA results in tremendous lung inflammation, cellular recruitment, and airway constriction, modeling asthma progression in humans. From results in Chapter 5, we hypothesized that NP-OVA formulations would produce an ever stronger immune response. However, delivery of anionic NP-OVA formulations by this same route of administration at the same OVA masses did not produce the same response (Figure 6.3B) H&E staining of lung sections following particle-OVA challenge yielded sections which resembled PBS-treated lungs, with only small pockets of cellular infiltrates. This is in stark contrast to lungs challenged with soluble OVA, which were overwhelmed with cellular infiltrates.

202

Figure 6.3. OVA allergy model. (A) Model timeline. (B) Light microscopy images of H&E stained sections of lung following challenge.

These findings are further corroborated by a lower serum IgE antibody response in NP-OVA challenged mice. In this model, OVA-specific IgE is an indicator of asthma allergy (Figure 6.4).

203

Figure 6.4. OVA-specific serum IgE levels following instillation challenge (n=3).

While our initial hypothesis that NP formulations would increase the IgE production and overall allergy response, these data suggest the opposite. From the literature, it appears that, in the absence of a danger signal or adjuvant, functionalization of antigen-fixed leukocytes or microparticles with antigens can encourage a tolerance response.34, 35 The NPs used in this study were anionic and their non-stimulatory properties were supported by results in Chapter 5. These initial findings highlight the ability of well-designed particulate systems to skew the immune system. Here, the typical Th2 asthma response was ameliorated by NP- challenge. We hypothesize that anionic NP-OVA formulations tolerized the immune system to OVA, without producing a corresponding increase in Th1 or Th17 immune profiles. From these initial studies however, it is unclear if this decreased response was merely due to restricted OVA exposure through the concentration on NP surfaces. A follow-up study is required to again sensitize via i.p. administration of OVA and alum, first challenge with NP- OVA formulations and then with soluble OVA. Our hypothesis of NP-OVA induced tolerance would be supported if reduction of the response is again observed.

204 This study potentially represents a NP-mediated change to an immune response. From the hypothesis presented in the case here, Th1, Th2, or Th17 mediated autoimmune diseases may be treated through NP-induced tolerance. Other applications of NP-mediated immune therapeutics could involve the skewing from one immune status to another in order to elevate symptoms. This has profound applications ranging from allergy, cancer and infectious diseases. Future studies would need to dramatically improve the understanding of NP and immune cell interaction. Work presented in this thesis demonstrated immune changes through interaction and phagocytosis of APCs, however future immune modulation approaches may involve non-phagocytosed particles which provide sufficient signaling directly to T cells through surface functionalization and cytokine secretion. With ever increased immunological understanding, the materials-mediated modulation of the immune system will be realized.

6.3.Outlook In an era where the lines continue to blur between engineering, materials science, pharmacy, and immunology, the potential for novel therapeutics to advance patient care is limitless.36, 37 This work highlights a tiny effort of how a well-engineered material can have dramatic therapeutic benefits in an innovative application. Advances in pulmonary drug delivery, mucosal vaccines, and immunoengineering have only just begun to scratch the surface of what is possible. Pulmonary drug delivery will continue to be studied due to the tremendous potential patient benefit; the need for improvement in device efficiency and delivery of potent biologics for lung-specific conditions remains strong. The advent of particle platforms, such as PRINT, could dramatically increase the standard of delivered dose and certainly offers opportunities to efficiently deliver expensive therapeutics, such as protein biologics and vaccines. Particle designs will continue to be advanced through precision engineering as the degree of therapeutic complexity evolves and the regional and cellular targets narrow.

205 Pulmonary vaccination is also continued to gain momentum. The importance of mucosal immunity can no longer be underestimated, and novel routes of vaccine administration will need to be employed to engender local protection. Increasing studies have shown efficacy from pulmonary administration of spray-dried vaccine antigens, and the opportunities and successes are finally starting to outweigh the initial skepticism.38, 39 We anticipate as the advantages of designed micro- and nanoparticles become increasingly apparent through systemic studies like the one presented here, these formulations too will make their way into clinical evaluation. Precisely engineered particles designed to impart a specific immune response will not be limited to vaccine designs. The work presented here is at the forefront of a new wave of immunoengineering, a new term to describe the use of designer materials to harness and redirect the immune system. Micro- and nano-particles naturally interact with the immune system and are rapidly internalized by phagocytic cells; immunoengineering approaches are poised to exploit this biological inevitability to retool those same immune cells for a new purpose. The potential implications of such an approach will be revolutionary, changing the way clinicians treat cancer, asthma, autoimmune diseases and inflammation. As mentioned in section 6.2.5, recent findings in the literature indicate that otherwise inflammatory antigens presented on a particle with no inherent danger signals can produce a tolerance response, offering a potential to treat asthma, inflammation and autoimmune diseases.34 An inverse approach has been suggested for cancer immunotherapy; tumor cells naturally suppress T cell function and skewing the immune system to Th1 response instead can work to eliminate the cancer.40 Instead of requiring particles themselves to be the therapeutic stimuli, particles will be designed to reprogram the immune response so that the appropriate inherent biological mechanisms will be activated to ameliorate disease. In the age of convergent science, patient care will see tremendous benefits through these and countless other advances in medical technology. We are excited to contribute in this field and look towards a bright research future ahead.

206 6.4.Acknowledgements The author would like to thank Reid Roberts, Greg Robbins, Ben Maynor, Pete Mack, Tammy Shen, Marc Kai, Stu Dunn, Chris Luft, and Joe DeSimone for useful discussions throughout the years advising these conclusions. Additional thanks goes to Coy Allen and Sorin Mitran for providing some of the preliminary data presented in this chapter. Grateful acknowledgement to Liquidia Technologies and funding sources: the NIH Pioneer Award to J.M.D. (1DP1OD006432) and DTRA (HDTRA1-13-1-0045).

6.5.References 1. Garcia A, Mack P, Williams S, Fromen C, Shen T, Tully J, Pillai J, Kuehl P, Napier M, Desimone JM, Maynor BW. Microfabricated engineered particle systems for respiratory drug delivery and other pharmaceutical applications. J Drug Deliv. 2012, 2012, 941243.

2. Vehring R. Pharmaceutical particle engineering via spray drying. Pharm Res. 2008, 25, 999-1022.

3. Controlled pulmonary drug delivery. New York: Springer; 2011.

4. Kleinstreuer C, Seelecke S. Inhaler system for targeted maximum drug-aerosol delivery. In: USPTO, ed.: North Carolina State University, 2011.

5. Kleinstreuer C, Zhang Z, Donohue JF. Targeted drug-aerosol delivery in the human respiratory system. Annu Rev Biomed Eng. 2008, 10, 195-220.

6. Kleinstreuer C, Zhang Z, Li Z, Roberts WL, Rojas C. A new methodology for targeting drug-aerosols in the human respiratory system. International Journal of Heat and Mass Transfer. 2008, 51, 5578-5589.

7. Blank F, Stumbles PA, Seydoux E, Holt PG, Fink A, Rothen-Rutishauser B, Strickland DH, von Garnier C. Size-dependent uptake of particles by pulmonary antigen-presenting cell populations and trafficking to regional lymph nodes. Am J Respir Cell Mol Biol. 2013, 49, 67-77.

8. von Andrian UH, Mempel TR. Homing and cellular traffic in lymph nodes. Nat Rev Immunol. 2003, 3, 867-878.

207 9. Nembrini C, Stano A, Dane KY, Ballester M, van der Vlies AJ, Marsland BJ, Swartz MA, Hubbell JA. Nanoparticle conjugation of antigen enhances cytotoxic t-cell responses in pulmonary vaccination. Proc Natl Acad Sci U S A. 2011, 108, E989-997.

10. Stano A, Nembrini C, Swartz MA, Hubbell JA, Simeoni E. Nanoparticle size influences the magnitude and quality of mucosal immune responses after intranasal immunization. Vaccine. 2012, 30, 7541-7546.

11. Li AV, Moon JJ, Abraham W, Suh H, Elkhader J, Seidman MA, Yen M, Im E, Foley MH, Barouch DH, Irvine DJ. Generation of effector memory t cell-based mucosal and systemic immunity with pulmonary nanoparticle vaccination. Science Translational Medicine. 2013, 5, 204-130.

12. Christensen D, Agger EM, Andreasen LV, Kirby D, Andersen P, Perrie Y. Liposome- based cationic adjuvant formulations (caf): Past, present, and future. J Liposome Res. 2009, 19, 2-11.

13. Henderson A, Propst K, Kedl R, Dow S. Mucosal immunization with liposome-nucleic acid adjuvants generates effective humoral and cellular immunity. Vaccine. 2011, 29, 5304-5312.

14. Korsholm KS, Agger EM, Foged C, Christensen D, Dietrich J, Andersen CS, Geisler C, Andersen P. The adjuvant mechanism of cationic dimethyldioctadecylammonium liposomes. Immunology. 2007, 121, 216-226.

15. Ma Y, Zhuang Y, Xie X, Wang C, Wang F, Zhou D, Zeng J, Cai L. The role of surface charge density in cationic liposome-promoted dendritic cell maturation and vaccine- induced immune responses. Nanoscale. 2011, 3, 2307-2314.

16. Gupta NK, Tomar P, Sharma V, Dixit VK. Development and characterization of chitosan coated poly-(varepsilon-caprolactone) nanoparticulate system for effective immunization against influenza. Vaccine. 2011, 29, 9026-9037.

17. Koppolu B, Zaharoff DA. The effect of antigen encapsulation in chitosan particles on uptake, activation and presentation by antigen presenting cells. Biomaterials. 2013, 34, 2359-2369.

18. Slutter B, Jiskoot W. Dual role of cpg as immune modulator and physical crosslinker in ovalbumin loaded n-trimethyl chitosan (tmc) nanoparticles for nasal vaccination. J Control Release. 2010, 148, 117-121.

208 19. van der Lubben IM, Verhoef JC, Borchard G, Junginger HE. Chitosan and its derivatives in mucosal drug and vaccine delivery. Eur J Pharm Sci. 2001, 14, 201-207.

20. Kwon YJ, Standley SM, Goh SL, Frechet JM. Enhanced antigen presentation and immunostimulation of dendritic cells using acid-degradable cationic nanoparticles. J Control Release. 2005, 105, 199-212.

21. Wilson JT, Keller S, Manganiello MJ, Cheng C, Lee C, Opara C, Convertine A, Stayton PS. Ph-responsive nanoparticle vaccines for dual-delivery of antigens and immunostimulatory oligonucelotides. ACS Nano. 2013, 7, 3912-3925.

22. Varkouhi AK, Scholte M, Storm G, Haisma HJ. Endosomal escape pathways for delivery of biologicals. J Control Release. 2011, 151, 220-228.

23. Tubo NJ, Jenkins MK. Cd4+ t cells: Guardians of the phagosome. Clin Microbiol Rev. 2014, 27, 200-213.

24. Lai SK, O'Hanlon DE, Harrold S, Man ST, Wang YY, Cone R, Hanes J. Rapid transport of large polymeric nanoparticles in fresh undiluted human mucus. Proc Natl Acad Sci U S A. 2007, 104, 1482-1487.

25. Dendritic cells: Controllers of adaptive immunity. Available at: nature.com/nri/posters/dendriticcells.

26. Safety of tuberculosis vaccine, mva85a, administered by the aerosol route and the intradermal route. Available at: clinicaltrials.gov/ct2/show/NCT01497769?term=aerosol+vaccine&rank=1. Accessed: July 7, 2014.

27. Ballester M, Nembrini C, Dhar N, de Titta A, de Piano C, Pasquier M, Simeoni E, van der Vlies AJ, McKinney JD, Hubbell JA, Swartz MA. Nanoparticle conjugation and pulmonary delivery enhance the protective efficacy of ag85b and cpg against tuberculosis. Vaccine. 2011, 29, 6959-6966.

28. Yu F, Wang J, Dou J, Yang H, He X, Xu W, Zhang Y, Hu K, Gu N. Nanoparticle-based adjuvant for enhanced protective efficacy of DNA vaccine ag85a-esat-6-il-21 against mycobacterium tuberculosis infection. Nanomedicine. 2012, 8, 1337-1344.

209 29. Garcia-Contreras L, Wong YL, Muttil P, Padilla D, Sadoff J, Derousse J, Germishuizen WA, Goonesekera S, Elbert K, Bloom BR, Miller R, Fourie PB, Hickey A, Edwards D. Immunization by a bacterial aerosol. Proc Natl Acad Sci U S A. 2008, 105, 4656-4660.

30. Moon JJ, Huang B, Irvine DJ. Engineering nano- and microparticles to tune immunity. Adv Mater. 2012, 24, 3724-3746.

31. General information about small cell lung cancer. Available at: cancer.gov/cancertopics/pdq/treatment/small-cell-lung/Patient/page1. Accessed: July 8, 2014.

32. Nardelli-Haefliger D, Dudda JC, Romero P. Vaccination route matters for mucosal tumors. Science Translational Medicine. 2013, 5.

33. Guzman MG, Halstead SB, Artsob H, Buchy P, Farrar J, Gubler DJ, Hunsperger E, Kroeger A, Margolis HS, Martinez E, Nathan MB, Pelegrino JL, Simmons C, Yoksan S, Peeling RW. Dengue: A continuing global threat. Nat Rev Microbiol. 2010, 8, S7-16.

34. Getts DR, Martin AJ, McCarthy DP, Terry RL, Hunter ZN, Yap WT, Getts MT, Pleiss M, Luo X, King NJ, Shea LD, Miller SD. Microparticles bearing encephalitogenic peptides induce t-cell tolerance and ameliorate experimental autoimmune encephalomyelitis. Nat Biotechnol. 2012, 30, 1217-1224.

35. Smarr CB, Hsu CL, Byrne AJ, Miller SD, Bryce PJ. Antigen-fixed leukocytes tolerize th2 responses in mouse models of allergy. J Immunol. 2011, 187, 5090-5098.

36. Chow AH, Tong HH, Chattopadhyay P, Shekunov BY. Particle engineering for pulmonary drug delivery. Pharm Res. 2007, 24, 411-437.

37. The global burden of disease 2004 update. Available at: who.int/healthinfo/global_burden_disease/GBD_report_2004update_full.pdf.

38. Neutra MR, Kozlowski PA. Mucosal vaccines: The promise and the challenge. Nat Rev Immunol. 2006, 6, 148-158.

39. Gillies RJ. In vivo molecular imaging. J Cell Biochem Suppl. 2002, 39, 231-238.

40. Kline JN. Eat dirt: Cpg DNA and immunomodulation of asthma. Proc Am Thorac Soc. 2007, 4, 283-288.

210