Self-Emulsifying Adjuvant Systems Containing Alpha-Tocopherol for Subunit Vaccines

Self-Emulsifying Adjuvant Systems Containing Alpha-Tocopherol for

Thesis Presented by

Rushit Nakul Lodaya

The Bouvé Graduate School of Health Sciences in Partial Fulfillment of the Requirements for the Degree of DOCTOR OF PHILOSOPHY in Pharmaceutical Sciences with specialization in Pharmaceutics and Drug Delivery Systems

NORTHEASTERN UNIVERSITY

BOSTON, MASSACHUSETTS

June 13th, 2019

Northeastern University Graduate School of Bouvé College of Health Sciences Dissertation Approval

Dissertation title: SELF-EMULSIFYING ADJUVANT SYSTEMS CONTAINING

ALPHA-TOCOPHEROL FOR SUBUNIT VACCINES

Author: Rushit Nakul Lodaya

Program: Pharmaceutical Sciences

Approval for dissertation requirements for the Doctor of Philosophy in Pharmaceutical Sciences with Specialization in Pharmaceutics and Drug Delivery

Dissertation Committee (Chairman):

Date

Other Committee Members:

Date

Dean of Bouvé College Graduate School:

Date

ABSTRACT

Recombinant DNA technology has revolutionized vaccine antigen production yielding highly purified synthetic antigens that offer significant advantages such as higher yield, lower production costs and improved safety. However, their higher purity resulted in reduced immunogenicity compared to conventional vaccines, like whole-inactivated or live-attenuated, due to absence of inherent immunomodulatory components such as diverse pathogen-associated molecular patterns (PAMPs). Vaccine adjuvants are components that enhance the immune response to antigens with low immunogenicity. These adjuvants represent a wide range of materials from synthetic molecules to heterogenous extract of natural products and include particulates such as emulsions, nanoparticles and liposomes.

Among the particulate adjuvants, emulsion adjuvants have been used since early

1930s to improve immune responses to vaccine antigens. Emulsion adjuvants used in clinic such as MF59 (used in Fluad®) and AS03 used (in Pandemrix®) have helped in reducing the antigen dose as well as the number of immunizations required to achieve the required immune response. These attributes make emulsion adjuvants favorable for use, especially with influenza antigens. Most of the emulsion adjuvants currently available in clinic or in development use high shear and energy consuming processes such as high-pressure homogenization and microfluidization. Such complex processes, requiring expensive equipment and strict manufacturing requirements, make it a limitation for production in developing countries leading to unavailability in these less-privileged regions of the world.

The 2009 H1N1 pandemic was an example where not enough doses of vaccine and the adjuvant were available to dose the entire population. Thus, there certainly exists a need to

3 make the adjuvant production simple and less-expensive to aide in efficient storage, distribution and availability in all parts of the world.

Like MF59 and AS03, several other emulsion adjuvants have squalene oil as main component along with different surfactants to stabilize the oil droplets. AS03 also has α- tocopherol added as an immunomodulator. α-Tocopherol has been used as an immune supplement for several decades now and has been an active component of veterinary vaccine adjuvants. AS03 has shown better immune response profile compared to squalene- only emulsion adjuvants in preclinical (using Hepatitis B antigen in BALB/c mice) as well as clinical studies (using H7N9 antigen and comparison with MF59) suggesting role of α- tocopherol in providing additional immunostimulation. More recently, it was shown that

AS03 elicited a rapid and transient downregulation of lipid metabolism-related genes in the draining lymph node. They showed that the adjuvant effect elicited by AS03 may be due to the metabolic changes induced by AS03 via stress kinase receptor IRE1-α in the monocytic cells. Despite these evidences, use of α-tocopherol has only been in AS03 hitherto. This work might serve as an established route for discovery of more potent novel adjuvants containing α-tocopherol. Developing simpler emulsion adjuvants containing α- tocopherol may allow to understand its role even better at a molecular and cellular level.

In this project, we use self-emulsification, as an alternative simple and low energy process to microfluidization, to formulate α-tocopherol containing emulsion adjuvant using components of AS03. We first demonstrate that this new self-emulsified adjuvant system

(SE-AS) shows immune response similar to AS03 when administered with inactivated

Quadrivalent Influenza Vaccine (QIV) in female BALB/c mice. We then used a soluble recombinant antigen, CMV wild-type pentamer to demonstrate that SE-AS shows

4 significantly higher humoral response compared to SEA160 (a squalene oil only emulsion) in C57BL/6 mice. This confirmed that emulsion adjuvants containing α-tocopherol improved the immune response to a soluble recombinant antigen compared to emulsion adjuvants without α-tocopherol.

Adjuvanted vaccines are distributed as a two or three vial presentation which heavily relies on cold-chain for storage and release. Cold-chain distribution not only gets expensive but also makes it difficult to distribute vaccines in developing countries. As mentioned previously, there are constant efforts to improve storage and distribution of vaccines in less-privileged countries and creating a single vial vaccine with help eliminating the cold-chain making it less expensive to distribute vaccines in these countries. In this project, we used lyophilization to create a single vial SE-AS adjuvanted

CMV pentamer vaccine for reconstitution. We demonstrated that a single vial freeze-dried vaccine can be created and showed that it generates comparable immune responses to the bed-side mixed liquid vaccine.

ACKNOWLEDGEMENT

I’d like to begin my list of appreciations with my advisors Dr. Mansoor A Amiji and Dr. Derek T O’Hagan. They were the constant pillars who made this PhD fellowship program possible. I’d like to express immense gratitude to Dr. Amiji for teaching me to persevere through difficult situations and for constantly supporting and encouraging my ideas throughout the PhD. I’m also very grateful to Dr. O’Hagan for teaching me pragmatism and for guiding me to find the right answers to every question. I am fortunate to have gotten the opportunity to work under their guidance.

I would also like to extend my thanks to Dr. Kim and Dr. Khaw for their invaluable input to my project as a part of my thesis committee. Also, I am grateful to Dr. Sylvie

Bertholet from GSK for her instrumental support and guidance through this project, especially her contribution towards the immunological aspects. I’d like to thank the members of GSK and Northeastern to make this fellowship program possible through their

“backstage” support with special mention to George Balaconis from Northeastern and Greg

Son from GSK.

Additionally, I’d like to thank the talented scientists at GSK for their educational and emotional support through this PhD program. I’d like to thank Dr. Corey Mallett for continuous intellectual discussions, Dr. Kelly Forney-Stevens for her immense support with lyophilization studies and Amey Kanitkar, who I have known since I came to the US, as a roommate, a friend, a confidant and finally a talented scientist who has helped me in each step of my career. Also, many thanks to Dawn Henson, Kristian Friedrich, Kunal

Tungare, Ellen Kuta, Lauren Hirao, Sandra Nuti, Asma Ashraf, Douty Bamba, and Kate

Luisi from the immunology and molecular biology teams at GSK for their support. Finally,

6 cheers to the drug product group at GSK who made me smile through difficult experiments and shared happiness for productive data, especially Neha Sahni, Kanwaldeep Gill,

Andrzej Pitek, Lindsey Sharpe, Jessica Cohen, Elizabeth Zecca, Jimmy Adeoti, Jinjin

Zhang and Sonia Gregory.

Without my crew at legacy Novartis vaccines, I’d not have enjoyed and appreciated vaccines research as much and for that I’d like to thank Dr. Luis Brito for embarking that curiosity in me to learn more about vaccines and Padma Malyala, and Dr. Sushma

Kommareddy. Special thanks to Dr. Ruchi Shah for setting a precedent with her work on

SEA160 and being a great friend and guide through my PhD project.

In my last set of appreciation, I’d like to dedicate this work to my family and friends who have understood my frustration, mourned with me during my failures and celebrated my little victories. Special mention to my best buddies Jonathan Ly, Fenil Gandhi and

Shravan Sriraman for being the “ears” I needed to rant and for blindly supporting me through everything. I’ve had unwavering love and support from my sister-in-law Neha and brother Ankit and I’d like to thank them for giving me the ultimate happiness in the form of my nephew Yuvaan. My brother has been my biggest pillar of support at every stage of my career, so I thank him specially for always being there. And finally, my biggest thank you to my parents for never losing hope in me, and providing me with a solid upbringing, making me capable of this achievement today. My dad always told me he wanted to see that “Dr.” title preceding my name, and I’m elated to fulfil his wish.

Last, but certainly not the least, I couldn’t thank my wife enough for her irrevocable and solid backing for the past 5 or so years of knowing her. Thank you, Megha for being my ultimate strength and for patience you have shown throughout this PhD.

TABLE OF CONTENTS

ABSTRACT ...... 3

ACKNOWLEDGEMENT ...... 6

OBJECTIVES AND SPECIFIC AIMS ...... 13

LIST OF TABLES ...... 16

LIST OF FIGURES ...... 17

CHAPTER 1 ...... 20

Vaccine Adjuvants and Alternative Formulation Strategies for Emulsion Adjuvants ...... 20

1.1 Vaccine Adjuvants ...... 20

1.1.1 Classification of Vaccine Adjuvants ...... 21

1.1.2 Mechanisms of Adjuvant Action ...... 24

1.1.3 Specific Types of Adjuvants ...... 25

1.1.4 Future Considerations in the Development of Adjuvants ...... 29

1.2 Emulsion Adjuvants ...... 30

1.3 Emulsion Adjuvants in Influenza Vaccines ...... 33

1.4 Formulation of Emulsion Adjuvants: Current and Alternative Strategies ...... 35

1.4.1 Current Formulation and Manufacturing of Emulsion Adjuvants ...... 35

1.4.2 Alternative Formulation Strategies for Emulsion Adjuvants ...... 36

1.5 Conclusions ...... 38

CHAPTER 2 ...... 40

Alpha-Tocopherol (Vitamin E) as an Immunostimulant ...... 40

2.1 Introduction: Vitamins in vaccines ...... 40

2.2 History of vitamin E: Immune Supplement and Veterinary Vaccines ...... 42

2.2.1 Vitamin E as an immune supplement ...... 43

2.2.2 Vitamin E in veterinary vaccines ...... 44

2.3 AS03 mechanism of action and role of α-tocopherol as an immunostimulant ...... 44

2.4 Pharmaceutical properties of α-tocopherol ...... 50

2.5 Conclusion ...... 51

CHAPTER 3 ...... 52

Lyophilization of Antigen-in-Emulsion Adjuvants to Create a Single Vial Vaccine Product ...... 52

3.1 Introduction ...... 52

3.1.1 Rationale for single vial adjuvanted vaccine ...... 52

3.1.2 Why lyophilize? ...... 54

3.2 Lyophilization of Nanoemulsions: feasibility and considerations ...... 56

3.3 Conclusion ...... 60

CHAPTER 4 ...... 61

Formulation Design, Optimization and In Vivo Evaluations of an Alpha- Tocopherol-Containing Self-Emulsified Adjuvant System using Inactivated Influenza Vaccine ...... 61

4.1 Introduction ...... 61

4.2 Materials and Methods ...... 64

4.2.1 Formulation Materials ...... 64

4.2.2 Design-of-Experiment (DoE) Analysis for SE-AS Optimization ...... 65

4.2.3 Effect of different surfactants on SE-AS...... 66

4.2.4 Effect of Surfactant Reduction on SE-AS Development ...... 67

4.2.5 Stability Study of SE-AS without Antigens at Higher Temperatures...... 67

4.2.6 Short-term Stability Analysis of Combination OVA and SE-AS...... 68

4.2.7 In Vivo Evaluations of SE-AS with Inactivated Influenza Vaccine Antigens . 68

4.2.8 Data Analysis and Statistics ...... 74

4.3 Results ...... 74

4.3.1 DoE Analysis for SE-AS Formulations ...... 74

4.3.2 Effect of Surfactant Reduction on SE-AS Formulation ...... 82

4.3.3 Stability of combination OVA Antigen and SE-AS ...... 84

4.3.4 In Vivo Potency of SE-AS with QIV Antigens ...... 85

4.4 Discussion ...... 89

4.5 Conclusion ...... 94

CHAPTER 5 ...... 96

Self-Emulsified Adjuvant System (SE-AS): Formulation Optimization and In Vivo Immunogenic Effects with Soluble Subunit CMV Pentamer Antigen ...... 96

5.1 Introduction ...... 96

5.2 Materials and Methods ...... 98

5.2.1 Formulation Materials ...... 98

5.2.2 Sterile filtration of emulsions ...... 98

5.2.3 Percent content of squalene and α-tocopherol in emulsions ...... 99

5.2.4 Optimization of SE-AS ...... 99

5.2.5 Cryo-electron microscopy (Cryo-EM) imaging of SE-AS 44 ...... 100

5.2.6 Stability study to enable delivery of stable emulsions for in vivo evaluations ...... 100

5.2.7 In vivo evaluations using CMV wild-type pentamer antigen ...... 100

5.3 Results ...... 107

5.3.1 Optimizing SE-AS for sterile filtration ...... 108

5.3.2 Stability assessment of SE-AS at elevated temperatures ...... 112

5.3.3 In vivo potency evaluation using CMV subunit protein ...... 115

5.4 Discussion ...... 122

5.5 Conclusion ...... 125

CHAPTER 6 ...... 126

Lyophilization of SE-AS with CMV Antigen to Develop a Single Vial Vaccine for Reconstitution ...... 126

6.1 Introduction ...... 126

6.2 Materials and Methods ...... 128

6.2.1 Formulation materials ...... 128

6.2.2 Freeze-drying of SE-AS with and without antigen ...... 128

6.2.3 In vivo comparison of liquid bed-side mixed and lyophilized single vial adjuvanted CMV vaccine ...... 131

6.3 Results ...... 138

6.3.1 Single-vial vaccine feasibility on VirTis bench-scale freeze dryer ...... 138

6.3.2 Lyophilization cycle development and excipient screening using SEA160. .. 139

6.3.3 SE-AS 44 formulation optimizations for lyophilization...... 143

6.3.4 In vivo comparison of potency between bed-side mixed liquid and single-vial lyophilized vaccine ...... 145

5.4 Discussion ...... 149

CONCLUDING REMARKS ...... 153

REFERENCES ...... 155

OBJECTIVES AND SPECIFIC AIMS

Adjuvants have enhanced the overall immune response to highly purified and synthetic antigens. Emulsion adjuvants, especially have shown dose-sparing effect when used with influenza antigens and have helped reduce the number of immunizations required to get the desired level of antibody titers. However, during the 2009 H1N1 pandemic, only about half of the total number of vaccine doses were deployed due to low antigen and adjuvant stock and availability (1). Expensive and complex emulsion adjuvant manufacturing process requires high-maintenance equipment and facility and hence are difficult and expensive to produce in less-privileged countries. And this could be one of the reasons for unavailability of vaccines in poor countries especially during pandemic situations. Thus, there is a need for simpler less-expensive processes to manufacture emulsion adjuvants to allow for production in any well-established GMP facility.

These adjuvanted vaccines are generally distributed in two or three vials with antigen and adjuvant (and sometimes buffer for dilution) to be mixed at the dosing site.

This process heavily relies on cold-chain to stock-pile and distribute vaccines which makes it an expensive process and raise the costs making it difficult to provide in less-privileged countries (2). Antigens and adjuvants are not stored in a single vial due to potential interactions between the two leading to possible loss in efficacy of the vaccine; however, lyophilization of a single vial adjuvanted vaccine has been done to show stability at elevated temperatures and comparable immune responses to liquid vaccine (3, 4).

Thus, our hypothesis was that using self-emulsification, an emulsion adjuvant can be created containing squalene and α-tocopherol which would yield immune responses similar to AS03 and better than emulsion adjuvants without α-tocopherol. We also

13 postulated that lyophilized single vial adjuvanted vaccine would show comparable immune responses to bed-side mixed liquid vaccine in mice.

Our first objective was to formulate emulsion adjuvants using low-shear, low- energy process that is comparatively simple than high-pressure homogenization and microfluidization. Squalene and α-tocopherol have different structures and hydrophilic- lipophilic balance (HLB) values and hence, our starting aim was to incorporate α- tocopherol in a squalene oil-in-water (o/w) emulsion using self-emulsification. Our goal was to get an SE-AS which was close in composition to AS03 to be able to compare immune responses in vivo using BALB/c mice. We also aimed at optimizing the SE-AS and comparing the immune responses with an emulsion adjuvant without containing α- tocopherol. Finally, our last objective was to conduct a proof-of-concept study to lyophilize

SE-AS adjuvant with a soluble recombinant antigen and test the potency of lyophilized single vial vaccine compared to the bed-side mixed vaccine.

To fulfil the objectives, the following specific aims were proposed:

Specific Aim 1: Incorporating α-tocopherol in squalene oil-in-water emulsions using

Design of Experiments to obtain SE-AS

Setting the initial criteria size and polydispersity index (PdI) as 200nm and 0.3, respectively, several emulsion mixture ratios were screened to identify if a stable SE-AS close to composition of AS03 could be prepared. Our size criteria enabled us to get an SE-

AS in nanoemulsion size range close to size of AS03 (~155nm) and PdI would help in getting a homogenous emulsion which is a very important factor for stability. Our exhaustive formulation screening resulted in two novel SE-AS with lower amounts of tocopherol which were then evaluated in vivo for comparison with AS03

Specific Aim 2: Evaluating the potency of novel SE-AS in vivo using inactivated QIV in BALB/c mice

Novel SE-AS with lower amounts of α-tocopherol were compared for their potency with

AS03 using influenza vaccine antigens due to wide use of emulsion adjuvants in flu vaccines. Our results confirmed the adjuvant nature of the novel SE-AS and comparable immune responses to AS03

Specific Aim 3: Optimizing SE-AS and evaluating its potency in vivo using a soluble recombinant CMV pentamer antigen in C57BL/6 mice

We optimized SE-AS from aim 2 to improve physicochemical stability and tested the optimized SE-AS in vivo to compare its potency with an emulsion adjuvant without α- tocopherol to demonstrate the role of α-tocopherol in the emulsion adjuvant in enhancing the overall immune response.

Specific Aim 4: Lyophilization of SE-AS with CMV pentamer antigen to create single vial vaccine for reconstitution and administration and its in vivo evaluation and comparison with liquid bed-side mixed vaccine in C57BL/6 mice.

We transferred and optimized freeze-drying cycle from VirTis bench scale lyo to LyoStar3 for single vial adjuvanted vaccine. We compared in vivo, using the same antigen as in Aim

3, immune responses to liquid vaccine and used AS03 as a positive control.

LIST OF TABLES

1.1 Specific examples of adjuvants used in commercial vaccines as well as those in clinical development……………………………………………………………….21 4.1 List of oil- and water-soluble surfactants evaluated with SE-AS………………….67 4.2 Hydrodynamic droplet size and PdI from DoE-1 measured using DLS……………77 4.3 Hydrodynamic droplet size and PdI from DoE-2 measured using DLS……………79 4.4 Size and PdI from surfactant study oil and water-soluble surfactants………………79 4.5 Reduction in Polysorbate 80 content in SE-AS 6…………………………………..82 5.1 Size of the emulsions before and after filtration and % content loss after filtration for SE-AS 22 and 36………………………………………………………………….109 5.2 % v/v in oil: surfactant mixture for novel SE-AS combinations and size and PdI using DLS……………………………………………………………………………….110 6.1 Freeze-drying cycle developed on VirTis bench-scale Lyophilizer……………….129 6.2 Size and PdI from freezing ramp rate evaluation…………………………………141 6.3 Size and PdI measurements from annealing experiments with mannitol…………142 6.4 Final Lyophilization cycle used to develop single vial vaccine…………………..143

LIST OF FIGURES

1.1 Key developments and milestones in vaccine adjuvants and a timeline for approval and marketing of the vaccines containing adjuvants………………………………..21 1.2 Putative mechanism of action of adjuvants after an intramuscular or subcutaneous injection…………………………………………………………………………….25 1.3 Manufacturing of emulsion adjuvants……………………………………………...35 1.4 Self-emulsification process ex-vivo as an alternative method to create emulsion adjuvants…………………………………………………………………………...38 2.1 α-Tocopherol in AS03 modulates immune cell numbers and antigen loading in monocytes in the dLNs……………………………………………………………..46 2.2 Anti-HbsAg antibody titers comparing AS03 with AS03 w/o tocopherol………….48 2.3 Hemagglutination Inhibition Antibody titers using H7N9 and either no adjuvant, MF59, AS03 or combination of these………………………………………………49 3.1 The five main stages of lyophilization include formulation, freezing, primary (1°) drying, secondary (2°) drying, and stoppering……………………………………..57 4.1 Ternary Plot showing application of design of experiment (DoE) by identifying possible ratios of squalene, α-tocopherol and polysorbate 80 via mixture design…66 4.2 Gating strategy to obtain antigen-specific cytokine positive T cells……………….74 4.3 Robustness and stability of SE-AS and effect of process change on size and PdI…75 4.4 Mixture profile analysis using contour mapping and ternary plots…………………78 4.5 Experimental space to operate for mixture ratios for the three components based on contour analysis…………………………………………………………………….81 4.6 Effect of buffers on size and PdI of SE-AS………………………………………...82 4.7 SE-AS stability without the antigen at elevated temperatures……………………..84 4.8 Gel Electrophoresis to assess ovalbumin (OVA) antigen stability with SE-AS……85 4.9 Hemagglutination Inhibition titers post 1st and 2nd immunization in BALB/c mice………………………………………………………………………...………87 4.10 Anti-HA IgG subclass antibody titers in mouse serum post 2nd immunization……90 4.11 Intracellular cytokine staining (T-cell response)…………………………………...91 5.1 Study design, dosing groups and immunization schedule to compare in vivo SE-AS and SEA160………………………………………………………………………102

5.2 Example of gating strategy for ICS…………………………………………….…106 5.3 Example of gating strategy for Memory markers…………………………………107 5.4 Model chromatograms showing squalene and α-tocopherol peaks……………….108 5.5 Standard curves for tocopherol and squalene for concentration calculation in the emulsions…………………………………………………………………………109 5.6 Bimodal size distribution after filtration through 0.22µ PES filter……………….110 5.7 Size distribution for SE-AS 44 showing overlay of size distribution graphs before (red) and after (green) filtration…………………………………………………...112 5.8 pH and Osmolality for emulsions up to 10 weeks…………………………………113 5.9 Size and PdI for emulsions up to 10 weeks……………………………………….113 5.10 % Squalene and Tocopherol content for emulsions up to 2 weeks……………….114 5.11 Cryo-EM images of AS03 and SE-AS 44 showing comparable droplet morphology……………………………………………………………………….115 5.12 Neutralizing antibody titers in serum obtained three weeks post 1st, 2nd and 3rd immunization, against CMV TB40 strain of virus………………………………..117 5.13 Anti-CMV Penta IgG antibody titers in serum obtained three weeks post 2nd and 3rd immunization……………………………………………………………………..118 5.14 Frequencies of overall antigen-specific CD4+ T cells classified as either Central memory, effector memory or effector cells……………………………………….120 5.15 Frequencies of antigen-specific Th2-type CD4+ T cells classified as either Central memory, effector memory or effector Th2 cells…………………………………..121 5.16 Frequencies of antigen-specific effector memory cells gated from Quadrant 3 after gating for memory markers……………………………………………………….121 5.17 Frequencies of antigen-specific effector cells gated from Quadrant 4 after gating for memory markers…………………………………………………………………..122 6.1 Study design, dosing groups and immunization schedule to compare liquid bed-side mixed formulation with Lyophilized single vial formulation…………………….132 6.2 Example of gating strategy for ICS……………………………………………….137 6.3 Freeze-drying SEA160 on VirTis lyophilizer……………………………………..139 6.4 A DSC thermogram showing a regular cool-heat cycle for the pre-lyo formulation of SEA160 with 10% w/v sucrose……………………………………………………140

6.5 Measuring collapse temperature using FDM……………………………………..141 6.6 SE-AS 44 lyophilization…………………………………………………………..144 6.7 Optimized SE-AS 44 Lyophilization with CMV pentamer……………………….145 6.8 nAb titers in serum obtained three weeks post 1st, 2nd and 3rd immunization, against CMV TB40 strain of virus………………………………………………………..146 6.9 Anti-CMV Penta IgG antibody titers in serum obtained three weeks post 2nd and 3rd immunization……………………………………………………………………..147 6.10 Antigen specific CD4+ T cells using ICS assay…………………………………...148

CHAPTER 1 Vaccine Adjuvants and Alternative Formulation Strategies for Emulsion Adjuvants 1.1 Vaccine Adjuvants

Next to clean water, vaccines have had the most important significance in terms of prevention and treatment of diseases, becoming the most successful medical invention in the past Century. Vaccines have evolved significantly since the development of first successful smallpox vaccine in 1796 (5). Live-attenuated and whole-inactivated vaccines form the basis of most vaccines approved for clinical use. With attenuated or inactivated virus, there is a constant risk of reversion of the killed and/or attenuated pathogen in these vaccines to its virulent form, leading to a pressing need for safer vaccines. Advances in the field of biotechnology and recombinant DNA technology in the early 1980s allowed for engineering of subunit protein antigens (6). These synthetic antigens offered several prominent advantages over conventional vaccines, such as improved safety, higher yield, improved stability, and lower cost of production(5, 7). Higher purity and safety of these antigens came with a drawback of being less potent and immunogenic compared to attenuated/inactivated vaccines.

Thus, the need to provide an additional or “external” innate immune activation along with subunit antigens was evident. This led to addition of immunostimulatory components to the vaccine antigen, called adjuvants (in Latin, it means “to help”), to enhance immunogenicity of the immunogen (8). Over the past two decades, research on vaccine adjuvants has grown remarkably, with several new generations of adjuvants being included in licensed vaccine products against infectious diseases such as malaria, influenza, shingles and many more.

1.1.1 Classification of Vaccine Adjuvants

Vaccine adjuvants have evolved in the past Century from only Alum and emulsion based adjuvants administered clinically between 1920s-2000s to next generation adjuvants that include combination of first generation and newer molecular specific adjuvants (9).

Figure 1.1 illustrates a timeline of key milestones in the development of adjuvants in vaccine products. Vaccine adjuvants are of a diverse family, and hence, cannot be defined using a single integrated structure. They comprise of several naturally occurring or synthetic materials that boost the immunological effect of the antigen (10).

Figure 1.1: Key developments and milestones in vaccine adjuvants and a timeline for approval and marketing of the vaccines containing adjuvants. Top of the arrow indicates the licensure of vaccines and the bottom of the arrow shows key stages in clinical development.

There are several vaccine adjuvants that are approved and used in commercial products and other currently in clinical development (Table 1.1). These vaccine adjuvants can be broadly classified into two main classes based on their mechanism of interaction with innate immune system – First class are adjuvants that effectively present the antigen to the innate immune cells at the site of injection to facilitate rapid uptake of the antigen for enhanced immune response (5). These first generation are also called as particulate adjuvants; more so because of their physicochemical properties and particle size that would mimic “pathogens” and act as efficient delivery vehicles for antigen also providing for added antigen stability (11). Alum salts, emulsions, liposomes are examples of this class of adjuvants. Alum salts have been used as adjuvants from as early as the 1930s and are considered as the “gold standard” in the field of adjuvants (12). Second class would be immune potentiators that provide specific innate immune signals through pattern recognition receptors (PRRs) such as Toll-like receptors (TLRs). These immune potentiators represent an array of molecules from natural products, to semi-synthetic and synthetic molecules. in most instances, these molecules when injected alone, manifest unwanted and thus undesired pharmacological effects. The newer generation adjuvants are essentially combination adjuvants that consist of a particulate/delivery vehicle carrying antigen and or immune potentiator.

Table 1.1: Specific examples of adjuvants used in commercial vaccines as well as those in clinical development

Adjuvant Name Formulation Vaccine for Specific Disease Target Adjuvants in Licensed Vaccines Alum (aluminum Stable suspension of hydroxide, Included in several salts) phosphate or hydroxy-phosphate routine vaccines for sulfate salts children such as DTaP

O/W emulsions Squalene oil in water emulsions Products for seasonal stabilized with non-ionic Flu (such as Fluad) or surfactant(s); could include other Pandemic Flu (such as oils and stabilizers Pandemrix) Virosomes Dispersed lipid vesicles including Flu vaccine (Infexal) viral membrane proteins and HAV vaccine (Epaxal) AS04 TLR4 agonist MPL adsorbed onto HPV (Cervarix) and Alum HBV (Fendrix) MPL A naturally derived TLR4 ligand For pollen allergies on named patient basis in Europe (Pollinex) RC-529 Synthetic TLR4 ligand adsorbed to Was licensed in Alum Argentina in Hepatitis B vaccine Supervax AS01 Liposome encapsulating MPL and Malaria (RTS,S) QS21 (Saponin) vaccine, Herpes Zoster vaccine (HZ/su) and Shingles vaccine (Shingrix) CpG ODN (1018 A soluble oligonucleotide serving as HBV vaccine (Heplisav- ISS) a ligand to TLR9 B) in adults 18 or older ISA51 – o/w Oil dispersed nanoemulsion with Influenza vaccine emulsion nonionic surfactant Adjuvants in Clinical Development –Phase III & II Topical cream Topical formulation for intradermal Influenza (phase III) with TLR7 vaccine of TLR7 ligand R848 ligand (Imiquimod) ISCOMs Phospholipid and cholesterol Influenza vaccine (ISCOMATRIX immune Nano-complexes with (ISCOM) or Matrix-M purified saponins Maternal RSV-F GEN-003 for genital herpes GLA-SE Squalene-oil in water emulsion with Tuberculosis vaccine, tocopherol as antioxidant with GLA RSV and Leishmania as TLR4 ligand vaccine IC31 Cationic peptide complexed with Tuberculosis vaccine TLR9 ligand (oligonucleotide) VAX2012Q, TLR5 ligand protein (flagellin Influenza vaccine VAX125 linked to antigen Poly I:C Double stranded RNA polymer Rabies Vaccine and (Ampligen, analog and TLR ligand Influenza vaccine rintatolimod) PIKA

VCL-HB01 Cationic Liposome Genital Herpes vaccine (Vaxfectin) (both therapeutic and prophylactic)

1.1.2 Mechanisms of Adjuvant Action

Several factors affect the adjuvant immune action once administered, such as antigen-adjuvant interaction, mechanism of adjuvant action, choice of adjuvant, toxicity or local/systemic reaction of the adjuvant. Different adjuvants have several proven or proposed mechanisms of action (Figure 1.2) to elicit cellular and/or humoral immune response. Thus, prediction of an immune response post adjuvant administration becomes rather difficult. Janeway (13) described adjuvant mechanism of action as “immunologists dirty little secret”. Schijns (14) has systematically summarized several theories explaining critical pathways of adjuvanticity that gives an overall understanding of key immunological routes paving a path for novel adjuvant development. These theories explain the general mechanism; however, no adjuvant follows a single pathway of enhancing the immunogenicity. The mechanism of adjuvant action remains a “treasure” that keeps unfolding with advances in research and analytical tools for characterization. In general, purified vaccine antigens require co-localization at injection site to facilitate pick up by antigen presenting cells (APCs). Depending on the type of antigen, and adjuvant, spatial and temporal availability of antigen and/or immune potentiator changes and which eventually is crucial to achieve desired immune response of sufficient magnitude and quality. Combination of adjuvants (addressed in a later section) that provide synergy in two or more pathways of adjuvant action could provide for a more potent adjuvant. In the next few sections, we have described the status of vaccine adjuvants prominently used in licensed vaccines.

Figure 1.2: Putative mechanism of action of adjuvants after an intramuscular or subcutaneous injection (15).

1.1.3 Specific Types of Adjuvants

1.1.3.1 Aluminum Salts

Around 1920s, in attempt to purify tetanus and diphtheria vaccine antigens, aluminum salts (referred to as “Alum” in this chapter) were added to precipitate them from the growth media (16). Later, it was discovered that these precipitated antigens were more immunogenic than the soluble ones. Since then, Alum-based adjuvants have been used extensively in clinical vaccines providing for a vast database for safety and efficacy of vaccine adjuvants, as well as providing for a gold standard for comparison for all future novel adjuvants (17). Ironically, very little is known about molecular and cellular mechanism of adjuvant action for Alum, although several theories have only recently started to surface (12). Some prominent ones are acting via nlrp3 inflammasome pathway

(18-20), producing depot effect to provide antigen at the injection site for longer period.

Although, Alum alone is still used in some licensed vaccines, in the past 2 decades, advances in adjuvant research has allowed for more potent vaccine adjuvants to be used for more virulent pathogens. Alum has now been used extensively as a “delivery” adjuvant, to co-deliver antigen and immune potentiating molecules (21) such as monophosphoryl lipid A MPL (AS04 adjuvant used in Cervarix). Promising proof-of-concept studies have shown the feasibility of lyophilizing Alum formulations and could potentially be an alternative to stabilize Alum formulations and create a single vial vaccine making it independent of cold chain distribution (22, 23). Although Alum salts are extensively used in clinical vaccines, optimizing Alum adjuvant using formulation science, can potentially lead to more potent versions of this “grandfather” adjuvant.

1.1.3.2 Emulsion Adjuvants

Emulsion adjuvants have also been around for as long as Alum in the form of

Freund’s complete (FCA) and Freund’s incomplete (FIA) adjuvants made with mineral oil.

Most of the clinically available emulsion adjuvants such as MF59 and AS03 contain biodegradable squalene oil as one of the major components. These adjuvants have shown substantial improvement in influenza pandemic, pre-pandemic and seasonal vaccines (24).

Emulsion adjuvants are discussed in detail in the next section.

1.1.3.3 Other Lipid-Based Adjuvants

Liposomes are spherical nanoparticles comprising of a phospholipid bilayer with an aqueous core and are used in vaccines for delivery of both antigen and additional immune potentiators. Based on the choice of the phospholipids and other components, liposomes can be used as a carrier by encapsulating in the core, incorporation in the bilayer,

26 or by adsorbing on the surface. CAF01 is one of the oldest cationic liposome-containing adjuvants with dimethyldioctadecylammonium (DDA) and trehalose dibehenate (TDB), is now being investigated in development of tuberculosis (TB) vaccine (25, 26). Recent formulation advances have allowed for a successful single vial vaccine development using

CAF01 and H56, which is a multistage TB antigen by spray drying the two to obtain a single powder that can be reconstituted at the time of administration (27). Another prominent lipid-based adjuvant is ISCOMATRIX (28) comprising of immunostimulating complexes (ISCOMS) and QuilA adjuvant (now replaced with refined Saponin preparations). These liposomal adjuvants offer strong CD4+ T-helper cell 1 type immune response as well as high CD8+ immune response (29) More recently, a combination adjuvant from GSK, AS01 a liposome delivering two synergistically acting immune potentiators – QS21 and MPL, was developed for Shingrix, a shingles vaccine in the United

States (30).

1.1.3.4 Immune Potentiators

PRRs like TLRs (31) were found to be activated via pathogen-associated molecular pattern (PAMPs) provided by TLR agonists (TLRa). These TLRa molecules that are typically present in older vaccines have resurfaced much later for their immune potentiation activity. This has allowed for a better understanding of how adjuvants work, and how preferred immune responses can be induced (32). One of the first molecules to be identified as a TLRa was monophosphoryl lipid A (MPL-A) a lipopolysaccharide from gram negative bacteria. Unfortunately, lipopolysaccharides molecules like MPL from gram negative bacteria (33), are typically large and complex molecules, which bring significant formulation challenges. Recognition of the receptor systems such as TLRs involved in

27 adjuvant immune potentiation pathways has allowed the discovery of small molecule immune potentiators (SMIPs) (34, 35). SMIPS for melanoma such as Resiquimod and

Imiquimod that are administered topically, have shown strong immune responses in combating the tumor cells. These molecules have been used as a reference to build similar

TLR agonists that can be used for prophylactic vaccines. Encouragingly, SMIPs have been shown to be more potent than the large biologic molecules (36, 37) and as most immune potentiators, they also have potential for inducing unwanted systemic inflammatory responses when allowed to diffuse away from the site of injection. Hence, a formulation approach is necessary to efficiently deliver the SMIP to local immune cells, while restricting the ability of the SMIP to diffuse from the site of injection (38). This perceptively led to next generation adjuvants involving delivery vehicles/adjuvants that limit these immune potentiators spatially and temporally to allow for desired adjuvant action.

1.1.3.5 Combination Adjuvants

These are “next” generation adjuvants that essentially utilize the knowledge from existing adjuvants, mainly the immunological pathways and immune system activation, to create novel combinations that enhance the immunogenicity of the antigen in general.

Depending on the adjuvant combination, the overall immune response can be shifted. For example, Alum when used alone promotes a Th2 dominant immune response, however when used to deliver MPL (as in AS04) there is significant increase in antibody titers as well as the cellular response is skewed towards Th1. AS01 is a classic example for not two but three adjuvants – liposomes, MPL and QS21, that are delivered along with the antigen.

These adjuvants are presumed to act via separate pathways to provide an enhanced immune

28 response. Recently, AS01 was licensed to use in Shingrix in the elderly, a vaccine for

Shingles. AS01 has shown significantly higher cellular immune response compared to

AS02 (an emulsion formulated of 3D-MPL and QS21 i.e. an emulsion equivalent of AS01) when administered with a tuberculosis antigen M72. The frequency of polyfunctional Th1 cells was particularly significantly higher for AS01 groups compared to AS02. In another scenario, Alum poses as the “best delivery candidate” for immune potentiators for several reasons such as vast database of proven safety and efficacy profile, easy to characterize and optimize the formulation and easy to manipulate the molecule and Alum for adsorption purposes. As previously mentioned, Wu, et al., used the rational design to develop a synthetic small molecule immune potentiator and carefully modified to include the phosphate group for efficient adsorption to Alum facilitating the final Alum/TLR7 adjuvant. A recent study with emulsion adjuvants shows the importance of formulation in stabilizing a lipophilic TLR7 agonist and facilitating its formulation with squalene-based emulsions to obtain Adjuvant nanoemulsion (ANE). These ANE formulations showed higher titers and favorable pharmacokinetic profile of SMIP compared to TLR7 SMIP or

ANE alone (39).

1.1.4 Future Considerations in the Development of Adjuvants

A plethora of published research with adjuvants thus far clearly point at two major considerations: first, the type of immune response elicited by the adjuvants majorly depends upon the co-administered antigen and second, evaluate the need for an adjuvant with the antigen, and then based on the desired immune response, use most appropriate adjuvant to improve vaccine immunogenicity. There are several theories on the mechanisms of currently approved adjuvants. These adjuvants, however, have naturally-

29 derived components that are difficult to characterize and standardize like squalene oil (from shark liver) in oil-in-water emulsions viz. MF59, AS03, and GLA-SE. The focus now should be on finding the synthetic alternatives for these components with inspiration drawn from SMIPs that have shown comparable or even better immune responses compared to naturally derived TLRa. Furthermore, use of high throughput techniques can facilitate the adjuvant research mainly finding novel molecules as well as combinations for targeting

TLRs and newer targets such as stimulator of interferon genes (STING), RIG-1-like receptors (RLRs), and Nod-like receptors (NLRs). Along with the discovery of novel immune potentiators, as seen with the summary above, formulation of these immune potentiators will play a pivotal role in shaping the immune response. Another key approach in adjuvant discovery should be the idea of simplicity meaning one should ask if the adjuvant is really needed in the vaccine, and if yes, would Alum alone generate the desired immune response. Thus, one should carefully map the desired immune response and use the existing knowledge for further adjuvant/vaccine development.

After aluminum salts, emulsion adjuvants are the oldest family of adjuvants used in marketed vaccines. Although, they have evidently proven its advantage over unadjuvanted vaccines, we believe there is scope in refining these adjuvants and understanding the underlying mechanisms of action to be able to develop better, cos-effective adjuvants for future use. In this next section, we discuss a brief history and current development of emulsion adjuvants.

1.2 Emulsion Adjuvants

Emulsions are biphasic systems comprising of a hydrophilic and hydrophobic phase: they can be classified as oil-in-water (O/W), water-in-oil (W/O), or multiple

30 emulsions like water-in oil-in water (W/O/W) or oil-in water-in oil (O/W/O). These formulations use surfactants to render thermodynamic stability and avoid phase separation.

Freund’s Complete Adjuvant (FCA) and Freund’s Incomplete Adjuvant (FIA) were the first emulsions-based (W/O) adjuvants developed (40), with the use of FCA reported as far back as 1937. Although no longer in clinical use owing to its reactogenicity (local reactions in humans; carcinogenicity in mice) and because of the subsequent issue of tolerability,

FCA is still used in research (41). FCA was also the first emulsion-based adjuvant to be studied for prophylaxis. Its severe reactogenicity, deriving from the quality of mineral oil used for formulation (42), leads to granuloma formation and nerve involvement.

Montanides, which are like FIA, are W/O emulsion adjuvants manufactured by Seppic since 1992 have also shown local reactions, and are expensive to produce as each antigen requires an extensive emulsification procedure. Another W/O emulsion, Adjuvant 65, showed good adjuvanticity in influenza vaccine with no severe adverse effects. Although this adjuvant reduced the antigen load by four times, the main component of Adjuvant 65 being peanut oil, dampened further use of this adjuvant due to peanut allergies (43).

Use of O/W emulsion adjuvants has also been explored within the context of utilizing the immunogenic properties of lipids and reducing the adverse effects by decreasing the amount of oil in the adjuvant. It was discovered that using oils in small amounts to make O/W emulsions provided the adjuvant response while diminishing the reactogenicity. Squalene oil used in Ribi Adjuvant Systems (RAS) and squalane

(hydrogenated form of squalene) in Syntex Adjuvant Formulation (SAF) showed little to no reactogenicity (44, 45). These O/W emulsion adjuvants use a metabolizable oil and hence are safe compared to FCA and FIA. Thus, O/W emulsions are researched extensively

31 owing to their benefits and limited reactogenicity. Although many emulsion adjuvants are under development, currently only two emulsion adjuvants - MF59 (Novartis Vaccines &

Diagnostics) and AS03 (GlaxoSmithKline Biologicals) are included in commercially available preventative vaccines. Both emulsions are vital components of seasonal as well as pandemic and prepandemic influenza vaccines.

MF59, a squalene only o/w emulsion adjuvant was first approved for human use in

Europe in 1997. Being the first approved adjuvant and over 30 million doses administered, it has a vast safety and efficacy database (46). MF59 was developed as a delivery vehicle for MTP-PE (muramyl tripeptide phosphatidylethanolamine), however, successive studies revealed that it was better tolerated and provided similar response without MTP-PE and thus was removed from its formulation. This is indicative of the importance of simplicity in adjuvant development. Although squalene is available from several resources, shark liver oil derived squalene is used in both MF59 and AS03. Being a precursor for cholesterol synthesis and easily biodegradable, squalene became an extensively researched oil for creating emulsion adjuvants (47). MF59 showed the first evidence of dose-sparing and reducing the number of immunizations when administered with influenza antigens such as

H5N1 and H1N1 (40). There have been several attempts at understanding the mechanism of action of this adjuvant. Seubert et al. discovered that unlike Alum, MF59 does not operate via depot effect and nlp3 inflammasome, but is dependent on MyD88 pathway (48,

49). MF59 and the co-administered antigen were found to be cleared independently from the injection site. Additionally, it was discovered that monocytes MF59 creates an immunocompetent environment at the injection site by recruiting several immune cells like monocytes, granulocytes, macrophages, etc. The differentiation of these immune cells and

32 consequent antigen presentation occur in TLR-independent mechanism. This enhanced recruitment of immune cells result in improved antigen presentation to naïve T cells, thus priming a potent immune response (50). The enriched recruitment of immune cells leads to release of several chemokines and pro-inflammatory cytokines creating a self-limiting positive immune feedback loop. More recently, an interesting mechanistic study showed that in squalene emulsion adjuvants, the activation of immune cells might occur due to production of reactive oxygen species (ROS) due to interaction between squalene and the immune cells such as DC’s (51). Interestingly, they observed that the effect of squalene in immune cell activation was diminished in presence of an anti-oxidant, confirming their hypothesis of ROS-mediated immune cell activation by squalene based adjuvants.

Overall, mechanistic studies using the well-established adjuvants such as MF59 and

AS03 will certainly help in developing novel emulsion adjuvants with defined adjuvant action to help identify its application towards appropriate viral or bacterial target. We discuss the detailed research on mechanistic pathways of action for AS03 and the possible role of α-tocopherol in the next chapter.

1.3 Emulsion Adjuvants in Influenza Vaccines

Worldwide, several million cases of influenza are reported every year (1), with recent numbers from WHO reporting 290000 to 650000 deaths due to infection followed by respiratory failure (52). The most vulnerable population to infection by seasonal or pandemic Influenza viruses are children less than 5 years of age, pregnant women, adults with chronic medical condition, as well as adults over 65 years. Seasonal Influenza vaccine is supplied from different manufacturing companies based on CDC’s evaluation of the circulating strains of influenza virus. Most common types of Flu viruses are the A and B

33 type, and within type A, A/H1N1 and A/HrN2 are the common ones. Adjuvants have been used with seasonal and pandemic Flu vaccines such as Fluad (MF59 adjuvanted trivalent influenza vaccine), Pandemrix (AS03 adjuvanted A/H1N1 vaccine), and more. Emulsion adjuvants have shown dramatic increase in Flu vaccine potency, allowing to reduce the antigen dose dramatically, as well as reducing the number of immunizations required for complete protection (1, 40, 53-56). Overall mechanistic studies with influenza vaccine show that these emulsion adjuvants act by inducing proinflammatory cytokines and improve cell recruitment independent of TLR pathway or Interferon type-1 pathway. Thus, influenza antigens make an ideal candidate for novel emulsion adjuvants evaluation.

The 2009 H1N1 Pandemic showed shameful statistics of Flu vaccine available only in rich countries as enough doses for worldwide distribution were not available. This is partly due to both limitation of both antigen and adjuvant manufacturing capabilities (54).

There were several important lessons from the 2009 vaccine pandemic, most important being the pre-pandemic readiness for even more pathogenic strains of Flu virus such as

H5N1. H5N1, in controversial research, has shown transmission in ferrets (1, 57, 58). In the 2009 pandemic, the first case infected individual reported was in March and the manufacturing for bulk doses didn’t begin until June, with first doses shipped as late as

October. Adjuvants can help double or even quadruple the doses of flu vaccine (54).

Although AS03 adjuvanted A/H1N1 was used in the 2009 pandemic, sufficient doses of the adjuvant could not be produced. Additionally, due to manufacturing constraints, these adjuvant doses could be distributed only in rich countries (1). It would be prudent to improve the manufacturing process of antigen, but also the adjuvant. Perhaps, simpler and comparatively inexpensive methods of formulating the emulsion adjuvants, requiring less

34 stringent site specifications could allow for bulk manufacturing of these “novel” emulsions in developing and poor countries, allowing for a uniform distribution of influenza doses to children and adults worldwide.

1.4 Formulation of Emulsion Adjuvants: Current and Alternative Strategies

In this section, several different was of formulating emulsion adjuvant have been reviewed with the possible advantages and disadvantages of each.

1.4.1 Current Formulation and Manufacturing of Emulsion Adjuvants

Figure 1.3 shows the current methods of emulsion manufacturing. AS03, MF59, and SE are manufactured using high pressure homogenization (HPH) and microfluidization subjecting the oils to high shear. AF03, from Sanofi Pasteur, is formulated using a Phase inversion temperature (PIT) method (59). These methods subject the oils and surfactants through high shear, pressure and/or temperature changes.

Figure 1.3: Manufacturing of emulsion adjuvants using (a) microfluidization or High- Pressure Homogenization (HPH) or (b) Phase inversion temperature (PIT). Figures taken from Haensler J. Manufacture of Oil-in-Water Emulsion Adjuvants. Methods in molecular biology (Clifton, NJ). 2017; 1494:165-80.

Advantages: The resulting droplet size from such processes is usually in the submicron size of 100-180nm and homogenous (PdI ˂0.2). The process yields robust and repeatable doses and have shown to produce millions of doses.

Disadvantages: In general, these processes are complex, and often involves several crucial steps needing GMP facilities with rigorous specifications. Such facilities come with a great expense and thus, are rarely found in developing countries. Figure 3 shows the steps involved in manufacturing these emulsions, from raw materials, to final vaccine. These methods are expensive, since they require several different steps and in-process characterization. In addition to that, these bulky instruments require laborious cleaning and maintenance adding to the cost of the already expensive and complex process. These processes are unable to generate emulsions with particle sizes as low as 20nm, to understand size-related differences in adjuvant potency in vivo.

Fox et al. describe in detail the process and costs in transferring technology to manufacture oil-in-water emulsions to a facility in Romania (60). Additionally, they also show in vivo data with split virion H5N1 to establish successful technology transfer of antigen and adjuvant (61). Although they acknowledge the need for transferring manufacturing capabilities of adjuvant and vaccine in developing countries, the process and costs highlighted might be the major hurdle in overcoming the problem.

1.4.2 Alternative Formulation Strategies for Emulsion Adjuvants

Since the process of emulsification using microfluidization, HPH or PIT is so convoluted, one of the strategies for making novel emulsion adjuvants would be to simplify the process. Work done by Shah et al, used a self-emulsification process to generate

36 emulsion adjuvants with sizes as low as 20nm - SEA20 and as high as 160nm - SEA160

(62). Self-emulsified adjuvant systems have been researched and used since several decades where the idea is to use oils and lipids to solubilize lipophilic drugs with the help of surfactants and deliver orally via capsules. These capsules then self-emulsify in gastric or enteric environment making the delivery of lipophilic drugs possible (63). Figure 1.4 shows a possible alternative strategy of formulating emulsions via self-emulsification. This process can be used to formulate emulsion adjuvants “ex-vivo” by mixing oils and surfactants and then generating emulsions by heating the isotropic mixture of oils and surfactants at body temperature i.e. 35-40°C. Such a process yielded squalene o/w emulsions with droplet size as low as 20nm and as high as 160nm which is similar to the size of MF59 and AS03. In comparison with MF59, SEA160 showed similar immune responses in vivo when administered with antigens such as ovalbumin, Influenza A/H1N1, and HIV gp120. Hence self-emulsification process provides for an alternative platform, which is less complex and inexpensive, to further explore development of novel emulsion adjuvants.

Advantages: This process is simple and requires minimum shear and pressure in yielding emulsions with similar droplet size as microfluidization or PIT. The process on a laboratory scale requires magnetic stir plate and stir bars, a heat block and conical tubes and glass vials. Although this process hasn’t been scaled up, the anticipated instrument in manufacturing would be jacketed vessels with heat and temperature controls and standard manufacturing scale mixing equipment. Thus, this process fairly and comparatively inexpensive suggesting the possibility of manufacturing in any simple and well-established manufacturing facility.

Disadvantages: Sterile filtration at lab scale cannot often be translated well to a manufacturing scale emulsion, and thus, issues during sterile filtration may occur.

Appropriate ratio of surfactants to oils is needed to produce stable emulsions.

Figure 1.4: Self-emulsification process ex-vivo as an alternative method to create emulsion adjuvants

1.5 Conclusions

Emulsion adjuvants have shown an exponential difference in influenza vaccines in terms of reducing the antigen dose, thus making them pivotal in seasonal and pandemic influenza vaccine programs. Although adjuvants like MF59 and AS03 improved the 2009

A/H1N1 pandemic situation, the total number of doses available especially in developing and poor countries were insufficient by a large margin. The primary reason was lack of manufacturing capability of adjuvants and antigen in these countries in a short period, as well as the cost of shipping several billion doses across countries. One of the major take- home from the experience during A/H1N1 pandemic, was the need to be prepared for such events, and have adequate capability to provide for the required doses. An alternative,

38 simple and inexpensive method of formulation for emulsion adjuvants may be plausible solution.

Although several other novel methods of formulation of emulsion adjuvants is encouraged (64), along with building next generation adjuvants by delivering immune potentiators; focus and efforts should also be placed in finding ways to cost-effectively produce adjuvants for storage and deployment in less privileged countries.

CHAPTER 2 Alpha-Tocopherol (Vitamin E) as an Immunostimulant

2.1 Introduction: Vitamins in vaccines

It has been a known fact that vitamins are essential part of the body required for normal growth of the organs, especially since they are not synthesized in sufficient quantities by the human body. The indirect effect of vitamin-deficient diet on immune system can be traced in research back to 1900s. Most, if not all vitamins have been used as immune supplements (65). Fat soluble vitamins like A, D, and E have shown more influence directly or indirectly on immune system, as a supplement or as an adjuvant in vaccine. Recently use of B-complex vitamins and their metabolites in addition to vitamins

A, D and E have shown promise in improving immune response of influenza vaccine (66).

Retinol or retinoic acid (RA), otherwise known as vitamin A is obtained from preformed retinoids and carotenoids. Retinoic acid or retinol has often been associated with reduced immunity to infectious agents especially eye-related infections (67). Immune supplementation with vitamin A is recommended by WHO in developing countries, despite the controversial benefits offered by them. vitamin A included in emulsions as adjuvants were compared to Alum in a TdaP (Tetanus, Diphtheria and Pertussis) vaccine in mice and showed higher antibody response correlated with higher Th2 type CD4+ response (68). In a unique mustard-oil emulsion adjuvant, vitamin A showed mucosal immunity as well as high antibody production (69). In general there is evidence that RA increases the B-cell maturation and differentiation along with antibody class switching, making it a potential adjuvant for mucosal immune response against pathogens like HIV and tuberculosis (70).

Additionally, RA and its derivatives might also have an effect on regulating the Th17 and

40 regulatory T cell response based on the dose used (71). Preclinical data thus far has shown promise in vitamin A as an adjuvant, however more standardized studies with uniform doses and appropriate comparators are needed to justify its use as an immune potentiator.

Research on molecular mechanism of action might help in use of RA indications as an adjuvant.

Vitamin D has a biologically active form - 1α,25-dihydroxy vitamin D3 that has receptors on several innate immune cells rendering them important in regulating immune responses. It is known mostly to enhance IL4, IL5 regulating Th2 type responses (72).

Additionally, when this active form of vitamin D was co-administered with inactivated polio virus vaccine, higher amounts of IgA antibody in addition to IgG antibodies was observed, suggesting its role in mucosal immunity. In another study with intramuscular injection in pigs with vitamin D showed antibody isotype switching improving IgA, IgG and IgM responses (73). Evidence of APC migration to lymphatic organs as well as enhanced antibody response after administering vitamin D as an adjuvant suggests its active role in improving both cellular and humoral immune responses. Although the research conducted thus far looks promising, detailed preclinical as well as clinical studies might be needed to understand the exact effect of vitamin D as an adjuvant (70).

Water soluble vitamins, like vitamin C and some of the B-complex vitamins have been believed to “boost” immune response, however not a lot of preclinical or clinical data is available to confirm this. Several research studies for veterinary vaccines have been conducted with vitamin C as an immune or dietary supplement, while the exact mechanism is still unknown (70). However, there has been recent focus on use of B-complex vitamins as potential adjuvants. Folic Acid (vitamin B9) and riboflavin (vitamin B2) metabolites

41 when used as antigens have showed presentation to Mucosal-associated invariant T

(MAIT) cells via MHC class I-like molecule MR1 cells that are found abundantly in humans (74). This study implies that MAIT cells use these metabolites to detect bacterial infection and thus, these metabolites could potentially act as adjuvants in improving antigen presentation. Additionally, a comprehensive evaluation of several adjuvants including vitamin B2 and B9 by Quintilio, et al. showed that Alum with vitamin B2 showed most promising immune response in enhancing overall immune response of the A/H1N1

A/California/07/2009 (66). In the same study, vitamin E or α-tocopherol showed the highest increase in HAI titers as well as IgG1 and IgG2a titers.

Vitamin E is considered as one of the most valuable vitamins due to its extent of biological activity as an antioxidant and the abundant research in animal models regarding its use as immune supplement. We have discussed the use of vitamin E in detail in the next sections.

2.2 History of vitamin E: Immune Supplement and Veterinary Vaccines

Vitamin E, originally discovered in 1922, is one of the major lipid soluble antioxidants found in cells. It exists in 8 isotropic forms – four tocopherols and four tocotrienols. α-tocopherol is interchangeably used with vitamin E as it’s the most biologically active isomer of vitamin E for its antioxidant properties. It was discovered as an important component for fertility in rats, however was later found to be most important in preventing peroxidation of phospholipids in cell membrane (75). The exact mechanisms of α-tocopherol as an immune supplement are not clear however there are several animal and clinical studies that manifest its role in boosting the immune system. Alternatively, vitamin E has also shown anti-inflammatory responses contradicting its immune action.

Most of the biological activity of vitamin E stems from its antioxidant properties to prevent peroxidation of cell membrane, by free radicals, especially in immune cells. For example macrophages contain high amounts of vitamin E to prevent oxidation of their phospholipids

(76). A direct correlation of antioxidant activity and immune potentiation by vitamin E is unclear and remains to be established.

2.2.1 Vitamin E as an immune supplement

Vitamin E impacts on several important proteins of arachidonic acid metabolism.

Arachidonic acid is the precursor of many different molecules like prostaglandin, thromboxane, or leukotriene involved in inflammation and immune functions. Inhibition of phospholipase, lipoxygenase, or cyclooxygenase leads to decreased prostaglandins, thromboxane, and leukotriene levels (77). All these prostanoids play an important role in the regulation of the cross talk between innate and adaptive immunity. Some of the recent studies show that supplemental vitamin E influences early signaling proteins involved in the immune synapse of naïve T cells, resulting in enhanced IL2 transcriptions beginning the T cell proliferation (78). Several studies have shown that age-related shift of immune responses from Th1 to Th2 due to lesser IL2 activation and hence T-cell proliferation can be alleviated with vitamin E supplementation (79, 80). vitamin E supplementation has shown resistance against several pathogens by enhancing cellular and humoral immune responses (65, 81). Particularly in the elderly, vitamin E supplementation has shown an impact on immunostimulatory activity by possibly restoring age-induced decline in immune responses. In a particular study in 2005, vitamin E supplementation was directly correlated with improved immune responses from Influenza vaccine in elderly population

(82). Although there is plenty of evidence on vitamin E supplementation and immune

43 response, there has been a lack of consistency in studies in terms of end-point, dosage, use of isomers other than α-tocopherol (83-85). Thus, probability of human trials with α- tocopherol or combination of antioxidants has largely been negative. A lot of differences between responses have genetic root cause and thus, instead of vitamin E supplementation, co-administration of vitamin E was researched.

2.2.2 Vitamin E in veterinary vaccines

There is large amount of preclinical evidence of use of vitamin E, either as supplement or directly as an adjuvant, in preclinical models (86-88). The first evidence of use of vitamin E in veterinary adjuvants was in immunization of chickens to substitute the mineral oil component of Freund’s Adjuvant. After incorporating vitamin E into vaccines for sheep, dogs, and mice, enhanced immune responses were observed compared to the conventional adjuvants used in veterinary vaccines (84, 85, 87, 89). These responses were higher than mere immune supplementation of vitamin E. Hence, subcutaneous or intramuscular delivery of vitamin E was found to be advantageous. Despite of plenty of evidence showing the effect of vitamin E supplementation as well as injecting vitamin E with the vaccine, it’s use in human vaccines was limited. This learning of α-tocopherol in preclinical models could be used in creating novel adjuvants for humans containing vitamin

2.3 AS03 mechanism of action and role of α-tocopherol as an immunostimulant

AS03 was described in detail in Chapter 1 as an emulsion adjuvant. Although the exact mechanism of action of AS03 or α-tocopherol in AS03 is not known, there have been several preclinical studies that point at role of vitamin E in enhancing the immune response of AS03 (90). The scientists at GSK also discovered that AS03 induces release of local

44 cytokines and chemokines that may be responsible for higher antibody response after AS03 administration (Figure 2.1). It was also observed that induction of these cytokines peaks at about 6 hours post injection and then the levels started to drop, suggesting a fast clearance of the emulsion from injection site. They also manifested the importance of co- administration of adjuvant with the antigen for optimal immune response. In comparison with Alum, AS03 showed enhanced recruitment of antigen positive monocytes and DCs by tracking their migration to draining lymph nodes (dLNs) when administered with fluorescently labeled OVA. Other than monocytes, granulocytes like eosinophils and neutrophils were found in dLNs compared to DCs and macrophages. Induction of cytokines in muscles corroborated the results obtained from dLNs, thus, suggesting that AS03 strongly induces recruitment of APCs compared to Alum, and can may be correlated with higher antibody responses post AS03 administration with antigen. To understand the role of α-tocopherol in AS03, they compared AS03 with an emulsion without α-tocopherol in terms of cytokine, cellular and humoral responses. In vivo cytokine kinetics at the injection site in muscle revealed that AS03 induced significantly higher amounts of most chemokines and cytokines except IL-6, IL1β, CCL5 and CCL3, which were higher in emulsion without α-tocopherol. This data suggested that AS03 enhances recruitment of monocytes more than macrophages or Eosinophils in contrast to AS03 without α- tocopherol. When tested with fluorescent OVA and analyzing cell population in the DLNs, it was observed that antigen loading in monocytes and granulocytes was significantly better than AS03 without α-tocopherol (Figure 1). After analyzing the OVA+ cell population in dLNs, AS03 showed significantly higher OVA+ monocytes compared to DCs, manifesting monocytes as the primary cell target for recruitment for AS03 at the injection site.

Figure 2.1: From Morel et. Al. (90) α-Tocopherol in AS03 modulates immune cell numbers and antigen loading in monocytes in the dLNs. (a) Number of monocytes, DCs, eosinophils and neutrophils/dLN and cell numbers were measured by flow cytometry in samples taken at 3, 24, 48 and 72 h after the injection of 5µg of fluorescent Ova, either alone (no adjuvant), adjuvanted with AS03 or AS03 without-tocopherol (AS03 w/o α- tocopherol). Each data point describes the value determined from a pool of dLNs from 6 mice. Vertical bars describe the geometric means of 3 pools. Significant differences between AS03 and AS03 w/o α-tocopherol groups are indicated above the vertical bars. (b) Fluorescent Ova (5µg), alone or adjuvanted with AS03, AS03 without α-tocopherol

(AS03 w/o α-tocopherol) or aluminum hydroxide (Alum), was injected and the amount of Ova/cell (mean fluorescent intensity; MFI) was measured by flow cytometry at 24 h after injection (corresponding to the peak of cell recruitment) in monocytes and DCs. Each data point describes the value determined from a pool of dLNs from 4 mice. Vertical bars describe the geometric means of 3 pools. Significant differences between groups are indicated above the vertical bars. Comparisons were made by ANOVA, followed by Tukey’s test; *P < 0.05; **P < 0.01; ***P < 0.001.

In an in vivo study using Hepatitis B antigen (HbsAg) AS03 showed about 6-fold higher antibody response compared to AS03 without α-tocopherol (Figure 2.2). This difference however, was lower than that between Alum and AS03. This shows that induction of some early pro-inflammatory cytokines like IL-6, CSF3 and CXCL1 along with enhanced antigen loading of monocytes contributes to higher antibody responses for emulsions in general compared to Alum. Although comparison was made with AS03 without tocopherol, the data thus far suggests the role of tocopherol in enhancing the cell recruitment and directly influencing the antibody production. In summary, there is not much preclinical data directly comparing the role of α-tocopherol in emulsion adjuvants as an immunostimulant.

Figure 2.2: Anti-HbsAg antibody titers comparing AS03 with AS03 w/o tocopherol. Mice (n = 10/group) were immunized at day 0 and day 14 with 2g of HBsAg, alone or adjuvanted with 2, 4 or 10µl of AS03, or 2, 4 or 10µl of AS03 without α-tocopherol (AS03 w/o α- tocopherol) in injection volumes made up to 50µl with PBS. Anti-HBsAg titers determined by ELISA on blood samples taken at day 28 (0.3g/mL was the limit of detection). Vertical bars describe the geometric means and error bars describe 95% CIs. Groups were compared by ANOVA, followed by Tukey’s test. *P < 0.05; ***P < 0.001.

Clinically, there hasn’t yet been comparison between AS03 and emulsion adjuvant without α-tocopherol. However, there is one clinical study where AS03 was compared with

MF59 (squalene only adjuvant). The goal of the study was to compare the two adjuvants in different permutations and combinations with H7N9 influenza antigen, to test whether either of the adjuvants can be used during a pandemic situation (91). In this study, two doses of AS03, three weeks apart with 3.75 µg inactivated H7N9 showed the highest hemagglutination inhibition antibody titers compared to either unadjuvanted vaccine, or two doses of MF59, or alternating doses of MF59 and AS03 (Figure 2.3).

Figure 2.3: Hemagglutination Inhibition Antibody titers using H7N9 and either no adjuvant, MF59, AS03 or combination of these. Number of participants were randomized such that total number of participants in each age group: 19-34 y (n = 454), 35-49 y (n = 216), 50-64 y (n = 276)

Recently, some more work was done to identify the underlying plausible mechanism for higher antibody avidity post AS03 administration. Givord et al in their study found that AS03 acts via the endoplasmic reticulum stress kinase receptor IRE1α in the APCs such as monocytes by sensing the metabolic changes at the local site, thus contributing towards enhanced cytokine release and eventually higher antigen-specific antibody titers (92). They show that these IRE1α in myeloid cells sense the change in lipid metabolism after AS03 administration, which may enhance the production of IL6, improving the T-follicular helper (TFH) cell response. There was a further commentary on this work speculating the role of individual components in elucidating the adjuvant action

(93). They also reviewed the role of unfolded protein response on other cytokine gene transcription factors. Thus, this work sheds light on the existing knowledge on the emulsion adjuvants led mechanism of innate immune response enhancement. This will pave the path to identify novel emulsion adjuvants that can be used to enhance the antigen-specific

49 immune response. It also asks the question on role of squalene and/or α-tocopherol and the specific formulation strategies and amounts needed to carry out the desired immune response.

2.4 Pharmaceutical properties of α-tocopherol

vitamin E has a wide application as antioxidant in several oral as well as parenteral pharmaceutical and cosmetic products. It’s 8 isoforms provide a wide range of pre- formulation characteristics that can be used with several active agents. Mostly, vitamin E is used to dissolve lipophilic drugs and eventually emulsify to deliver the active ingredient in vivo. α-Tocopherol is poorly water soluble, and hence the formulation into a drug product has always been a challenge (94, 95). One of the popular applications of α-

Tocopherol and it’s several salt forms such as acetate or succinate is to dissolve lipophilic drugs and deliver orally in soft gelatin capsules to form self-emulsified formulations and release the drug via lymphatic system (96). Several different formulations have been used to make soft gelatin capsules for self-emulsification and often surfactants are used to stabilize α-Tocopherol emulsions (97-99). An ex-vivo self-emulsifying formulation of α- tocopherol was evaluated and compared to a soft gelatin capsule in terms of stability to find no significant differences in bioavailability (96), thus proving that α-tocopherol can be self-emulsified externally; however, the purpose in this case was in situ emulsification, and hence feasibility of “external” self-emulsification and stability were not evaluated. α- tocopherol has often been formulated by itself as a single oil component, mostly as a carrier to lipophilic drugs. It has also been used in its several other forms such as succinate or acetate. There are few instances of use of α-tocopherol with other oils to form a stable emulsion formulation. AS03 is one such example where high pressure homogenization and

50 microfluidization is used to formulate the emulsion adjuvant (100). vitamin E – succinate or tocopheryl polyethylene glycol succinate (TPGS) is another derivative of α-tocopherol that has shown little to no biological property but is often used as a surfactant in formulations (98). It is still unknown whether different salts and isoforms of α-tocopherol show similar biological property; however, most of these isoforms and salts are used to substitute α-tocopherol, mainly as an antioxidant in pharmaceutical drug products.

Among the currently researched vaccine adjuvants, α-tocopherol is used in two emulsion adjuvant formulations. It is used in stable emulsion (GLA-SE) from IDRI as an antioxidant in small amounts (~0.01% v/v), whereas in AS03, a GSK adjuvant, it is used at a concentration of 2.5% v/v (47) and was shown to add to the immune responses via

AS03 (90). Both emulsion adjuvant formulations use high shear and pressure to emulsify the oil components. Thus, a stable self-emulsifying formulation of α-Tocopherol does not exist hitherto, especially formulated with another oil component squalene for use as a vaccine adjuvant.

2.5 Conclusion

Despite several years of using vitamin E as an immune supplement and manifesting its advantage as an adjuvant in veterinary vaccines, the exact molecular mechanism of action of α-tocopherol as unknown. However, the work done hitherto has shown a promise in α- tocopherol as an immune stimulant, especially in a vaccine adjuvant. It would be revolutionary, for the field of vaccine adjuvants, to discover specific molecular and cellular mechanism of action of α-tocopherol. First step towards this could be formulating emulsion adjuvants containing α-tocopherol in a simpler manner to make it easy for high throughput formulation evaluation.

CHAPTER 3 Lyophilization of Antigen-in-Emulsion Adjuvants to Create a Single Vial Vaccine Product

3.1 Introduction

Recombinant subunit vaccine antigens are evolving quickly and replacing the conventional vaccines such as live attenuated or whole-inactivated virus; owing to higher safety and purity of these synthetic antigens. Adjuvants such as aluminum salts, oil-in- water emulsion adjuvants, liposomes etc. have been used with these antigens and are explained in more detail in chapter 1. The usual presentation of these vaccines is a two or three vial with bed-side mixing of the components prior to administration. The adjuvant is usually stored and shipped at refrigerated temperatures i.e. 2-8°C. The subunit antigen is usually stored and shipped at frozen temperatures (-60 to -80°C) or in many instances the antigen is also usually lyophilized to obtain a dry powdered cake. For several vaccine drug products there’s a third vial containing buffer or diluent for either the antigen or adjuvant or both. Although this presentation has worked for most clinical vaccines, there are several steps involved along with several chains for shipment and storage of vaccines. In this chapter, and next two sections, the motivation to build a single vial vaccine especially by means of freeze-drying are discussed in detail.

3.1.1 Rationale for single vial adjuvanted vaccine

Cold chain distribution of vaccines has hitherto driven the modern vaccine and adjuvant research. The drug product is developed to prove stability from the time of manufacturing to bed-side vaccine administration, depending heavily on cold-chain storage and distribution (101). Even during cold-chain storage, vaccine damage and reduced

52 vaccine efficacy has been observed due to failure in cooling systems, especially in developing countries and in developed countries after epidemics or natural calamities

(102). In general, the two-vial presentation of vaccines in less-privileged countries demands cold-chain storage leading to increased costs, eventually leading to high price mark-ups for these products making them less affordable (103, 104). Vaccine wastage due to temperature instability is a common occurrence, thus several vaccines are lyophilized and either reconstituted with a buffer or adjuvant. Although lyophilization of vaccine antigen has helped stabilize and preserve the efficacy of the antigen, newer strategies are required to improve distribution and administration of adjuvanted vaccines, ultimately improving the cost-effectiveness (105). A lot of effort and focus today is on stabilizing the adjuvanted vaccines by eliminating the need for cold chain distribution by directly or indirect means (106, 107). Eliminating cold-chain and possibility of making potent adjuvanted vaccine available globally can be achieved by developing single vial adjuvanted vaccine. Since most of the currently used vaccine antigens are lyophilized, for reasons beyond stability as well (108, 109), freeze-drying of antigen with the adjuvant may seem like a possible solution. However, lyophilizing the antigen with the adjuvant may pose a different set of stability concerns that would need addressing (4). Adjuvants like Alum salts and oil-in-water emulsions have several attributes, like size, conformation, content etc., that can affect the potency of the adjuvant or antigen, and thus the efficacy of vaccine.

Attempts have been made to lyophilize either Alum (22) or emulsion adjuvants with vaccine antigen to make a dry single vial vaccine (3, 4, 110). Stable single vial adjuvanted vaccine can revolutionize the distribution and administration of vaccines facilitating delivery of vaccines in every part of the world.

3.1.2 Why lyophilize?

Dry powdered vaccines have gained a lot of interest, especially due to continuous research for nasal route of vaccine administration for higher mucosal immune responses

(111, 112). The appropriate formulation of antigens in a dry, solid state can result in improved stability, thereby removing cold-chain storage complications associated with conventional liquid-based vaccines making them ideal for pulmonary vaccination (113).

The particulate nature of dry powder vaccines could also induce a better immune response.

Drying of adjuvant with the antigen poses a few stability challenges for both antigen and adjuvant. Two commonly using drying techniques in pharmaceutical world have been used to research dried powdered vaccines – spray drying and lyophilization.

Spray drying of vaccine antigen with stabilizing polymers have been demonstrated in several different studies (111, 112). Spray drying of adjuvanted vaccines occurs by forcing the well-mixed drug product through desired nozzle size into a pre-heated vessel collector. Also, spray freeze drying has been used in which instead of atomizing the antigen solution into a preheated vessel of the spray dryer, the antigen solution with polymers and/or stabilizers are atomized into a vessel containing liquid nitrogen which is then

“dried” using a lyophilizer or freeze-dryer (114, 115). Inspired by spray drying of antigens,

Thakur et al. also generated a single vial spray dried adjuvanted vaccine with tuberculosis antigen and adjuvant CAF01 (27). Even though spray drying has been used extensively for creating dry powdered vaccines specifically for inhalation, it might not be the most effective method to generate a single vial vaccine for parenteral administration. Spray drying uses heat from hot gas stream for solvent evaporation and atomizing the solution into spray dried powder. Most proteins undergo irreversible thermal denaturation upon

54 exposure to high temperatures (116). Thus, stability of the antigen as well as individual adjuvant components as well as interactions between different components of the adjuvanted vaccine at the high spray-drying temperatures may be unpredictable and limiting step in this process. Even in spray-freeze drying, the delicate settings of nozzle size, atomization rate and pressure may be detrimental to the antigen stability. Another issue would be the sterilization of the spray dried material as well as the yield, in which radiation sterilization or autoclaving may be an issue. Lyophilization or freeze-drying can be the alternate solution to spray drying since it is also extensively used, not only in pharmaceutical world but also, in producing dry powdered vaccine or vaccine antigens for reconstitution (108, 109).

In any single vial adjuvanted vaccines, it is imperative that both antigen and adjuvant maintain their efficacy and potency, respectively, after drying. Lyophilization or freeze-drying helps in arresting the Brownian motion of molecules in a matrix atmosphere with diluents by facilitating sublimation of ice. For most nanoparticulate adjuvants where size is an important factor, freeze-drying might help preserve droplet size and limit the interaction between antigen and adjuvant in a dried drug product cake. Orr et al from IDRI showed that lyophilizing co-vialed antigen with emulsion adjuvant (GLA-SE) was possible with a slight increase in size post lyophilization that did not affect the in vivo potency of the adjuvant or efficacy of the antigen (4). They also showed that at elevated temperatures the lyophilized single vial vaccine maintained its size upon reconstitution as well as showed comparable immune responses to not only lyophilized vials stored at 4C but also the fresh formulated liquid formulation. Also, adjuvanted tuberculosis vaccine was successfully lyophilized in a cartridge without significant changes in physicochemical properties of all

55 components of the vaccine (117). Iyer et al also successfully showed lyophilization of emulsion adjuvants along with the protein antigen, without changes in physicochemical properties of adjuvant or antigen. Although, they observed a size increase in the final formulation post reconstitution, they found out that this increase did not affect the in vivo performance of the vaccine using two protein antigens (3). In the next section, the lyophilization process is discussed in detail in terms of nanoemulsion formulations, Mainly the impact of the process on the physicochemical properties of the formulation and considerations to achieve an ideal product will be discussed.

3.2 Lyophilization of Nanoemulsions: feasibility and considerations

Lyophilization or freeze-drying is essentially used for improving stability of a biopharmaceutical product and essentially facilitate long-term storage and shipping of the drug product (118). Optimization of a freeze-drying cycle plays a very important part in making the manufacturing of the drug-product more cost-effective. Typically, freeze- drying process consists of five steps (Figure 3.1). It is important to test the feasibility of single-vial liquid formulation and short-term stability during the lyophilization process.

For emulsion adjuvanted vaccine, it is important that upon reconstitution, the concentration of adjuvant and antigen should be the same as for bedside mixed vaccine. A pre-screening of stability of both adjuvant and antigen in same vial with the final buffer and diluents should be performed before proceeding with the freezing step to evaluate vital physicochemical properties of the formulation.

Figure 3.1: The five main stages of lyophilization include formulation, freezing, primary (1°) drying, secondary (2°) drying, and stoppering. In liquid formulation, excipients are added to the active ingredient and then vialed and half stoppered. Samples are then frozen. During primary drying, a vacuum is pulled to lower pressure and heat is applied by increasing the shelf temperature. This allows sublimation to occur, removing unbound water from the samples. During secondary drying, temperature is increased and bound water is removed from the samples. Samples are then brought back to atmospheric pressure, blanketed with an inert gas, and stoppered. (119)

The process of freeze drying begins with freezing of the formulation typically between -50 to -40°C. For emulsified systems, this is a crucial step since ice crystal formation may cause the droplets to aggregate or swell and temperature driven crystallization of buffer salts may cause changes in pH eventually altering drug product stability (118). The freezing ramp should be carefully determined based on the formulation as well as the nanoparticulate system used. The increase in size in nanoemulsions, post lyophilization, observed in a few studies carried out (3, 4, 120) manifests that the freezing step is critical and might lead to disruption of nanoemulsion if not carefully designed. In some instances, to cause uniform ice crystal development, the annealing step where for a certain period, the drug product is maintained at the subfreezing temperature above the Tg

(glass transition temperature) but below the eutectic temperature of the diluent. Usually mannitol is the diluent of choice along with either trehalose or sucrose in the formulation for annealing. Annealing might accelerate and/or improve the growth of ice crystals leading to quicker primary drying (118). However, annealing might not be the solution for several formulations where size is an important factor. With nanoemulsions, annealing might cause aggregation of oil droplets upon reconstitution. The next step after freezing is primary drying where the ice crystals formed during the freezing step sublimate causing pores in the frozen solid cake. Evaporation of the ice usually starts from the top and makes its way to the bottom of the frozen liquid. Primary drying is induced by lowering the pressure and increasing the temperature. Collapse temperature of the formulations should be determined before the freeze-drying process of nanoemulsions, and the primary drying should occur below the collapse temperature. This avoids the collapse of the drug product into a rubbery amorphous state (121). Primary drying usually depends on the combination of higher temperature and lower pressures and can take from several hours to several days. It is very important to complete this step which marks the evaporation of all the ice crystals from the frozen cake. The next step is secondary drying which can occur at higher temperatures and marks the removal of residual water to render a dry lyophilized cake. In this step, the bound water that remains in the frozen cake after primary drying, is removed by increasing the temperature. The higher the temperature, the faster is the removal of water. For nanoemulsions, the temperature should be carefully selected to avoid degrading any components of the emulsions due to exposure at higher temperatures. Ensuring complete removal of water is essential for reducing mobility between molecules and thus improving the shelf life stability of the drug product.

For nanoemulsions, these steps should be carefully designed based on the pre- formulation characteristics of the final adjuvanted formulation. Not only the adjuvant nanoemulsion stability, but also antigen stability before and after lyophilization should be monitored. With several changes in parameters such as pressure and temperature throughout the freeze-drying process, the interaction between adjuvant and antigen or instability of individual components of the single vial drug product is highly likely. Thus, freeze drying of nanoemulsion systems should be approached case by case, keeping in mind the critical quality attributes of the final drug product (122).

Feasibility of freeze-drying nanoemulsions: There are several ways and methods that can be used to determine the feasibility of lyophilizing nanoemulsion systems. One of the first and most important one is the size and polydispersity of the nanoemulsion before and after lyophilization. Dynamic light scattering can be used to determine droplet size changes before and after lyophilization. Several studies as mentioned before have shown a marginal increase in size after lyophilization; however, the exact mechanism of this increase is unknown. This size increase has also shown to not affect the in vivo performance of the nanoemulsions. Other orthogonal techniques such as nanoparticle tracking might be sensitive to particle concentration in solution, as well as static light scattering that might show larger aggregates. These techniques help evaluate the changes in size which is most crucial step in determining feasibility of lyophilizing nanoemulsions.

To lyophilize nanoemulsions with the antigen, first the properties of the drug product to be lyophilized should be evaluated with appropriate diluents. Collapse temperature and glass transition temperature must be determined to set the freezing and primary drying temperatures. Before lyophilization of single vial vaccine, feasibility of freezing of the

59 nanoemulsions and evaluation of different diluents can be conducted by simple freeze thaw studies with size and PdI as important parameters. Other than size, the content of individual components in adjuvant can be tested using chromatographic techniques, and the stability and integrity of the antigen can be tested using size, or in vitro potency assays, or even chromatography assays. It is also very important to test the stability of the lyophilized single vial vaccine to compare with a liquid single vial vaccine, as well as a freshly prepared standard protein.

3.3 Conclusion

Single vial adjuvanted vaccines can revolutionize storage, distribution and administration of protein subunit vaccines especially in developing countries by eliminating the dependency on cold-chain maintenance. Liquid single vial adjuvanted vaccine might augment unwanted interactions between adjuvant and antigen in long term, causing instability of the drug product; thus, several drying techniques are currently researched to provide extended shelf life. Although there are several drying techniques out there, for a single vial emulsion adjuvanted vaccine, spray drying or lyophilization may be of more relevance. Lyophilization is in general preferred for ease of storage, sterilization, shipping and distribution of the drug product; although it is more expensive compared to spray drying. Lyophilized single vial vaccine may additionally provide for long term shelf stability rendering the process even more cost-efficient and allow for long term storage of the drug product along with simplifying the shipping and distribution.

CHAPTER 4

Formulation Design, Optimization and In Vivo Evaluations of an Alpha- Tocopherol-Containing Self-Emulsified Adjuvant System using Inactivated Influenza Vaccine 4.1 Introduction

Advances in the fields of molecular biotechnology and recombinant DNA technology in the early 1980s allowed for the development of subunit protein antigens for vaccines (6). These synthetic antigens offered several prominent advantages over conventional vaccines composed mainly of inactivated pathogens, such as improved safety, higher titer yields, improved stability, and lower cost of production (5, 7). Higher purity and safety of these antigens came with a drawback of being less potent and immunogenic as compared to attenuated or inactivated vaccines, thus requiring additional or “external” innate immune activation with an adjuvant along with subunit antigen.

Vaccine adjuvants were used to enhance overall immunogenicity of the subunit antigen by acting as a particulate delivery system or directly acting on immune cell receptors to stimulate innate immune cells (8). Over the past two decades, research on vaccine adjuvants has grown remarkably, with several new generation of adjuvants being included in vaccines that are either licensed or in advanced development against infectious diseases, such tuberculosis, human papilloma virus, influenza, shingles and many more.

Vaccine adjuvants are of a diverse family, and hence, cannot be defined using a single integrated structure. They comprise several naturally occurring or synthetic materials that boost the immunological effect of the antigen. Overall, adjuvants provide advantages such as antigen dose sparing, reduction in the number of vaccinations, improving the quality of the immune response, and in some cases potentially increasing the stability of the final

61 vaccine product (10). Among these adjuvants, squalene oil-based emulsion adjuvants, such as MF59 and AS03, have shown significant enhancement of immune responses to influenza vaccine antigens, especially reducing the dose of vaccine antigen as well as reducing the number of immunizations (24, 55, 56).

AS03 is a squalene oil-based oil-in-water emulsion adjuvant containing α- tocopherol as an additional immunostimulant that has shown a dose sparing effect and a general enhancement in immune responses when administered with influenza antigens

(100, 123-125). AS03-adjuvanted vaccines have elicited more robust immune responses as compared to squalene oil-only adjuvanted vaccines in preclinical mouse models (90) as well as in humans (91).

For commercial scale, oil-in-water emulsions like MF59 and AS03 have been formulated using microfluidization, which creates stable emulsions using high shear and pressure. The complex process of manufacturing these emulsion adjuvants often requires specialized and high maintenance facilities that are scarce and found mostly in developed countries. There has always been an attempt at reducing cost in scaling up vaccines and manufacturing them in developing countries (126) for a sole purpose of making vaccines available at the right time and to the most needful population. Doses of influenza vaccines can be doubled with the use of adjuvants (54), however, the cost of manufacturing these antigens and adjuvants has limited the production and distribution to only affluent countries

(1). A potential solution would be to make available a simpler, cost-effective method to manufacture these vaccine adjuvants, so that they can be scaled up in developing countries to be made available for seasonal vaccination, as well as aid towards pandemic preparedness.

Self-emulsification is a technique used to dissolve lipophilic drug with oils and surfactants for self-emulsification in gastric environment post oral administration has been well-studied (63, 127). More recently, self-emulsification process was used to make an emulsion adjuvant similar in composition to MF59 (62), and was shown to have an adjuvant effect. The essence of this process is low energy and simplicity, that has a potential to be scaled up, and manufactured in established facilities in poorer countries.

There has not yet been an attempt to incorporate the immunostimulant - α-tocopherol in these self-emulsifying formulations to create a novel adjuvant like AS03.

α-Tocopherol, also generically referred to as vitamin E, is a widely used antioxidant in several oral pharmaceutical and cosmetic products. It is highly lipid soluble and exists in seven other isoforms. α-tocopherol is poorly water soluble, and hence the formulation into a drug product has always been a challenge (94, 95). α-Tocopherol has been widely used in oral drug products, mostly to dissolve lipophilic drugs and dosed orally in soft gelatin capsules to form self-emulsified formulations and release the drug via lymphatic system (96). Attempts have been made to formulate α-tocopherol for self-emulsification via soft gelatin capsules and often with surfactants (97-99). A self-emulsifying formulation of α-tocopherol was evaluated and compared to a soft gelatin capsule to find no significant differences in bioavailability (96), thus proving that α-tocopherol can be self-emulsified; however, the purpose in this case was in situ emulsification, and hence feasibility of

“external” self-emulsification and stability were not evaluated. α-tocopherol has often been formulated by itself as a single oil component, mostly as a carrier to lipophilic drugs. It has also been used in its several other forms such as succinate or acetate. There are few instances of use of α-tocopherol with other oils to form a stable emulsion formulation.

AS03 is one such example where high pressure homogenization and microfluidization is used to formulate the emulsion adjuvant (100).

In this study, we have adopted a simple self-emulsification formulation strategy to create a stable emulsion-based immune adjuvant containing squalene oil and α-tocopherol.

Our main objective was to employ self-emulsification to attempt to incorporate α- tocopherol in squalene o/w emulsion adjuvant to create an “AS03-like” emulsion. We have demonstrated the challenges and the use of formulation science and design of experiments

(DoE) to create a stable self-emulsified adjuvant system (SE-AS) containing α-tocopherol.

We investigated its potency in vivo using QIV and measured humoral and cellular immune responses to evaluate its potency and compare it with AS03-adjuvanted responses to the same vaccine strains.

4.2 Materials and Methods

4.2.1 Formulation Materials

Squalene oil was obtained from J.X. Nippon Oil Trading Company (Tokyo, Japan) and dl-α-tocopherol was obtained from DSM Nutritional Products (Heerlen, Netherlands).

Polysorbate 80 (PS80) was obtained from JT Baker (Center Valley, PA). Hypure water for injection (WFI) like quality water and phosphate buffered saline (PBS) were obtained from

HyClone Laboratories (Logan, UT). Tocopheryl poly-ethylene glycol sulphate (TPGS) was obtained from Antares Healthcare Products, Inc. (St. Charles, IL). Ovalbumin (OVA) was obtained from InvivoGen (San Diego, CA).

4.2.2 Design-of-Experiment (DoE) Analysis for SE-AS Optimization

Mixture design to generate 23 unique mixture ratios of squalene, PS80 and α- tocopherol was created using JMP® software. Each of these SE-AS (Figure 4.1) were self- emulsified by mixing the oils and surfactants at the respective ratios overnight at RT (RT) and mixing this premix with warm PBS at a ratio of 1:10 and heated with intermittent mixing at 35-40°C for one hour. The resultant emulsion was diluted 500x to measure size and polydispersity index (PdI) using dynamic light scattering (DLS). Repeatability studies

(for emulsions that fit the criteria) were conducted in a similar fashion, with additional pH and osmolality measurements of the formulations. pH measurement was performed using an Orion 3-star pH meter from Thermo Scientific (Waltham, MA) and osmolality was measured on advanced instruments osmometer model 2020 (Norwood, MA). Process optimization studies were carried out to modify formulation temperature and mixing speeds for oil-surfactant mixture. Data from twenty-three SE-AS was analyzed on JMP using mixture profile analysis and ternary plot to predict SE-AS ratios to fit the criteria. Fit model criteria were used to predict a design space based on size and PdI predictions. 12 new mixture ratios were generated and were formulated in a similar way as in the previous DoE.

Figure 4.1. Ternary Plot showing application of design of experiment (DoE) by identifying possible ratios of squalene, α-tocopherol and polysorbate 80 via mixture design. Twenty-three different ratios were generated and analyzed for their ability to self-emulsify, with a goal of reaching a composition similar to AS03 (“□” in the ternary diagram). Each side of the ternary diagram represents a component and each “x” represents an SE-AS

4.2.3 Effect of different surfactants on SE-AS

A list of surfactants was evaluated without or in combination with PS80 (Table

4.1). These surfactants were either previously used with tocopherol only for self- emulsifying oral formulations or based on the required HLB value for a combination of squalene and tocopherol in a ratio of SE-AS. Water soluble surfactants were added in addition to the 100% oil-surfactant mixture and generally 1:10 of the PS80 content. SE-AS were formulated in a manner previously described. For purposes of evaluation, SE-AS 5,

6 and 25 were picked to cover a complete range of tocopherol. Oil and water-soluble surfactants were evaluated in two separate experiments.

Table 4.1: List of oil- and water-soluble surfactants evaluated with SE-AS Surfactant HLB value Obtained from Oil soluble surfactants Span85 1.8 Span80 4.3 Sigma life sciences (St. Louis, MO) Span20 8.6 Aqueous surfactants Polysorbate 80 15 JT Baker (Center Valley, PA) Polysorbate 20 16.7 Emprove (Billerica, MA) TPGS 13-13.2 Antares Health pdts. (St. Charles, IL) Poloxamer 188 29 Spectrum Chemicals (New Brunswick, NJ) Poloxamer 407 22

4.2.4 Effect of Surfactant Reduction on SE-AS Development

In this experiment, the percent content [% (v/v)] of polysorbate 80 was reduced from 16 to 7 in presence of TPGS (from previous surfactant evaluations), using SE-AS 6 which showed consistent and repeatable data. Emulsions were formulated using a similar self-emulsification technique and size and PdI were measured using DLS. Repeatability experiments were conducted using the protocol previously described.

4.2.5 Stability Study of SE-AS without Antigens at Higher Temperatures.

SE-AS 22 and SE-AS 36 were subjected to a 2-week long stability study at temperatures 4C, 25C and 37C. SE-AS were formulated at Day 0 and divided into 5 vials for each time-point. The time points analyzed for size, PdI, pH and Osmolality were 0,

1day, 2day, 7day and 18days.

4.2.6 Short-term Stability Analysis of Combination OVA and SE-AS

OVA at a dose of 20 µg was mixed with SE-AS 22 or SE-AS 36 in a 1:1 ratio to mimic bed-side mixing in an adjuvanted vaccine. pH, Osmolality, size and PdI of the resulting vaccines stored at RT were measured at time 0, 2 hours, 6 hours, 24 hours, and

48 hours. At all the time points, protein integrity was measured using Sodium Dodecyl

Sulphate Polyacrylamide Gel Electrophoresis (SDS-PAGE). Materials required for this experiment were obtained from Invitrogen (Carlsbad, CA). Briefly OVA in PBS was mixed with the adjuvants at respective time-points and emulsion droplets were disrupted by centrifugation. The subnatant was collected and mixed with 4X sample buffer containing dithiothreitol (DTT) and bromophenol blue. These samples were boiled at 95°C and centrifuged at 12,000 RPM for 5 minutes. All samples were loaded on to 4-12% Bis-

Tris gel with one well containing Mark 12 Unstained Protein Standard obtained from

Invitrogen. The gel was run at 200 volts for 35-45 minutes and then placed in the Imperial protein stain obtained from Thermo Scientific overnight at RT. The next day the gel was de-stained with deionized (DI) water for 3-4 hours to remove the non-specific staining and was scanned with a Chemi-doc MP imaging system from Bio-Rad and analyzed using the

ImageLab image processing software.

4.2.7 In Vivo Evaluations of SE-AS with Inactivated Influenza Vaccine Antigens

4.2.7.1 Ethics statement: All studies were conducted in accordance with the GSK

Policy on the Care, Welfare and Treatment of Laboratory Animals and were reviewed by the Institutional Animal Care and Use Committee by the ethical review process at the institution where the work was performed. All studies were executed in compliance with provisions of the USDA Animal Welfare Act, the Public Health Service Policy on Humane

Care and Use of Laboratory Animals and the U.S. Interagency Research Animal

Committee Principles for the Utilization and Care of Research Animals.

4.2.7.2 In vivo study design and immunization regimen: Cell-line derived CMV pentamer was used to test the in vivo potency of the novel SE-AS in mice. The dose of antigen in each mouse was 0.05µg. The antigen content was determined by UV/visible spectrophotometry as well as reverse phase HPLC. Sample size of 10 animals per group was calculated such that it provided a power of 80% to detect a 3-fold difference between any 2 groups with 95% confidence. 6-8 weeks old female C57BL/6 mice from Charles

River laboratories (Gaithersburg, MD) were used. The study design comprised of three immunizations three weeks apart, where 50µL of the vaccine was injected intramuscularly in the gastrocnemius muscle such that at each time-point alternate limb was used. The groups in this study were: Physiological saline (as negative control), unadjuvanted CMV,

CMV with SE-AS 44, CMV with AS03 and CMV with SEA160. The vaccine was formulated to mimic bed-size immunizations, with 1:1 mixing of antigen and adjuvant, to reach the final antigen concentration. All the formulations were characterized for pH,

Osmolality, size and PdI and endotoxin before immunization. Endotoxin was measured using the Endosafe NexGen – PTS from Charles river laboratories (Wilmington, MA) and limulus amoebocyte lysate (LAL) cartridges with test range 10-0.1 EU/mL. It was ensured that the vaccine dosed for each group had endotoxin lower than 1 EU per dose. Bleeds were collected 3 weeks post 1st (3wp1), 3 weeks post 2nd (3wp2) and 3 weeks post 3rd (3wp3) immunization and the processed sera was used to test the humoral immune responses by neutralizing antibody assay (nAb) and IgG titers by ELISA. Spleens from 5 animals per

69 group were harvested 3wp3 and 4wp3, and used to measure T cell immune responses via intracellular cytokine staining.

4.2.7.3 In Vivo Studies: Four egg-derived inactivated split influenza antigens from the licensed seasonal QIV (GSK, Dresden, Germany) were used to test the in vivo potency of the novel SE-AS in mice. The vaccine antigens were as follows:

A/Singapore/GP1908/2015 (H1N1), A/HongKong/4801/2014 (H3N2),

B/Brisbane/60/2008 and B/Phuket/3073/2013. The hemagglutinin (HA) content in these vaccine antigens was confirmed and determined by single radial immunodiffusion as recommended by regulatory authorities. These antigens were used at 2 doses 0.01µg HA and 0.1µg HA each making it 0.04µg and 0.4µg total HA, respectively per group. Sample size of 8 animals per group was calculated such that it provided a power of 80% to detect a 3-fold difference between any 2 groups with 95% confidence interval. 5-7 weeks old female BALB/c mice from Charles River laboratories (Gaithersburg, MD) were used and the study was run at Smithers Avanza laboratories in Gaithersburg, MD. The study design comprised of two immunizations three weeks apart, where 50µL of the vaccine was injected intramuscularly in the gastrocnemius muscle such that 25µL was injected in each leg at both time-points. The groups in this study were: Physiological saline (as negative control), unadjuvanted QIV, QIV with SE-AS 36, QIV with SE-AS 22 and QIV with AS03.

The vaccine was formulated to mimic bed-size immunizations, with 1:1 mixing of antigen and adjuvant, to reach the final antigen concentration. All the formulations were characterized for pH, osmolality, size and PdI and endotoxin before immunization.

Endotoxin was measured using the Endosafe NexGen – PTS from Charles river laboratories (Wilmington, MA) and limulus amoebocyte lysate (LAL) cartridges with test

70 range 10-0.1 EU/mL. It was ensured that the vaccine dosed for each group had endotoxin lower than 1 EU per dose. Bleeds were collected 3 weeks post 1st (3wp1) and 2 weeks post

2nd (2wp2) immunization and the processed sera were used to test the humoral immune responses by HAI and IgG ELISA assays. Spleens from 5 animals per group were harvested

2wp2, and used to measure T cell immune responses via intracellular cytokine staining.

4.2.7.4 Hemagglutination Inhibition Assay Titers: The primary serology readout to assess potency of novel adjuvants using influenza antigens was the HAI assay, and was performed on serum from each mouse obtained at 3wp1 and 2wp2. Before the assay, sera were treated with receptor destroying enzyme (RDE) to eliminate non-specific inhibitors of hemagglutination. To determine the viral titer, antigens used for immunization were used for A/H1N1 and both B strains; whereas, for A/H3N2, inactivated virus was used. We used Greiner Bio-One 96-Well non-treated Polystyrene “v-bottom’ microplates, to determine viral and HAI titers. Influenza HA (detergent-split vaccine antigen or heat inactivated virus) diluted to 8 HA Units /50 µL were added to pre-diluted RDE-treated serum samples and incubated at RT for 50 mins, followed by an incubation for 40-60 min with 50 µL 0.5% v/v chicken erythrocytes. The samples in each well were then read visually as either agglutinated in which red blood cells (RBCs) formed a pattern whereas non-agglutinated RBCs formed a tear drop in the center of the “V-bottom”. HAI titer was defined as the reciprocal of the last dilution that showed agglutination. Each sample was run in duplicate. Each plate had a pooled sample from the negative dosing group

(physiological saline). Titers were normalized to Log2 scale for further analysis.

4.2.7.5 IgG Subtype Antibody Titers: A multiplex assay was performed to evaluate titers of influenza specific antibodies in mouse serum (2wp2) from immunized mice.

Luminex microspheres (MagPlex microspheres, Luminex Corp from Austin, TX) of different classification were coupled with detergent split HA antigens by chemical coupling according to manufacturer instructions. In 96 well white plates, 1000 microspheres/well for each classification/ HA antigen were mixed in 50 μL/well of PBS + 1% bovine serum albumin (BSA) + 0.05% sodium azide (assay buffer) with five-fold serial dilutions of serum down each column and incubated on an orbital shaker, in the dark, at RT for 60 minutes. After washing three times with 200 μl/well of PBS containing 0.05% (w/v)

Tween 20 (wash buffer), R-Phycoerythrin (RPE) conjugated anti-mouse IgG subclasses

IgG1, IgG2a and IgG2b was added separately to each sample (for a total of three tests per sample), 50 μl/well in assay buffer, and incubated on an orbital shaker, in the dark, for 1 h. at RT. After a final wash with wash buffer and an incubation with PBS on an orbital shaker for 20 – 120 minutes, fluorescence intensity was measured using a Luminex FlexMap 3D

(Life Technologies model FM3D000, Carlsbad CA). End-point titers were calculated and used for plots and analyses.

4.2.7.6 Intracellular Cytokine Staining Assay: T-cell responses were analyzed

2wp2 by intracellular cytokine staining of in vitro antigen-stimulated splenocytes. Spleens from individual animals were processed to single-cell suspensions, followed by treatment with RBC lysis buffer (Ebioscience, Thermo Fisher Waltham, MA). These splenocytes were stimulated at one million cells per well with anti-CD3 (positive control), media

(negative control), and individual QIV antigens. For A/H1N1, the detergent-split antigen was used, whereas recombinant HA proteins obtained from Protein Sciences (Meriden,

CT) were used for A/H3N2 and B strains. Anti-CD28 antibody was added to each well as a co-stimulant and brefeldin A (BFA) was added two hours after stimulation at 1 µg/ml

72 concentration as golgi stop for blocking cytokine secretion. The cells were stimulated overnight and stained with live/dead reagent (Near IR, EX 633/EM 750) to discriminate live cells from dead. The cells were fixed and permeabilized using Cytofix/Cytoperm reagent. To avoid non-specific binding, Fc block (BD Biosciences, San Jose CA) was added. Single-step staining was performed with CD3 conjugated with BV711, IL-17F conjugated with AF647 obtained from BioLegend (San Diego, CA), CD4 conjugated with

BUV395, CD8 conjugated with BB700, CD44 conjugated with PEFC594, Interleukin 2

(IL-2) conjugated with APCR700, Interferon γ (IFN-γ) conjugated with BV786, tissue necrotic factor α (TNF-α) conjugated with BV650, IL-17A conjugated with BV421 obtained from BD Biosciences (San Jose, CA) and IL-13 and IL-4 conjugated with AF488 obtained from Thermo fisher Scientific (Waltham, MA). Since most of the anti-mouse antibodies used are rat or hamster derived; anti-rat anti-hamster Ig, κ/Negative control compensation particles from BD Biosciences (San Jose CA) stained with all the above fluorochrome conjugated antibodies including an unstained control for preparing compensation controls. The samples were acquired on a BD FortessaX20 SORP flow cytometer from BD Biosciences (San Jose, CA) followed by analysis with FlowJo software

(Ashland, OR). We defined our gating strategy (Figure 4.2) in FlowJo where, the live cells were first differentiated from dead and were then used to differentiate singlets. From the singlets, we identified CD3+ T cells and used them to gate for CD4 and CD8 T cells.

Antigen specific cells were identified by gating on upregulated CD44 cells. Individual cytokine gates were then established on these antigen specific CD4 and CD8 T cells.

Figure 4.2. Gating strategy to obtain antigen-specific cytokine positive T cells.

4.2.8 Data Analysis and Statistics

GraphPad Prism software (San Diego, CA) was used to analyze and plot data from the in vivo immune responses. For humoral responses, one-way analysis of variance

(ANOVA) followed by Tukey’s Multiple comparisons test was used to evaluate differences in immune responses from individual animals in the dosing groups. Non-inferiority compared to AS03 for HAI titers was tested by running Dunnett’s test post one-way

ANOVA. For ICS, a nonparametric Kruskal-Wallis test was run followed by Dunn’s multiple comparisons test for comparison within different dosing groups.

4.3 Results

4.3.1 DoE Analysis for SE-AS Formulations

The goal of the initial DoE (DoE-1) study was to obtain a stable SE-AS similar in percent composition to AS03 (Figure 4.1). SE-AS from 23 different ratios of PS80 with the two oils – squalene and α-tocopherol were formulated and analyzed for size and PdI by

DLs. Different PS80 to oils ratios showed differential size and homogeneity in droplet concentration (Table 2). SE-AS 5, with a percent composition like AS03, showed higher droplet size and polydispersity index manifesting instability. In general, we observed that lower α-tocopherol content was favored in terms of fitting the desired criteria of size

˂200nm and PdI ≤ 0.3. Among the emulsions that fit the criteria with 10% α-tocopherol, i.e. SE-AS 6, 14, and 21; the increase in size was inversely proportional to surfactant content. SE-AS 6, 12 and 22 showed size like AS03, and had about 1/4th (SE-AS 6) or 1/3rd

(SE-AS 12 & 22) the amount of α-tocopherol. We evaluated repeatability of these SE-AS by conducting independent stability tests for 1 week at RT and 4°C and measured size and

PdI for these emulsions (Figure 4.3a). All emulsions were stable for 1 week and showed consistent size and PdI across all temperature and in two independent experiments.

Additionally, modifying the process parameters did not affect the size and PdI of these emulsions (Figure 4.3b).

Figure 4.3: Robustness and stability of SE-AS and effect of process change on size and PdI Figure a) shows the size over 1 week at 4C and room temperature for 6 emulsions, and two independent runs. Figure b) shows two independent changes in self-emulsification process;

75 first (two graphs) with the heating of the emulsion at different times, and second (bottom graphs) with changing the mixing speed for oils and surfactant. For each process change size and PdI were measured.

We analyzed data from DoE-1 (Table 4.2) to better understand the trend and the interaction of the components in obtaining a stable emulsion via self-emulsification. We used “fit model” analysis and contour mapping in a ternary plot to predict the mixture ratios and percent amounts of the three components that would allow for a stable SE-AS fitting the

DoE criteria (Figure 4.4). The contours were plotted for both size and PdI based on data obtained in Table 1. Each line of the contour represents a specific size and PdI, respectively.

This analysis allowed for mapping an area of the ternary diagram to design SE-AS composition that would ideally fit the criteria as shown in DOE-2 (Figure 4.5). Within this experimental space, twelve unique ratios of components were generated and emulsified in a similar fashion seen in previous DoE to generated twelve new SE-AS. Results from DoE-

2 are shown in Table 4.3. SE-AS with α-tocopherol content ≤ 16% showed size less than

200nm, and those with higher percentage of α-tocopherol did not fit the criteria. Even after evaluating different buffers, the size and PdI for SE-AS containing higher than 15% α- tocopherol could not be improved (Figure 4.6). Thus, despite using prediction tools, the amount of α-tocopherol and its interaction with the right amount of surfactant and squalene is crucial to obtaining stable “AS03-like” emulsions with self-emulsification.

Table 4.2. Hydrodynamic droplet size and PdI from DoE-1 measured using DLS

SE-AS PS80 Squalene α-Tocopherol Size (nm) PdI

1 10 16 74 474.0 0.748 2 10 35 55 1363.4 0.902 3 10 50 40 656.6 0.750

4 10 64 26 315.2 0.642 5 16 42 42 702.8 0.784 6 16 74 10 170.6 0.208 7 18 24 58 1642.7 0.860 8 19 57 24 517.7 0.647 9 19 23 58 474.1 0.748 10 21 10 69 1563.7 0.940 11 26 37 37 279.6 0.871 12 27 57 16 129.6 0.357 13 28 46 26 960.4 0.804 14 35 55 10 87.3 0.122 15 35 21 44 1182.4 0.903 16 36 10 54 234.7 0.619 17 39 41 20 410.3 0.786 18 41 28 31 761.2 0.851 19 50 10 40 444.7 0.920 20 50 30 20 375.7 0.569 21 50 40 10 46.2 0.259 22 15 70 15 130.5 0.326 23 63.6 27.3 9.1 31.6 0.402 Twenty-three ratios (Figure 1) of PS80, squalene and α-tocopherol were mixed and emulsified to form emulsions and evaluated for size and PdI on DLS to identify SE-AS that fit the criteria of size ˂200nm and PdI ≤ 0.3.

Several different surfactants have been used for oral and parenteral formulations containing α-tocopherol, including its isotropic forms and derivatives. We compiled a list of such surfactants (Table 4.1) and tested them with or without polysorbate 80 in ratios from both the previous DoEs. We used a required hydrophilic-lipophilic balance (HLB) calculation to determine amounts of surfactant used. For surfactant study, we used SE-AS

5 for higher α-tocopherol content, SE-AS 6 for its repeatable and stable data as well as SE-

AS 25. Based on the data obtained (Table 4.4), our previous findings were validated. The mixture of α-tocopherol and squalene required a specific ratio of oils and surfactants to get a stable SE-AS. However, based on these surfactant studies, we discovered that water soluble surfactants such as TPGS may provide stability as secondary surfactants.

Figure 4.4. Mixture profile analysis using contour mapping and ternary plots. Mixture profile and contour mapping were used in JMP software to analyze the trends in size and PdI, to better predict the mixture ratio that can be the closest to percent composition of AS03 as well as fit the criteria defined in DoE-1; (A) size analysis. The contour lines each represent a size 200nm, 400nm or 800nm, and the area to the left of the line entails SE-AS with a z-average below the respective size represented by the line. (B) PdI analysis. The contour is set for 0.3 and the area to the left of the contour represents SE- AS with PdI of 0.3 or less.

Table 4.3. Hydrodynamic droplet size and PdI from DoE-2 measured using DLS

SE-AS PS80 Squalene α-Tocopherol Size (nm) PdI

24 7 82 11 189.2 0.475

25 12 62 26 644.7 0.725 26 20 60 20 563.2 0.878 27 8 68 24 548.3 0.540 28 15 79 6 180.6 0.155 29 5 75 20 932.9 0.728 30 6 89 5 204.0 0.322 31 17 67 16 154.3 0.531 32 6 64 30 2132.0 0.780 33 19 72 9 159.1 0.178 34 11 71 18 406.3 0.695 35 13 74 13 144.27 0.275 Twelve new ratios (Figure 3) of PS80, squalene and α-tocopherol were mixed and emulsified to form emulsions and evaluated for size and PdI on DLS to identify SE-AS that fit the criteria of size ˂200nm and PdI ≤ 0.3 and allow for increase in α-tocopherol content to match AS03.

Table 4.4: Size and PdI from surfactant study oil and water-soluble surfactants.

Surfactant Component(s) Squalene Tocopherol SE-AS Surfactant A % B % Size (nm) PdI (% vol) (% vol) Surfactant B A vol vol*

5a 42 42 PS80 9.7 Span 80 6.3 250.87 0.508

5b 42 42 PS20 7.3 Span 85 4.7 1116.13 0.947

5c 42 42 PS20 9.7 Span 20 6.3 441.57 0.592

5d 42 42 PS80 16 TPGS 1.6 167.53 0.382

Poloxamer 5e 42 42 PS80 16 1.6 201.53 0.472 188

Poloxamer 5f 42 42 PS80 16 1.6 229.70 0.574 407

25a 62 26 PS80 9.7 Span 80 6.3 184.73 0.546

25b 62 26 PS20 7.3 Span 85 4.7 426.40 0.618

25c 62 26 PS20 9.7 Span 20 6.3 329.50 0.694

25d 62 26 PS80 12 TPGS 1.2 162.60 0.420

Poloxamer 25e 62 26 PS80 12 1.2 157.80 0.416 188

Poloxamer 25f 62 26 PS80 12 1.2 174.97 0.458 407

6a 74 10 PS80 9.7 Span 80 6.3 100.06 0.379

6b 74 10 PS20 7.3 Span 85 4.7 151.23 0.251

6c 74 10 PS20 9.7 Span 20 6.3 160.30 0.222

6d 74 10 PS80 16 TPGS 1.6 160.87 0.181

Poloxamer 6e 74 10 PS80 16 1.6 160.93 0.178 188

Poloxamer 6f 74 10 PS80 16 1.6 160.47 0.186 407 SE-AS 5, 25 and 6 were used with six different surfactant combinations and are labeled from a-f for each SE-AS. For formulations d-f in each of the three SE-AS, the composition of secondary surfactant was excluded from the 100% oil-surfactant ratio. In this case, the secondary surfactant was added in the aqueous phase.

Figure 4.5. Experimental space to operate for mixture ratios for the three components based on contour analysis Twelve new ratios of PS80, squalene and α-tocopherol were formulated in a similar fashion as in the previous DoE (Figure 1) with the same criteria of attaining size ˂200nm and PdI ≤ 0.3. Each dot represents an SE-AS with a new ratio of the three components.

Figure 4.6: Effect of buffers on size and PdI of SE-AS The left graph shows effect on size after emulsification in PBS, citrate and phosphate buffers, whereas the right graph shows the effect on PdI.

4.3.2 Effect of Surfactant Reduction on SE-AS Formulation

Although the quality of surfactants manufactured and used in emulsion adjuvants have improved significantly after adverse reactions seen in some early adjuvants (128), lower surfactant content is generally preferred in emulsion adjuvants. Since all the formulation optimization attempts did not allow for increasing the α-tocopherol content in

SE-AS to match that in AS03, we attempted at reducing the PS80 content using SE-AS 6 as a model emulsion. We reduced the PS80 content in SE-AS 6 while keeping the proportions of squalene and α-tocopherol consistent to generate five new SE-AS with

TPGS as secondary surfactant. We observed that even when PS80 content was reduced to

7% (SE-AS 36), the size and PdI remained like SE-AS 6 (Table 4.5).

Table 4.5. Reduction in Polysorbate 80 content in SE-AS 6.

SE-AS PS80 Squalene α-Tocopherol Size (nm) PdI

36 7 82 11 137.9 0.243 37 9 80 11 143.5 0.210 38 11 79 11 154.3 0.210 39 13 77 10 155.6 0.212 40 14 76 10 155.7 0.206 6 16 74 10 155.0 0.211 Five new formulations where % v/v squalene and α-tocopherol were kept relatively similar as SE-AS 6 while reducing the polysorbate 80 content, in presence of 2% w/v TPGS as a secondary surfactant.

Emulsions like SE-AS 14 and SE-AS 21 showed lower size and PdI but were not considered further. It was previously demonstrated that a size ~ 160nm was favored in vivo compared to lower sized droplets (62). SE-AS 36 and SE-AS 22 from previous optimization experiments were subjected to a short-term stability for about 18 days to observe changes in basic properties like size & PdI (Figure 4.7), and pH & osmolality

(data not shown). These studies were carried out to ensure stability during shipping, and off-site in vivo studies. Both SE-AS emulsions were stable for up to 18 days at all temperatures.

Figure 4.7: SE-AS stability without the antigen at elevated temperatures Size (top two graphs) and PdI (bottom two graphs) for SE-AS 22 and 36 for the emulsions stored at 4C, 25C or 37C at time 0, 1 day, 2 days, 7 days or 18

4.3.3 Stability of combination OVA Antigen and SE-AS

OVA antigen was used as a model antigen to demonstrate short term stability of the antigen with the adjuvant stored at RT for up to 48 hours. Figure 4.8 show gels for SE-AS

36 and SE-AS 22 respectively. The general smearing observed in the adjuvanted OVA vaccine is due to the possibility of remnants from disruption of oil droplets via vacuum- driven centrifugation and can be seen in the emulsion only lane 4 in both gels. 0.4mg/mL

(20µg dose) fresh OVA standard was comparable to bands in both emulsions at all time- points thus proving that protein remained stable in the emulsion adjuvanted vaccine for

48h at RT.

Figure 4.8. Gel Electrophoresis to assess ovalbumin (OVA) antigen stability with SE-AS.

OVA was mixed with both SE-AS 22 and SE-AS 36 and stored for 48 hours at RT. OVA protein bands are shown at time-points 0h, 2h, 6h, 24h and 48h. SE-AS alone was added as a negative control and Ovalbumin at 20 µg dose per mice was added as fresh protein standard. Values on the left indicate the molecular weight (kDa) of the corresponding bands of the protein ladder on the gel. The left gel shows OVA with SE-AS 36 and the right gel shows OVA with SE-AS 22. Individual lanes in both gels are as follows: lane 1 is ladder, lane 2 is fresh protein standard, lane 3 is emulsion only, lanes 4-8 are short term stability samples at time 0h, 2h, 6h, 24h and 48h, respectively.

4.3.4 In Vivo Potency of SE-AS with QIV Antigens

The immunoadjuvant potency of SE-AS was evaluated in vivo in BALB/c mice using QIV and measuring QIV-specific humoral and cellular immune responses at different time-points after immunization. 8 female mice per group were immunized twice, three weeks apart with the different formulation groups described in methods section.

HAI assays have been used to measure functional serum antibodies against the different influenza vaccine strains and is considered an important read-out with influenza

85 vaccines (129). We measured HAI titers at 3wp1 and 2wp2 (Figure 4.9). An increase in

HAI titers was observed between two immunizations demonstrating a boosting response following the second immunization. Overall, across all 4 strains, unadjuvanted QIV showed significantly lower antibody titers after both immunizations than both novel SE-

AS adjuvants at both antigen doses. Additionally, both novel SE-AS induced titers that were not significantly different than AS03. The difference between the two SE-AS was also not significant at both time-points and doses. On comparing novel SE-AS and AS03 using Dunnett’s test, we observed that the overall HAI titers at low dose, especially after

3wp1 were non-inferior compared to AS03.

Figure 4.9. Hemagglutination Inhibition titers post 1st and 2nd immunization in BALB/c mice HAI antibody titers in serum obtained post 1st and 2nd immunization, against all strains of QIV and at two QIV doses. (a-d) log2 HAI titers against A/Singapore, A/Hong Kong, B/Brisbane and B/Phuket, respectively, at 3wp1 (white bars) and 2wp2 (pattern bars). Each bar represents geometric mean titers (GMT) with 95% confidence interval (CI) from n=8 animals per group. Statistics was performed using one-way ANOVA followed by Tukey’s multiple comparisons test to compare difference between each group. Significant differences are marked on the graph. The values shaded represent comparison with AS03 whereas the values not shaded are comparison with QIV only for respective dose groups. where, ns = not significant, *= p˂0.05, **= p˂0.005, ***= p˂0.0005 and ****= p˂0.00005.

To characterize the 2wp2 humoral response further, IgG titers against each strain in QIV were evaluated in terms of subclasses IgG1, IgG2a and IgG2b (Figure 4.10).

Unadjuvanted QIV, at both doses showed overall lower total IgG response compared to adjuvant groups demonstrating adjuvant effect by novel adjuvants. IgG1 was the major subclass in the total IgG response. Overall, across all the strains, the novel SE-AS adjuvants induced significantly higher total IgG and IgG1 titers compared to unadjuvanted QIV at both doses, while the titers were not significantly different when compared to AS03. Titers for both SE-AS emulsions were comparable across all strains and both QIV doses. Overall, this data is consistent in trend with the HAI titers, and showed that the novel adjuvants with lower amounts of α-tocopherol formulated using self-emulsification showed higher humoral immune responses compared to unadjuvanted vaccine and elicited similar antibody titers compared to AS03 adjuvanted QIV.

Frequencies of QIV antigen-specific CD4+ and CD8+ polyfunctional T cells were measured using ICS. We have reported CD4+CD44+ T-cell response at 2wp2 (Figure

4.11) as frequency of antigen specific polyfunctional cells and classified into T-helper (Th) cell subtypes - Th0, Th1, Th2 and Th17, after Boolean gating based on the individual cytokine responses. The magnitude of response to A/H1N1 was greater compared to the other three vaccine strains. Overall, trend between different groups remained similar to humoral immune response i.e. QIV only groups showed lower antigen -specific CD4 T cells compared to adjuvanted groups at both doses. Among the CD4 T cell subtypes, Th0 and Th2 responses were dominant across all groups. Th0 and Th2 responses were in general higher for adjuvanted groups compared to no adjuvant (QIV only). Among the adjuvanted groups, both the novel SE-AS showed comparable responses to AS03.

4.4 Discussion

α-tocopherol has been used as an immune supplement (65, 81) and has been used as an adjuvant in veterinary vaccines for a few decades now (87, 88). α-tocopherol is also a component in AS03 adjuvant (part of GSK’s previously licensed H1N1 pandemic influenza vaccine - Pandemrix®) and has shown improved potency compared to squalene only emulsions (90, 92). Despite its use in veterinary vaccines and AS03, the role of α- tocopherol as an immunostimulant is not completely characterized and evaluated at a molecular and cellular level. A recent work by Givord, et al reports role of lipids in AS03 on improving the innate immune response via building stress on the endoplasmic reticulum

(92), altering the lipid metabolism in monocytic cells. The recruitment of immune cells after AS03 administration might be enhanced due to stressing the macrophages and monocytes; thus, eventually affecting the magnitude of antibody avidity and T follicular helper (Tfh) cell response. While it was known that emulsion adjuvants alter the lipid metabolism of the innate cells, this work showed a downstream molecular signaling pathway to demonstrate the change in lipid metabolism. More work is needed to understand the exact molecular mechanism of action of α-tocopherol in emulsion adjuvants.

Development of novel adjuvants containing α-tocopherol and their evaluation in vivo may help understand the benefits of α-tocopherol in an emulsion.

Figure 4.10. Anti-HA IgG subclass antibody titers in mouse serum post 2nd immunization The sera from the same in vivo potency evaluation study as Figure 9 was used to determine the IgG antibodies against each strain in the QIV. Figures a-d represent log10 antibody titers against A/Singapore, A/HongKong, B/Brisbane and B/Phuket respectively at time-point 2wp2. Each bar represents GMT (within each subclass) with 95% CI values from n=8 animals per group. Statistics was performed on total IgG as well as IgG sub-type responses from individual animals using one-way ANOVA followed by Tukey’s multiple comparisons test to compare difference between each group. Significant differences are marked on the graph for total IgG response. The values shaded represent comparison with AS03 whereas the values not shaded are comparison with QIV only for respective dose groups. where, ns = not significant, *= p˂0.05, **= p˂0.005, ***= p˂0.0005 and ****= p˂0.00005.

Figure 4.11. Intracellular cytokine staining (T-cell response) Spleens from individual animals were collected and processed and restimulated with antigen, to get cellular immune responses. Polyfunctional CD4+ immune responses were characterized further into antigen specific subtypes of Th immune cells. Based on the cytokines released by the splenocytes, the T- cell response was Boolean gated into Th1, Th2, Th17 and Th0 responses. Figures a-d represent frequency of antigen specific CD4+ T cells against A/Singapore, A/HongKong, B/Brisbane and B/Phuket respectively at time- point 2wp2. Each bar represents the mean (within each subtype) with standard deviation values from n=5 animals per group. Statistics shown was performed on overall antigen specific CD4+ T cells using Kruskal-Wallis test followed by Dunn’s test for multiple comparisons. *= p˂0.05 The main goal of this study was to develop a novel self-emulsified adjuvant containing α-tocopherol and test its in vivo potency compared to AS03. AS03 is manufactured using high pressure homogenization and microfluidization. We attempted

91 formulating “AS03-like” emulsion adjuvant without using high pressure or shear, but rather using a simple heating and mixing process. α-tocopherol and squalene have different physicochemical properties and required HLB values, thus making it a challenge to self- emulsify the mixture of these two oils. We started off with designing a DoE and screening several mixture ratios with an objective to get an SE-AS similar in composition and size to

AS03 The criteria for selection was set to essentially obtain a nanoemulsion that can be eventually sterile filtered during development and also to be as close to the size of AS03

(~155nm) as possible (100). It was evident that the right amount of surfactant and lower amounts of α-tocopherol were necessary to obtain stable emulsions under 200nm in size.

To be able to compare the potency of these SE-AS, we attempted to increase the α- tocopherol content in them to match AS03. Our fit model prediction analysis also revealed ratios that had lower α-tocopherol content to obtain emulsions with lower size and PdI. We also attempted to use other surfactants that were used with several combinations based on the literature, however, we couldn’t increase the α-tocopherol content to match AS03.

Thus, our exhaustive formulation studies resulted in SE-AS 22 with approximately 1/3rd the amount of α-tocopherol in AS03.

Lower surfactant content is generally a desired attribute in emulsions and so we reduced the polysorbate 80 content to a considerable extent to achieve a stable emulsion.

This resulted in SE-AS 36. Although there has been evidence of using fractional doses of

AS03 (for example, AS03B with ~ 5.93mg α-tocopherol) in clinical trials with A/H5N1 or

A/H1N1 influenza antigens in children (53, 130), there has been no direct comparison on using lower amounts of α-tocopherol while keeping squalene content consistent. Thus, we decided to use the novel SE-AS 22 and 36 to compare with AS03. Emulsion adjuvants are

92 predominantly known to be advantageous in influenza vaccines primarily for their attribute of dose sparing as well as number of immunization (24, 56, 91). They are also found to be compatible with the excipients in these detergent-split influenza vaccines. Our choice of antigen was quadrivalent Influenza vaccine (QIV) with containing detergent-split antigens for 4 strains of influenza virus and evaluated the humoral and cellular immune responses against each strain. We picked BALB/c mice based on extensive literature evidence on evaluation of emulsion adjuvants in preclinical models (49, 90). Our main aim was to test and compare humoral and cellular - immune responses post immunization with novel adjuvants and AS03. HAI titers as well as QIV-specific IgG antibody titers showed that novel SE-AS elicited statistically similar immune responses to AS03 and higher to unadjuvanted QIV vaccine. The non-inferiority post 1st immunization in HAI titers between the novel SE-AS and AS03 further supports our conclusions. We saw higher IgG1 responses, which is a characteristic of emulsion adjuvants and this also paralleled with cellular responses where we saw higher polyfunctional and antigen specific Th0/Th2 cells compared to other subtypes of CD4 T cells. Between the two SE-AS no striking differences were observed across all assays and immune responses. Therefore, the novel SE-AS adjuvants contribute in enhancing the immune response of an antigen confirming their adjuvant action.

Self-emulsification of two oils, squalene and α-tocopherol has not been performed before, and data from this study as well as previous studies with MF59-like self- emulsifying adjuvants (62) suggests that emulsion adjuvants can be formulated using low energy, low pressure, simple and relatively inexpensive process that if scaled up, can be manufactured at well-established development sites, even in most developing countries.

This would significantly impact the cost of production, clinical development, as well as make sufficient doses available during influenza virus pandemic situations such as A/H1N1 in 2009 (1, 126).

In this study, we have formulated an optimized “AS03-like” emulsions using self- emulsification to provide additional proof that we can use simpler manufacturing methods as an alternative to complex, high energy equipment, requiring expensive and specific manufacturing infrastructure. We faced a formulation challenge in incorporating α- tocopherol in squalene oil in water emulsions and used design of experiment methods and meticulous formulation optimization to incorporate ~ 2.5 and 3.5 mgs of α-tocopherol in

SE-AS 36 and 22 respectively compared to 11.8 mgs in a single human dose of AS03 adjuvanted vaccine. The overall similarity in immune responses in all four strains of QIV between two novel SE-AS and AS03 suggests that lower amounts of α-tocopherol may be enough to provide an immunostimulatory effect, however, more studies would be required to fully explore this promising finding.

4.5 Conclusion

Formulation research is vital in the developing novel adjuvants and delivery systems for different types of vaccine antigens. We have shown that using simple self-emulsification strategy, emulsion adjuvants showing comparable immune response to AS03 in BALB/c mice, can be prepared. Emulsion adjuvants have been evaluated for co-delivery of antigen and immune potentiators earlier and have faced several formulation challenges (39, 131).

Thus, simple and novel formulation strategies for emulsion adjuvants may not only pave a path for less expensive manufacturing globally, but also provide for a useful platform for building combination adjuvants. Although more studies may be needed to evaluate the role

94 of α-tocopherol in SE-AS, our work provides a proof of concept that simple, inexpensive formulation strategies can be used to create novel emulsion adjuvants containing α- tocopherol.

CHAPTER 5 Self-Emulsified Adjuvant System (SE-AS): Formulation Optimization and In Vivo Immunogenic Effects with Soluble Subunit CMV Pentamer Antigen

5.1 Introduction

As elaborately described in the previous chapters, emulsion adjuvants offer strong adjuvant properties in boosting immune responses for protein antigens, especially influenza. Nanoemulsions are prone to instability due to constant Brownian motion of the oil droplets and hence it is important to maintain size and homogeneity, which are the two important parameters for product development as well as potency of the emulsion adjuvant.

For most pharmaceutical drug products, eventually meant for parenteral use, sterilization is mandatory. For most nanoemulsions typical methods of sterilization such as radiation, autoclaving are inconvenient (132). Thus, terminal sterilization by filtration is not only convenient, but also cost effective for these formulations. However, for sterile filtration, developing a stable product with homogenous size distribution is important. Thus, we attempt at optimizing the SE-AS to improve PdI, and evaluate laboratory scale feasibility for sterile filtration of these emulsions.

In the previous chapter, we achieved our goal of creating a novel emulsion with composition similar to AS03 using a simpler, low-energy process. These self-emulsifying adjuvant systems (SE-AS) showed an adjuvant effect when administered with inactivated

Quadrivalent Influenza vaccine (QIV) in BALB/c mice and additionally, showed immune responses like AS03 in the same study. SE-AS 22 and SE-AS 36, used in this study had approximately 1/3rd and 1/4th of α-tocopherol compared to AS03; therefore, leading to a plausible conclusion that higher amounts of α-tocopherol may not be needed for higher

96 immune responses. Considering the prospects of developing a product, our next objective was to improve stability of the SE-AS. Thus, we optimized SE-AS formulation to improve

PdI, and evaluate laboratory scale feasibility for sterile filtration of these emulsions.

We previously observed that the components of SE-AS affect the size and homogeneity depending on the concentration used. SE-AS 22 and SE-AS 36 showed similar immune responses in the previous chapter compared to each other; however, they differed in composition such that SE-AS 36 had TPGS as an additional surfactant. SE-AS

22 was used for further optimization due it’s to the same components of AS03. Inactivated

QIV comprises of four split influenza virus antigens. Although, Flu is the choice of antigen for emulsion adjuvants due to the vast history of clinical use and compatibility (56), the minute difference between different adjuvants may not be evident in such detergent-split antigens (133). Adjuvants offer a greater advantage for poorly immunogenic antigens that are mostly synthetic in nature. In this chapter we decided to use Cytomegalovirus (CMV) pentamer antigen, which is a soluble protein, that comprises of a pentameric complex with

5 subunits gH/gL/Ul128/UL130/UL131 (134). gH/gL are part of the virion that form a pentameric complex with the ULs to facilitate virus entry into mostly the epithelial or endothelial cells (135). CMV in humans is a type of β-herpesvirus and is mainly responsible for cross placental infection leading to several neurological defects.

Additionally, according to CDC, over half of adults by age 40 have been infected with

CMV, and the virus can reactivate due to any trigger (136). Thus, a potent vaccine against

CMV is an unmet medical need. We use a soluble pentamer antigen produced using

Chinese hamster ovary (CHO) cells, and thus is poorly immunogenic and would be an ideal target to see an adjuvant effect and differences between different adjuvant groups.

“MF59-like” self-emulsifying adjuvant SEA160(under journal review for publication in scientific reports) is a well-characterized squalene oil containing emulsion adjuvant. SEA160 has shown similar immune responses to MF59 using several antigens such as ovalbumin, A/H1N1 flu egg-derived antigen, and HIV gp120 antigen. Thus,

SEA160 made an ideal candidate to compare with an optimized SE-AS to establish the role of α-tocopherol in these self-emulsifying emulsion adjuvants

5.2 Materials and Methods

5.2.1 Formulation Materials

Squalene oil was obtained from JX Nippon Oil Trading Company (Tokyo, Japan) and dl-α-tocopherol was obtained from DSM Nutritional Products (Heerlen, Netherlands).

Polysorbate 80 (PS80) was obtained from JT Baker (Center Valley, PA). Hypure water for injection (WFI) like quality water and phosphate buffered saline (PBS) were obtained from

HyClone Laboratories (Logan, UT). Tocopheryl poly-ethylene glycol sulphate (TPGS) was obtained from Antares Healthcare Products, Inc. (St. Charles, IL). CMV wild-type pentamer protein was provided by GSK Vaccines (Rockville, MD).

5.2.2 Sterile filtration of emulsions

Formulation of SE-AS was carried out as previously described in chapter 4 section

4.2.2. Emulsions were filtered using a 33mm, 022µm pore size polyethersulfone (PES) membrane syringe filter from Millipore (Burlington, MA). The resultant filtered emulsions were characterized for size, PdI and % content of squalene and α-tocopherol using reverse phase ultra-high-performance liquid chromatography (RP-UPLC).

5.2.3 Percent content of squalene and α-tocopherol in emulsions

For percent content measurements an Xterra C18 column from Waters® (Milford,

MA) was used. Mobile phase was 95:5 Methanol: Acetonitrile. The run time was 15 mins at 1mL/min flow rate. The column was heated at 37C during elution and PDA detector was used to record the eluting peaks. The retention time for tocopherol was ~ 4.3 mins and squalene was ~7.4 mins. A standard curve of squalene and tocopherol mixture was run before each run with concentrations ranging from 600 µg/mL to 2.34 µg/mL. Using the slope and intercept from this standard curve, concentration of squalene and tocopherol in emulsion samples was determined. Emulsions were disrupted by dissolving in IPA such that the squalene concentration in each emulsion was 200µg/mL. 10µL in duplicate was injected in the UPLC.

5.2.4 Optimization of SE-AS

Keeping the α-tocopherol content constant at 15% v/v in oil: surfactant mixture, the surfactant content in SE-AS emulsions was increased to obtain an SE-AS with better PdI.

The goal of this experiment was to obtain an SE-AS which maintains size and PdI post filtration. The emulsions were formulated similarly as previously described. pH measurement was performed using an Orion 3-star pH meter from Thermo Scientific

(Waltham, MA) and osmolality was measured on advanced instruments osmometer model

2020 (Norwood, MA). Size and PdI were measured using DLS and % content of the oils was measured as described in section 5.2.2

5.2.5 Cryo-electron microscopy (Cryo-EM) imaging of SE-AS 44

To understand the morphology of the oil droplets in SE-AS 44 compared to AS03, we collaborated with Creative Biostructure (Shirley, NY) to obtain cryo-EM images and size analysis for SE-AS 44 and AS03. In summary, FEI Talos F200C Cryo-Transmission

Electron Microscope was used for imaging and analysis purpose. Undiluted emulsion was placed on a thin copper grid that had been glow discharged and the sample was then loaded to the freezing chamber. The sample is rapidly frozen by plunging into a cryogen (liquid ethane cooled by liquid nitrogen). This frozen sample stored in liquid nitrogen is then used for imaging and size distribution analysis.

5.2.6 Stability study to enable delivery of stable emulsions for in vivo evaluations

An extended stability study for SE-AS 44 was carried out at temperatures 5°C, 25°C and 50°C. SE-AS 22 and SE-AS 36 were used for comparison. Emulsions were formulated and divided into separate vials and stored at respective temperatures and at each time-point a vial was removed for characterization. pH, Osmolality, size, PdI and % oil content was measured. pH, osmolality, size and PdI were measured up to 10 weeks as previously described and % content was measured using UPLC-PDA for up to 2 weeks.

5.2.7 In vivo evaluations using CMV wild-type pentamer antigen

5.2.7.1 Ethics statement: All studies were conducted in accordance with the GSK

100

Care and Use of Laboratory Animals and the U.S. Interagency Research Animal

Committee Principles for the Utilization and Care of Research Animals.

5.2.7.2 In vivo study design and immunization regimen: Cell-line derived CMV pentamer was used to test the in vivo potency of the novel SE-AS in mice. The dose of antigen in each mouse was 0.05µg. The antigen content was determined by UV/visible spectrophotometry as well as reverse phase HPLC. Sample size of 10 animals per group was calculated such that it provided a power of 80% to detect a 3-fold difference between any 2 groups with 95% confidence interval. 6-8 weeks old female C57BL/6 mice from

Charles River laboratories (Gaithersburg, MD) were used. The study design comprised of three immunizations three weeks apart, where 50µL of the vaccine was injected intramuscularly in the gastrocnemius muscle such that at each time-point alternate limb was used. The groups in this study were: Physiological saline (as negative control), unadjuvanted CMV, CMV with SE-AS 44, CMV with AS03 and CMV with SEA160

Figure 5.1 shows the summary of the study design and immunization schedule.

101

Figure 5.1: Study design, dosing groups and immunization schedule to compare in vivo SE-AS and SEA160.

The vaccine was formulated to mimic bed-size immunizations, with 1:1 mixing of antigen and adjuvant, to reach the final antigen concentration. All the formulations were characterized for pH, Osmolality, size and PdI and endotoxin before immunization.

Endotoxin was measured using the Endosafe NexGen – PTS from Charles river laboratories (Wilmington, MA) and limulus amoebocyte lysate (LAL) cartridges with test range 10-0.1 EU/mL. Additionally, protein integrity was confirmed via gel electrophoresis

(SDS-PAGE). It was ensured that the vaccine dosed for each group had endotoxin lower than 1 EU per dose. Bleeds were collected 3 weeks post 1st (3wp1), 3 weeks post 2nd (3wp2) and 3 weeks post 3rd (3wp3) immunization and the processed sera was used to test the humoral immune responses by neutralizing antibody assay (nAb) and IgG titers by ELISA.

102

Spleens from 5 animals per group were harvested 3wp3 and 4wp3, and used to measure T cell immune responses via intracellular cytokine staining.

5.2.7.3 Determination of neutralizing antibody (nAb) titers: Retinal pigment epithelial cell line (ARPE-19) was used since TB40 (a CMV virus strain) was known to infect these cells in this assay. On day 1, 100uL of ARPE-19 cells were plated in 96 well flat bottom plates in complete growth medium i.e. DMEM + 10% fetal bovine serum (FBS)

+ 1% Penicillin-Streptomycin. Plated were incubated in 37C overnight ~ 24 hours. On day

2, Tecan – liquid handling robot was used to perform serum dilutions. Different starting dilutions were used for different time-points depending on the expected titers. A positive control from Sera care known to neutralize TB40 virus was used in every plate at a constant

1:50 dilution. In each plate, 75uL of serum dilutions were prepared using Tecan and then

75uL of TB40 virus was added to each well to make a total of 150 in each plate. This was enough for duplicates of each sample. This virus-serum mixture was incubated at 37C, 5%

CO2 for 2 hours. The cells plates (duplicate for each group) were removed from the incubator. Media was taken out from each well and 50uL of virus-serum cocktail was added. These plates were incubated at 37C, 5% CO2 for at least 20 hours. On day 3, the cells were fixed using 4% paraformaldehyde and incubated at RT for 20 mins following by

1 wash using 1XPBS and then permeabilized using 0.1% TritonX-100 and incubated for another 10 mins. Primary antibody (anti-mouse anti-CMV IE monoclonal antibody) was added immediately and incubated for 1 hour in 37C, 5% CO2 incubator. Cells were washed twice and then secondary antibody (anti-mouse AlexaFlour488 antibody) was added and incubated for another 1 hour. Post incubation cells were washed 3 times and 1X PBS was

103 added. These plates were then read using high content imaging – CX7 (by selecting to read

10-20 fields per well). Interpolated titers are then calculated at 50% fluorescence intensity.

5.2.7.4 IgG antibody ELISA: Antibody titers were determined in serum obtained from each animal at 3wp2 and 3wp3. To determine the CMV pentamer specific binding

IgG antibody titers, sandwich ELISA was used. 96 well Nunc-immuno Maxisorp F96 plates were used to coat 100µL of 1µg/mL CMV pentamer antigen per well overnight at

4°C. Antigen coated plates were washed with 1X phosphate buffered saline (PBS) &

0.05%w/v Tween20 and blocked with 1% w/v bovine serum albumin (BSA) solution in

PBS. Serum from immunized animals was added in the first row of the plate such that well

A1 received positive control (serum from previous CMV study that showed consistently higher titers) and well A12 received sample buffer as negative control. The serum was prediluted before adding 10µL to row1. Serial dilution was the performed down the plate from row A to H. Serum incubation was allowed for one hour before washing the plates and adding horse radish peroxidase (HRP) conjugated goat anti-mouse IgG from Jackson

Immunoresearch (West Grove, PA) for another one-hour incubation at room temperature.

Substrate was added quickly after washing plates again, for 15 mins and then immediately stop solution was added. Plates were read using EnVision 2105 Multimode plate reader from Perkin Elmer (Waltham, MA). Titers were calculated at 50% interpolated optical density (OD) value obtained from the plate reader.

5.2.7.5 Intracellular Cytokine Staining Assay: T-cell responses were analyzed

104 peptides gH, gL, UL128, UL130 and UL131 from GeneScript (were used for stimulation of splenocytes. These splenocytes were stimulated at one million cells per well density with anti-CD3 from BD Biosciences (San Jose CA) used as positive control, media was used as negative control, and peptide pool was prepared for antigen stimulation condition.

Anti-CD28 antibody from BD Biosciences was added to each well as a co-stimulant and brefeldin A (BFA) from BD Biosciences was added two hours after stimulation at 1 µg/ml concentration for blocking cytokine secretion. The cells were stimulated overnight and stained with live/dead reagent (Near IR, EX 633/EM 750). Before the cells were fixed and permeabilized using Cytofix/Cytoperm reagent, Fc block was added to avoid extracellular non-specific binding, followed by memory marker staining using CD62L conjugated with

BV510 and CD127 conjugated with BV421 from BD Biosciences. Fc block was again added to avoid intracellular non-specific binding before single-step staining with CD3 conjugated with BV711, IL-17F conjugated with AF647 from BioLegend (San Diego,

CA), CD4 conjugated with BUV395, CD8 conjugated with BB700, CD44 conjugated with

PEFC594, Interleukin 2 (IL-2) conjugated with APCR700, Interferon γ (IFN-γ) conjugated with BV786, tissue necrotic factor α (TNF-α) conjugated with BV650, IL-17A conjugated with BV421 from BD Biosciences, and IL-13 and IL-4 conjugated with AF488 obtained from Thermo fisher Scientific (Waltham, MA). Since most of the anti-mouse antibodies used are rat or hamster derived; anti-rat anti-hamster Ig, κ/Negative control compensation particles from BD Biosciences stained with all the above fluorochrome conjugated antibodies including an unstained control for preparing compensation controls. The samples were acquired on a BD FortessaX20 SORP flow cytometer from BD Biosciences

(San Jose, CA) followed by analysis with FlowJo software (Ashland, OR). The gating

105 strategy is described in the supplementary section. We defined our gating strategy (Figure

5.2) in FlowJo where, the live cells were first differentiated from dead and were then used to differentiate singlets. From the singlets, we identified CD3+ T cells and used them to gate for CD4 and CD8 T cells. Antigen specific cells were identified by gating on upregulated CD44 cells. Individual cytokine gates were then established on these antigen specific CD4 and CD8 T cells. Memory markers were used to identify antigen-specific transitional, central memory, effector memory, and effector populations. Individual cytokine gates were established on these memory populations (Figure 5.3).

Figure 5.2: Example of gating strategy for ICS. First live cells are differentiated from dead, and then separated based on morphology to get single cell lymphocytes, differentiated based on CD3 marker. CD4+ and CD8+ T cells are then gated based on CD3 gates followed by identifying antigen specific CD4+ or CD8+ T cells.

106

Figure 5.3: Example of gating strategy for Memory markers. CD3- cells are used to gate four quadrants of T lymphocyte cells based on CD62L and CD127 markers. Q1 represents transitional T cells, Q2 represents Central Memory T cells, Q3 represents Effector memory and Q4 represents Effector cells.

5.2.7.6 Statistics and Data analysis: GraphPad Prism software (San Diego, CA) was used to analyze and plot data from the in vivo immune responses. For humoral responses, one-way analysis of variance (ANOVA) followed by Tukey’s Multiple comparisons test was used to evaluate differences in immune responses from individual animals in the dosing groups. Non-inferiority compared to AS03 for HAI titers was tested by running Dunnett’s test post one-way ANOVA. For ICS, a nonparametric Kruskal-Wallis test was run followed by Dunn’s multiple comparisons test for comparison within different dosing groups.

5.3 Results

Studies from previous chapter revealed that even with lower α-tocopherol content compared to AS03, SE-AS can be generating statistically non-significantly different immune responses. The SE-AS generated were not optimized for PdI, for sterile filtration,

107 as the goal previously was to obtain a novel emulsion and compare its immune response in vivo. In this chapter, an attempt to optimize the PdI of the novel SE-AS was made, to generate a stable, sterile filterable SE-AS.

5.3.1 Optimizing SE-AS for sterile filtration

Ideally sterile filtration of adjuvant emulsions is preferred to avoid other forms of terminal sterilization. Although SE-AS 22 and SE-AS 36 were comparable to AS03 in terms of immune responses, they showed an overall PdI of 0.2-0.4, which might be an issue during sterile filtration. We developed a chromatographic method to evaluate the emulsions and the content loss before and after filtration. Figure 5.4 shows the general peaks for squalene and α-tocopherol in a single run, as described in the method explained earlier.

And figure 5.5 shows the standard curves with regression analysis upon running a mixture of a range of concentrations of squalene mixed with tocopherol in IPA. These graphs delivered the formula to calculate the content of squalene and tocopherol in the emulsions.

Figure 5.4: Model chromatograms showing squalene and α-tocopherol peaks. Retention times for α-tocopherol and squalene are 4.4 and 7.5 in the standard curve at concentration 150ug/mL. x-axis shows the time in minutes and the y axis showed the absorbance intensity in light scattering units (LSU)

108

Figure 5.5: Standard curves for tocopherol and squalene for concentration calculation in the emulsions. R2 is the regression calculated for curve fit. X-axis determines the concentration of squalene and tocopherol in µg/mL and y-axis represents the absorbance in LSU.

Table 5.1 and figure 5.6 obtained from this experiment indicated that polydispersity of the emulsions increase after filtration showing a bimodal size distribution, as well as significant loss of oil content after filtration. % loss was calculated using the following formula:

[(Conc. Before filtration - Conc. After filtration) / Conc. Before filtration] * 100

Thus, it is essential that these SE-AS be optimized to improve the droplet dispersity

Table 5.1: Size of the emulsions before and after filtration and % content loss after filtration for SE-AS 22 and 36.

Before filtration After filtration % content loss after filtration SE-AS Size PdI Size PdI Squalene Tocopherol

22 135.2 0.241 230.1 0.484 31.12 32.38

36 173.3 0.283 371.3 0.619 45.12 61.02

109

Figure 5.6: Bimodal size distribution after filtration through 0.22µ PES filter. Graphs in red & green for SE-AS 36 are before filtration, and those in blue and black are post filtration. For SE-AS 22, red is before and green is after filtration.

In the previous DOEs explained in chapter 4, the formulation experiments conducted manifested that a balance of surfactant content is needed for optimal dispersity of droplets. However, surfactant content was very high ~ 35-55% v/v in the oil surfactant mixture showed particle to give a PdI of less than 0.15 but also a size of less than 100nm.

Our goal was to optimize the SE-AS while keeping the size fairly similar to AS03 or SE-

AS 22/SE-AS 36 i.e. 120nm-160nm. A series of new emulsion combinations (Table 5.2) were evaluated with varying polysorbate 80 concentrations, keeping tocopherol content between 10-18%. The data showed a trend such that keeping squalene and tocopherol somewhat constant and increasing the surfactant content shows reduction in size and more monodisperse emulsion. SE-AS 44 and 45 show size and PdI slightly better and do not have very high polydispersity. These emulsions were used for a similar filtration study as described previously with SE-AS 22 & 36.

Table 5.2: % v/v in oil: surfactant mixture for novel SE-AS combinations and size and PdI using DLS. SE-AS 22 is shown for comparison.

SE-AS Tween80 Squalene α-Tocopherol Avg (Z.ave) PdI

14 35 55 10 79.76 0.113

41 33 55 12 94.13 0.166

110

42 30 55 15 120.73 0.219

43 30 52 18 134.20 0.360

44 25 60 15 131.73 0.192

45 20 65 15 139.10 0.215

22 15 70 15 129.07 0.212

SE-AS 45 showed bimodal distribution after filtration, and a rise in PdI and hence was not tested on HPLC (Figure 5.7). On the other hand, SE-AS 44 retained its size and

PdI after filtration. There was still some content loss in squalene and α-tocopherol; however, it remained within the acceptable range of 80%-120% recovery by RP-UPLC based on method variation. Further optimizations using SE-AS 44 were carried out using poloxamer 188 (P188) as well as TPGS that helped improve the PdI in previous experiments, as well as process changes. However, little to no difference in size or PdI was observed (data not shown). Thus far, SE-AS 44 showed stable and repeatable size and PdI, maintained the size and PdI post filtration as well as showed minimal loss in oil content on filtration.

111

Figure 5.7: Size distribution for SE-AS 44 showing overlay of size distribution graphs before (red) and after (green) filtration. Table below shows size and PdI for SE-AS 44 & 45, as well as % content of oils for SE-AS 44.

5.3.2 Stability assessment of SE-AS at elevated temperatures

Physicochemical stability SE-AS 44 was measured to facilitate in vivo studies. To serve this objective, SE-AS 44 as well as SE-AS 22 and SE-AS 36 were subjected to three temperature conditions i.e. 5°C, 25°C and 50°C. pH, osmolality, size and PdI were measured up to 10 weeks. However, % content data was only measured up to 2 weeks due to an issue with the column. Osmolality and pH remained in the range for all three emulsions (Figure 5.8); however, pH for all emulsions stored at 50C seemed to drop ~0.3 units. Osmolality remained the same.

Size and PdI for SE-AS 44 remained consistent over 10 weeks at all temperatures.

Size for SE-AS 22 and 36 also remained within a ± 10nm range (Figure 5.9). For SE-AS

22 at 10 weeks, the size and PdI increased. The sample was repeated twice and the data

112 looked the same. PdI for both SE-AS 22 and 36 kept fluctuating thus confirming the high polydispersity of these emulsions.

Figure 5.8: pH and Osmolality for emulsions up to 10 weeks. Emulsions were stored at 4C (blue), 25C (red) and 50C (green).

Figure 5.9: Size and PdI for emulsions up to 10 weeks. Emulsions were stored at 4C (blue), 25C (red) and 50C (green). Size and PdI for SE-AS 44 remained consistent compared to SE- AS 22 and SE-AS 36.

UPLC-PDA was used to monitor the % squalene and tocopherol content throughout the stability study. However, at 4 weeks’ time-point, the column was found to be clogged

113 and needed to be cleaned. The data thus, only available up to 2 weeks at all three temperatures. Since the method was not developed entirely, we see several ups and downs in the % content (Figure 5.10). SE-AS 44 showed a pretty consistent recovery compared to T=0 at all temperatures indicative of the stability of the optimized novel emulsion. We also used Cryo-EM to compare morphology of SE-AS 44 with AS03 (Figure 5.11). The droplet morphology of SE-AS 44 was similar at a 500nm scale to AS03. The size distribution and analysis for SE-AS 44 showed that majority of the droplets (~87% considered in the analysis) were less than 150nm in size.

Figure 5.10: % Squalene and Tocopherol content for emulsions up to 2 weeks. Emulsions were stored at 4C (blue), 25C (red) and 50C (green). The blue lines represent the usual acceptable limits of % content from an RP-UPLC experiment i.e. 100% ± 20%.

114

Figure 5.11: Cryo-EM images of AS03 and SE-AS 44 showing comparable droplet morphology. The oil droplets in AS03 are slightly bigger than SE-AS 44 as expected.

All the data obtained thus far suggests that SE-AS 44 is a stable self-emulsified formulation with composition like AS03 and can be sterile filtered without altering any physicochemical properties of the emulsion. To confirm the results from QIV study and investigate whether SE-AS 44 remains non-significantly different to AS03 as well as if we observe any significant differences compared to SEA160, in vivo evaluation using a different antigen was conducted.

5.3.3 In vivo potency evaluation using CMV subunit protein

The goal of this in in vivo study was to compare the optimized SE-AS with a well characterized squalene only emulsion SEA160 using a soluble subunit protein with poor inherent immunogenicity. AS03 was used as a control. The groups are as described on the in the materials and methods section.

The primary readout for this assay was the neutralizing antibody titers. We also performed a total IgG assay to assess the anti-CMV IgG antibodies present in the serum.

115

Together these assays will show humoral immune response. We used spleen four weeks post 3rd immunization to measure T cell response by performing intracellular cytokine staining (ICS).

The neutralizing antibody titers are shown in figure 5.12. 3wp1 titers were low overall. This was expected since the antigen itself is weak, overall titers were expected to be low. Adjuvanted groups show a slight improvement in titers compared to the unadjuvanted titers. At 3wp2, there was a considerable increase in titers for adjuvanted groups compared to unadjuvanted or negative control. SE-AS 44 showed significantly higher titers compared to unadjuvanted. The titers were not significantly different compared to AS03. Most importantly, significant difference was observed between SE-AS

44 and SEA160 titers at 3wp2. At 3wp3, the trend remained similar to titers from 3wp2.

The titers in the negative control and unadjuvanted groups did not increase at all, but considerably increased for all adjuvant groups. SE-AS 44 showed significantly better titers compared to the unadjuvanted group. Mean titers for SE-AS 44 were higher compared to

SEA160. Titers compared to AS03 at 3wp3 were not significantly different.

116

Figure 5.12: Neutralizing antibody titers in serum obtained three weeks post 1st, 2nd and 3rd immunization, against CMV TB40 strain of virus. Each bar represents geometric mean titers (GMT) with 95% confidence interval (CI) from n=10 animals per group. Statistics was performed using one-way ANOVA followed by Tukeys test to compare all groups with each other followed by Dunnett’s multiple comparisons test to compare each group with SE-AS 44. Significant differences are marked on the graph. Comparison with CMV alone is shown in blue, with AS03 is shown in red and with SEA160 is shown in green; where, ns = not significant, *= p˂0.05, **= p˂0.005, ***= p˂0.0005 and ****= p˂0.00005.

Binding antibody titers in sera obtained from 3wp2 and 3wp3 time-points were measured by IgG ELISA using CMV pentamer protein (Figure 5.13). The assay was a secondary read-out to evaluate the humoral immune response from the adjuvanted vaccine.

Mean titers from individual animals in each group are shown in figure 10. The overall trend between the groups remained similar to the trend observed in nAb titers. All adjuvants showed significantly higher titers compared to CMV alone. AS03 gave the highest titers

117 post 3rd immunization. However, at 3wp2, significant difference in titers from SE-AS 44 adjuvanted group were obtained compared to SEA160. Thus, overall humoral response favored adjuvanted groups with SE-AS 44 showing evidence of higher titers post 2nd immunization and overall higher mean IgG and nAb titers compared to SEA160. This data suggests that α-tocopherol may be playing a role in improving the humoral immune response compared to squalene only emulsion adjuvants.

Figure 5.13: Anti-CMV Penta IgG antibody titers in serum obtained three weeks post 2nd and 3rd immunization. Each bar represents geometric mean titers (GMT) with 95% confidence interval (CI) from n=10 animals per group. Statistics was performed using one- way ANOVA followed by Dunnett’s multiple comparisons test to compare difference between each group with SE-AS 44. Significant differences are marked on the graph. Comparison with CMV alone is shown in blue, with AS03 is shown in red and with SEA160 is shown in green; where, ns = not significant, *= p˂0.05, **= p˂0.005, ***= p˂0.0005 and ****= p˂0.00005.

118

Fresh spleens from 10 animals per group were split into 5 animals each at 3wp3 and 4wp3 to ensure data collection at least one time-point. Data from 4wp3 time-point is shown here. To evaluate the cellular immune response, we used the memory markers

CD62L and CD127 to distinguish the CD4+ and CD8+ T cells into transitional, central memory, Effector memory and Effector cells. Prophylactic vaccines aim at developing efficient memory T cell response against any pathogen that can eventually help the effector response upon infection with that pathogen. Fresh spleens four weeks post 3rd immunization, were used to evaluate the memory T cell immune response. Splenocytes were stained for both CD4+ and CD8+ T cells; however, CD8+ T cells were little to negligent and hence only CD4+ T cell data was characterized. The gating strategy was like that shown in Chapter 4, with additional memory marker gating as explained in Figure 2b in materials and methods section.

Upon our evaluation and gating, it was discovered that there were little to no transitional T cells but we saw frequency of cells in the other three quadrants. In Figure 5.14, we have stacked the total antigen specific CD4+ T cells based on the memory response. Effector memory T cells were the majority population, there were also some effector cells, as expected. Additionally, after Boolean gating cytokine+ cells into Th1, Th2, Th0 and Th17 cells, we observed that the main response was Th0/Th2 type. Th2-type response is a characteristic of most emulsion adjuvants, and seen in the previous chapter with QIV antigens, hence we also characterized the Th2 response across all CD4+ population after gating for memory markers. SE-AS 44 showed the highest overall Th2 cells compared to other groups (Figure 5.15). This corroborated with previously seen data with influenza antigens. Figures 5.16 and 5.17 show the CD4+ subtype response for effector memory

119 and effector cells respectively. SE-AS 44 shows a slightly lower effector memory population compared to SEA160; however, that decrease in effector memory cells seems to have converted into effector cells. Since the overall frequency of T cells was low, this effect of conversion from effector memory to effector cells may not be prominent. Overall, the trend for ICS remained like humoral immune response.

A n tig e n s p e c ific C D 4 + T c e ll re s p o n s e

T E ffe c to r Q 4

+ 0 .6

4 E ffe c to r m e m o ry Q 3 4

D C e n tra l m e m o ry Q 2

+ 0 .4

f o

0 .2

e u

q 0 .0

e S P Figure 5.14: Frequencies of overall antigen-specific CD4+ T cells classified as either Central memory, effector memory or effector cells.

120

A n tig e n s p e c ific T h 2 -ty p e C D 4 + re s p o n s e

0 .4

s E ffe c to r Q 4

l e

c E ffe c to r m e m o ry Q 3

0 .3 2

h C e n tra l m e m o ry Q 2

f o

0 .2

n e

u 0 .1

r F

0 .0

e S P Figure 5.15: Frequencies of antigen-specific Th2-type CD4+ T cells classified as either Central memory, effector memory or effector Th2 cells.

A n tig e n - s p e c if ic E ff e c t o r m e m o r y C D 4 + c e lls

l l

e T h 17

T T h2

0 .4 +

4 T h1

4 D

C T h0

D C

0 .2

e r

F 0 .0

n e P Figure 5.16: Frequencies of antigen-specific effector memory cells gated from Quadrant 3 after gating for memory markers

121

A n t ig e n - s p e c if ic E f f e c t o r C D 4 + c e lls s

l 0 .1 0 l

e T h 17

T T h2

4 T h1

4 D

C T h0

4 0 .0 5

e r

F 0 .0 0

n e P Figure 5.17: Frequencies of antigen-specific effector cells gated from Quadrant 4 after gating for memory markers.

Overall, SE-AS 44 adjuvanted group shows a better humoral and cellular immune response compared to SEA160, insinuating at a possible role of α-tocopherol in providing additional immunostimulatory effect.

5.4 Discussion

As elaborated in chapters 1-3, simpler methods for developing emulsion adjuvants could be one of the key ways to improve adjuvant production, storage and distribution in poor and developing countries. Additionally, it could aid in pandemic preparedness in case of epidemics and prepandemic availability of vaccines. In the previous chapter, we established a proof of concept by using design of experiments and formulation science to incorporate α-tocopherol in a squalene oil containing self-emulsifying formulation to develop two novel SE-AS formulations. These were compared in vivo with AS03 using

122 inactivated influenza vaccine to show similar immune responses and confirm its adjuvant activity. AS03 has shown superior humoral response compared to squalene oil only emulsion adjuvants such as MF59 in preclinical and clinical studies (90, 91), suggesting the role of α-tocopherol in enhancing the potency of these o/w emulsions.

Here in this chapter, we first optimized the SE-AS by improving the PdI.

Throughout our DoE experiments in chapter 4, we observed that the surfactants have a considerable effect on the droplet size and distribution, in presence of a corresponding amounts of squalene and α-tocopherol. Our initial objective was to achieve an SE-AS as close in composition to AS03 as possible to be able to compare the immune responses in vivo. Although the size for both novel SE-AS was less than 200nm, the PdI varied. For sterile filtration and long-term stability, it is essential to have a homogeneous droplet distribution to avoid flocculation (reversible separation of components) or creaming of emulsion. Both the novel SE-AS showed loss in content and heterogeneity in droplet size distribution post sterile filtration through a 0.22µm syringe filter suggesting that there might not be enough surfactant to stabilize the droplets in these emulsions. Thus, our main criteria for optimizing the SE-AS was the retention of size and % oil content before and after filtration for these SE-AS. Our data showed that having 1/3rd the amount of α- tocopherol did not affect the humoral and cellular immune response compared to AS03 using inactivated QIV in BALB/c mice. Hence, for optimization of the SE-AS we kept the

α-tocopherol content consistent at 15% v/v in oil-surfactant mixture and increased the surfactant content, and evaluated the resulting SE-AS based on the feasibility of sterile filtration. We observed that 25% v/v PS80 resulted in SE-AS44 and was the only emulsion that maintained size and PdI after sterile filtration. Additionally, the % content of the oils

123 was within the 80%-120% recovery using UPLC-PDA. Upon subjecting SE-AS44 to long term stability at accelerated conditions, we reported that the size and PdI can be maintained for up to 10 weeks and the % content remained in the acceptable range for up to 2 weeks.

Thus, our formulation optimization studies resulted in an SE-AS that can be sterile filtered on a laboratory scale and remained stable for up to 10 weeks in terms of size and PdI, which were the most important attributes.

The in vivo study was designed to compare humoral and cellular immune responses

SE-AS 44 with a squalene only emulsion adjuvant SEA160 and AS03. We used CMV pentamer antigen which is a soluble, purified subunit protein having inherently lower immunogenicity but makes a great candidate for neutralizing antibody especially against the CMV entry into epithelial and endothelial cells. Our primary read-out was neutralizing antibody titers where we saw a significant increase in titers 3weeks post 2nd immunization for all the adjuvant groups compared to negative control (saline) and unadjuvanted protein group, confirming the adjuvant effect for SE-AS 44. Additionally, SE-AS 44 showed significantly higher titers compared to SEA160, confirming our initial hypothesis for a potential role of α-tocopherol as an immunomodulator. The increase in titers 3 weeks post

3rd immunization was not as dramatic suggesting the saturation of antibody titers at the

CMV dose used. However, the mean titers were still higher for SE-AS 44 compared to

SEA160. The results from nAb titers were corroborated by the functional antibody IgG titers with similar trend and comparison between different groups. Thus, the overall humoral response was better for SE-AS 44 especially after the second immunization, compared to squalene only emulsion adjuvant. We performed ICS with memory marker

CD127 (IL7 receptor α) and naïve cell marker CD62L (137). The goal was to evaluate the

124 memory CD4+ T cell population in immunized mice as well as the effector population

(CD62L-CD127-). Overall frequencies of antigen-specific CD4+ T cells was low. As observed in the previous chapter with inactivated QIV, emulsion adjuvants showed an overall higher Th0/Th2 population. We characterized the CD4+ responses in terms of memory responses in Figure 10a to show that there was a high amount of effector memory response in adjuvanted groups 4 weeks post 3rd immunization. Since Th2-type response is characteristic of emulsion adjuvants, we showed overall antigen specific Th2 cells in figure

10b that showed higher Th2 cells for SE-AS 44 compared to other groups. In 10c and 10d we showed Boolean gated antigen specific CD4+ T cells in effector memory and effector

T cells. The group adjuvanted with SE-AS 44 shows lower effector memory response but higher effector response.

Overall, SE-AS 44 showed higher immune responses compared to unadjuvanted and similar immune response to AS03. The humoral immune response was also significantly better than SEA160, proposing the role of α-tocopherol in improving the antibody titer response against CMV.

5.5 Conclusion

In summary, in this chapter we could provide data to support the second aspect of our hypothesis that adding α-tocopherol would provide with enhanced immune response compared to squalene only emulsion adjuvants. We also optimized the SE-AS to make it as an ideal candidate for scale-up. In our next chapter, staying on track with your central goal on making vaccines available in underprivileged countries as well as staying prepared for pandemic situations, we explore creating a single vial lyophilized vaccine for reconstitution using SE-AS 44 as our lead adjuvant candidate.

125

CHAPTER 6 Lyophilization of SE-AS with CMV Antigen to Develop a Single Vial Vaccine for Reconstitution

6.1 Introduction

Most adjuvanted vaccines, especially with emulsion adjuvants, are mixed bed-side at the site of injection and depend on cold-chain maintenance and distribution. Typically, the protein antigen and the emulsion must be mixed together just before administration to obtain the optimal efficacy. Single vial liquid vaccine containing of both antigen and adjuvant have been reported to show some unwanted hydrophobic interactions of the antigen with adjuvant eventually leading to reduced immunogenicity of the vaccine (138).

The two-vial or sometimes needs additional formulation of either antigen and/or adjuvant leading to a potential third vial in the on-site mixing of the vaccine. This not only increases the costs of storing, shipping and distributing these vaccines but also makes it a complex process. Furthermore, according to the World Health Organization, vaccines vialed in multi-dose packaging waste between 15 and 50% of the actual vaccine (139). Maintaining cold-chain is tedious but also expensive, which might lead to lesser availability in less- privileged countries (2, 101, 102). Even when the cold chain is maintained, freezing from overexposure to cooling systems can result in vaccine damage and reduced efficacy or failure (103, 104). One of the potential solutions to reduce interactions between antigen and adjuvant and improve stability could be removal of water. There have been reports published to achieve this in terms of making a single vial dry vaccine for reconstitution using spray drying (27, 115) or lyophilization (3, 4). These dried vaccines have shown stability at elevated temperatures and thus, building a single vial dried vaccine for

126 reconstitution could be a potential solution at improving vaccine distribution in developing countries by simultaneously making storage and distribution less expensive.

Several aspects of emulsion adjuvant and antigen need to be taken into consideration before co-vialing and freeze-drying these formulations. Droplet size distribution is one of the most important aspects of an emulsion adjuvant contributing to its potency and might lead to suboptimal immune responses to an antigen if the size is changed post lyophilization. Maintaining antigen integrity is also vital to maintain the immunogenicity post lyophilization.

In this chapter, we have evaluated the feasibility of freeze-drying SE-AS 44, our lead candidate without the antigen to optimize the lyophilization cycle, and the use of optimal excipients for a stable product. Our main criterion of evaluation was to maintain the emulsion droplet size before and after lyophilization. We also evaluated several excipients based on their use in literature as a lyo-protectant. We used SEA160, a squalene oil-in-water emulsion previously described and performed lyophilization cycle development work. We finalized a lyophilization cycle and used sucrose as a lyoprotectant to make a single vial adjuvanted vaccine for reconstitution with CMV pentamer antigen.

The rationale for use of CMV pentamer antigen was explained in chapter 5, which also reports the adjuvant effect of SE-AS 44 with CMV pentamer. We performed an in vivo potency evaluation of the liquid and lyophilized SE-AS 44 adjuvanted CMV pentamer vaccine to understand the comparison between the two forms.

127

6.2 Materials and Methods

6.2.1 Formulation materials

Squalene oil was obtained from JX Nippon Oil Trading Company (Tokyo, Japan) and dl-α-tocopherol was obtained from DSM Nutritional Products (Heerlen, Netherlands).

Polysorbate 80 (PS80) was obtained from JT Baker (Center Valley, PA). Hypure water for injection (WFI) like quality water and phosphate buffered saline (PBS) were obtained from

6.2.2 Freeze-drying of SE-AS with and without antigen

6.2.2.1 Formulation before lyophilization (Pre-lyo): Pre-lyo formulation was necessary to include the diluent in the formulation. In general, per 400µL fill volume for each vial, 100µL adjuvant was used with 100µL of antigen (at 2X concentration) and finally diluted with 200µL of the lyo-protectant or diluent. Emulsions (SEA160 or SE-AS

44) were formulated as described in chapter 4. For pre-lyo formulation without protein, buffer was added as a substituent. The goal was to get the adjuvant and antigen at the same concentration as liquid bed-side mixed vaccine upon reconstitution with water. The pre- lyo formulation was characterized for size and PdI using dynamic light scattering (DLS) as well as for pH and Osmolality as described in chapter 4.

6.2.2.2 Lyophilization cycle development: We used information from some preliminary work performed on bench-scale VirTis lyophilizer using SEA160 as a starting point for lyo cycle development. This lyo cycle in VirTis lyo was long and had very low

128 primary and secondary drying pressures. Typically, lyo-cycle was set up based on parameters in Table 6.1. This table describes variable sin letters, A, B, C, X, Y and Z. In

VirTis lyo the vacuum was applied at 10mTorr; whereas, transferring this cycle to

LyoStar3, we changed the pressure to either 55mTorr or 70mTorr.

Table 6.1: Freeze-drying cycle developed on VirTis bench-scale Lyophilizer

Ramp Vacuum Step Temp °C Hold (hours) (˚/min) (mTorr)

5 1 0.5 -

Freezing -5 1 0.5 -

X A 2 -

Primary Drying Y B C 10,55 or 70

Secondary Drying Z 0.1 6 10,55 or 70

Hold 5 1 Until taken out

For additional lyo-cycle development work, we used differential scanning calorimetry and freeze-dried microscopy (explained in detail in the next section) to determine a primary drying temperature as well as shorten the time of the drying by increasing the vacuum during the drying phase.

6.2.2.3 Glass transition temperature and collapse temperature evaluation: For measuring the collapse temperature, we used a simple Olympus microscope connected to linkam software and QiCam imaging. One drop of the SEA160emulsion with 10% sucrose was place in the middle of the stage and a coverslip was slowly placed on it and the camera

129 was set to take pictures every 2 seconds. The emulsion was cooled to -55°C and then heated at a ramp rate of 1°C/min.

To determine the Tg’ of the emulsion, TA® instrument’s DSC Q2000 was used to run a cool-heat DSC cycle. An aluminum hermetic pan was used and ~15mgs of emulsion was weighed and sealed with hermetic lid. The sample was cooled to -80C and the heated gradually at a rate of 1C/min and modulated to see minute differences in heat exchange.

6.2.2.4 Protein integrity using SDS-PAGE: CMV pentamer at a dose of 0.05 µg was mixed with SE-AS 44 and pre- and post lyo formulation were tested for protein integrity using SDS-PAGE. After lyophilization, the lyo cake was reconstituted with 200

µL water and emulsion droplets were disrupted by centrifugation. The subnatant was collected and mixed with 4X sample buffer containing dithiothreitol (DTT) and bromophenol blue. These samples were boiled at 95°C and centrifuged at 12,000 RPM for

5 minutes. All samples were loaded on to 4-12% Bis-Tris gel with one well containing

Mark 12 Unstained Protein Standard obtained from Invitrogen. The gel was run at 200 volts for 35-45 minutes and then stained with silver stain following the procedure in kit obtained from Thermo Scientific. The next day the gel was de-stained with deionized (DI) water for 3-4 hours to remove the non-specific staining and was scanned with a Chemi-doc

MP imaging system from Bio-Rad and analyzed using the ImageLab image processing software.

6.2.2.5 Determination of percent water content using Karl-Fisher apparatus:

Moisture content was determined using a Metrohm 851 (Herisau, Switzerland) Karl Fisher

(KF) moisture analyzer and Tiamo software. The vial with lyo cake is weighed and then

N, N-Dimethylformamide (DMF) was used as an anhydrous water standard and to dissolve

130 the lyo cake to form a homogenous liquid. The vial cap and stopper are then saved and dried in an oven for a final and accurate cake weight. After blanking the instrument with

DMF for n=3 times, the sample is also run n=3 times. The water abs (in µg) value is taken from the software and using the exact cake weight and the average content in the blank samples, the water content in the lyo cake is measured using the equation below:

6.2.3 In vivo comparison of liquid bed-side mixed and lyophilized single vial adjuvanted CMV vaccine

6.2.3.1 Ethics statement: All studies were conducted in accordance with the GSK

Humane Care and Use of Laboratory Animals and the U.S. Interagency Research Animal

Committee Principles for the Utilization and Care of Research Animals.

6.2.3.2 In vivo study design and immunization regimen: Cell-line derived CMV pentamer was used to test the in vivo potency of the novel SE-AS in mice. The dose of antigen in each mouse was 0.05µg. The antigen content was determined by UV/visible spectrophotometry as well as reverse phase HPLC. Sample size of 13 animals per group was calculated such that it provided a power of 80% to detect a 3-fold difference between any 2 groups with 95% confidence interval. 6-8 weeks old female C57BL/6 mice from

131

Charles River laboratories (Gaithersburg, MD) were used. The study design comprised of three immunizations three weeks apart, where 50µL of the vaccine was injected intramuscularly in the gastrocnemius muscle such that at each time-point alternate limb was used. The groups in this study were: unadjuvanted CMV, CMV with SE-AS 44, CMV with AS03 and CMV with SE-AS 44 lyophilized as a single vial vaccine. Figure 6.1 shows the summary of the study design and immunization schedule.

Figure 6.1: Study design, dosing groups and immunization schedule to compare liquid bed-side mixed formulation with Lyophilized single vial formulation.

The vaccine was formulated to mimic bed-size immunizations, with 1:1 mixing of antigen and adjuvant, to reach the final antigen concentration. The lyophilized vaccine was reconstituted with 200µL of sterile water-for-injection like (WFI-like) water All the formulations were characterized for pH, Osmolality, size and PdI and endotoxin before immunization. Endotoxin was measured using the Endosafe NexGen – PTS from Charles

132 river laboratories (Wilmington, MA) and limulus amoebocyte lysate (LAL) cartridges with test range 10-0.1 EU/mL. Additionally, protein integrity was confirmed via gel electrophoresis (SDS-PAGE) as shown in Figure 2. It was ensured that the vaccine dosed for each group had endotoxin lower than 1 EU per dose. Bleeds were collected 3 weeks post 1st (3wp1), 3 weeks post 2nd (3wp2) and 3 weeks post 3rd (3wp3) immunization and the processed sera was used to test the humoral immune responses by neutralizing antibody assay (nAb) and IgG titers by ELISA. Spleens from 5 animals per group were harvested

3wp3 and 4wp3, and used to measure T cell immune responses via intracellular cytokine staining.

6.2.3.3 Determination of neutralizing antibody (nAb) titers: Retinal pigment epithelial cell line (ARPE-19) was used since TB40 (a CMV virus strain) was known to infect these cells in this assay. On day 1, 100uL of ARPE-19 cells were plated in 96 well flat bottom plates in complete growth medium i.e. DMEM + 10% fetal bovine serum (FBS)

+ 1% Penicillin-Streptomycin. Plated were incubated in 37C overnight ~ 24 hours. On day

1:50 dilution. In each plate, 75uL of serum dilutions were prepared using Tecan and then

75uL of TB40 virus was added to each well to make a total of 150 in each plate. This was enough for duplicates of each sample. This virus-serum mixture was incubated at 37C, 5%

133 cells were fixed using 4% paraformaldehyde and incubated at RT for 20 mins following by

10-20 fields per well). Interpolated titers are then calculated at 50% fluorescence intensity.

6.2.4.4 IgG antibody ELISA: Antibody titers were determined in serum obtained from each animal at 3wp2 and 3wp3. To determine the CMV pentamer specific binding

IgG antibody titers, sandwich ELISA was used. 96 well Nunc-immuno Maxisorp F96 plates were used to coat 100µL of 1µg/mL CMV pentamer antigen per well overnight at

4°C. Antigen coated plates were washed with 1X phosphate buffered saline (PBS) &

0.05%w/v Tween20 and blocked with 1% w/v bovine serum albumin (BSA) solution in

PBS. Serum from immunized animals was added in the first row of the plate such that well

Immunoresearch (West Grove, PA) for another one-hour incubation at room temperature.

Substrate was added quickly after washing plates again, for 15 mins and then immediately stop solution was added. Plates were read using EnVision 2105 Multimode plate reader

134 from Perkin Elmer (Waltham, MA). Titers were calculated at 50% interpolated optical density (OD) value obtained from the plate reader.

6.2.4.5 Intracellular cytokine staining assay: T-cell responses were analyzed

4wp3 by intracellular cytokine staining of in vitro antigen-stimulated splenocytes. Spleens from individual animals were processed to single-cell suspensions, followed by treatment with RBC lysis buffer (Ebioscience, Thermo Fisher Waltham, MA). CMV pentamer peptides gH, gL, UL128, UL130 and UL131 from GeneScript (were used for stimulation of splenocytes. These splenocytes were stimulated at one million cells per well density with anti-CD3 from BD Biosciences (San Jose CA) used as positive control, media was used as negative control, and peptide pool was prepared for antigen stimulation condition.

CA), CD4 conjugated with BUV395, CD8 conjugated with BB700, CD44 conjugated with

135 from Thermo fisher Scientific (Waltham, MA). Since most of the anti-mouse antibodies used are rat or hamster derived; anti-rat anti-hamster Ig, κ/Negative control compensation particles from BD Biosciences stained with all the above fluorochrome conjugated antibodies including an unstained control for preparing compensation controls. The samples were acquired on a BD FortessaX20 SORP flow cytometer from BD Biosciences

(San Jose, CA) followed by analysis with FlowJo software (Ashland, OR). The gating strategy is described in Figure 6.2. We defined our gating strategy (Figure 2) in FlowJo where, the live cells were first differentiated from dead and were then used to differentiate singlets. From the singlets, we identified CD3+ T cells and used them to gate for CD4 and

CD8 T cells. Antigen specific cells were identified by gating on upregulated CD44 cells.

Individual cytokine gates were then established on these antigen specific CD4 and CD8 T cells. Memory markers were used to identify antigen-specific transitional, central memory, effector memory, and effector populations. Individual cytokine gates were established on these memory populations.

136

Figure 6.2: Example of gating strategy for ICS. First live cells are differentiated from dead, and then separated based on morphology to get single cell lymphocytes, differentiated based on CD3 marker. CD4+ and CD8+ T cells are then gated based on CD3 gates followed by identifying antigen specific CD4+ or CD8+ T cells.

6.2.4.6 Data analysis and statistics: GraphPad Prism software (San Diego, CA) was used to analyze and plot data from the in vivo immune responses. For humoral responses, one-way analysis of variance (ANOVA) followed by Tukey’s Multiple comparisons test was used to evaluate differences in immune responses from individual animals in the dosing groups. For nAb titers, Dunnett’s test post one-way ANOVA was used to compare SEA160 with SE-AS 44. For ICS, a nonparametric Kruskal-Wallis test was run followed by Dunn’s multiple comparisons test for comparison within different dosing groups.

137

6.3 Results

6.3.1 Single-vial vaccine feasibility on VirTis bench-scale freeze dryer

Figure 6.3 represents the lyo cycle, the formulation before lyophilization as well as the size and the PdI comparison before and after lyo. The cycle used on VirTis lyo was conventional with low vacuum pressure and longer drying times. The increase in size was approximately 20-40nm but the PdI remained similar across several experiments and repeats. This lyophilization cycle was used by another PhD student to formulate single vial lyophilized vaccine with ovalbumin as well as an HIV glycoprotein gp120. The antigen integrity was tested via SDS-PAGE and the lyophilized vaccine was tested in vivo in

BALB/c mice and was compared to liquid vaccine in terms of antibody responses (data not shown). The data showed that lyophilized SEA160 showed comparable responses to liquid vaccine for both antigens. Thus, this provided the basis for further development of the lyophilization cycle as well as the formulation.

We evaluated the same cycle with higher pressures 55mTorr and 70mTorr (data not shown) in LyoStar3 and selected 55mTorr for further cycle development. All the preliminary optimization work was done using SEA160 and then applied to SE-AS 44. In the next section we discuss the cycle development and optimization.

138

Figure 6.3: Freeze-drying SEA160 on VirTis lyophilizer. The table in red above shows the formulation of pre-lyo single vial vaccine. The table on the right bottom shows the lyophilization cycle. The graph on the right shows the size distribution before (red) and after (green) lyophilization.

6.3.2 Lyophilization cycle development and excipient screening using SEA160.

To set the primary drying temperature, and modify the cycle for LyoStar3, we performed DSC and FDM experiments to measure glass transition temperature (Tg’) and collapse temperature (Tc) respectively. Glass transition temperature is important to set the primary drying temperature to avoid leading formulation into a “glassy state” and change its physicochemical properties. The Tg’ for SEA160 with 10% sucrose as well as 10% trehalose was found to be approximately -33.5°C (Figure 6.4). We decided to keep our primary drying at -35°C slightly below the Tg’. We also looked at the collapse temperature of the pre-lyo formulation for SEA160 with 10% w/v sucrose on FDM and found out that the emulsion adjuvant showed a collapse temperature of about 27°C, with beginning of collapse at 32°C and complete collapse at 22°C (Figure 6.5). When setting up the cycle in

139

LyoStar 3, we used the cycle as described in table 2. Size for SEA160 post lyophilization showed a consistent 20nm increase with little to no change in PdI. We then evaluated different freezing ramp temperatures as the freezing rate may define the formation of ice crystals between the oil droplets and thus affect the size of the emulsion. We evaluated a slow (0.1°C/min) and a fast-freezing rate (5°C/min). The data shown in table 6.2 suggests that fast freezing rate resulted in higher size distribution and increase in size. There wasn’t a clear difference between 1°C/min and 0.1°C/min. We decided to use 0.1°C/min for further optimization work with the excipients.

Figure 6.4: A DSC thermogram showing a regular cool-heat cycle for the pre-lyo formulation of SEA160 with 10% w/v sucrose. The left Y-axis showed the heat flow and the right y axis shows the reversible heat flow. The change in reversible heat flow is recoded at the range of temperature and the average is considered as the Tg’

140

Figure 6.5: Measuring collapse temperature using FDM. Images under FDM where the pre-lyo formulation of SEA160 with 10% w/v sucrose is subjected to a cool-heat ramp to observe the collapse of the lyo cake

Table 6.2: Size and PdI from freezing ramp rate evaluation

Size, increase in size (nm) and PdI for SEA160 formulations with varying amounts of sucrose and trehalose after lyophilization as measured on DLS. Suc = sucrose and Treh = trehalose

141

Additional to sucrose, we also evaluated trehalose in the above experiment with freezing ramp rates. Trehalose has been used in emulsion adjuvant freeze-drying before

(3). We tested sorbitol and glycerol in combination with sucrose as solid-state stabilizers, however, adding these excipients led to increase in size and PdI post lyophilization. We also tried annealing using mannitol. Annealing is applied during freezing where the product is held at a temperature above the final freezing temperature for a defined period to crystallize the potentially crystalline components (in this case, mannitol) in the formulation during the freezing stage (118). Annealing is often applied to allow for efficient crystallization ensuring uniform sublimation of ice crystals during the drying phase. Data in Table 6.3 shows size and PdI for SEA160 when freeze-dried with sucrose and trehalose using mannitol.

Table 6.3: Size and PdI measurements from annealing experiments with mannitol

Increase SEA160 Sugar Size PdI in size control N/A 173.03 0.149 filtered N/A 178.70 0.121 filtered 5% suc + 2% Mann 248.55 69.85 0.314 filtered 5% Treh + 2% Mann 264.85 86.15 0.562 Size, increase in size (nm) and PdI for SEA160 formulations with varying amounts of sucrose or trehalose with mannitol after lyophilization as measured on DLS. Suc = sucrose and Treh =- trehalose

Overall, our freeze-drying cycle development as well as excipient screening on

Lyostar3 yielded 5-10% sucrose as the most optimal lyoprotectant with minimal increase in droplet size (~20nm increase).

142

6.3.3 SE-AS 44 formulation optimizations for lyophilization

The lyo cycle developed for SEA160 was applied to SE-AS 44 (Table 6.4 & Figure

6.6). There was a slight increase in size observed as expected, but the PdI remained similar.

However, one major observation was the increase in osmolality after lyophilization which can be attributed to the additional salt content in SE-AS44 due to the salt content in PBS.

Also, a slight cake shrinkage was observed with the SE-AS44 post-lyo, that did not seem the affect the size and PdI after reconstitution. Thus, SE-AS 44 was reformulated without salt in 10mM Potassium phosphate buffer at pH 6.8-7.0. The osmolality of the resulting formulation was reduced to approximately 330 mOsm/kg and the size and PdI were comparable to the previous formulation. Additionally, the cake shrinkage was addressed too suggesting that it might have been due to the salt content. The water content of the final formulation was also well under 1% w/v as measured by KF. The SE-AS 44 when formulated with CMV pentamer showed similar bands as standard on gels using SDS-

PAGE. Figure 6.7 shows protein integrity and size distribution for optimized SE-AS 44 formulations with CMV pentamer. Collectively, this data suggested the feasibility of co- vialing and lyophilizing a single-vial emulsion adjuvanted CMV vaccine.

Table 6.4: Final Lyophilization cycle used to develop single vial vaccine.

Temp Time Ramp/Hold Vacuum Step °C (min) (R/H) (mTorr)

5 20/30 R/H N/A

Freezing -5 10/30 R/H N/A

-45 400/120 R/H N/A

Primary -35 100 R 55 Drying

143

-35 1680 H 55

25 600 R 55 Secondary Drying 25 600 H 55

Figure 6.6: SE-AS 44 lyophilization. Top left table represents the post lyophilization characterization data. The image on the right shows shrinkage of the lyo cake. The graph in bottom shows size distribution before and after lyophilization.

144

Figure 6.7: Optimized SE-AS 44 Lyophilization with CMV pentamer. Left image is of the gel showing pre- and post-lyophilization protein bands, the graph in the middle shows size distribution before and after lyo and the vial image on the right shows the single vial vaccine lyo cake.

6.3.4 In vivo comparison of potency between bed-side mixed liquid and single-vial lyophilized vaccine

The goal of this in vivo study was to compare potency of the traditional bed-side mixed adjuvanted vaccine with the single-vial lyophilized vaccine. AS03 was used as a control. The groups are as described on the top left of figure 1, with details in the bottom and study schedule on the right side of the figure. The primary readout for this assay was determining nAb. We also performed a total IgG assay to assess the anti-CMV IgG antibodies present in the serum. We used spleen four weeks post 3rd immunization to measure T cell response by performing intracellular cytokine staining (ICS).

The nAb titers shown in figure 6.8 were calculated from serum three weeks post each immunization. Increase in titers after every immunization was observed, especially

3wp2 time-point, where adjuvant effect was observed. SE-AS 44 adjuvanted liquid and bed-side formulations showed no significant difference in nAb titers. Additionally, both

145 the formulations were comparable to AS03. This data provides evidence that single vial lyophilized vaccine can be developed while maintaining the potency of the adjuvanted vaccine.

Figure 6.8: nAb titers in serum obtained three weeks post 1st, 2nd and 3rd immunization, against CMV TB40 strain of virus. Each bar represents geometric mean titers (GMT) with 95% confidence interval (CI) from n=13 animals per group. Statistics was performed using one-way ANOVA followed by Tukeys test to compare all groups with each other followed by Dunnett’s multiple comparisons test to compare each group with SE-AS 44. Significant differences are marked on the graph. Comparison with CMV alone is shown in blue, with AS03 is shown in red and with lyophilized single vial is shown in green; where, ns = not significant, *= p˂0.05, **= p˂0.005, ***= p˂0.0005 and ****= p˂0.00005.

To additionally characterize the humoral immune response, we calculated the total

IgG antibody titers against CMV pentamer at time-points 3wp2 and 3wp3 (Figure 6.9).

We could harmonize the IgG titer data with nAb titers in terms of trend between the groups the adjuvanted groups showed increase in titers at both time-points with SE-AS 44 showing

146 significantly higher titers compared to unadjuvanted. The difference in IgG antibody titers between liquid and lyophilized vaccine was not significant. Therefore, the overall humoral response supports the hypothesis of developing a single vial lyophilized vaccine using an

SE-AS and CMV pentamer.

Figure 6.9: Anti-CMV Penta IgG antibody titers in serum obtained three weeks post 2nd and 3rd immunization. Each bar represents geometric mean titers (GMT) with 95% confidence interval (CI) from n=10 animals per group. Statistics was performed using one-way ANOVA followed by Tukeys multiple comparisons test to compare difference between each group with SE-AS 44. Significant differences are marked on the graph. Comparison with CMV alone is shown in blue, with AS03 is shown in red and with SEA160 is shown in green; where, ns = not significant, *= p˂0.05, **= p˂0.005, ***= p˂0.0005 and ****= p˂0.00005.

147

C D 4 + C D 4 4 + T c e lls

s 1 .0

l l

e T h 17

T T h2

4 T h1

4 D

C T h0

4 0 .5

e r

F 0 .0

)

(

- E S Figure 6.10: Antigen specific CD4+ T cells using ICS assay Spleens from individual animals were collected and processed and restimulated with antigen, to get cellular immune responses. Polyfunctional CD4+ immune responses were characterized further into antigen specific subtypes of Th immune cells. Based on the cytokines released by the splenocytes, the T- cell response was Boolean gated into Th1, Th2, Th17 and Th0 responses. Frequencies of antigen-specific CD4+ T cells at time-point 4wp2. Each bar represents the mean (within each subtype) with standard deviation values from n=5 animals per group. Statistics was performed on overall antigen specific CD4+ T cells using non-parametric Kruskal-Wallis test followed by Dunn’s test for multiple comparisons.

To understand the cellular immune response, we conducted a ICS experiment using fresh splenocytes from immunized mice at 4wp3 and the data is shown in figure 6.10. The experiment was set up as described in the materials and methods section earlier in this chapter and was like the set-up in Chapter 4. We characterized for both antigen-specific

CD4+ and CD8+ T cell response however, based on the previous study in chapter 5, we learned that with emulsion adjuvants and CMV pentamer, we get lesser CD8+ T cells compared to CD4+ cells, and hence data for only CD4+ T cells is shown in terms of Th

148 subtype response. The results showed a dominant Th0/Th2 response as seen in chapter 5 with SE-AS and CMV pentamer antigen. Overall the trend was similar to humoral immune response where adjuvanted groups showed higher overall frequencies of antigen-specific

CD4+ T cells than unadjuvanted CMV pentamer alone. Lyophilized and liquid vaccine also showed no significant difference with each other and with AS03.

Thus, we achieved our goal to provide a proof-of-concept to develop a single vial adjuvanted lyophilized vaccine that maintained its potency in vivo compared to bed-side mixed formulations

5.4 Discussion

The continuous research on adjuvants will always require efficient ways of formulating and presenting the adjuvanted vaccine in clinic. The main reason why adjuvant and antigen are not stored in a single vial is mainly due to the interactions between the components of antigen and adjuvant in the liquid form. These interactions of the antigen with adjuvant components like surfactants or buffers could render them less immunogenic.

However, if the water is removed from these co-vialed antigen and adjuvant formulation, it might be possible to prevent these interactions. There are a few ways of drying liquid formulations such as spray drying, freeze-drying, or spray freeze drying as discussed briefly in chapter 3. In this chapter, we have tried to lyophilize the novel adjuvant SE-AS

44 to evaluate the feasibility of creating a single vial dried vaccine for reconstitution without losing the immunological properties of the antigen and adjuvant. The importance of this work is crucial since it would revolutionize the formulation, storage, distribution and administration of adjuvanted vaccines reducing the overall cost of the vaccine thus making it convenient to manufacture, store and distribute in developing countries.

149

Here, we used the SP Scientific Lyostar3 which is a laboratory scale smart lyophilization system representative of a large scale GMP freeze-dryer. We used SEA160 for some preliminary lyoprotectant evaluation and lyo cycle development work. Work previously done on bench scale lyo with SEA160 showed approximately 20nm increase in size however, it did not affect the antibody titers when co-vialed and lyophilized with HIV gp120 antigen compared to liquid adjuvanted vaccine. This suggested that lyophilization might be causing limited swelling of the droplet in the adjuvant emulsion which does not affect the in vivo potency of the adjuvant. There’s also some published work on lyophilized adjuvanted vaccine that report similar increase in droplet size in adjuvant emulsions (3, 4,

110). The eventual goal was to develop a cycle with shortened drying times, and thus short overall freeze-drying time to reduce costs, in a freeze-dryer that can be tech transferred for scale-up. As discussed in results, we continued our cycle development using SEA160 especially evaluating the freezing ramp rates which might be important for nanoparticulate materials. Our data suggested that fast freezing at 5°C/min led to increase in size post lyophilization; whereas, 0.1°C/min and 1°C/min, maintained the PdI and the size increase comparable to previous cycles. This may have been due to several reasons such as insufficient crystal formation during the fast freezing, or large and non-uniform crystal formation leading to irreversible oil-droplet disruption or aggregation. Slower freezing ramps extend the cycle however are ideal option for optimization work and hence we used

0.1°C/min for all further work. We also evaluated excipients such as trehalose at concentrations 5-10% w/v. Trehalose is a very common disaccharide alternative to sucrose as a lyoprotectant with a similar property. Upon DSC evaluation, both sucrose and trehalose reveal the Tg’ with SEA160 to be around 33.5°C. However, our data suggested

150 that trehalose led to bimodal distribution in size compared to sucrose. Possibility of annealing during freezing to ensure and improve crystallization was evaluated using mannitol with either sucrose and trehalose. However, our data showed that annealing step caused a higher increase in size and PdI. Glycerol and Sorbitol were tested in lower concentrations along with sucrose to improve solid state stability. However, the size post lyophilization increased again, suggestion little to no compatibility with these additional excipients. Finally, after all optimization and cycle development, we used 10% w/v sucrose as a final lyoprotectant and applied the optimized cycle to SE-AS 44

SE-AS 44 is formulated in a different buffer – PBS, and hence is already isotonic upon formulation. After preparing the pre-lyo formulation with 10% sucrose, the tonicity post lyo is essentially increased and thus SE-AS 44 was reformulated using 10mM

Potassium Phosphate buffer. The reformulation with 10mM buffer did not essentially cause any change in physicochemical property of SE-AS 44 and addressed the osmolality issue.

When co-vialed with CMV, we tested the integrity of the protein post lyophilization and found comparable gel bands with the control showing conservation of the protein integrity post lyophilization. Thus, the next step was to compare in vivo the potency of a conventional bed-side mixed liquid CMV vaccine with single vial lyophilized vaccine reconstituted prior to injection. Our primary nAb assay showed comparable titers at all time points between lyophilized and liquid vaccine. This data concurred with the anti-CMV IgG antibody titer data leading to a conclusion that lyophilized single vial vaccine shows similar humoral immune response compared to liquid vaccine. The ICS data also showed no significant difference between lyophilized and liquid vaccine.

151

In summary, data in this chapter supports our objective of lyophilizing SE-AS along with a CMV pentamer protein antigen to create a single vial lyophilized vaccine for reconstitution that showed comparable immune responses in vivo in C57BL/6 mice.

152

CONCLUDING REMARKS

There exists a constant need to make vaccines available in every part of the world, especially the developing and less-privileged countries. Thus, it is important to make these vaccines less-expensive and easily accessible particularly during pandemic situations and for prepandemic preparedness. Adjuvants have helped reduced number of doses required as well as antigen dose rendering in reduced costs; however, most of these adjuvants are also complex and expensive to produce and store. One of the potential ways to reduce costs of vaccines and adjuvants is to provide alternative simpler methods than the ones currently available enabling manufacturing and storage of these vaccines adjuvants in developing countries. Cold-chain storage and maintenance of these adjuvanted vaccines is another factor that adds to the cost of vaccines due to two-vial presentation. Making a single vial stable lyophilized vaccine could potentially help in eliminating cold-chain dependence providing another way to make adjuvanted vaccines more accessible across the world.

AS03 is an emulsion adjuvant containing squalene and α-tocopherol, manufactured using high pressure homogenization and microfluidization. AS03 has shown higher immune response compared to emulsions without α-tocopherol in preclinical and clinical models. Thus, we created novel SE-AS with components of AS03 to first demonstrate immune responses significantly higher than unadjuvanted antigens as well as not significantly different to AS03 using inactivated QIV. Then, we reported significantly higher humoral response to CMV pentamer antigen, especially after the 2nd immunization, using SE-AS compared to SEA160 (no α-tocopherol) control. We finally showed successful lyophilization of adjuvant with the CMV antigen and demonstrated that lyophilized single vial vaccine yields comparable immune responses to liquid vaccine

153 mixed bed-side. We have reported results that support our initial hypotheses, and created a novel emulsion adjuvant using an alternative, simple, and low-energy process that is comparable in immune responses to AS03, better in humoral immune responses to SEA160 and that be lyophilized with an antigen to yield a single vial vaccine showing comparable immune response in vivo to liquid vaccine.

To conclude, this work implies that using alternative and simple methods of emulsification can result in novel emulsion adjuvants that can be used in different aspects of adjuvant research due to the ease of formulation. The simplicity of the formulation process makes it an ideal platform to build next generation adjuvants and support novel adjuvant discovery and research.

154

REFERENCES

1. Rappuoli R, Dormitzer PR. Influenza: options to improve pandemic preparation. Science. 2012;336(6088):1531-3. 2. Kristensen DD, Lorenson T, Bartholomew K, Villadiego S. Can thermostable vaccines help address cold-chain challenges? Results from stakeholder interviews in six low- and middle-income countries. Vaccine. 2016;34(7):899-904. 3. Iyer V, Cayatte C, Marshall JD, Sun J, Schneider-Ohrum K, Maynard SK, et al. Feasibility of Freeze-Drying Oil-in-Water Emulsion Adjuvants and Subunit Proteins to Enable Single-Vial Vaccine Drug Products. J Pharm Sci. 2017;106(6):1490-8. 4. Orr MT, Kramer RM, Barnes L, Dowling QM, Desbien AL, Beebe EA, et al. Elimination of the cold-chain dependence of a nanoemulsion adjuvanted vaccine against tuberculosis by lyophilization. Journal of Controlled Release. 2014;177(Supplement C):20-6. 5. Shah RR, Hassett KJ, Brito LA. Overview of Vaccine Adjuvants: Introduction, History, and Current Status. In: Fox CB, editor. Vaccine Adjuvants: Methods and Protocols. New York, NY: Springer New York; 2017. p. 1-13. 6. Plotkin SA, Plotkin SL. The development of vaccines: how the past led to the future. Nat Rev Microbiol. 2011;9(12):889-93. 7. O'Hagan DT, Fox CB. New generation adjuvants--from empiricism to rational design. Vaccine. 2015;33 Suppl 2:B14-20. 8. Coffman RL, Sher A, Seder RA. Vaccine Adjuvants: Putting Innate Immunity to Work. Immunity. 2010;33(4):492-503. 9. O’Hagan DT, Friedland LR, Hanon E, Didierlaurent AM. Towards an evidence based approach for the development of adjuvanted vaccines. Current Opinion in Immunology. 2017;47:93-102. 10. Schijns VEJC, Lavelle EC. Trends in vaccine adjuvants. Expert Review of Vaccines. 2011;10(4):539-50. 11. Brito LA, Malyala P, O’Hagan DT. Vaccine adjuvant formulations: A pharmaceutical perspective. Seminars in Immunology. 2013;25(2):130-45. 12. Ennio DG, Elaine T, Rino R. Alum adjuvanticity: Unraveling a century old mystery. European Journal of Immunology. 2008;38(8):2068-71. 13. Janeway CA. Approaching the Asymptote? Evolution and Revolution in Immunology. Cold Spring Harbor Symposia on Quantitative Biology. 1989;54:1-13. 14. Schijns VEJC. 1 - Unraveling “the immunologist's dirty little secret”. In: Schijns VEJC, O'Hagan DT, editors. Immunopotentiators in Modern Vaccines. London: Academic Press; 2006. p. 1-16. 15. Reed SG, Orr MT, Fox CB. Key roles of adjuvants in modern vaccines. Nature medicine. 2013;19(12):1597-608. 16. T. GA, G. PC, Hilda W, U. W. Immunological notes. XVII–XXIV. The Journal of Pathology and Bacteriology. 1926;29(1):31-40. 17. Malyala P, Laera D, Cianetti S, Bufali S, Aggravi M, Ianni E, et al. The Preparation and Physicochemical Characterization of Aluminum Hydroxide/TLR7a, a Novel Vaccine Adjuvant Comprising a Small Molecule Adsorbed to Aluminum Hydroxide. Journal of Pharmaceutical Sciences. 2018;107(6):1577-85.

155

18. Kool M, Soullié T, van Nimwegen M, Willart MAM, Muskens F, Jung S, et al. Alum adjuvant boosts adaptive immunity by inducing uric acid and activating inflammatory dendritic cells. The Journal of Experimental Medicine. 2008;205(4):869-82. 19. Eisenbarth SC, Colegio OR, O’Connor W, Sutterwala FS, Flavell RA. Crucial role for the Nalp3 inflammasome in the immunostimulatory properties of aluminium adjuvants. Nature. 2008;453:1122. 20. Martinon F, Pétrilli V, Mayor A, Tardivel A, Tschopp J. Gout-associated uric acid crystals activate the NALP3 inflammasome. Nature. 2006;440:237. 21. Garcon N, Van Mechelen M. Recent clinical experience with vaccines using MPL- and QS-21-containing adjuvant systems. Expert Rev Vaccines. 2011;10(4):471-86. 22. Hassett KJ, Vance DJ, Jain NK, Sahni N, Rabia LA, Cousins MC, et al. Glassy-State Stabilization of a Dominant Negative Inhibitor Anthrax Vaccine Containing Aluminum Hydroxide and Glycopyranoside Lipid A Adjuvants. Journal of Pharmaceutical Sciences. 2015;104(2):627-39. 23. Clausi AL, Morin A, Carpenter JF, Randolph TW. Influence of protein conformation and adjuvant aggregation on the effectiveness of aluminum hydroxide adjuvant in a model alkaline phosphatase vaccine. Journal of Pharmaceutical Sciences. 2009;98(1):114-21. 24. Wilkins AL, Kazmin D, Napolitani G, Clutterbuck EA, Pulendran B, Siegrist C-A, et al. AS03- and MF59-Adjuvanted Influenza Vaccines in Children. Frontiers in Immunology. 2017;8:1760. 25. Christensen D, Korsholm KS, Andersen P, Agger EM. Cationic liposomes as vaccine adjuvants. Expert Review of Vaccines. 2011;10(4):513-21. 26. van Dissel JT, Joosten SA, Hoff ST, Soonawala D, Prins C, Hokey DA, et al. A novel liposomal adjuvant system, CAF01, promotes long-lived Mycobacterium tuberculosis-specific T-cell responses in human. Vaccine. 2014;32(52):7098-107. 27. Thakur A, Ingvarsson PT, Schmidt ST, Rose F, Andersen P, Christensen D, et al. Immunological and physical evaluation of the multistage tuberculosis subunit vaccine candidate H56/CAF01 formulated as a spray-dried powder. Vaccine. 2018;36(23):3331-9. 28. Drane D, Gittleson C, Boyle J, Maraskovsky E. ISCOMATRIX™ adjuvant for prophylactic and therapeutic vaccines. Expert Review of Vaccines. 2007;6(5):761- 72. 29. Didierlaurent AM, Laupèze B, Di Pasquale A, Hergli N, Collignon C, Garçon N. Adjuvant system AS01: helping to overcome the challenges of modern vaccines. Expert Rev Vaccines. 2017;16(1):55-63. 30. Lecrenier N, Beukelaers P, Colindres R, Curran D, De Kesel C, De Saegher J-P, et al. Development of adjuvanted recombinant zoster vaccine and its implications for shingles prevention. Expert Review of Vaccines. 2018:1-16. 31. Medzhitov R, Preston-Hurlburt P, Janeway CA. A human homologue of the Drosophila Toll protein signals activation of adaptive immunity. Nature. 1997;388(6640):394-7. 32. Dowling JK, Mansell A. Toll-like receptors: the swiss army knife of immunity and vaccine development. Clin Transl Immunology. 2016;5(5):e85.

156

33. Wang Y, Su L, Morin MD, Jones BT, Whitby LR, Surakattula MMRP, et al. TLR4/MD-2 activation by a synthetic agonist with no similarity to LPS. Proc Natl Acad Sci U S A. 2016;113(7):E884-93. 34. Testerman TL, Gerster JF, Imbertson LM, Reiter MJ, Miller RL, Gibson SJ, et al. Cytokine induction by the immunomodulators imiquimod and S-27609. J Leukoc Biol. 1995;58(3):365-72. 35. Wagner TL, Ahonen CL, Couture AM, Gibson SJ, Miller RL, Smith RM, et al. Modulation of TH1 and TH2 cytokine production with the immune response modifiers, R-848 and imiquimod. Cellular immunology. 1999;191(1):10-9. 36. Honegr J, Soukup O, Doležal R, Malinak D, Penhaker M, Prymula R, et al. Structural Properties of Potential Synthetic Vaccine Adjuvants - TLR Agonists. Curr Med Chem. 2015;22(29):3306-25. 37. Wu TY-H. Strategies for designing synthetic immune agonists. Immunology. 2016;148(4):315-25. 38. Brito LA, O'Hagan DT. Designing and building the next generation of improved vaccine adjuvants. Journal of Controlled Release. 2014;190:563-79. 39. Lodaya RN, Brito LA, Wu TYH, Miller AT, Otten GR, Singh M, et al. Stable Nanoemulsions for the Delivery of Small Molecule Immune Potentiators. Journal of Pharmaceutical Sciences. 40. O'Hagan DT. MF59 is a safe and potent vaccine adjuvant that enhances protection against influenza virus infection. Expert Rev Vaccines. 2007;6(5):699-710. 41. Stewart-Tull DE. Freund's complete and incomplete adjuvants, preparation, and quality control standards for experimental laboratory animals use. Methods in molecular biology (Clifton, NJ). 2010;626:59-72. 42. Stills HF, Jr. Adjuvants and antibody production: dispelling the myths associated with Freund's complete and other adjuvants. ILAR journal. 2005;46(3):280-93. 43. Smith JW, Fletcher WB, Peters M, Westwood M, Perkins FJ. Response to influenza vaccine in adjuvant 65-4. J Hyg (Lond). 1975;74(2):251-9. 44. Ribi E, Cantrell JL, Takayama K, Qureshi N, Peterson J, Ribi HO. Lipid A and immunotherapy. Rev Infect Dis. 1984;6(4):567-72. 45. Ribi E, Meyer TJ, Azuma I, Parker R, Brehmer W. Biologically active components from mycobacterial cell walls. IV. Protection of mice against aerosol infection with virulent mycobacterium tuberculosis. Cellular immunology. 1975;16(1):1-10. 46. Ott G, Barchfeld GL, Chernoff D, Radhakrishnan R, van Hoogevest P, Van Nest G. MF59 Design and Evaluation of a Safe and Potent Adjuvant for Human Vaccines. In: Powell MF, Newman MJ, editors. Vaccine Design: The Subunit and Adjuvant Approach. Boston, MA: Springer US; 1995. p. 277-96. 47. Fox CB. Squalene emulsions for parenteral vaccine and drug delivery. Molecules (Basel, Switzerland). 2009;14(9):3286-312. 48. Seubert A, Calabro S, Santini L, Galli B, Genovese A, Valentini S, et al. Adjuvanticity of the oil-in-water emulsion MF59 is independent of Nlrp3 inflammasome but requires the adaptor protein MyD88. Proc Natl Acad Sci U S A. 2011;108(27):11169-74. 49. O’Hagan DT, Ott GS, De Gregorio E, Seubert A. The mechanism of action of MF59 – An innately attractive adjuvant formulation. Vaccine. 2012;30(29):4341-8.

157

50. Seubert A, Monaci E, Pizza M, O’Hagan DT, Wack A. The Adjuvants Aluminum Hydroxide and MF59 Induce Monocyte and Granulocyte Chemoattractants and Enhance Monocyte Differentiation toward Dendritic Cells. The Journal of Immunology. 2008;180(8):5402-12. 51. Huang CH, Huang CY, Huang MH. Unsaturated Squalene Content in Emulsion Vaccine Adjuvants Plays a Crucial Role in ROS-Mediated Antigen Uptake and Cellular Immunity. Molecular pharmaceutics. 2018;15(2):420-9. 52. (WHO) WHO. Influenza (Seasonal) - Factsheet 2018. Available from: https://www.who.int/news-room/fact-sheets/detail/influenza-(seasonal). 53. Carmona A, Omeñaca F, Tejedor JC, Merino JM, Vaman T, Dieussaert I, et al. Immunogenicity and safety of AS03-adjuvanted 2009 influenza A H1N1 vaccine in children 6–35 months. Vaccine. 2010;28(36):5837-44. 54. Clark TW, Pareek M, Hoschler K, Dillon H, Nicholson KG, Groth N, et al. Trial of 2009 influenza A (H1N1) monovalent MF59-adjuvanted vaccine. N Engl J Med. 2009;361(25):2424-35. 55. Tetsutani K, Ishii KJ. Adjuvants in influenza vaccines. Vaccine. 2012;30(52):7658- 61. 56. Vogel FR, Caillet C, Kusters IC, Haensler J. Emulsion-based adjuvants for influenza vaccines. Expert Review of Vaccines. 2009;8(4):483-92. 57. Perez DR. Public health and biosecurity. H5N1 debates: hung up on the wrong questions. Science. 2012;335(6070):799-801. 58. Kawaoka Y. H5N1: Flu transmission work is urgent. Nature. 2012;482(7384):155. 59. Haensler J. Manufacture of Oil-in-Water Emulsion Adjuvants. Methods in molecular biology (Clifton, NJ). 2017;1494:165-80. 60. Fox CB, Huynh C, O'Hara MK, Onu A. Technology transfer of oil-in-water emulsion adjuvant manufacturing for pandemic influenza vaccine production in Romania. Vaccine. 2013;31(12):1633-40. 61. Stavaru C, Onu A, Lupulescu E, Tucureanu C, Rasid O, Vlase E, et al. Technology transfer of oil-in-water emulsion adjuvant manufacturing for pandemic influenza vaccine production in Romania: Preclinical evaluation of split virion inactivated H5N1 vaccine with adjuvant. Human vaccines & immunotherapeutics. 2016;12(4):1009-26. 62. Shah RR, Dodd S, Schaefer M, Ugozzoli M, Singh M, Otten GR, et al. The Development of Self-Emulsifying Oil-in-Water Emulsion Adjuvant and an Evaluation of the Impact of Droplet Size on Performance. Journal of Pharmaceutical Sciences. 2015;104(4):1352-61. 63. Gursoy RN, Benita S. Self-emulsifying drug delivery systems (SEDDS) for improved oral delivery of lipophilic drugs. Biomedicine & pharmacotherapy = Biomedecine & pharmacotherapie. 2004;58(3):173-82. 64. Staats HF, Burkhart DJ. Softness makes it better. Nature Materials. 2018;17:113. 65. Meydani SN, Barklund MP, Liu S, Meydani M, Miller RA, Cannon JG, et al. Vitamin E supplementation enhances cell-mediated immunity in healthy elderly subjects. The American journal of clinical nutrition. 1990;52(3):557-63. 66. Quintilio W, de Freitas FA, Rodriguez D, Kubrusly FS, Yourtov D, Miyaki C, et al. Vitamins as influenza vaccine adjuvant components. Arch Virol. 2016;161(10):2787- 95.

158

67. Stephensen CB. Vitamin A, infection, and immune function. Annual review of nutrition. 2001;21:167-92. 68. Dias WO, Horton DSPQ, Cainelli Gebara VCB, Furuyama N, Risoléo L, Ferreira VRF, et al. Vitamin A as adjuvant to the mouse immune response to pertussis, tetanus and diphtheria vaccines. Biotechnology Letters. 2002;24(18):1515-8. 69. Yu M, Vajdy M. A novel retinoic acid, catechin hydrate and mustard oil-based emulsion for enhanced cytokine and antibody responses against multiple strains of HIV-1 following mucosal and systemic vaccinations. Vaccine. 2011;29(13):2429-36. 70. Vajdy M. Immunomodulatory properties of vitamins, flavonoids and plant oils and their potential as vaccine adjuvants and delivery systems. Expert opinion on biological therapy. 2011;11(11):1501-13. 71. Mora JR, Iwata M, von Andrian UH. Vitamin effects on the immune system: vitamins A and D take centre stage. Nature Reviews Immunology. 2008;8:685. 72. Ivanov AP, Dragunsky EM, Chumakov KM. 1,25-dihydroxyvitamin d3 enhances systemic and mucosal immune responses to inactivated poliovirus vaccine in mice. The Journal of infectious diseases. 2006;193(4):598-600. 73. Van der Stede Y, Cox E, Van den broeck W, Goddeeris BM. Enhanced induction of the IgA response in pigs by calcitriol after intramuscular immunization. Vaccine. 2001;19(15-16):1870-8. 74. Kjer-Nielsen L, Patel O, Corbett AJ, Le Nours J, Meehan B, Liu L, et al. MR1 presents microbial vitamin B metabolites to MAIT cells. Nature. 2012;491(7426):717-23. 75. Evans HM, Bishop KS. ON THE EXISTENCE OF A HITHERTO UNRECOGNIZED DIETARY FACTOR ESSENTIAL FOR REPRODUCTION. Science. 1922;56(1458):650-1. 76. Coquette A, Vray B, Vanderpas J. Role of vitamin E in the protection of the resident macrophage membrane against oxidative damage. Archives internationales de physiologie et de biochimie. 1986;94(5):S29-34. 77. Douglas CE, Chan AC, Choy PC. Vitamin E inhibits platelet phospholipase A2. Biochim Biophys Acta. 1986;876(3):639-45. 78. Marko MG, Pang HJ, Ren Z, Azzi A, Huber BT, Bunnell SC, et al. Vitamin E reverses impaired linker for activation of T cells activation in T cells from aged C57BL/6 mice. The Journal of nutrition. 2009;139(6):1192-7. 79. Burns EA. Effects of aging on immune function. J Nutr Health Aging. 2004;8(1):9- 18. 80. Miller RA. The aging immune system: primer and prospectus. Science. 1996;273(5271):70-4. 81. Han SN, Meydani SN. Vitamin E and infectious diseases in the aged. The Proceedings of the Nutrition Society. 1999;58(3):697-705. 82. Hara M, Tanaka K, Hirota Y. Immune response to influenza vaccine in healthy adults and the elderly: association with nutritional status. Vaccine. 2005;23(12):1457-63. 83. Gardner EM, Bernstein ED, Popoff KA, Abrutyn E, Gross P, Murasko DM. Immune response to influenza vaccine in healthy elderly: lack of association with plasma beta- carotene, retinol, alpha-tocopherol, or zinc. Mechanisms of ageing and development. 2000;117(1-3):29-45.

159

84. Tengerdy RP, Ameghino E, Riemann H. Serological responses of rams to a Brucella ovis-vitamin E adjuvant vaccine. Vaccine. 1991;9(4):273-6. 85. Tengerdy RP, Lacetera NG. Vitamin E adjuvant formulations in mice. Vaccine. 1991;9(3):204-6. 86. Tengerdy RP, Meyer DL, Lauerman LH, Lueker DC, Nockels CF. Vitamin E- enhanced humoral antibody response to Clostridium perfringens type D in sheep. Br Vet J. 1983;139(2):147-52. 87. Franchini A, Bertuzzi S, Tosarelli C, Manfreda G. Vitamin E in viral inactivated vaccines. Poult Sci. 1995;74(4):666-71. 88. Franchini A, Canti M, Manfreda G, Bertuzzi S, Asdrubali G, Franciosi C. Vitamin E as adjuvant in emulsified vaccine for chicks. Poult Sci. 1991;70(8):1709-15. 89. Afzal M, Tengerdy RP, Ellis RP, Kimberling CV, Morris CJ. Protection of rams against epididymitis by a Brucella ovis-vitamin E adjuvant vaccine. Veterinary immunology and immunopathology. 1984;7(3-4):293-304. 90. Morel S, Didierlaurent A, Bourguignon P, Delhaye S, Baras B, Jacob V, et al. Adjuvant System AS03 containing alpha-tocopherol modulates innate immune response and leads to improved adaptive immunity. Vaccine. 2011;29(13):2461-73. 91. Jackson LA, Campbell JD, Frey SE, et al. Effect of varying doses of a monovalent h7n9 influenza vaccine with and without as03 and mf59 adjuvants on immune response: A randomized clinical trial. JAMA. 2015;314(3):237-46. 92. Givord C, Welsby I, Detienne S, Thomas S, Assabban A, Lima Silva V, et al. Activation of the endoplasmic reticulum stress sensor IRE1α by the vaccine adjuvant AS03 contributes to its immunostimulatory properties. npj Vaccines. 2018;3(1):20. 93. Orr MT, Fox CB. AS03 stresses out macrophages: Commentary on 'Activation of the endoplasmic reticulum stress sensor IRE1α by the vaccine adjuvant AS03 contributes to its immunostimulatory properties'. NPJ Vaccines. 2018;3:27. 94. Abla MJ, Banga AK. Formulation of tocopherol nanocarriers and in vitro delivery into human skin. International journal of cosmetic science. 2014;36(3):239-46. 95. Bouchemal K, Briancon S, Perrier E, Fessi H. Nano-emulsion formulation using spontaneous emulsification: solvent, oil and surfactant optimisation. International journal of pharmaceutics. 2004;280(1-2):241-51. 96. Julianto T, Yuen KH, Noor AM. Improved bioavailability of vitamin E with a self emulsifying formulation. International journal of pharmaceutics. 2000;200(1):53-7. 97. Bateman NE, Uccellini DA. Effect of formulation on the bioavailability of retinol, D-alpha-tocopherol and riboflavine. The Journal of pharmacy and pharmacology. 1984;36(7):461-4. 98. Guo Y, Luo J, Tan S, Otieno BO, Zhang Z. The applications of Vitamin E TPGS in drug delivery. European journal of pharmaceutical sciences : official journal of the European Federation for Pharmaceutical Sciences. 2013;49(2):175-86. 99. M. Waghmode MK, A. Shruti. Pharmaceutical profile of alpha-tocopherol - A brief review. International Journal of Pharmaceutical and Chemical Sciences. 2012;1(3):17. 100. Garcon N, Vaughn DW, Didierlaurent AM. Development and evaluation of AS03, an Adjuvant System containing alpha-tocopherol and squalene in an oil-in-water emulsion. Expert Rev Vaccines. 2012;11(3):349-66. 101. Savage N. Logistics: Keeping cool. Nature. 2014;507(7490):S8-9.

160

102. Watson JT, Gayer M, Connolly MA. Epidemics after natural disasters. Emerg Infect Dis. 2007;13(1):1-5. 103. Nelson C, Froes P, Dyck AM, Chavarria J, Boda E, Coca A, et al. Monitoring temperatures in the vaccine cold chain in Bolivia. Vaccine. 2007;25(3):433-7. 104. Techathawat S, Varinsathien P, Rasdjarmrearnsook A, Tharmaphornpilas P. Exposure to heat and freezing in the vaccine cold chain in Thailand. Vaccine. 2007;25(7):1328-33. 105. Zhang J, Pritchard E, Hu X, Valentin T, Panilaitis B, Omenetto FG, et al. Stabilization of vaccines and antibiotics in silk and eliminating the cold chain. Proc Natl Acad Sci U S A. 2012;109(30):11981-6. 106. Douglas A, Dulal P, Ashfield R, Ritchie A, Kowal K, Russell S, et al. Thermostabilization of adenovirus-vectored vaccines, removing the need for continual cold-chain storage. 2018. 107. Shrivastava S, Shrivastava P, Ramasamy J. Encouraging real-time tracking of vaccines to reduce the risk of stock-outs: World Health Organization. Journal of Medical Society. 2018;32(3):239-40. 108. Kasper JC, Winter G, Friess W. Recent advances and further challenges in lyophilization. Eur J Pharm Biopharm. 2013;85(2):162-9. 109. Wang W. Lyophilization and development of solid protein pharmaceuticals. International journal of pharmaceutics. 2000;203(1-2):1-60. 110. Kramer RM, Archer MC, Orr MT, Dubois Cauwelaert N, Beebe EA, Huang P-WD, et al. Development of a thermostable nanoemulsion adjuvanted vaccine against tuberculosis using a design-of-experiments approach. Int J Nanomedicine. 2018;13:3689-711. 111. Garmise RJ, Mar K, Crowder TM, Hwang CR, Ferriter M, Huang J, et al. Formulation of a dry powder influenza vaccine for nasal delivery. AAPS PharmSciTech. 2006;7(1):E131-E7. 112. LiCalsi C, Christensen T, Bennett JV, Phillips E, Witham C. Dry powder inhalation as a potential delivery method for vaccines. Vaccine. 1999;17(13-14):1796-803. 113. Sou T, Meeusen EN, de Veer M, Morton DAV, Kaminskas LM, McIntosh MP. New developments in dry powder pulmonary vaccine delivery. Trends in Biotechnology. 2011;29(4):191-8. 114. Amorij J-P, 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(52):8707-17. 115. Saluja V, Amorij J, Kapteyn J, de Boer A, Frijlink H, Hinrichs W. A comparison between spray drying and spray freeze drying to produce an influenza subunit vaccine powder for inhalation. Journal of Controlled Release. 2010;144(2):127-33. 116. Abdul‐ Fattah AM, Kalonia DS, Pikal MJ. The challenge of drying method selection for protein pharmaceuticals: product quality implications. Journal of pharmaceutical sciences. 2007;96(8):1886-916. 117. Barnes VL, Fedor DM, Williams S, Dowling QM, Archer MC, Cloutier S, et al. Lyophilization of an Adjuvanted Mycobacterium tuberculosis Vaccine in a Single- Chamber Pharmaceutical Cartridge. AAPS PharmSciTech. 2017;18(6):2077-84.

161

118. Tang X, Pikal MJ. Design of freeze-drying processes for pharmaceuticals: practical advice. Pharm Res. 2004;21(2):191-200. 119. Chan MY, Dutill TS, Kramer RM. Lyophilization of Adjuvanted Vaccines: Methods for Formulation of a Thermostable Freeze-Dried Product. Methods in molecular biology (Clifton, NJ). 2017;1494:215-26. 120. Gupta S, Chavhan S, Sawant KK. Self-nanoemulsifying drug delivery system for adefovir dipivoxil: design, characterization, in vitro and ex vivo evaluation. Colloids and surfaces A: physicochemical and engineering aspects. 2011;392(1):145-55. 121. Ingvarsson PT, Yang M, Nielsen HM, Rantanen J, Foged C. Stabilization of liposomes during drying. Expert Opin Drug Deliv. 2011;8(3):375-88. 122. Morais AR, Alencar Edo N, Xavier Junior FH, de Oliveira CM, Marcelino HR, Barratt G, et al. Freeze-drying of emulsified systems: A review. International journal of pharmaceutics. 2016;503(1-2):102-14. 123. Baras B, Stittelaar KJ, Kuiken T, Jacob V, Bernhard R, Giannini S, et al. Longevity of the protective immune response induced after vaccination with one or two doses of AS03A-adjuvanted split H5N1 vaccine in ferrets. Vaccine. 2011;29(11):2092-9. 124. Moris P, van der Most R, Leroux-Roels I, Clement F, Dramé M, Hanon E, et al. H5N1 influenza vaccine formulated with AS03 A induces strong cross-reactive and polyfunctional CD4 T-cell responses. J Clin Immunol. 2011;31(3):443-54. 125. Yam KK, Gupta J, Winter K, Allen E, Brewer A, Beaulieu É, et al. AS03-Adjuvanted, Very-Low-Dose Influenza Vaccines Induce Distinctive Immune Responses Compared to Unadjuvanted High-Dose Vaccines in BALB/c Mice. Front Immunol. 2015;6:207. 126. Bishai D, McQuestion M, Chaudhry R, Wigton A. The costs of scaling up vaccination in the world's poorest countries. Health affairs (Project Hope). 2006;25(2):348-56. 127. Singh B, Bandopadhyay S, Kapil R, Singh R, Katare O. Self-emulsifying drug delivery systems (SEDDS): formulation development, characterization, and applications. Critical reviews in therapeutic drug carrier systems. 2009;26(5):427- 521. 128. Salk JE, Contakos M, Laurent AM, Sorensen M, Rapalski AJ, Simmons IH, et al. Use of adjuvants in studies on influenza immunization. III. Degree of persistence of antibody in human subjects two years after vaccination. Journal of the American Medical Association. 1953;151(14):1169-75. 129. Serum cross-reactive antibody response to a novel influenza A (H1N1) virus after vaccination with seasonal influenza vaccine. MMWR Morbidity and mortality weekly report. 2009;58(19):521-4. 130. Díez-Domingo J, Baldó J-M, Planelles-Catarino MV, Garcés-Sánchez M, Ubeda I, Jubert-Rosich A, et al. Phase II, randomized, open, controlled study of AS03- adjuvanted H5N1 pre-pandemic influenza vaccine in children aged 3 to 9 years: follow-up of safety and immunogenicity persistence at 24 months post-vaccination. Influenza and other respiratory viruses. 2015;9(2):68-77. 131. Baudner BC, Ronconi V, Casini D, Tortoli M, Kazzaz J, Singh M, et al. MF59 Emulsion Is an Effective Delivery System for a Synthetic TLR4 Agonist (E6020). Pharmaceutical Research. 2009;26(6):1477-85. 132. Lidgate DM, Trattner T, Shultz RM, Maskiewicz R. Sterile Filtration of a Parenteral Emulsion. Pharmaceutical Research. 1992;9(7):860-3.

162

133. O'Hagan DT, Fox CB. Are we entering a new age for human vaccine adjuvants? Expert Rev Vaccines. 2015;14(7):909-11. 134. Chandramouli S, Malito E, Nguyen T, Luisi K, Donnarumma D, Xing Y, et al. Structural basis for potent antibody-mediated neutralization of human cytomegalovirus. Science immunology. 2017;2(12). 135. Hahn G, Revello MG, Patrone M, Percivalle E, Campanini G, Sarasini A, et al. Human cytomegalovirus UL131-128 genes are indispensable for virus growth in endothelial cells and virus transfer to leukocytes. Journal of virology. 2004;78(18):10023-33. 136. CDC. Cytomegalovirus and congenital CMV infection. 2018. 137. Bachmann MF, Wolint P, Schwarz K, Jager P, Oxenius A. Functional properties and lineage relationship of CD8+ T cell subsets identified by expression of IL-7 receptor alpha and CD62L. Journal of immunology (Baltimore, Md : 1950). 2005;175(7):4686-96. 138. Ott G, Radhakrishnan R, Fang J-H, Hora M. The adjuvant MF59: a 10-year perspective. Springer; 2000. p. 211-28. 139. Wolfson LJ, Gasse F, Lee-Martin SP, Lydon P, Magan A, Tibouti A, et al. Estimating the costs of achieving the WHO-UNICEF Global Immunization Vision and Strategy, 2006-2015. Bulletin of the World Health Organization. 2008;86(1):27-39.

163