Altered esophageal epithelial differentiation in EoE

and regulation of IL-1 cytokine responses by

chromatin

A dissertation submitted to

the Graduate School of the University of Cincinnati

in partial fulfillment of the requirements for the degree of

Doctor of Philosophy (Ph.D.)

In the Immunology Graduate Program

of the College of Medicine

2017

by Jared Bernard Travers

B.S., Purdue University, 2011

Advisory Committee:

Marc E. Rothenberg, M.D., Ph.D. (Chair)

Simon Hogan, Ph.D.

Susanne Wells, Ph.D.

Raphael Kopan, Ph.D.

George Deepe, M.D.

i

ABSTRACT

The molecular and cellular etiology of eosinophilic esophagitis (EoE), an emerging tissue-specific allergic disease, involves dysregulated gene expression in esophageal epithelial cells. We assessed the expression of esophagus-specific genes in esophageal biopsy specimens from patients with EoE. We identified that 39% of esophagus-specific genes have altered expression in EoE (termed Eso-EoE genes).

Notably, Eso-EoE genes are enriched for multiple IL-1 family members, proteases, and protease inhibitors. Additionally, we found dysregulated esophageal epithelial differentiation in patients with EoE. These findings demonstrated a profound loss of esophageal tissue identify in EoE and suggested that esophagus-specific, imbalanced activity of proteases and IL-1 family members is integral to the pathogenesis of EoE.

We then assessed the esophageal expression of the IL-1 family member IL-33, a pro- allergic, chromatin-binding alarmin. IL-33 protein was markedly overexpressed within the nuclei of a subpopulation of basal layer esophageal epithelial cells in patients with active EoE compared to control individuals. These IL-33–positive basal epithelial cells expressed markers consistent of being undifferentiated, quiescent epithelial progenitor cells. Because strong nuclear expression of IL-33 was observed, we then aimed to uncover the characteristics and functions of the nuclear localization and chromatin binding of IL-33. Epithelial overexpression of IL-33 did not change gene expression, as assessed by RNA-sequencing. Following cellular necrosis, wild-type (WT) IL-33 demonstrated decreased extracellular release as high molecular weight species with increased intracellular retention compared to truncated IL-33. Time-lapse microscopy revealed retention of WT IL-33, but not truncated IL-33, after membrane dissolution. WT

ii

IL-33 displayed a slow, linear release over time compared with non-chromatin binding

IL-33 and IL-1α. Direct interaction between released WT IL-33 and histone H2B was detected by co-immunoprecipitation. Notably, histones and IL-33 synergistically activated ST2-mediated responses. We propose that chromatin binding finely regulates the extracellular activity of IL-33 by limiting its ability to be released while simultaneously increasing both the duration of its release and ST2-mediated bioactivity. Together, our studies further our understanding of both the role of esophagus-specific genes, notably

IL-1 family members, in pathogenesis of EoE and the unique regulation of the extracellular activity of IL-1 cytokines.

iii

iv

ACKNOWLEDGEMENTS

I undoubtedly would not have been able to perform this work without the support and assistance of too many people to name. First and foremost, I cannot give enough thanks to my family, including my parents Joan and Jeff, my brother Jeffrey, and my sisters Justine, Juliet, and Jessica, for all of their love and support. I am also very grateful and appreciative of the excellent mentorship and guidance I have received from my thesis advisor, Dr. Marc Rothenberg. I would also like to give special thanks to Dr.

Mark Rochman, Dr. Rahul D’Mello, Cora Miracle, and everyone else in the Rothenberg laboratory for all of their assistance and support over the years. I would also like to thank my committee members (Dr. George Deepe, Dr. Simon Hogan, Dr. Susanne

Wells, and Dr. Rafi Kopan) and my senior physician-scientist mentor, Dr. John Harley, for all of their feedback, advice, and support.

I would also like to thank everyone in the Medical Scientist Training Program, the

Immunology Graduate Program, and the Allergy & Immunology family. I feel very privileged to have trained in such an excellent environment. I would especially like to thank my MSTP classmates Zubin, Aynara, Jon, Caitlin, and Arya, who have been there for me every step of the way since we all began our MSTP training.

Finally, I would also like to thank my previous advisors I have had from earlier stages of my training, including Dr. Mark Kaplan at Indiana University School of Medicine and Dr.

Jean-Christophe Rochet at Purdue University, for helping to foster my initial interest in research and medicine.

v

TABLE OF CONTENTS

ABSTRACT ...... ii

ACKNOWLEDGEMENTS ...... v

TABLE OF CONTENTS ...... vii

LIST OF FIGURES AND TABLES ...... x

LIST OF ABBREVIATIONS ...... xiii

PUBLICATIONS & CONFERENCE PROCEEDINGS ARISING FROM THIS WORK ... xvi

CHAPTER 1: INTRODUCTION ...... 1

Allergic Diseases ...... 1

Introduction ...... 1

Pathogenesis of Allergic Diseases ...... 2

Esophageal Epithelial Cells ...... 3

Introduction ...... 3

Differentiation ...... 4

Keratins ...... 5

Epithelial Stem Cells ...... 6

Eosinophilic Esophagitis ...... 7

Introduction ...... 7

Pathogenesis ...... 8

Epithelial Changes in EoE ...... 9

Treatment ...... 10

vi

Genetic Basis ...... 10

Chromatin ...... 13

Chromatin Structure ...... 13

Nucleosome Acidic Patch-binding Proteins ...... 14

Alarmins ...... 14

HMGB1 ...... 15

IL-1α ...... 15

IL-33 Introduction and Expression...... 16

IL-33 Nuclear Localization and Chromatin Binding ...... 17

IL-33 Release Mechanisms ...... 19

IL-33 Receptor ...... 20

IL-33 Post-Translational Modification ...... 21

IL-33 Involvement in EoE Pathogenesis ...... 22

Summary ...... 23

References ...... 24

CHAPTER 2: PROFOUND LOSS OF ESOPHAGEAL TISSUE DIFFERENTIATION IN

EOSINOPHILIC ESOPHAGITIS ...... 47

Abstract ...... 49

Introduction ...... 53

Materials and Methods ...... 56

Results ...... 60

Discussion ...... 68

vii

References ...... 72

Figures and Figure Legends ...... 78

Supplementary Figure and Figure Legend ...... 87

CHAPTER 3: IL-33 IS INDUCED IN UNDIFFERENTIATED, QUIESCENT

ESOPHAGEAL EPITHELIAL CELLS IN EOSINOPHILIC ESOPHAGITIS ...... 88

Abstract ...... 89

Introduction ...... 90

Results ...... 91

Discussion ...... 93

Materials and Methods ...... 95

References ...... 99

Figures and Figure Legends ...... 103

CHAPTER 4: CHROMATIN REGULATES THE MAGNITUDE AND KINETICS OF IL-33

RELEASE AND BIOACTIVITY ...... 107

Abstract ...... 108

Introduction ...... 109

Results ...... 112

Discussion ...... 118

Materials and Methods ...... 122

References ...... 133

Figures and Figure Legends ...... 141

viii

Supplementary Figure and Figure Legend ...... 152

CHAPTER 5: GENERAL DISCUSSION AND SUMMARY ...... 153

Introduction ...... 153

Dysregulation of Epithelial Differentiation in EoE ...... 155

Functions of IL-33-expressing Basal Layer Esophageal Epithelial Cells in EoE ...... 158

Regulation of IL-33 Expression in EoE ...... 160

Potential Roles of IL-33 in the Pathogenesis of EoE ...... 164

Comparison of IL-33 to other Nuclear Alarmins ...... 167

Chromatin-mediated Regulation of IL-33-ST2 bioactivity ...... 168

Reflections and Conclusions ...... 171

References ...... 172

Figures and Figure Legends ...... 182

ix

LIST OF FIGURES AND TABLES

CHAPTER 1: INTRODUCTION

Figure 1.1: Pathogenesis of eosinophilic esophagitis (EoE) ...... 43

Figure 1.2: IL-33 protein structure...... 44

Figure 1.3: Interaction of alarmins with chromatin ...... 45

Table 1.1: Post-translational modifications and alternate forms of IL-33Chromatin ... 46

CHAPTER 2: PROFOUND LOSS OF ESOPHAGEAL TISSUE DIFFERENTIATION IN

EOSINOPHILIC ESOPHAGITIS

Figure 2.1: Altered transcription of esophagus-specific genes in EoE ...... 78

Figure 2.2: Functional enrichment analysis of esophagus-specific Eso-EoE genes

altered in EoE ...... 79

Figure 2.3: Effect of differentiation on expression of esophagus-specific Eso-EoE

genes ...... 80

Figure 2.4: Effect of IL-13 treatment on expression of esophagus-specific Eso-EoE

genes at the ALI...... 81

Figure 2.5: Protein expression, protease activity, and expression of differentiation

markers in esophageal biopsy specimens of patients with EoE ...... 83

Figure 2.6: Model of esophagus-specific allergic inflammation ...... 84

Table 2.1: List of Eso-EoE genes with mutations identified by using WES in 33

unrelated patients with EoE ...... 85

Figure 2.E1: Effect of IL-1α replenishment on ALI ...... 87

x

CHAPTER 3: IL-33 IS INDUCED IN UNDIFFERENTIATED, QUIESCENT

ESOPHAGEAL EPITHELIAL CELLS IN EOSINOPHILIC ESOPHAGITIS

Figure 3.1: Subcellular localization of IL-33 in esophageal epithelial cells...... 103

Figure 3.2: Epithelium-specific protein expression in esophageal tissue ...... 104

Figure 3.3: Cell cycle and differentiation status in vivo in esophageal tissue ...... 105

Figure 3.4: Cell cycle status of IL-33-expressing esophageal cells ex vivo ...... 106

CHAPTER 4: CHROMATIN REGULATES THE MAGNITUDE AND KINETICS OF IL-33

RELEASE AND BIOACTIVITY

Figure 4.1: Effect of nuclear IL-33 on gene expression ...... 141

Figure 4.2: IL-33 association with chromatin ...... 143

Figure 4.3: Heterochromatic localization of IL-33 ...... 144

Figure 4.4: Dynamics of IL-33 binding to chromatin ...... 145

Figure 4.5: Effect of chromatin binding on IL-33 release during necrosis ...... 146

Figure 4.6: Chromatin binding dynamics regulate the kinetics of release of IL-33 ... 147

Figure 4.7: Characterization of released IL-33 species ...... 148

Figure 4.8: Proposed model of chromatin binding-mediated regulation of IL-33

extracellular activity ...... 150

Figure 4.S1: Nuclear localization of IL-33 in esophageal epithelial cells...... 152

CHAPTER 5: GENERAL DISCUSSION AND SUMMARY

Figure 5.1: Loss of esophageal epithelial differentiation in EoE ...... 182

Figure 5.2: Markers of IL-33-positive esophageal epithelial cells in EoE ...... 183

Figure 5.3: Route of immune cell infiltration into the esophagus in EoE ...... 184 xi

Figure 5.4: Regulation of the ST2-mediated bioactivity of IL-33 through post- translational modification ...... 185

Figure 5.5: Changes in epithelial differentiation in EoE and regulation of IL-33 ...... 186

Table 5.1: Comparison of active secretion of nuclear alarmins ...... 187

xii

LIST OF ABBREVIATIONS

53BP1 p53-binding protein 1 ALI Air-liquid interface ANOVA Analysis of variance ATP Adenosine triphosphate BMP Bone morphogenetic proteins BRCA1 Breast cancer 1 CAPN14 Calpain-14 CBD Chromatin binding domain CCL26 CC Chemokine ligand 26 CD Cluster of differentiation CRNN Cornulin CTD Connective tissue disorders DAMP Damage-associated molecular patterns DAPI 4',6-Diamidino-2-Phenylindole, Dilactate DC Dendritic cells DNA Deoxyribonucleic acid DSG1 Desmoglein-1 EDC Epidermal differentiation complex EG Eosinophilic gastritis EGFR Epidermal growth factor receptor EGID Eosinophilic gastrointestinal diseases ELISA Enzyme-linked immunosorbent assay EoE Eosinophilic esophagitis Eso-EoE Esophagus-specific genes with altered expression in EoE EtOH Ethanol FDA Food and Drug Administration FFPE Formalin-fixed, paraffin-embedded FLG Fillagrin FPKM Fragments per kilobase per million reads FRAP Fluorescence recovery after photobleaching GERD Gastroesophageal reflux disease GFP Green fluorescence protein GI Gastrointestinal GO Gene ontology analysis analysis GWAS Genome-wide association studies HMG High mobility group

xiii

HMW High molecular weight hpf High-power field Hsp90 Heat shock protein 90 IBF Impaired barrier function IFN Interferon Ig Immunoglobulin IL Interleukin IL-1RAcP IL-1 receptor accessory protein IL1RL1 IL-1 receptor-like 1 ILC2 Type 2 innate lymphoid cells iNKT Invariant natural killer T cells IRAK IL-1R-associated kinase IVL Involucrin KRT Keratin LANA Latency-associated nuclear antigen LMW Low molecular weight LPS Lipopolysaccharide LRRC31 Leucine-rich repeat-containing protein 31 MAPK Mitogen-activated protein kinases miR microRNA MMP Matrix metalloproteinase mRNA Messenger RNA MYD88 Myeloid differentiation primary response protein 88 NF-HEV Nuclear factor from high endothelial venules NFκB Nuclear factor kappa-light-chain-enhancer of activated B cells NK Natural killer NLS Nuclear localization sequence OVA Ovalbumin p75NTR P75 neurotrophin receptor PAR2 Proteinase-activated receptor 2 PBS Phosphate-buffered saline PCNA Proliferating cell nuclear antigen pH3 Phosphorylated histone-H3 PHTS PTEN hamartoma syndromes PMA Phorbol myristate acetate PTM Post-translational modifications RNA-seq RNA-sequencing ROI Region of interest RPKM Reads per kilobase per million reads

xiv

SERPIN Serine peptidase inhibitor SNP Single-nucleotide polymorphism SPINK Serine protease inhibitor Kazal-type SPRR Small proline-rich protein sST2 Soluble ST2 ST2 Suppressor of tumorigenicity 2 STAT6 Signal transducer and activator of transcription 6 TGF-β Transforming growth factor- β TGM Transglutaminase Th1 T helper type 1 Th2 T helper type 2 TLR Toll-like receptor TNF Tumor necrosis factor TRAF TNF receptor-associated factor Trunc Truncated TSLP Thymic stromal lymphopoietin WES Whole-exome sequencing WT Wild-type

xv

LIST OF PUBLICATIONS AND CONFERENCE PROCEEDINGS ARISING FROM THIS WORK

Published Reviews and Editorials: Travers J, Rochman M, Miracle C, Cohen J, Rothenberg ME. Interleukin-33: Linking impaired skin barrier function to esophageal allergic inflammation. Journal of Allergy and Clinical Immunology. 2016; 138(5):1381-1383. PMID: 27664378. Editorial.

Travers J, Rothenberg ME. Eosinophils in mucosal immune responses. Mucosal Immunology. 2015; 8(3):464-75. PMID: 25807184. Review.

Published Research Article: Rochman M*, Travers J*, Azouz N, Caldwell J, Kiran KC, Sherrill JC, Davis BP, Rymer J, Kaufman K, Rothenberg ME. Profound loss of esophageal tissue differentiation in eosinophilic esophagitis. Journal of Allergy and Clinical Immunology. 2017; pii:S0091- 6749(17)30036-2. Note: * denotes co-first author contributions.

Research Articles in Progress: Travers J, et al. IL-33 is induced in undifferentiated, quiescent esophageal epithelial cells in eosinophilic esophagitis. Not submitted, will submit May 2017.

Travers J, et al. Chromatin regulates the magnitude and kinetics of IL-33 release and bioactivity. Not submitted, will submit June 2017.

PRESENTATIONS Travers, J. et al. “Dynamic binding to chromatin regulates the extracellular release of interleukin-33 (IL-33).” Presented (poster) at the American Academy of Allergy, Asthma & Immunology Annual Meeting, Atlanta, GA, 2017.

xvi

Travers, J, et al. “IL-33 is selectively overexpressed by esophageal basal layer epithelial cells during allergic inflammation”. Presented (poster) at the American Academy of Allergy, Asthma & Immunology Annual Meeting, Los Angeles, CA, 2016.

Travers, J, et al. “IL-33 is selectively expressed by esophageal epithelial progenitor cells during allergic inflammation”. Presented (poster) at the 9th Biennial Symposium of the International Eosinophil Society, Chicago, IL, 2015.

xvii

CHAPTER 1: INTRODUCTION

1.1. Allergic Diseases

1.1.1. Introduction

Allergy involves inappropriately strong immune responses against allergens, or foreign antigens that would otherwise be harmless1. The number of allergens that trigger hypersensitivity reactions in individual patients can vary. Some patients can exhibit hypersensitivity reactions to single allergens, such as individual foods or medications, whereas atopic individuals react to a diverse array of allergens. Hypersensitivity responses can exhibit diverse symptomology, with variable kinetics, magnitude, and tissue distribution. Reactions range from severe, immediate responses (e.g. anaphylaxis) to delayed, mild responses (e.g. hives or pruritus). Some allergic patients have diverse chronic inflammatory disorders that result from sustained allergic responses to multiple allergens. These disorders can occur in diverse tissues, such as the lungs (asthma), skin (atopic dermatitis), GI tract (eosinophilic gastrointestinal diseases [EGID]), and nasal region (allergic rhinitis]. Although these disorders can be co-morbid, typically these disorders are tissue-specific for currently unclear reasons.

In the developed Western world, allergy has become a major health burden as its prevalence has reached almost epidemic proportions2,3. Currently, allergic diseases are commonly treated with glucocorticoid steroids. However, some patients have only partial or temporary responses to steroid treatment. Furthermore, patients who do successfully respond to steroids still often have to deal with dramatic side effects,

1 including obesity and risk for metabolic diseases such as type 2 diabetes4. Therefore, a strong need exists to develop new therapeutic agents in order to treat chronic allergic diseases. Strategically-designed, biologic therapeutic agents, such as antibodies directed against pathogenic disease-causing cytokines or receptors, are in development and clinical testing. Some progress has been made as a few of these therapies have shown clinical benefit for individual allergic diseases. For instance, the anti-IL-5 humanized antibodies mepolizumab and reslizumab recently received FDA approval for treatment of eosinophilic asthma5. However, currently is no FDA-approved treatment for some allergic diseases, such as eosinophilic esophagitis (EoE)6. There is still a need to develop better treatments, which will depend on better characterization of the cellular and molecular mechanisms of allergy and identification of novel biomarkers and therapeutic targets.

1.1.2. Pathogenesis of Allergic Diseases

Allergic disorders classically begin during early childhood7 through the development of hypersensitivity to certain allergens, such as aeroallergens, antibiotics or food. This development of hypersensitivity can be split into the sensitization and challenge phases.

During sensitization, allergens are presented by dendritic cells to naïve CD4+ T helper cells. Notably, there often are also environmental insults or mechanical disruption which causes the release of the innate epithelium-derived cytokines as thymic stromal lymphopoietin (TSLP) and interleukin 33 (IL-33). TSLP and IL-33 directly activate dendritic cells8 to promote differentiation of naïve CD4+ T helper cells into type 2 T

2 helper (Th2) cells. The Th2 cells secrete Th2 cytokines, such as IL-4 and IL-13, to activate allergen-reactive B cells to produce and secrete IgE antibodies9 with specificity towards the allergen. These IgE molecules then cross-link on the surface of immune cells that express the surface IgE receptor, most notably mast cells and basophils.

During the challenge phase, which occurs during subsequent exposures to the triggering allergen, the allergen-specific surface IgE then induces basophil and mast cell degranulation and release of a plethora of soluble mediators, including Th2 cytokines, histamines, lipid mediators, and bioactive proteases1. These mediators then induce acute and chronic phase reactions and symptoms, which occur on the scale of minutes and hours, respectively. In addition to promoting IgE-mediated activation of basophils and mast cells, Th2 cytokines also cause eosinophil infiltration. IL-4 and IL-13 induce the release of eosinophil chemoattractants, such as eotaxins, from epithelial cells. IL-5 promotes eosinophil survival and activation. Activated eosinophils release cytotoxic granule proteins that damage epithelial cells, leading to release of TSLP and IL-33 and exacerbated allergic responses.

1.2. Esophageal Epithelial Cells

1.2.1. Introduction

Epithelia constitute the outermost layers of the surface of the body and act as a shield to separate the underlying mucosa from the outside world. The epithelium lining the esophagus is stratified squamous and non-keratinized layer in humans. Herein, we

3 focus on the differentiation and organ-specific gene expression of esophageal epithelium.

1.2.2. Differentiation

Epithelial differentiation allows the esophageal epithelium to self-renew in order to replace damaged or dead epithelial cells that accumulate over the course of life. It is a highly-ordered process that is required to maintain an intact, functioning barrier. In the esophagus, proliferation occurs within the basal zone10,11. Upon commitment to differentiate, these basal cells cease proliferation and move towards the esophageal lumen12. Once epithelial cells reach the upper layers of the esophagus, they enter into a terminally differentiated, quiescent state. The molecular mechanisms regulating esophageal epithelial differentiation are only beginning to be elucidated. Sequential activation of members of the Notch pathway, a cell-cell contact system with crucial roles in dictating organ development and cell fate throughout the body13, regulate esophageal epithelial differentiation14. At the initiation of differentiation, Notch1 becomes activated in basal esophageal epithelial cells and increases Notch3 expression. This leads to expression of differentiation markers, such as keratin 13, and maintenance of an organized epithelium. Additionally, loss of expression of the transcription factor p63 in ex vivo-cultured esophageal epithelial progenitor cells decreases clonal growth capacity and increases the expression of suprabasal genes15.

4

1.2.3. Keratins

The differentiation status of epithelial cells can be assessed through the expression of keratins, which are epithelial-cell specific and can be divided into type I (acidic) and type

II (basic to neutral) keratins16. Type I and type II keratins heterodimerize in equimolar proportions17 to form long, unbranched cytoskeletal intermediate filaments with a diameter of approximately 10 nanometers18. These filaments increase the structural integrity of the epithelium and maintain keratinocyte mechanical properties19. Basal keratins include keratins 5 and 14. In contrast, keratins 4, 10, and 13 are exclusively expressed by differentiated cells18. Filaggrin, a structural protein that interacts with keratins, can also be used as a marker of differentiated keratinocytes. During terminal differentiation, profilaggin, a large precursor protein with a molecular weight of over 400 kDa, is processed into free filaggrin monomers20. These monomers then bind keratin intermediate filaments and are cross-linked to help a protective barrier21.

1.2.4. Epithelial Stem Cells

Classically, the constant renewal of epithelial cells is mediated through stem cells or unipotent progenitor cells, which undergo occasional divisions to give rise to transit amplifying cells. The transit-amplifying cells then expand for a finite amount of time before undergoing differentiation22. A question that has remained unanswered in terms of esophageal epithelial biology is the existence of epithelial stem cells in the esophagus10-12,15,23-25. This question has been resolved to a striking degree in other tissues22, including the skin and portions of the GI tract such as the intestine and

5 stomach. A diverse combination of studies have been used to delineate the stem cell capacities of different cell types in different tissues, including lineage tracing, colony/sphere forming assays, and determination of expression of progenitor markers.

However, the esophageal epithelium does not contain structures, such as crypts or glands, which are home to stem cells in other tissues. Detailed studies utilizing elegant gene-editing techniques have identified that there is no long-lived quiescent epithelial progenitor or stem cell population in the murine esophagus23,24. However, there is difficulty in translating this finding to humans because of marked differences between the human and mouse esophagus, including keratinization, lack of papillae, and functional involvement in the process of digestion due to different diets between the two species. Furthermore, suggestive evidence for the existence of quiescent stem cells in the human esophagus has been obtained. Analysis of formalin-fixed, paraffin- embedded esophageal biopsy specimens from healthy individuals revealed that the interpapillary basal layer, the layer in contact with the basement membrane in between papillae, had rare mitoses that typically rise to one basal cell and one suprabasal daughter cells10. Furthermore, ex vivo culture of isolated interpapillary basal layer esophageal epithelial cells showed that they had a higher clonogenic capacity than other esophageal epithelial cell populations10. In contrast, one study performed idoxuridine (IdU) incorporation studies in normal resections of patients with esophageal cancer and found label-retaining cells in the basal layer at the tips of the papillae26. An additional study using spheroid culture of primary human esophageal epithelial cells

hi low 15 found that sphere-forming cells are enriched among integrin α6 CD24 cells .

Furthermore, p75NTR, the low-affinity nerve growth factor receptor, has been identified

6 as a marker for ex vivo-cultured primary human esophageal epithelial cells with the highest proliferation capacity. Collectively, the exact location of human esophageal epithelial progenitor cells and their exact contributions to tissue homeostasis and differentiation, remain unclear.

1.3. Eosinophilic Esophagitis

1.3.1. Introduction

Eosinophilic gastrointestinal disorders (EGID) are disorders that are characterized by abnormal accumulation of eosinophils in the GI tract and include eosinophilic esophagitis (EoE), eosinophilic gastritis, eosinophilic gastroenteritis and eosinophilic colitis. Of these, EoE is the most prevalent by a considerable margin and is the most studied. The prominent histological features of EoE include eosinophil accumulation and basal layer expansion in the epithelium and fibrosis of the lamina propria. The eosinophilic infiltrate is notable, as the esophagus is the only GI segment that is free of eosinophils under homeostatic conditions. Clinical features of this disorder include atopy, dysphagia, food impaction and proton pump inhibitor unresponsiveness27. Food allergens are critical drivers of the disease, as the most effective therapies are strict elimination diets and the disease reoccurs upon food re-introduction28,29. Although food- specific IgE is readily detectable in EoE30, most patients do not have food anaphylaxis31. Despite the lack of IgE involvement, it has recently been proposed that food-specific IgG4 may be a key pathoetiological component32.

7

1.3.2. Pathogenesis

A critical driver of the pathogenesis of EoE is IL-1333 (Figure 1.1), which is derived from several sources. One source of IL-13 is Th2 cells, whose development from naïve CD4+

T cells is promoted by DC under the regulation of TSLP34 and the microRNA miR-21, which is up-regulated in EoE, is primarily restricted in its expression to DCs and macrophages and promotes Th2 differentiation by silencing transcription of IL-12p3535.

Furthermore, TSLP and IL-33 released from the epithelium directly activate basophils to produce IL-4, which also promotes the Th2-polarization of naïve CD4+ T cells. Other proposed sources of IL-13 in EoE include invariant natural killer T (iNKT) cells36 and IL-

13–secreting FoxP3+ T cells37. The critical importance of this cytokine to the disease is highlighted by the fact that IL-13 stimulation results in an EoE-like transcriptome in ex vivo cultures of primary esophageal epithelial cells33. Two critical downstream targets of

IL-13 are CCL26 (eotaxin-3) and periostin. In contrast with asthma, which is primarily driven by increased eotaxin-1 and eotaxin-2 expression, EoE features CCL26, which is the predominant eotaxin family member with increased expression38 and is strongly induced in esophageal epithelial cells by IL-1333. Periostin increases eosinophil adhesion in vitro and likely facilitates eosinophil infiltration into the esophagus39. It is secreted from fibroblasts in response to IL-13 and TGF-β39. Because eosinophils can both secrete TGF-β directly and induce TGF-β secretion from mast cells through the release of MBP, eosinophils have the capacity to auto-amplify their own infiltration.

Moreover, decreased epithelial expression of the intercellular cadherin desmoglein 1

(DSG1) by IL-13 stimulation or overexpression of the cysteine protease calpain 14

(CAPN14)40 results in increased periostin expression, as well as impaired barrier

8 function41. It is interesting to note that periostin levels predict responsiveness to biological agents for asthma (such as anti-IgE and anti–IL-13)42, highlighting the key relationship between these factors in a variety of human atopic diseases. Esophageal biopsies procured from patients with EoE typically have evidence of eosinophil degranulation43,44, suggesting that direct damage to esophageal epithelial cells by granule proteins may contribute to the pathogenesis of the disorder. Notably, IL-33 is a potent activator eosinophil degranulation45. Similar to their proposed role in asthma, eosinophils contribute to subepithelial fibrosis and dysmotility present in the EoE by secretion of TGF-β, of which they are the main source in the lamina propria46. TGF-β promotes fibrosis by inducing myofibroblast activation and proliferation, increased extracellular matrix synthesis and dysmotility by inducing esophageal smooth muscle hyperplasia. Thus, the pathogenesis of EoE involves a complex interplay between innate epithelial cell-derived cytokines (IL-33, TSLP), Th2 cytokines (IL-13), and immunocytes such as eosinophils.

1.3.3. Epithelial Changes in EoE

There are numerous changes of the architecture of the epithelium patients with EoE.

Notably, there is expansion of the basal zone due to hyperplasia, elongation of the papillae and dilated intercellular spaces. These changes are known to be due in part to dysregulation of the activity of the bone morphogenetic proteins (BMP)47, which are members of the TGF-β cytokine superfamily. BMP activation decreases the proliferation of basal progenitor cells and initiates squamous differentiation. Additionally, in esophageal biopsies from patients with EoE there is decreased BMP signaling and

9 increased expression of the BMP antagonist follistatin. Additionally, in EoE there is down-regulation of numerous epidermal differentiation complex genes, such as filaggrin, because of the actions of IL-1348.

1.3.4. Treatment

Despite being effective in decreasing esophageal eosinophil infiltration, the humanized,

IL-5–neutralizing antibodies mepolizumab and reslizumab have only shown limited clinical improvement49,50. A preliminary phase II trial with the IL-13–neutralizing antibody

QAX576 showed trends for clinical improvement in EoE51. In addition, molecular analysis of esophageal biopsies procured from participants in this trial showed that IL-

13 neutralization markedly corrected the EoE transcriptome, as there was altered expression of genes in several pathways known to be dysregulated in EoE, including eosinophil recruitment, mastocytosis, tissue fibrosis, epithelial barrier function, and epithelial differentiation. Although larger clinical trials still need to be performed, these collective data suggest that IL-13 is important in the development of EoE and has significant eosinophil-independent contributions to the pathogenesis of the disease.

1.3.5. Genetic basis

There has been considerable progress with elucidating the genetic basis of EoE on the basis of candidate gene approaches, studies of EoE-associated Mendelian disorders and genome-wide association studies (GWAS). These interrogations have confirmed that EoE is a multifactorial disorder driven by dysregulation of several biological

10 processes, including eosinophil recruitment, epithelial barrier function, tissue remodeling and innate immune activation. Using a candidate gene approach, the first single- nucleotide polymorphism (SNP) to be associated with EoE (rs2302009) was found within the 3’ untranslated region of CCL2638. Though the functional effect of this SNP has not yet been determined, it likely promotes eosinophil recruitment to the esophagus by increasing CCL26 mRNA transcription. Another candidate gene study found an association between EoE and a loss-of-function mutation of the epithelial barrier gene filaggrin (FLG), which results in reduced epithelial barrier function and increased allergen sensitization52. Genetic studies have also implicated TGF-β in the pathogenesis of the disorder. Candidate gene studies have shown that there is a genetic variant in the TGFB promoter, C509T, which increases transcription by creating a new binding site for the transcription factor YY153. A preliminary study indicated that homozygous expression of the minor 509T allele increases the number of TGF-β– expressing cells in the esophageal lamina propria54. Additionally, EoE is often a comorbid condition in patients with connective tissue disorders (CTD)55, such as Marfan syndrome, Loeys-Dietz syndrome and Ehlers-Danlos syndrome, which are associated with increased TGF-β signaling. EoE can also be seen in patients with PTEN hamartoma syndromes (PHTS)56, which are caused by loss-of-function mutations in

PTEN that result in dysregulation of cell proliferation and epithelial hyperplasia.

Interestingly, loss of PTEN is known to cooperate with TGF-β in the induction of colon cancer57, suggesting that such interactions could also be important for the epithelial hyperproliferation seen in EoE. In many ways, TGF-β is an important contributor to disease pathogenesis in EoE. The first GWAS identified a strong association at 5q22,

11 which encode the TSLP and WDR36 genes58. On the basis of the known biology of

TSLP and its overexpression in the esophagus of patients with EoE, it is the likely gene responsible for the genetic association. Indeed, genetic variants in the TSLP receptor

CRLF2 have also been linked with EoE using a candidate gene approach59. In addition to TSLP’s known effects on DC and eosinophil function described earlier, a recent functional study has demonstrated that the 5q22 genotype affects basophil responses, which have been proposed to contribute to a murine model of EoE60. Notably, the 5q22 locus (TSLP) has been linked with other atopic diseases, so it is unlikely to explain the tissue-specific nature of EoE. Although not yet studied in EoE, a humanized anti-TSLP antibody (AMG 157) has been shown to lower eosinophil levels in patients with asthma61. A larger, recent GWAS study has identified a major susceptibility locus for

EoE at 2p23, wherein the esophagus-specific protein CAPN14 is encoded62.

Interestingly, CAPN14 is selectively expressed in esophageal epithelial cells, induced by IL-13 and located in an epigenetic hotspot regulated by IL-1362. Although the substrate(s) for CAPN14 have not yet been identified, it is notable that other calpain family members proteolytically regulate the activity of STAT663 and intracellular IL-3364.

Because of the multifactorial nature of EoE, further work is needed to expand the number of known causal variants and their mechanistic functions. To this end, the recent development of a 96-gene EoE diagnostic panel65, based on analysis of esophageal biopsies, provides deep information concerning the contribution of individual genes to the pathogenesis of EoE, especially on a patient-to-patient basis.

This diagnostic panel differentiates EoE from controls including gastroesophageal reflux disease (GERD), can also distinguish patients with active and inactive disease and

12 identify glucocorticoid exposure, providing substantial clinical value in providing personalized medicine. In addition, it should prove helpful in the further elucidation of the pathogenesis of mucosal eosinophilic disorders.

1.4. Chromatin

1.4.1. Chromatin Structure

In eukaryotic organisms, DNA is present in a proteinaceous complex referred to as chromatin. The basic unit of chromatin is the nucleosome, which is comprised of 147 base pairs of DNA wound around an octamer histone core containing two copies each of histones H2A, H2B, H3, and H466. The very basic nature of histones causes them to maintain strong interactions with the negatively-charged DNA. Nucleosomes can then interact with each to form higher-order structures not only condense DNA to allow it to fit within the nucleus but also alter access of RNA polymerase and transcription factors to different regions of DNA66,67. Higher-order chromatin structures are dependent on interactions between the H4 tail domain from one nucleosome with the nucleosome acidic patch of neighboring nucleosomes68. This acidic patch is formed by six negatively-charged amino acid residues from H2A and two from H2B. In addition to chromatin compaction, the nucleosome acidic patch has been shown to have important roles in regulating the expression of different histone modifications, such as ubiqutination69, and the recruitment of the DNA repair proteins70.

13

1.4.2. Nucleosome Acidic Patch-binding Proteins

In line with the importance of the nucleosome acidic patch and its critical roles in chromatin compaction and structure, there are very few proteins that bind in this region.

Instead, transcription factors typically bind to DNA nucleotides directly. Examples of mammalian nucleosome acidic patch binding proteins include IL-33 and and HMGN268.

The Kaposi sarcoma virus protein latency-associated nuclear antigen (LANA) also binds the nucleosome acidic patch, but it has a very similar chromatin binding domain as IL-

33 and is thought to have derived from IL-33 over the course of evolution71. LANA has been shown to mediate episome attachment through binding to the nucleosome acidic patch72,73 and has been shown to have diverse effects on infected cells, promoting transformation and proliferation74-76. HMGN2 has been shown to be a transcriptional regulator77. Further knowledge of IL-33 will be summarized later.

1.5. Alarmins

Alarmins are cytokines that are present intracellularly, typically in the nucleus, and are released extracellularly during cellular stress, damage, and/or necrosis78. They are damage-associated molecular patterns (DAMPs) that serve to alert the immune system that the body is under attack and initiate and propagate immune responses. However, at times there can be excessive alarmin release, which can lead to subsequent inflammation and allergy and/or autoimmunity79. Classic nuclear alarmins include

HMGB1, IL-1α, and IL-3379.

14

1.5.1. HMGB1

High mobility group Box 1 (HMGB1) exhibits direct binding to DNA via HMG-box domains80. HMGB1 has very transient interactions with DNA81,82 that mediate transcriptional regulatory roles79. In addition to this intracrine function, HMGB1 is released extracellularly after necrosis and can be actively secreted from living cells under the appropriate conditions. Upon cellular activation, HMGB1 undergoes post- translational modifications, including hyperacetylation of certain amino acid residues, to cause shuttling to the cytoplasm83,84 and subsequent extracellular secretion via nonclassical pathway containing vesicles. Triggers of this active secretion of HMGB1 include IFN-gamma, TNF-alpha, IL-1α, LPS, and activators of the NLRP3 inflammasome79. Upon extracellular release, HMGB1 activates surface receptors present on the plasma membrane of immune cells on dendritic cells and macrophages.

These receptors include TLR4 and the receptor for advanced glycation end-products

(RAGE). Through activation of these receptors, HMGB1 induces chemotaxis and proinflammatory cytokine production and has important contributions to sepsis, sterile injury, cancer, and systemic lupus erythematosus78.

1.5.2. IL-1α

IL-1α is expressed in structural cells in diverse tissues, including the esophagus, skin and GI tract79. Full-length IL-1 binds DNA and nuclear histone acetyltransferases85,86 and exhibits transcriptional regulatory properties78,79. Intracellular IL-1α increases cellular proliferation and expression of inflammatory cytokines such as IL-6 and IL-887.

15

Similar to HMGB1, IL-1α can be released during necrosis or actively secreted. During homeostasis, IL-1α is present bound to an intracellular receptor, IL-1 receptor 2 (IL-

1R2). During cellular activation, IL-1R2 is processed by caspase-188, which then allows calpain to cleave the full-length IL-1α into a mature form89 that is then secreted from the cytoplasm via a noncanonical vesicular pathway. While IL-1α and HMGB1 are both secreted via noncanonical vesicular pathways, no intracellular receptor for HMGB1 has been described. Upon extracellular release, IL-1α binds to the IL-1 receptor (IL-1R) present on epithelial and immune cells. Through this extracellular cytokine activity IL-1α has important roles in diverse inflammatory processes, including colitis, cancer development, and autoimmune disorders78,79.

1.5.3. IL-33 Introduction and Expression

The first description of IL-33 was as a nuclear protein termed “nuclear factor from high endothelial venules (NF-HEV)” that was found to be expressed in endothelial cells in the nucleus90. It is now known that in addition to endothelial cells IL-33 is expressed by fibroblasts, epithelial cells, and macrophages91-93. The regulation of IL-33 expression is not well-characterized in epithelial cells. IL-33 is a marker of endothelial cell quiescence as its expression is lost from endothelial cells upon initiation of proliferation94. However, lack of cell division was not the only requirement for IL-33 expression as IL-33 was not induced during cell cycle arrest95. Instead, Notch ligands causing activated Notch1 were responsible for both inducing IL-33 expression and maintain endothelial quiescence95.

The role of the Notch signaling in regulating IL-33 expression in other cell types has not

16 been examined. In fibroblasts, IL-33 protein expression is known to be induced by TNF- alpha96,97 or PMA98. However, there are contradictory reports as to whether these stimuli induce IL-33 protein expression in epithelial cells91. In fact, a few studies have indicated that TNF-alpha represses IL-33 protein expression, likely by promoting IL-33 proteolytic degradation, in skin keratinocytes99,100. However, these studies have not been repeated in primary epithelial cells derived from other tissues. A strong inducer of

IL-33 protein in epithelial cells is IFN-gamma91, which induces IL-33 in an EGFR- dependent manner101. Interestingly, IFN-gamma decreases IL-33 expression in fibroblasts by altering proteasome activity102. In macrophages, IL-33 can be induced by

PMA or LPS103-105. Additionally, recent studies have revealed that oncostatin M can induce IL-33 protein expression in murine type II pneumocytes106. Oncostatin M can also induce IL33 mRNA in liver endothelial cells107, but its effect on IL-33 protein in this cell type was not examined.

1.5.4. IL-33 Nuclear Localization and Chromatin Binding

The amino acid sequence of IL-33 does not contain a classical nuclear localization sequence (NLS) (Figure 1.2). Instead, the restriction of IL-33 to the nucleus is due to direct binding to the nucleosome acidic patch71,108. This binding site within the chromatin distinguishes IL-33 from the prototypical nuclear alarmins IL-1α and HMGB1 (Figure

1.3). An intriguing question that has remained unanswered is the purpose of the enigmatic nuclear localization and chromatin binding108-112, especially in light of the fact that these features are not present in the proallergic, innate cytokines IL-25 and

17

TSLP113. Intracellular nuclear functions have long been proposed for IL-33 on the basis of its direct interactions with chromatin and the fact that other nuclear alarmins, including IL-1α and HMGB1, have intracellular nuclear functions as outlined above.

Additionally, in cell-free in vitro systems recombinant IL-33 has been shown to bind nucleosome arrays and promote chromatin condensation71. Furthermore, transient transfection of IL-33 into HEK293T cells decreased transcription of a luciferase reporter, consistent with being a transcriptional repressor71,108. Since these seminal observations, preliminary studies have identified minor effects of nuclear IL-33 in endothelial cells, fibroblasts, and epithelial cells. A study found IL-33 affects expression of IL-6 in endothelial cells114, and another showed that IL-33 affects IL-6 and IL-8 expression in esophageal epithelial cells115,116. However, these studies are limited by the fact that the cell lines used also express the IL-33 receptor ST2. An additional study showed that IL-

33 overexpression in fibroblasts affected MMP expression97. However, none of these findings have been reproduced or rigorously tested. For instance, there have been no detailed molecular studies showing interactions of nuclear IL-33 with the promoters of regulated genes. Additionally, there have not been any studies demonstrating an effect of nuclear IL-33 on chromatin structure in cells. Therefore, the above findings may be due to indirect mechanisms other than chromatin binding, such as altering the activity of other transcription factors. In line with this potential mechanism, individual studies have demonstrated that IL-33 co-immunoprecipitates with methyltransferases such as

SUV39H1114 and transcription factors such as NFκB117. Moreover, a recent proteomic analysis performed after IL33 gene-silencing in unstimulated endothelial cells failed to

18 detect differentially expressed proteins118. However, no wide-scale study has been performed on cells under stimulatory conditions.

1.5.5. IL-33 Release Mechanisms

Consistent with being an alarmin, IL-33 is classically thought to be passively released extracellularly during necrosis due to loss of membrane integrity. As of now, there is no indication that there is an ordered secretion pathway for IL-33 that is induced in necrosis119. Additionally, no intracellular processing or changes in chromatin affinity, such as resulting from post-translational modifications, have been described for IL-33 during necrosis. A few reports have suggested that IL-33 can be actively secreted from live cells. It was found that Alternaria extract120 or uric acid121 caused increased IL-33 detection in supernatants of ex vivo-cultured human primary bronchial epithelial cells without measurable increases in necrosis. This increased release has been determined to involve activation of the proteinase-activated receptor 2 (PAR2), which causes ATP release which then bound purinergic receptors to induce intracellular calcium flux and

IL-33 release. However, these extracts contain active proteases, so it is not entirely clear whether the IL-33 was actually released in the absence of cellular damage.

Additionally, these stimuli were not found to cause active release of IL-33 from primary human skin keratinocytes101, although an unconfirmed explanation is tissue-specificity of this phenomenon. There also is no evidence for active secretion of IL-33 from live cells in vivo. Therefore, necrosis is considered to at least the main, if not the only,

19 inducer of IL-33 release in vivo. This contrasts IL-33 from IL-1α and HMGB1, which have well-described active secretion mechanisms (see sections 1.5.1. and 1.5.2.).

1.5.6. IL-33 Receptor

Extracellular IL-33 activates its receptor, ST2 (also known as IL-1 receptor like 1

[IL1RL1] or IL-33 receptor [IL-33R]). Upon ligation by IL-33, the ST2 receptor then interacts with the IL-1 receptor accessory protein (IL-1RAcP) to allow recruitment of the signaling molecules myeloid differentiation primary response protein 88 (MYD88), IL-

1R-associated kinases 1 and 4 (IRAK1 and IRAK4, respectively), and TNF receptor- associated factor 6 (TRAF6)109. There is then subsequent activation of important transcription factors and signaling molecules, including NFκB and mitogen-activated protein kinases (MAPK), including ERK, p38, and JNK)91. Through ST2, IL-33 potently activates a plethora of immune cells, including basophils, mast cells, eosinophils, type 2 innate lymphoid cells, CD4+ T cells, NK cells, and macrophages91,109. IL-33 was found to activate eosinophils more profoundly than IL-445 and IL-33 also provides eosinophils with a survival advantage to promote their trafficking to the lungs122. IL-33 also induces mast cell degranulation and release of potent soluble mediators123. The potency of IL-

33-induced activation through ST2 is demonstrated by the fact that knock-in of a truncated form of IL-33 that does not bind chromatin, exhibited at least a 10-fold higher

ST2-bioreactivity than wild-type IL-33, and was under control of the endogenous IL33 promoter, was lethal124. Additionally, the heterozygotes suffered from overwhelming systemic allergic inflammation. Additionally, IL-33 has been shown to have diverse roles

20 in allergic diseases, include allergic asthma, atopic dermatitis, allergic rhinitis, and atopic dermatitis109. There is genetic association between variants in IL33 and IL1RL1

(encodes ST2) and different allergic processes. In addition to allergy, IL-33 has important roles in diverse diseases including parasite infections, cancer, and inflammatory bowel disease.

1.5.7. IL-33 Post-Translational Modification

There are multiple mechanisms that profoundly modulate the ability of IL-33 to activate

ST2-expressing cells (Table 1.1). For instance, extracellular IL-33 is processed by proteolytic enzymes derived from neutrophils (e.g. elastase, cathepsin G)125 or mast cells (e.g. chymase, tryptase)126 into mature forms with enhanced abilities to activate

ST2. The cleavage sites for these enzymes lie within the central domain (Figure 1.2).

Additionally, upon extracellular release IL-33 is fairly rapidly converted into an inactive form via cysteine oxidation and formation of disulfide bonds that causes a conformation shift preventing binding to the ST2 receptor127. In allergic asthma there is up-regulation of IL33 mRNA splice variants with enhanced release in response to ionomycin treatment128. Importantly, bioreactive IL-33 is only released during necrosis as during apoptosis there is intracellular retention of IL-33 where it is cleaved within the IL-1-like cytokine domain by activated caspases 3 and 7 into unreactive forms103. Additionally, a decoy receptor, soluble ST2 (sST2), is also secreted extracellularly to neutralize IL-33.

Expression of this decoy receptor is often elevated in the context of inflammation129.

21

1.5.8. IL-33 Involvement in EoE Pathogenesis

Two recent studies have shown that esophageal biopsies from patients with EoE express higher levels of IL-33 protein than those from control individuals130,131.

Additionally, there is association between genetic variants in the IL33 locus and disease risk62. Finally, intraperitoneal injection of recombinant IL-33 induces esophageal eosinophilia, epithelial hyperplasia, and production of Th2-associated cytokines130.

Furthermore, IL-33 is required in the pathogenesis of experimental, murine, EoE-like disease132. Eosinophil infiltration and Th2-associated cytokine production were induced in the esophagus by repeated intranasal challenge with OVA after epicutaneous sensitization by disrupting the skin epithelial barrier through tape-stripping or genetic deficiency of FLG. Disrupting the skin epithelial barrier was required, as allergic sensitization did not occur in wild-type mice without tape-stripping of the skin. IL-33 was a critical mediator in this model, as esophageal eosinophil accumulation was not present in mice either genetically deficient in ST2 or given ST2-neutralizing antibodies.

Basophil depletion prevented disease induction, and experimental EoE was restored upon adoptive transfer of wild-type, but not ST2-deficient, basophils. This observation supports that basophils were the target of IL-33, at least in part. Collectively, the data show that IL-33–ST2 interactions in basophils and impaired skin epithelial barrier integrity mediate induction of EoE-like disease induced by epicutaneous allergen sensitization. Although this work provides strong evidence for the importance of the IL-

33–ST2–basophil axis in this experimental mouse model, concerns remain with regard to translating these findings to the human disease, as human biopsy specimens reveal little accumulation of basophils. In contrast, esophageal biopsies from patients with

22 active EoE are notable for infiltration of mast cells and eosinophils, which are both potently activated by IL-33. Therefore, it is likely that IL-33 activating these cells also contributes to the human disease. Additionally, the experimental antigen exposure is of supraphysiologic doses of a single antigen with one route of sensitization, whereas the human disease likely involves multiple allergens at more physiologic exposure doses and with several routes of sensitization. Finally, this study is limited by the fact that it did not directly test the requirement of IL-33 in the esophagus as it also induces allergic inflammation in the lungs, which is known to be IL-33-dependent133,134.

1.6. Summary

Within the past decade there has been significant progress in the elucidation of the pathogenesis of EoE, which demonstrates a complex interplay of the epithelium, innate and Th2 cytokines, and immune cells such as eosinophils, basophils, and mast cells. In this thesis work, we assessed the expression of esophagus-specific genes, with a particular focus on epithelial cells. We identified that the esophagus normally expresses

IL-1-related genes, which are dysregulated in the esophagus of patients with EoE. We subsequently identified up-regulation of the IL-1 family member IL-33, a potent pro- allergic innate cytokine, and characterized the regulation of the extracellular cytokine activities of IL-33 and IL-1α.

23

1.7. References

1 Ono, S. J. Molecular genetics of allergic diseases. Annu Rev Immunol 18, 347-

366, doi:10.1146/annurev.immunol.18.1.347 (2000).

2 Eder, W., Ege, M. J. & von Mutius, E. The asthma epidemic. N Engl J Med 355,

2226-2235, doi:10.1056/NEJMra054308 (2006).

3 Sicherer, S. H. & Sampson, H. A. Food allergy: recent advances in

pathophysiology and treatment. Annual review of medicine 60, 261-277,

doi:10.1146/annurev.med.60.042407.205711 (2009).

4 Rafii, B. et al. Glucocorticoids in laryngology: a review. The Laryngoscope 124,

1668-1673, doi:10.1002/lary.24556 (2014).

5 Rothenberg, M. E. Humanized Anti-IL-5 Antibody Therapy. Cell 165, 509,

doi:10.1016/j.cell.2016.04.020 (2016).

6 Kern, E. & Hirano, I. Emerging drugs for eosinophilic esophagitis. Expert opinion

on emerging drugs 18, 353-364, doi:10.1517/14728214.2013.829039 (2013).

7 Alduraywish, S. A. et al. Is there a march from early food sensitization to later

childhood allergic airway disease? Results from two prospective birth cohort

studies. Pediatric allergy and immunology : official publication of the European

Society of Pediatric Allergy and Immunology 28, 30-37, doi:10.1111/pai.12651

(2017).

8 Wang, Y. H. Developing food allergy: a potential immunologic pathway linking

skin barrier to gut. F1000Research 5, doi:10.12688/f1000research.9497.1

(2016).

24

9 Poulsen, L. K. & Hummelshoj, L. Triggers of IgE class switching and allergy

development. Annals of medicine 39, 440-456, doi:10.1080/07853890701449354

(2007).

10 Seery, J. P. & Watt, F. M. Asymmetric stem-cell divisions define the architecture

of human oesophageal epithelium. Current biology : CB 10, 1447-1450 (2000).

11 Seery, J. P. Stem cells of the oesophageal epithelium. Journal of cell science

115, 1783-1789 (2002).

12 Croagh, D., Thomas, R. J., Phillips, W. A. & Kaur, P. Esophageal stem cells--a

review of their identification and characterization. Stem cell reviews 4, 261-268,

doi:10.1007/s12015-008-9031-3 (2008).

13 Kopan, R. & Ilagan, M. X. The canonical Notch signaling pathway: unfolding the

activation mechanism. Cell 137, 216-233, doi:10.1016/j.cell.2009.03.045 (2009).

14 Ohashi, S. et al. NOTCH1 and NOTCH3 coordinate esophageal squamous

differentiation through a CSL-dependent transcriptional network.

Gastroenterology 139, 2113-2123, doi:10.1053/j.gastro.2010.08.040 (2010).

15 Jeong, Y. et al. Identification and genetic manipulation of human and mouse

oesophageal stem cells. Gut 65, 1077-1086, doi:10.1136/gutjnl-2014-308491

(2016).

16 Moll, R., Franke, W. W., Schiller, D. L., Geiger, B. & Krepler, R. The catalog of

human cytokeratins: patterns of expression in normal epithelia, tumors and

cultured cells. Cell 31, 11-24 (1982).

17 Quinlan, R. A., Cohlberg, J. A., Schiller, D. L., Hatzfeld, M. & Franke, W. W.

Heterotypic tetramer (A2D2) complexes of non-epidermal keratins isolated from

25

cytoskeletons of rat hepatocytes and hepatoma cells. Journal of molecular

biology 178, 365-388 (1984).

18 Moll, R., Divo, M. & Langbein, L. The human keratins: biology and pathology.

Histochemistry and cell biology 129, 705-733, doi:10.1007/s00418-008-0435-6

(2008).

19 Ramms, L. et al. Keratins as the main component for the mechanical integrity of

keratinocytes. Proc Natl Acad Sci U S A 110, 18513-18518,

doi:10.1073/pnas.1313491110 (2013).

20 McLean, W. H. Filaggrin failure - from ichthyosis vulgaris to atopic eczema and

beyond. The British journal of dermatology 175 Suppl 2, 4-7,

doi:10.1111/bjd.14997 (2016).

21 Sandilands, A., Sutherland, C., Irvine, A. D. & McLean, W. H. Filaggrin in the

frontline: role in skin barrier function and disease. Journal of cell science 122,

1285-1294, doi:10.1242/jcs.033969 (2009).

22 Blanpain, C., Horsley, V. & Fuchs, E. Epithelial stem cells: turning over new

leaves. Cell 128, 445-458, doi:10.1016/j.cell.2007.01.014 (2007).

23 Doupe, D. P. et al. A single progenitor population switches behavior to maintain

and repair esophageal epithelium. Science (New York, N.Y.) 337, 1091-1093,

doi:10.1126/science.1218835 (2012).

24 DeWard, A. D., Cramer, J. & Lagasse, E. Cellular heterogeneity in the mouse

esophagus implicates the presence of a nonquiescent epithelial stem cell

population. Cell reports 9, 701-711, doi:10.1016/j.celrep.2014.09.027 (2014).

26

25 Barker, N. Epithelial stem cells in the esophagus: who needs them? Cell stem

cell 11, 284-286, doi:10.1016/j.stem.2012.08.005 (2012).

26 Pan, Q. et al. Identification of lineage-uncommitted, long-lived, label-retaining

cells in healthy human esophagus and stomach, and in metaplastic esophagus.

Gastroenterology 144, 761-770, doi:10.1053/j.gastro.2012.12.022 (2013).

27 Miehlke, S. Clinical features of eosinophilic esophagitis. Digestive diseases

(Basel, Switzerland) 32, 61-67, doi:10.1159/000357011 (2014).

28 Kagalwalla, A. F. et al. Effect of six-food elimination diet on clinical and histologic

outcomes in eosinophilic esophagitis. Clinical gastroenterology and hepatology :

the official clinical practice journal of the American Gastroenterological

Association 4, 1097-1102, doi:10.1016/j.cgh.2006.05.026 (2006).

29 Kagalwalla, A. F. et al. Identification of specific foods responsible for

inflammation in children with eosinophilic esophagitis successfully treated with

empiric elimination diet. Journal of pediatric gastroenterology and nutrition 53,

145-149, doi:10.1097/MPG.0b013e31821cf503 (2011).

30 Erwin, E. A. et al. Serum IgE measurement and detection of food allergy in

pediatric patients with eosinophilic esophagitis. Annals of allergy, asthma &

immunology : official publication of the American College of Allergy, Asthma, &

Immunology 104, 496-502, doi:10.1016/j.anai.2010.03.018 (2010).

31 Sugnanam, K. K. et al. Dichotomy of food and inhalant allergen sensitization in

eosinophilic esophagitis. Allergy 62, 1257-1260, doi:10.1111/j.1398-

9995.2007.01454.x (2007).

27

32 Clayton, F. et al. Eosinophilic Esophagitis in Adults is Associated with IgG4 and

Not Mediated by IgE. Gastroenterology 147, 602-609,

doi:10.1053/j.gastro.2014.05.036 (2014).

33 Blanchard, C. et al. IL-13 involvement in eosinophilic esophagitis: transcriptome

analysis and reversibility with glucocorticoids. The Journal of allergy and clinical

immunology 120, 1292-1300, doi:10.1016/j.jaci.2007.10.024 (2007).

34 Ziegler, S. F. et al. The biology of thymic stromal lymphopoietin (TSLP).

Advances in pharmacology (San Diego, Calif.) 66, 129-155, doi:10.1016/b978-0-

12-404717-4.00004-4 (2013).

35 Lu, T. X., Munitz, A. & Rothenberg, M. E. MicroRNA-21 is up-regulated in allergic

airway inflammation and regulates IL-12p35 expression. Journal of immunology

(Baltimore, Md. : 1950) 182, 4994-5002, doi:10.4049/jimmunol.0803560 (2009).

36 Lexmond, W. S. et al. Involvement of the iNKT cell pathway is associated with

early-onset eosinophilic esophagitis and response to allergen avoidance therapy.

The American journal of gastroenterology 109, 646-657, doi:10.1038/ajg.2014.12

(2014).

37 Frischmeyer-Guerrerio, P. A. et al. TGFbeta receptor mutations impose a strong

predisposition for human allergic disease. Science translational medicine 5,

195ra194, doi:10.1126/scitranslmed.3006448 (2013).

38 Blanchard, C. et al. Eotaxin-3 and a uniquely conserved gene-expression profile

in eosinophilic esophagitis. The Journal of clinical investigation 116, 536-547,

doi:10.1172/jci26679 (2006).

28

39 Blanchard, C. et al. Periostin facilitates eosinophil tissue infiltration in allergic

lung and esophageal responses. Mucosal immunology 1, 289-296,

doi:10.1038/mi.2008.15 (2008).

40 Davis, B. P. et al. Eosinophilic esophagitis-linked calpain 14 is an IL-13-induced

protease that mediates esophageal epithelial barrier impairment. JCI insight 1,

e86355, doi:10.1172/jci.insight.86355 (2016).

41 Sherrill, J. D. et al. Desmoglein-1 regulates esophageal epithelial barrier function

and immune responses in eosinophilic esophagitis. Mucosal Immunol. 7, 718-

729, doi:10.1038/mi.2013.90 (2014).

42 Corren, J. et al. Lebrikizumab treatment in adults with asthma. The New England

journal of medicine 365, 1088-1098, doi:10.1056/NEJMoa1106469 (2011).

43 Mueller, S., Aigner, T., Neureiter, D. & Stolte, M. Eosinophil infiltration and

degranulation in oesophageal mucosa from adult patients with eosinophilic

oesophagitis: a retrospective and comparative study on pathological biopsy.

Journal of clinical pathology 59, 1175-1180, doi:10.1136/jcp.2005.031922 (2006).

44 Saffari, H. et al. Electron microscopy elucidates eosinophil degranulation patterns

in patients with eosinophilic esophagitis. The Journal of allergy and clinical

immunology 133, 1728-1734, doi:10.1016/j.jaci.2013.11.024 (2014).

45 Bouffi, C. et al. IL-33 markedly activates murine eosinophils by an NF-kappaB-

dependent mechanism differentially dependent upon an IL-4-driven

autoinflammatory loop. Journal of immunology (Baltimore, Md. : 1950) 191, 4317-

4325, doi:10.4049/jimmunol.1301465 (2013).

29

46 Aceves, S. S., Newbury, R. O., Dohil, R., Bastian, J. F. & Broide, D. H.

Esophageal remodeling in pediatric eosinophilic esophagitis. The Journal of

allergy and clinical immunology 119, 206-212, doi:10.1016/j.jaci.2006.10.016

(2007).

47 Jiang, M. et al. BMP-driven NRF2 activation in esophageal basal cell

differentiation and eosinophilic esophagitis. J Clin Invest 125, 1557-1568,

doi:10.1172/jci78850 (2015).

48 Blanchard, C. et al. Coordinate interaction between IL-13 and epithelial

differentiation cluster genes in eosinophilic esophagitis. J Immunol 184, 4033-

4041, doi:10.4049/jimmunol.0903069 (2010).

49 Spergel, J. M. et al. Reslizumab in children and adolescents with eosinophilic

esophagitis: results of a double-blind, randomized, placebo-controlled trial. The

Journal of allergy and clinical immunology 129, 456-463, 463.e451-453,

doi:10.1016/j.jaci.2011.11.044 (2012).

50 Straumann, A. et al. Anti-interleukin-5 antibody treatment (mepolizumab) in

active eosinophilic oesophagitis: a randomised, placebo-controlled, double-blind

trial. Gut 59, 21-30, doi:10.1136/gut.2009.178558 (2010).

51 Rothenberg, M. E. et al. Intravenous anti-IL-13 mAb QAX576 for the treatment of

eosinophilic esophagitis. The Journal of allergy and clinical immunology,

doi:10.1016/j.jaci.2014.07.049 (2014).

52 Blanchard, C. et al. Coordinate interaction between IL-13 and epithelial

differentiation cluster genes in eosinophilic esophagitis. Journal of immunology

(Baltimore, Md. : 1950) 184, 4033-4041, doi:10.4049/jimmunol.0903069 (2010).

30

53 Silverman, E. S. et al. Transforming growth factor-beta1 promoter polymorphism

C-509T is associated with asthma. American journal of respiratory and critical

care medicine 169, 214-219, doi:10.1164/rccm.200307-973OC (2004).

54 Aceves, S. S. et al. Resolution of remodeling in eosinophilic esophagitis

correlates with epithelial response to topical corticosteroids. Allergy 65, 109-116,

doi:10.1111/j.1398-9995.2009.02142.x (2010).

55 Abonia, J. P. et al. High prevalence of eosinophilic esophagitis in patients with

inherited connective tissue disorders. The Journal of allergy and clinical

immunology 132, 378-386, doi:10.1016/j.jaci.2013.02.030 (2013).

56 Henderson, C. J. et al. Increased prevalence of eosinophilic gastrointestinal

disorders in pediatric PTEN hamartoma tumor syndromes. Journal of pediatric

gastroenterology and nutrition 58, 553-560,

doi:10.1097/mpg.0000000000000253 (2014).

57 Yu, M. et al. Inactivation of TGF-beta signaling and loss of PTEN cooperate to

induce colon cancer in vivo. Oncogene 33, 1538-1547,

doi:10.1038/onc.2013.102 (2014).

58 Rothenberg, M. E. et al. Common variants at 5q22 associate with pediatric

eosinophilic esophagitis. Nature genetics 42, 289-291, doi:10.1038/ng.547

(2010).

59 Sherrill, J. D. et al. Variants of thymic stromal lymphopoietin and its receptor

associate with eosinophilic esophagitis. The Journal of allergy and clinical

immunology 126, 160-165.e163, doi:10.1016/j.jaci.2010.04.037 (2010).

31

60 Noti, M. et al. Thymic stromal lymphopoietin-elicited basophil responses promote

eosinophilic esophagitis. Nature medicine 19, 1005-1013, doi:10.1038/nm.3281

(2013).

61 Gauvreau, G. M. et al. Effects of an anti-TSLP antibody on allergen-induced

asthmatic responses. The New England journal of medicine 370, 2102-2110,

doi:10.1056/NEJMoa1402895 (2014).

62 Kottyan, L. C. et al. Genome-wide association analysis of eosinophilic

esophagitis provides insight into the tissue specificity of this allergic disease.

Nature genetics 46, 895-900, doi:10.1038/ng.3033 (2014).

63 Zamorano, J., Rivas, M. D., Setien, F. & Perez, G. M. Proteolytic regulation of

activated STAT6 by calpains. Journal of immunology (Baltimore, Md. : 1950) 174,

2843-2848 (2005).

64 Hayakawa, M. et al. Mature interleukin-33 is produced by calpain-mediated

cleavage in vivo. Biochemical and biophysical research communications 387,

218-222, doi:10.1016/j.bbrc.2009.07.018 (2009).

65 Wen, T. et al. Molecular diagnosis of eosinophilic esophagitis by gene expression

profiling. Gastroenterology 145, 1289-1299, doi:10.1053/j.gastro.2013.08.046

(2013).

66 Venkatesh, S. & Workman, J. L. Histone exchange, chromatin structure and the

regulation of transcription. Nature reviews. Molecular cell biology 16, 178-189,

doi:10.1038/nrm3941 (2015).

67 Li, B., Carey, M. & Workman, J. L. The role of chromatin during transcription. Cell

128, 707-719, doi:10.1016/j.cell.2007.01.015 (2007).

32

68 Kalashnikova, A. A., Porter-Goff, M. E., Muthurajan, U. M., Luger, K. & Hansen,

J. C. The role of the nucleosome acidic patch in modulating higher order

chromatin structure. Journal of the Royal Society, Interface 10, 20121022,

doi:10.1098/rsif.2012.1022 (2013).

69 Mattiroli, F., Uckelmann, M., Sahtoe, D. D., van Dijk, W. J. & Sixma, T. K. The

nucleosome acidic patch plays a critical role in RNF168-dependent ubiquitination

of histone H2A. Nature communications 5, 3291, doi:10.1038/ncomms4291

(2014).

70 Leung, J. W. et al. Nucleosome acidic patch promotes RNF168- and

RING1B/BMI1-dependent H2AX and H2A ubiquitination and DNA damage

signaling. PLoS genetics 10, e1004178, doi:10.1371/journal.pgen.1004178

(2014).

71 Roussel, L., Erard, M., Cayrol, C. & Girard, J. P. Molecular mimicry between IL-

33 and KSHV for attachment to chromatin through the H2A-H2B acidic pocket.

EMBO reports 9, 1006-1012, doi:10.1038/embor.2008.145 (2008).

72 Barbera, A. J., Ballestas, M. E. & Kaye, K. M. The Kaposi's sarcoma-associated

herpesvirus latency-associated nuclear antigen 1 N terminus is essential for

chromosome association, DNA replication, and episome persistence. Journal of

virology 78, 294-301 (2004).

73 Barbera, A. J. et al. The nucleosomal surface as a docking station for Kaposi's

sarcoma herpesvirus LANA. Science (New York, N.Y.) 311, 856-861,

doi:10.1126/science.1120541 (2006).

33

74 Komatsu, T., Barbera, A. J., Ballestas, M. E. & Kaye, K. M. The Kaposi' s

sarcoma-associated herpesvirus latency-associated nuclear antigen. Viral

immunology 14, 311-317, doi:10.1089/08828240152716565 (2001).

75 Liang, D. et al. Oncogenic herpesvirus KSHV Hijacks BMP-Smad1-Id signaling to

promote tumorigenesis. PLoS pathogens 10, e1004253,

doi:10.1371/journal.ppat.1004253 (2014).

76 Kumar, A. et al. Kaposi sarcoma herpes virus latency associated nuclear antigen

protein release the G2/M cell cycle blocks by modulating ATM/ATR mediated

checkpoint pathway. PloS one 9, e100228, doi:10.1371/journal.pone.0100228

(2014).

77 Postnikov, Y. & Bustin, M. Regulation of chromatin structure and function by

HMGN proteins. Biochimica et biophysica acta 1799, 62-68,

doi:10.1016/j.bbagrm.2009.11.016 (2010).

78 Rider, P., Voronov, E., Dinarello, C. A., Apte, R. N. & Cohen, I. Alarmins: Feel

the Stress. J Immunol 198, 1395-1402, doi:10.4049/jimmunol.1601342 (2017).

79 Bertheloot, D. & Latz, E. HMGB1, IL-1alpha, IL-33 and S100 proteins: dual-

function alarmins. Cellular & molecular immunology 14, 43-64,

doi:10.1038/cmi.2016.34 (2017).

80 Stros, M. HMGB proteins: interactions with DNA and chromatin. Biochimica et

biophysica acta 1799, 101-113, doi:10.1016/j.bbagrm.2009.09.008 (2010).

81 Mollica, L. et al. Glycyrrhizin binds to high-mobility group box 1 protein and

inhibits its cytokine activities. Chemistry & biology 14, 431-441,

doi:10.1016/j.chembiol.2007.03.007 (2007).

34

82 Catez, F. et al. Network of dynamic interactions between histone H1 and high-

mobility-group proteins in chromatin. Molecular and cellular biology 24, 4321-

4328 (2004).

83 Ito, I., Fukazawa, J. & Yoshida, M. Post-translational methylation of high mobility

group box 1 (HMGB1) causes its cytoplasmic localization in neutrophils. J Biol

Chem 282, 16336-16344, doi:10.1074/jbc.M608467200 (2007).

84 Bonaldi, T. et al. Monocytic cells hyperacetylate chromatin protein HMGB1 to

redirect it towards secretion. The EMBO journal 22, 5551-5560,

doi:10.1093/emboj/cdg516 (2003).

85 Buryskova, M., Pospisek, M., Grothey, A., Simmet, T. & Burysek, L. Intracellular

interleukin-1alpha functionally interacts with histone acetyltransferase

complexes. J Biol Chem 279, 4017-4026, doi:10.1074/jbc.M306342200 (2004).

86 Zamostna, B. et al. N-terminal domain of nuclear IL-1alpha shows structural

similarity to the C-terminal domain of Snf1 and binds to the HAT/core module of

the SAGA complex. PloS one 7, e41801, doi:10.1371/journal.pone.0041801

(2012).

87 Werman, A. et al. The precursor form of IL-1alpha is an intracrine

proinflammatory activator of transcription. Proc Natl Acad Sci U S A 101, 2434-

2439 (2004).

88 Zheng, Y., Humphry, M., Maguire, J. J., Bennett, M. R. & Clarke, M. C.

Intracellular interleukin-1 receptor 2 binding prevents cleavage and activity of

interleukin-1alpha, controlling necrosis-induced sterile inflammation. Immunity

38, 285-295, doi:10.1016/j.immuni.2013.01.008 (2013).

35

89 Kavita, U. & Mizel, S. B. Differential sensitivity of interleukin-1 alpha and -beta

precursor proteins to cleavage by calpain, a calcium-dependent protease. J Biol

Chem 270, 27758-27765 (1995).

90 Baekkevold, E. S. et al. Molecular characterization of NF-HEV, a nuclear factor

preferentially expressed in human high endothelial venules. Am J Pathol 163, 69-

79, doi:10.1016/s0002-9440(10)63631-0 (2003).

91 Martin, N. T. & Martin, M. U. Interleukin 33 is a guardian of barriers and a local

alarmin. Nat Immunol 17, 122-131, doi:10.1038/ni.3370 (2016).

92 Moussion, C., Ortega, N. & Girard, J. P. The IL-1-like cytokine IL-33 is

constitutively expressed in the nucleus of endothelial cells and epithelial cells in

vivo: a novel 'alarmin'? PloS one 3, e3331, doi:10.1371/journal.pone.0003331

(2008).

93 Pichery, M. et al. Endogenous IL-33 is highly expressed in mouse epithelial

barrier tissues, lymphoid organs, brain, embryos, and inflamed tissues: in situ

analysis using a novel Il-33-LacZ gene trap reporter strain. J Immunol 188, 3488-

3495, doi:10.4049/jimmunol.1101977 (2012).

94 Kuchler, A. M. et al. Nuclear interleukin-33 is generally expressed in resting

endothelium but rapidly lost upon angiogenic or proinflammatory activation. Am J

Pathol 173, 1229-1242, doi:10.2353/ajpath.2008.080014 (2008).

95 Sundlisaeter, E. et al. The alarmin IL-33 is a notch target in quiescent endothelial

cells. Am J Pathol 181, 1099-1111, doi:10.1016/j.ajpath.2012.06.003 (2012).

36

96 Kobori, A. et al. Interleukin-33 expression is specifically enhanced in inflamed

mucosa of ulcerative colitis. Journal of gastroenterology 45, 999-1007,

doi:10.1007/s00535-010-0245-1 (2010).

97 Kunisch, E., Chakilam, S., Gandesiri, M. & Kinne, R. W. IL-33 regulates TNF-

alpha dependent effects in synovial fibroblasts. International journal of molecular

medicine 29, 530-540, doi:10.3892/ijmm.2012.883 (2012).

98 Sanada, S. et al. IL-33 and ST2 comprise a critical biomechanically induced and

cardioprotective signaling system. J Clin Invest 117, 1538-1549,

doi:10.1172/jci30634 (2007).

99 Seltmann, J., Werfel, T. & Wittmann, M. Evidence for a regulatory loop between

IFN-gamma and IL-33 in skin inflammation. Experimental dermatology 22, 102-

107, doi:10.1111/exd.12076 (2013).

100 Meephansan, J., Tsuda, H., Komine, M., Tominaga, S. & Ohtsuki, M. Regulation

of IL-33 expression by IFN-gamma and tumor necrosis factor-alpha in normal

human epidermal keratinocytes. The Journal of investigative dermatology 132,

2593-2600, doi:10.1038/jid.2012.185 (2012).

101 Sundnes, O. et al. Epidermal Expression and Regulation of Interleukin-33 during

Homeostasis and Inflammation: Strong Species Differences. The Journal of

investigative dermatology 135, 1771-1780, doi:10.1038/jid.2015.85 (2015).

102 Kopach, P. et al. IFN-gamma directly controls IL-33 protein level through a

STAT1- and LMP2-dependent mechanism. J Biol Chem 289, 11829-11843,

doi:10.1074/jbc.M113.534396 (2014).

37

103 Luthi, A. U. et al. Suppression of interleukin-33 bioactivity through proteolysis by

apoptotic caspases. Immunity 31, 84-98, doi:10.1016/j.immuni.2009.05.007

(2009).

104 Nile, C. J., Barksby, E., Jitprasertwong, P., Preshaw, P. M. & Taylor, J. J.

Expression and regulation of interleukin-33 in human monocytes. Immunology

130, 172-180, doi:10.1111/j.1365-2567.2009.03221.x (2010).

105 Morris, M. C., Gilliam, E. A., Button, J. & Li, L. Dynamic modulation of innate

immune response by varying dosages of lipopolysaccharide (LPS) in human

monocytic cells. J Biol Chem 289, 21584-21590, doi:10.1074/jbc.M114.583518

(2014).

106 Richards, C. D. et al. Regulation of IL-33 by Oncostatin M in Mouse Lung

Epithelial Cells. Mediators of inflammation 2016, 9858374,

doi:10.1155/2016/9858374 (2016).

107 Arshad, M. I. et al. Oncostatin M induces IL-33 expression in liver endothelial

cells in mice and expands ST2+CD4+ lymphocytes. Am J Physiol Gastrointest

Liver Physiol 309, G542-553, doi:10.1152/ajpgi.00398.2014 (2015).

108 Carriere, V. et al. IL-33, the IL-1-like cytokine ligand for ST2 receptor, is a

chromatin-associated nuclear factor in vivo. Proc Natl Acad Sci U S A 104, 282-

287, doi:10.1073/pnas.0606854104 (2007).

109 Liew, F. Y., Girard, J. P. & Turnquist, H. R. Interleukin-33 in health and disease.

Nat Rev Immunol 16, 676-689, doi:10.1038/nri.2016.95 (2016).

38

110 Lefrancais, E. & Cayrol, C. Mechanisms of IL-33 processing and secretion:

differences and similarities between IL-1 family members. European cytokine

network 23, 120-127, doi:10.1684/ecn.2012.0320 (2012).

111 Ohno, T., Morita, H., Arae, K., Matsumoto, K. & Nakae, S. Interleukin-33 in

allergy. Allergy 67, 1203-1214, doi:10.1111/all.12004 (2012).

112 Oboki, K., Ohno, T., Kajiwara, N., Saito, H. & Nakae, S. IL-33 and IL-33

receptors in host defense and diseases. Allergology international : official journal

of the Japanese Society of Allergology 59, 143-160, doi:10.2332/allergolint.10-

RAI-0186 (2010).

113 Mitchell, P. D. & O'Byrne, P. M. Epithelial Derived Cytokines in Asthma. Chest,

doi:10.1016/j.chest.2016.10.042 (2016).

114 Shao, D. et al. Nuclear IL-33 regulates soluble ST2 receptor and IL-6 expression

in primary human arterial endothelial cells and is decreased in idiopathic

pulmonary arterial hypertension. Biochem Biophys Res Commun 451, 8-14,

doi:10.1016/j.bbrc.2014.06.111 (2014).

115 Shan, J. et al. Epithelial-derived nuclear IL-33 aggravates inflammation in the

pathogenesis of reflux esophagitis. Journal of gastroenterology 50, 414-423,

doi:10.1007/s00535-014-0988-1 (2015).

116 Shan, J. et al. Interferon gamma-Induced Nuclear Interleukin-33 Potentiates the

Release of Esophageal Epithelial Derived Cytokines. PloS one 11, e0151701,

doi:10.1371/journal.pone.0151701 (2016).

39

117 Ali, S. et al. The dual function cytokine IL-33 interacts with the transcription factor

NF-kappaB to dampen NF-kappaB-stimulated gene transcription. J Immunol 187,

1609-1616, doi:10.4049/jimmunol.1003080 (2011).

118 Gautier, V. et al. Extracellular IL-33 cytokine, but not endogenous nuclear IL-33,

regulates protein expression in endothelial cells. Scientific reports 6, 34255,

doi:10.1038/srep34255 (2016).

119 Zhao, W. & Hu, Z. The enigmatic processing and secretion of interleukin-33.

Cellular & molecular immunology 7, 260-262, doi:10.1038/cmi.2010.3 (2010).

120 Kouzaki, H., Iijima, K., Kobayashi, T., O'Grady, S. M. & Kita, H. The danger

signal, extracellular ATP, is a sensor for an airborne allergen and triggers IL-33

release and innate Th2-type responses. J Immunol 186, 4375-4387,

doi:10.4049/jimmunol.1003020 (2011).

121 Hara, K. et al. Airway uric acid is a sensor of inhaled protease allergens and

initiates type 2 immune responses in respiratory mucosa. J Immunol 192, 4032-

4042, doi:10.4049/jimmunol.1400110 (2014).

122 Wen, T., Besse, J. A., Mingler, M. K., Fulkerson, P. C. & Rothenberg, M. E.

Eosinophil adoptive transfer system to directly evaluate pulmonary eosinophil

trafficking in vivo. Proc Natl Acad Sci U S A 110, 6067-6072,

doi:10.1073/pnas.1220572110 (2013).

123 Joulia, R., L'Faqihi, F. E., Valitutti, S. & Espinosa, E. IL-33 fine tunes mast cell

degranulation and chemokine production at the single-cell level. J Allergy Clin

Immunol, doi:10.1016/j.jaci.2016.09.049 (2016).

40

124 Bessa, J. et al. Altered subcellular localization of IL-33 leads to non-resolving

lethal inflammation. Journal of autoimmunity 55, 33-41,

doi:10.1016/j.jaut.2014.02.012 (2014).

125 Lefrancais, E. et al. IL-33 is processed into mature bioactive forms by neutrophil

elastase and cathepsin G. Proc Natl Acad Sci U S A 109, 1673-1678,

doi:10.1073/pnas.1115884109 (2012).

126 Lefrancais, E. et al. Central domain of IL-33 is cleaved by mast cell proteases for

potent activation of group-2 innate lymphoid cells. Proc Natl Acad Sci U S A 111,

15502-15507, doi:10.1073/pnas.1410700111 (2014).

127 Cohen, E. S. et al. Oxidation of the alarmin IL-33 regulates ST2-dependent

inflammation. Nature communications 6, 8327, doi:10.1038/ncomms9327 (2015).

128 Gordon, E. D. et al. Alternative splicing of interleukin-33 and type 2 inflammation

in asthma. Proc Natl Acad Sci U S A 113, 8765-8770,

doi:10.1073/pnas.1601914113 (2016).

129 De la Fuente, M., MacDonald, T. T. & Hermoso, M. A. The IL-33/ST2 axis: Role

in health and disease. Cytokine & growth factor reviews 26, 615-623,

doi:10.1016/j.cytogfr.2015.07.017 (2015).

130 Judd, L. M. et al. Elevated IL-33 expression is associated with pediatric

eosinophilic esophagitis, and exogenous IL-33 promotes eosinophilic esophagitis

development in mice. Am J Physiol Gastrointest Liver Physiol 310, G13-25,

doi:10.1152/ajpgi.00290.2015 (2016).

131 Simon, D., Radonjic-Hosli, S., Straumann, A., Yousefi, S. & Simon, H. U. Active

eosinophilic esophagitis is characterized by epithelial barrier defects and

41

eosinophil extracellular trap formation. Allergy 70, 443-452, doi:10.1111/all.12570

(2015).

132 Venturelli, N. et al. Allergic skin sensitization promotes eosinophilic esophagitis

through the IL-33-basophil axis in mice. J Allergy Clin Immunol 138, 1367-

1380.e1365, doi:10.1016/j.jaci.2016.02.034 (2016).

133 Lee, H. Y. et al. Blockade of IL-33/ST2 ameliorates airway inflammation in a

murine model of allergic asthma. Experimental lung research 40, 66-76,

doi:10.3109/01902148.2013.870261 (2014).

134 Tjota, M. Y. et al. IL-33-dependent induction of allergic lung inflammation by

FcgammaRIII signaling. J Clin Invest 123, 2287-2297, doi:10.1172/jci63802

(2013).

42

Figure 1.1: Pathogenesis of eosinophilic esophagitis (EoE). TSLP and IL-33 are released from the epithelium and activate basophils and food antigen-presenting dendritic cells to induce Th2 polarization of naïve CD4+ T cells. Th2 polarization is aided by miR-21 that represses Th1 polarization by degradation of IL-12. These Th2 cells, in addition to invariant natural killer T (iNKT) cells and IL-13-producing FoxP3+ T cells, then secrete IL-13 that increases CCL26, calpain 14 (CAPN14), and periostin (POSTN) expression and decreases desmoglein1 (DSG1) expression in the epithelium. Decreased DSG1 levels impairs barrier function that forms a propagation loop by allowing further penetration of food antigen, and also leads to increased POSTN levels. The increased CCL26 and POSTN promote eosinophil recruitment from the bloodstream. The accumulating activated eosinophils further increase POSTN expression via the release of TGF-β and also cause epithelial cell cytotoxicity. CCL, CC-chemokine ligand; DC. Dendritic cell; ECP, eosinophil cationic protein; EDN, eosinophil-derived neurotoxin; Eos, eosinophil; EPO, eosinophil peroxidase; Fibro, fibroblast; IL, interleukin; MBP, major basic protein; miR, microRNA; TGF, transforming growth factor; Th0, naïve T helper cell; Th1, type 1 T helper cell; Th2, type 2 T helper cell; Treg, regulatory T cell; TSLP, thymic stromal lymphopoietin.

43

Figure 1.2: IL-33 protein structure. Restriction to the nucleus and direct interaction with chromatin is mediated by amino acid residues 44-58, termed the chromatin binding domain (CBD). No classical nuclear localization sequence is present. Amino acid residues 58-112 constitute the central domain, which contain cleavage sites for inflammatory proteases derived from mast cells and neutrophils. The IL-1-like cytokine domain (residues 112-270) mediates ST2 receptor binding and activation.

44

Figure 1.3: Interaction of alarmins with chromatin. Each nucleosome is composed of DNA nucleotides (black line) wrapped around a histone core (blue circle) of two copies each of histones H2A, H2B, H3, and H4. IL-33 (orange circle) directly binds H2A and H2B within the histone core. IL-1α (pink circle) and HMGB1 (gray circle) instead bind DNA nucleotides directly. DNA, deoxyribonucleic acid; HMGB1, high mobility group B 1; IL, interleukin.

45

Modification Site ST2 Bioactivity Mediators

Cleavage within central domain by Elastase, cathepsin G, Extracellular +++ inflammatory proteases chymase, and tryptase

Cleavage within IL-1-like cytokine Intracellular - Caspases 3 and 7 domain during apoptosis

Deletion of exons 3 or 4 (residues Intracellular +?* Alternative splicing 31-114)

Oxidation of cysteine residues 208, Extracellular - Unknown 227, 232, and 259

* These forms have been shown to activate ST2, but their bioactivity has not been compared to other forms of IL-33

Table 1.1. Post-translational modifications and alternate forms of IL-33. Table lists different post-translation modifications or alternate forms of IL-33, whether they occur before or after extracellular release, their ability to bind ST2 and initiate downstream signaling (“ST2 bioactivity”), and known mediators. It is currently not known which induces cysteine oxidation of extracellular IL-33.

46

CHAPTER 2: PROFOUND LOSS OF ESOPHAGEAL TISSUE DIFFERENTIATION IN

EOSINOPHILIC ESOPHAGITIS

J. Travers BS1#, M. Rochman PhD1#, C.E. Miracle1, M. C. Bedard BS1, T. Wen PhD1, N.P. Azouz PhD1, J.M. Caldwell PhD1, K. KC BS1, J.D. Sherrill PhD1, B.P. Davis MD PhD3, J.K. Rymer MSc1, K. M. Kaufman PhD2 and M.E. Rothenberg MD PhD1*

1Division of Allergy and Immunology, Cincinnati Children’s Hospital Medical Center, Cincinnati, Ohio 45229-3039, USA 2Center for Autoimmune Genomics and Etiology, Department of Pediatrics, Division of Rheumatology, Cincinnati Children's Hospital Medical Center, University of Cincinnati, Cincinnati VA Medical Center, Research Department, Cincinnati, Ohio, 45229-3039, USA. 3University of Iowa Hospitals and Clinics Department of Internal Medicine, Division of Immunology, 200 Hawkins Dr., Iowa City, IA 52242

# these authors equally contributed to the paper

*Corresponding author: Marc E. Rothenberg Division of Allergy and Immunology Cincinnati Children's Hospital Medical Center 3333 Burnet Avenue, MLC 7028 Cincinnati, OH 45229-3039 Phone: 513-803-0257 Fax: 513-636-3310 E-mail: [email protected]

Conflict of interest: M.E.R. has received payment from Immune Pharmaceuticals, Merck and Receptos and has an equity interest in PulmOne, NKT Therapeutics, Immune Pharmaceuticals and royalties from reslizumab (Teva Pharmaceuticals). M.E.R. is an inventor of several patents, owned by Cincinnati Children’s, and a set of these patents relates to molecular diagnostics of eosinophilic esophagitis. All of the other authors have no potential conflicts to disclose.

Funding support: This work was supported by National Institutes of Health R37 AI045898, R01 AI124355, U19 AI070235, U19 AI066738 (CoFAR supported by the

47

National Institute of Allergy and Infectious Diseases and National Institute of Diabetes and Digestive and Kidney Diseases), and P30 DK078392 (Gene and Protein Expression Core); the Campaign Urging Research for Eosinophilic Disease (CURED); the Buckeye Foundation; and the Sunshine Charitable Foundation and its supporters, Denise A. Bunning and David G. Bunning.

Acknowledgements: We thank Shawna Hottinger for editorial assistance.

Author contributions

M.R. and J.T. performed experiments, data analysis and wrote the paper; N.P.A., J.M.C., B.P.D., J.K.R, K.K, T.W, M.C.B, C.E.M and J.D.S. performed experiments and data analysis; M.E.R. supervised the study.

Running title: Loss of esophageal identity in eosinophilic esophagitis

48

Abstract

Background

A key question in the allergy field is to understand how tissue specific disease is manifested. Eosinophilic esophagitis (EoE) is an emerging tissue specific allergic disease whose pathogenesis remains unclear.

Objective

Herein, we tested the hypothesis that a defect in tissue specific esophageal genes is an integral part of EoE pathogenesis.

Methods

We interrogated the pattern of expression of esophagus-specific signature genes derived from the Human Protein Atlas in the EoE transcriptome and in EPC2 esophageal epithelial cells. Western blot and immunofluorescence were used for evaluating expression of esophageal proteins in control and active EoE biopsies.

Whole-exome sequencing (WES) was performed to identify mutations in esophagus- specific genes.

Results

We found that ~39% of the esophagus-specific transcripts were altered in EoE, with

~90% being downregulated. The majority of transcriptional changes observed in esophagus-specific genes were reproduced in vitro in esophageal epithelial cells

49 differentiated in the presence of IL-13. Functional enrichment analysis revealed keratinization and differentiation as the most affected biological processes, and identified IL-1 cytokines and serine peptidase inhibitors (SERPINs) as the most dysregulated esophagus-specific protein families in EoE. Accordingly, EoE biopsies evidenced a profound loss of tissue differentiation, decreased expression of keratin 4 and cornulin and elevated expression of keratin 5 and 14. Whole-exome sequencing of

33 unrelated EoE cases revealed 39 rare mutations in 18 esophagus-specific differentially expressed genes.

Conclusions

A tissue-centered analysis has revealed a profound loss of esophageal tissue differentiation as an integral and specific part of the pathophysiology of EoE, and implicated protease- and IL-1related activities as putative central pathways in disease pathogenesis.

50

Key messages:

 Esophagus-specific genes are enriched in IL-1 family members and a series of

proteases and protease inhibitors.

 Esophagus-specific transcripts have altered expression in EoE, with nearly all

(~90%) being downregulated.

 EoE involves a loss of tissue differentiation (e.g. loss of tissue identity).

 Organ-specific, imbalanced protease activity is an integral process in allergic

inflammation specific to the esophagus.

 Genetic profiling and assessing expression levels of esophagus-specific genes

may have diagnostic and prognostic value.

Capsule summary

There is a profound impairment of esophageal tissue differentiation including a loss of tissue identity genes in eosinophilic esophagitis, providing potential value for predictive medicine, diagnostics and treatment of this emerging disease.

Keywords: eosinophilic esophagitis,

IL-1 cytokines, protease activity,

IL-13, differentiation, whole-exome sequencing

51

Abbreviations used

EoE: eosinophilic esophagitis

EG: eosinophilic gastritis

ALI: air-liquid interface

WES: whole-exome sequencing

FPKM: fragments per kilobase per million reads

GWAS: genome-wide association study hpf: high-power field

EDC: epidermal differentiation complex

DSG1: Desmoglein-1

SERPIN: serine peptidase inhibitors

SPINK: serine protease inhibitor Kazal-type

GO analysis: gene ontology analysis

CAPN14: calpain 14

TGM: transglutaminase

IVL: involucrin

KRT: keratin

CRNN: cornulin

SPRR: small proline-rich protein

52

Introduction

Recent advances in human proteomics and transcriptomics have led to identification of tissue-specific genes that confer molecular identity to individual tissues, as highlighted by the Human Protein Atlas project that combines tissue-specific proteome data with global gene expression analysis for over 40 human tissues.1 These data sets allow examination of the contribution of the tissue-specific genes to disease pathogenesis, constituting a novel, tissue-centered approach for diagnostics and potentially for predictive purposes.

Eosinophilic esophagitis (EoE) is a chronic allergic inflammatory disease characterized by eosinophil-predominant, tissue-specific, mucosal inflammation. The disease is mediated in large part by an immune response to food allergens. This immune response involves the interaction of the innate immunity (epithelium and eosinophils) with a strong

T helper type 2 (Th-2) adaptive immune response characterized by IL-13 overproduction, resulting in a marked alteration in esophageal gene expression.

Notably, IL-13 directly stimulates the epithelium to produce a large subset of the EoE- associated transcriptome, including the eosinophil chemoattractant and activating factor

CCL26 (eotaxin-3), thus propagating the inflammatory loop.2, 3

Esophageal tissue integrity has a critical role in disease pathogenesis,4-10 as illustrated by profound dysregulation of the epidermal differentiation complex (EDC) genes 11 and decreased expression of desmoglein 1 (DSG1) in active EoE and in IL-13–stimulated

53 esophageal epithelial cells.4 In line with these findings, genes for structural proteins including keratin 6C, keratin 32 and the EDC gene cornulin (CRNN), are among the most highly expressed genes in the homeostatic esophagus on the basis of the Human

Protein Atlas. Recent genome-wide association study (GWAS) analyses identified the cysteine protease calpain 14 (CAPN14), another highly expressed gene in the homeostatic esophagus, as one of the strongest genetic risk factors associated with disease susceptibility,12, 13 suggesting that protease activity is critical for disease pathophysiology. Indeed, CAPN14, as well as serine peptidase inhibitor SERPINB13, were upregulated in EoE biopsies, and dysregulation of CAPN14 in esophageal epithelial cells resulted in impaired barrier function of differentiated epithelium.14

Herein, we interrogated changes in the esophagus-specific transcriptome defined by the

Human Protein Atlas (297 genes expressed > 5 fold higher in esophagus vs. any other human tissue 1) in EoE. We identified 117 esophagus-specific genes that were significantly dysregulated in EoE. Remarkably, ~90% of these genes were downregulated, demonstrating a profound loss of esophageal tissue gene signature in the allergic inflamed esophagus. Functional analysis of the esophagus-specific genes altered in EoE revealed loss of esophageal differentiation, identified proteolytic activity as a central pathway impaired in the diseased esophagus and implicated genes from the SERPIN and IL-1 families as those most significantly affected of the tissue-specific genes in EoE. Quantification of IL-36α, SERPINs B13 and B4 and CRNN proteins in control and inflamed esophageal biopsies corroborated transcriptional data.

Transcriptional changes observed in the inflamed esophagus were largely reproduced

54 in esophageal cells treated with IL-13 in vitro, which showed an impaired epithelial cell differentiation program. Whole-exome sequencing (WES) analysis of 33 unrelated EoE cases identified 39 rare mutations in 18 esophagus-specific genes that were transcriptionally altered in the disease. Functional enrichment analysis identified proteolytic activity and tissue differentiation as the most significant pathways associated with these genes. These data substantiate a potential functional link between the esophageal transcriptome and EoE pathogenesis. Collectively, our study reveals profound loss of esophageal epithelial differentiation as a hallmark of EoE and implicates protease- and IL-1related activities as putative, key, tissue-specific pathways involved in the disease. We propose that the loss of the esophagus-specific molecular signature in EoE is a disease-specific process that has potential value for predictive medicine, diagnostics and treatment of this disease.

55

Materials and Methods

Gene definitions

Definitions related to specificity of expression in the esophagus were from the Human

Protein Atlas (http://www.proteinatlas.org/humanproteome/esophagus). Tissue-enriched genes are expressed in the esophagus at least 5-fold higher than in any other tissue transcriptionally profiled by the Human Protein Atlas, group-enriched genes are expressed at least 5-fold higher in the group of 2-7 tissues, and tissue-enhanced genes are expressed at least 5-fold higher in the esophagus compared to the average of expression in all tissues.

Expression data and whole-exome sequencing

Publically available expression data were used for esophageal biopsies of patients with

EoE or controls3 (GSE58640), antrum or antrum/body biopsies of patients with EG or controls15 (GSE54043), TE-7 cells 42 (GSE57637) and EPC2 ALI cultures 21

(GSE58640, GSE65335). There was no statistical difference (Mann-Whitney U-test p =

0.26) in ages between patients in EoE and control groups used for RNA sequencing analysis. 3 Lists of esophagus- and stomach-specific genes were obtained from the

Human Protein Atlas (http://www.proteinatlas.org). Gene ontology (GO) enrichment analysis, which uses statistical methods to determine functional pathways and cellular processes associated with a given set of genes, was performed using ToppGene suite

(https://toppgene.cchmc.org/16); Venny

(http://bioinfogp.cnb.csic.es/tools/venny/index.html) was used to intersect gene lists;

56

Cluster 3.0 was applied for clustering genes, and Java TreeView was used for visualization of heat maps

(http://bonsai.hgc.jp/~mdehoon/software/cluster/software.htm). WES analysis was performed and analyzed as in Patel ZH et al.24 Briefly, only variant calls that had a read depth greater than 15 and a genotype quality score greater than 20 were included. In addition, variants were removed that had a minor allele frequency greater than 1% in the general population (based on 1000 genomes project

(http://www.1000genomes.org/) or the NHLBI Exome Sequencing Project

(https://esp.gs.washington.edu) and were required to alter the amino acid sequence of protein.

Western blot analysis

Proteins from control and EoE biopsies were isolated by Trizol, subjected to electrophoresis on a 4-12% protein gel and probed with anti-mouse HSP90 (BD

Bioscience), goat anti–IL-36α (R&D), rabbit anti-SERPINB13 (Sigma Aldrich), goat anti- cornulin (AF3607,R&D) and mouse anti-SERPINB4 (Santa Cruz) antibodies. Secondary

IRDye-conjugated antibodies were from Licor. Quantification of signal was performed using Image Studio Lite software

(http://www.licor.com/bio/products/software/image_studio_lite/). Signal from the tested proteins was normalized to the signal from the loading control (HSP90) following normalization to the level of the protein in one of the EoE samples. Statistical analysis was performed in GraphPad software.

57

Air-liquid interface (ALI) culture system

The ALI differentiation protocol was performed essentially as described 4, 21. Briefly,

EPC2 immortalized esophageal epithelial cells were seeded onto semi-permeable membranes (0.4 μm) and grown to confluence in standard low-calcium medium (KSFM with pituitary extract and EGF at 1 ng/ml, 0.09 mM calcium concentration). Epithelial differentiation was induced by the addition of KSFM with high calcium (1.8mM final concentration) over the course of five days (initial differentiation). Medium was replaced in both chambers every other day. Stratification was induced by removing the media from the upper chamber and exposing the cells to the air interface for a period of 5-6 days (terminal differentiation). Cells remained exposed to high calcium medium in the bottom chamber throughout terminal differentiation. Treatment with IL-13 (100 ng/mL) in the lower chamber started at the first day of exposure to the air interface (day 8), with fresh media with IL-13 added every other day. In some experiments, cultures were treated with IL-1α (10 ng/ml) for the last 2 days of ALI culture. At the last day of ALI culture, cells were collected for analysis.

Immunofluorescence analysis

Immunofluorescence staining was performed as previously described 4 on at least four distal esophageal biopsies from control individuals, distal esophageal biopsies with active inflammation from patients with active EoE, or proximal esophageal biopsies without active inflammation from the same patients with active EoE. The average ages

58 of EoE patients and control individuals were 13 and 11 years, respectively. Patients were also controlled for race and gender. The following primary antibodies were used at a final concentration of 1 g/mL: antikeratin 14 (PRB-155P, Covance), antikeratin 5

(ab24647, Abcam), antikeratin 4 (HPA034881, Sigma), or anti-cornulin (AF3607,

R&D). Nuclei were stained with 4’,6-diamidino-2-phenylindole (DAPI). Slides were blocked with phosphate-buffered saline with 10% goat or donkey serum. Secondary antibodies (1:500 dilution) used were donkey anti-rabbit Alexa Fluor 647 (A31573, Life

Technologies), goat anti-rabbit Alexa Fluor 647 (A21244, Life Technologies), or donkey anti-goat Alexa Fluor 488 (A11055, Life Technologies). Imaging was performed with a

Nikon A1 inverted confocal microscope.

59

Results

Specific loss of esophageal signature genes in EoE

To test the hypothesis that esophagus-specific genes contribute to EoE pathogenesis, we intersected genes transcriptionally altered in EoE 3 with various organ-specific transcriptomes obtained from the Human Protein Atlas, including esophagus, salivary gland, gastrointestinal tract organs, kidney, lung, skin and testis. Genes in the transcriptomes are classified as tissue-enriched (expressed at least 5 fold higher in the interrogated tissue than in any other tissue), group-enriched (at least 5 fold higher expression in the group of tissues compared to other tissues) and tissue-enhanced (at least 5 fold higher expression in the interrogated tissue compared to average level in all tissues). We found that the percentage of esophagus-specific genes altered in EoE was at least two fold higher than that of genes from other organ-specific transcriptomes

(Figure 1A). From 297 genes whose expression is specifically enriched in the esophagus relative to other tissues, we found 117 genes overlapping with the EoE transcriptome (designated as Eso-EoE genes), which represent 39% of esophageal signature genes and 7% of genes altered in EoE (Figure 1B). Notably, 89% of Eso-EoE genes (104 of 117 genes) were significantly downregulated in the esophageal tissue of patients with EoE (Figure 1B). Of the genes most highly expressed at the mRNA level in the homeostatic esophagus, 12 genes were expressed ~10-94 fold higher in the esophagus compared to other interrogated tissues. Of these, only CAPN14 and

SERPINB13 were significantly upregulated in EoE (Figure 1C). Overall, for the 117 tissue-enriched, tissue-enhanced and group-enriched genes, only 13 genes were significantly upregulated in EoE (Figure 1B). A similar comparison between stomach-

60 specific signature genes and the eosinophilic gastritis (EG) transcriptome 15 identified only 16 overlapping genes, representing 8% of the stomach transcriptome and approximately 1.5% of the transcripts dysregulated in EG (Figure 1D). These data suggest that loss of the tissue-specific molecular signature of the esophagus is an integral and specific part of the pathophysiology of EoE.

Functional enrichment analysis of Eso-EoE genes

By performing functional enrichment gene ontology (GO) analysis 16 of Eso-EoE genes, we identified endopeptidase inhibitor activity and keratinization as the most profoundly impaired molecular functions and biological processes (p<10-9 and p<10-14, respectively;

Figure 2A, B). Accordingly, a number of SERPIN-related genes were upregulated

(SERPINB2, B3, B4, B13), whereas genes from the serine protease inhibitor Kazal-type

(SPINK) family were downregulated (SPINK5, 7, 8). Notably, increased expression of

SERPINs, specifically SERPINE1 (PAI-1), has been previously reported in EoE and asthma and correlated with disease severity. 17, 18 The downregulated genes also included the differentiation markers keratin (KRT6B) and involucrin (IVL) and the small proline-rich proteins (SPRRs) that comprise members of the EDC previously implicated in EoE pathogenesis.11 Transglutaminases 1 and 3 (TGM1, TGM3), associated with pathogenesis of the rare cornification disease autosomal recessive congenital ichthyosis19, were also downregulated in EoE. In line with dysregulation of structural epithelial genes, the most significantly affected human phenotypes were abnormality of keratinization, epidermal thickening, hyperkeratosis and skin inflammation (Figure 2C).

61

These findings revealed a profound loss of esophageal tissue differentiation in EoE and a potentially critical role of esophagus-specific proteases in supporting the functional integrity of the epithelial barrier of the esophagus.

We next analyzed the properties of Eso-EoE genes on the basis of the common protein domains. The two top groups of genes most highly enriched by this analysis were serine protease inhibitors (SERPINs; SERPINB2, B3, B4, B13) and IL-1 family members

(IL36A, IL36RN, IL1A, IL1RN). SERPINs were upregulated (p<10-7), whereas IL-

1related genes were significantly downregulated in EoE (p<10-5; Figure 2D). Other protein domains enriched in Eso-EoE genes were SPINKs, keratins, members of the arginine deiminase family (PADI1, 3) and transglutaminases (TGM1, 3) (data not shown).

Eso-EoE genes are regulated during esophageal epithelial differentiation

We aimed to recapitulate the inflammatory process in a cell system in vitro to test whether esophagus-specific changes in transcription were driven by epithelial rather than a tissue-based, mixed cellular response. For this purpose, we utilized air-liquid interphase (ALI) culture of an esophageal epithelial cell line, EPC2 cells that undergoes squamous cell differentiation.20 Submerged EPC2 cells were grown in low calcium (0.09 mM) and subjected to initial differentiation by exposure to a high concentration of calcium (1.8 mM) followed by terminal differentiation at the ALI with and without IL-13 stimulation 21 (Figure 3A). Initial differentiation with high calcium triggered dramatic

62 transcriptional changes in submerged culture with 3146 genes changing expression.

Overlapping these genes with Eso-EoE genes revealed that 90 Eso-EoE genes (77%) were transcriptionally altered during initial differentiation (Figure 3B), with the vast majority being upregulated in the initial differentiation in submerged culture but downregulated in the biopsies of patients with EoE (Figure 3B, heat map). These results demonstrate the biological relevance of ALI EPC2 culture for studying esophageal epithelial differentiation.

IL-13 impacts the expression of Eso-EoE genes

We assessed the effect of IL-13 under conditions of terminal differentiation of epithelial cells at ALI culture (Figure 4A). We have previously shown that terminal differentiation of EPC2 cells in the presence of IL-13 leads to altered expression of epithelial differentiation genes.21 We identified 68 Eso-EoE genes (61% of Eso-EoE genes) affected in ALI culture: expression of 32 genes was altered under all conditions (Figure

4A, Group 1), 8 genes were altered in response to IL-13 stimulation but not by differentiation (Figure 4A, Group 2), and 28 genes were affected by the differentiation process but not by IL-13 (Figure 4A, Group 3). The gene expression pattern of Eso-

EoE genes in ALI cultures differentiated in the presence of IL-13 became strikingly similar to that of the biopsies of patients with EoE (32 genes, Group 1). For example,

CAPN14, SERPINB3 and SERPIN4 were upregulated in both IL-13–treated ALI cultures and biopsies, whereas SPINK5 and genes from IL-1 and SPRR families were downregulated in both. We observed similar patterns of gene expression in biopsy and

63 in ALI cultures differentiated in the presence of IL-13 independently of whether the genes were significantly affected by the differentiation process (8 genes, Group 2). The only exception was SERPINB2, which was downregulated in ALI culture. Epithelial differentiation led to changes of 28 genes that were refractory to IL-13 treatment (Group

3). Notably, expression of 16 genes from this group inversely correlated with expression in biopsies: the genes were upregulated in differentiated ALI culture but downregulated in the disease. We further delineated expression of IL-1 and SERPINB gene families in

ALI system compared to active EoE biopsies. Consistent with other Eso-EoE genes, the expression pattern for these genes was reproduced in ALI culture for 5 out of 8 genes with IL-1 family members downregulated and SERPINs upregulated in IL-13—treated cells (Figure 4B).

In light of downregulated expression of IL-1α in EoE biopsies and ALI cultures, we tested whether replenishment of IL-1α would reverse gene expression in IL-13—treated cells subjected to ALI culture. Addition of IL-1α to the cells subjected to ALI either alone or in the presence of IL-13 had no effect on the integrity of the culture, as measured by

TEER, and did not reverse expression of the prototypic IL-13 target genes CCL26 and

SERPINB4 (Supplemental Figure 1).

Additionally, in order to rule whether that this dysregulation of Eso-EoE genes is a common feature of esophageal disease in general, we assessed the changes in expression of Eso-EoE genes in gastroesophageal reflux disease (GERD), which also has epithelial damage and basal zone hyperplasia. Expression of none of 9 Eso-EoE genes tested was significantly changed in GERD (Figure 4C). Collectively, these results indicate that the downregulation of esophageal signature genes in EoE is likely

64 reflective of an intrinsic defect in esophageal epithelial cell differentiation in response to

IL-13.

Quantifying Eso-EoE proteins in esophageal biopsies

We sought to assess levels of expression of Eso-EoE proteins including structural proteins, proteins related to proteolytic activity and IL-1 cytokines in the lysates of esophageal biopsies isolated from unaffected controls and patients with active EoE

(maximum esophageal eosinophil count of >15 eosinophils/high-power microscopic field

(hpf)). Following functional enrichment analysis, we focused on serine protease inhibitors SERPINB4 and B13, structural protein CRNN, and IL-1 cytokine family member IL-36α. Western blot analysis revealed significant changes in the levels of all tested proteins, which paralleled gene expression patterns (Figure 5A, B), supporting changes in the differentiation, protease activity and IL-1 cytokines in the esophageal tissue in EoE.

Expression of epithelial differentiation markers in the biopsies of patients with active EoE

Expression of genes related to keratinocyte differentiation was most significantly enriched in the Eso-EoE gene set; all of these genes were downregulated in EoE. We therefore assessed the differentiation status of epithelial tissue in the biopsies of patients with active EoE. We applied immunofluorescence to test expression of keratins

65 as common markers of differentiation, as well as CRNN, a member of the EDC specifically expressed in esophagus (Figure 1B). In the biopsies of patients with active

EoE, we observed increased expression of keratin 5 and 14 linked to undifferentiated epithelium and a dramatic loss of keratin 4 associated with terminally differentiated tissue 23 (Figure 5C-E). Esophagus-specific structural gene CRNN was highly expressed in the suprabasal layers of normal esophageal tissue and was downregulated in EoE following expansion of the basal zone of the epithelium (Figure

5F). Furthermore, the increased expression of basal keratins and decreased expression of differentiated keratins does not occur throughout the entire length of the esophagus, but rather in areas of active inflammation. Collectively, these data are consistent with marked loss of differentiation of the esophageal tissue in EoE.

Whole-Exome Sequencing (WES) analysis of Eso-EoE genes

Since loss of esophagus-specific genes appeared to be a hallmark of EoE, we reasoned that the disease might be associated with an increased burden of rare, damaging mutations in Eso-EoE genes. To test this hypothesis, we performed WES of 33 unrelated EoE cases to identify variants with the potential to alter the biological function of any of the proteins in the list of the 117 Eso-EoE genes. We only considered rare variants (less than 1% minor allele frequency based on public databases) that passed a number of quality control measurements 24 and significantly altered the amino acid sequence (gaining or losing a stop codon, altering the open reading frame, disrupting the initiation methionine codon or altering splicing). In addition, we examined rare

66 variants present in at least two cases or genes with at least two different amino acidaltering variants in separate patients. We identified 39 variants that were located in

18 genes (Table 1). No variants were identified in 8 cases, 1 variant was identified in 12 cases, and 2 and 3 of these variants were present in 7 and 6 cases, respectively.

Among the genes identified as mutated in patients with EoE were those of the protease inhibitors SERPINB3 and SPINK5, CAPN14 and structural protein KRT6B. Interestingly, mutations in a multi-subunit chloride channel gamma-aminobutyric acid (GABA) A receptor, pi (GABRP), which was primarily studied in central nervous system, were identified in 4 out of 33 patients with EoE. Notably, functional enrichment analysis of these genes 16 identified the most significant molecular function as serine-type endopeptidase inhibitor activity (p= 8.6X10-5) and biological processes involved in epidermal cell differentiation (p=1.2X10-5). In order to test whether the number of mutations is correlated with disease severity we have analyzed a number of eosinophils per hpf in the biopsies of the patients with different numbers of variants. We were unable to stratify EoE patients based on the number of variants, probably due to a small number of patients in the groups (data not shown). Collectively, the WES data supported the view that there is a key role for esophagus-specific genes in esophageal differentiation and EoE pathogenesis.

67

Discussion

Herein we tested the hypothesis that dysregulation of esophagus-specific genes is an integral part of EoE pathogenesis. By comparing the esophageal gene signature with the transcriptome of the esophagus of EoE patients,3 we identified that ~39% of esophageal genes were altered in EoE (designated Eso-EoE genes); the vast majority of these genes were downregulated in the disease. Functional enrichment analysis of

Eso-EoE genes revealed major alterations in keratinization and differentiation, protease- related activities and IL-1related genes. We observed that the expression pattern of

Eso-EoE genes from the biopsies of active EoE patients and in ALI cultures differentiated in the presence of IL-13 were remarkably similar, providing evidence that loss of esophageal differentiation does not merely reflect a change in the cellular constituents or expansion of the basal epithelial zone. Exposure of epithelial cells to IL-

13 led to transcriptional and morphological changes in the epithelium often observed in allergic inflammation, including induction of eotaxin and loss of epithelial integrity.2, 4, 11,

25, 26 In this regard, our data reinforce the role of IL-13 as a molecular driver of EoE, consistent with recent findings from anti–IL-13 treatment of EoE, which affected tissue differentiation as demonstrated by normalized expression of desmoglein-1, keratin 14 and 16 and tissue-remodeling genes, such as periostin.29 While IL-13 is considered critical for EoE pathogenesis, a number of other cytokines and chemokines have been implicated in the disease responses.27 These molecules, such as the pro-inflammatory cytokine TNFα, likely contribute at least in part to EoE pathogenesis by altering the expression of esophageal-specific genes.28

68

Inflammation-driven epithelial responses, such as impaired barrier function, are common for epithelia of different origin and constitute a fundamental property of allergic inflammation.4, 30, 31 This is supported by GWAS, which identified genetic association of epithelial cell genes including TSLP/WDR36, IL33 and its receptor ST2 (IL1RL1) with atopic sensitization and various allergic diseases.32 GWAS identified filaggrin, a skin tissue-enriched gene, as strongly genetically associated with atopic dermatitis.33, 34 In the case of EoE, recent GWAS has identified genetic association with the esophagus- specific gene CAPN14.12 Analysis of genes within 25 kb of variants with combined association P-value < 1 × 10−4 revealed two additional esophagus-specific genessodium channel, non-voltage gated 1 beta subunit (SCNN1B) and short chain dehydrogenase/reductase family 9C, member 7 (SDR9C7). Notably, differential expression of these genes was the part of the signature that was sufficient to separate control from EoE cases.12 By identifying rare mutations in 18 Eso-EoE gene by WES analysis, including CAPN14, we provide additional supportive biological relevance for the role of esophagus-specific genes in the pathogenesis of EoE. Furthermore, functional analysis of Eso-EoE genes, as well as mutated Eso-EoE genes, has identified protease-related activity as a putative operational pathway in disease pathogenesis. Among the genes most significantly altered in EoE were those for the serine protease inhibitors SERPINBs and SPINKs, supporting a recently proposed critical role for the balance in serine proteases activity in EoE pathogenesis.35

69

An unexpected finding in our study was delineating IL-1 family members as esophagus- specific gene products. IL-36α directly activates naïve CD4 cells, inducing IL-2 secretion and thereby priming a Th1 inflammatory response.36 Given the potential role of IL- in the pathogenesis of inflammation,37 we assessed IL-36α protein levels in the biopsies of patients with active EoE and controls (Figure 3). We found that IL-36α protein was significantly decreased in EoE, implicating IL-36α in balancing Th1/Th2 responses in the disease. We have previously reported elevated levels of IL-1α protein in the blood and decreased esophageal expression of IL1RN mRNA in patients with active EoE.27

Additionally, decreased expression of IL-1α and IL-1Rα has been observed in active

EoE biopsies.38, 39 Collectively, these data implicate IL-1–related genes in the pathogenesis of EoE, presumably through modulating a balance of inflammatory and allergic signals in the context of esophageal tissue.40 A specific feature of the IL-1 family of cytokines is a requirement for N-terminal cleavage for enhanced biological activity.37

It has been reported that cysteine and serine proteases, such as calpains and elastase, can directly cleave IL-1 cytokines.41 Divergent expression of SERPINs and IL-1 cytokines in EoE biopsies suggests a functional link between proteolytic activity and IL-

1–related cytokine activity in disease pathogenesis. Taken together, our data implicate

IL-1–related proteins in disease progression. We speculate that IL-1 family members selectively expressed in the esophagus provide homeostatic regulation of the surface stimuli encountered by esophageal mucosa.

The finding that the esophagus normally expresses IL-1 family members, as well as protease inhibitors that likely curtail inflammation, suggests a previously unappreciated

70 role for basal innate immunity in the esophagus. We propose that the esophagus is normally equipped with these processes to respond successfully to its robust exposure to a variety of ingested foreign substances (e.g. food). An imbalance of esophageal homeostatic processes can contribute to the development of various diseases of the upper gastrointestinal tract and perhaps lead to systemic antigen sensitization, especially in view of the impaired barrier function that develops when epithelial differentiation is lost.4, 6 Our findings expand our understanding of the propagation of allergic inflammation on the level of tissue molecular identity and suggest that genetic profiling and assessing expression levels of esophagus-specific genes may have diagnostic and prognostic value. We propose that the altered esophageal transcriptome driven by organ-specific imbalanced protease activity leads to profound loss of esophageal differentiation, which constitutes an integral process in allergic inflammation specific to the esophagus (Figure 6).

71

References

1. Uhlen M, Fagerberg L, Hallstrom BM, Lindskog C, Oksvold P, Mardinoglu A, et

al. Proteomics. Tissue-based map of the human proteome. Science 2015;

347:1260419.

2. Blanchard C, Wang N, Stringer KF, Mishra A, Fulkerson PC, Abonia JP, et al.

Eotaxin-3 and a uniquely conserved gene-expression profile in eosinophilic

esophagitis. J Clin Invest 2006; 116:536-47.

3. Sherrill JD, Kiran KC, Blanchard C, Stucke EM, Kemme KA, Collins MH, et al.

Analysis and expansion of the eosinophilic esophagitis transcriptome by RNA

sequencing. Genes Immun 2014; 15:361-9.

4. Sherrill JD, Kc K, Wu D, Djukic Z, Caldwell JM, Stucke EM, et al. Desmoglein-1

regulates esophageal epithelial barrier function and immune responses in

eosinophilic esophagitis. Mucosal Immunol 2014; 7:718-29.

5. Simon D, Radonjic-Hosli S, Straumann A, Yousefi S, Simon HU. Active

eosinophilic esophagitis is characterized by epithelial barrier defects and

eosinophil extracellular trap formation. Allergy 2015; 70:443-52.

6. Simon D, Radonjic-Hosli S, Straumann A, Yousefi S, Simon HU. Active

eosinophilic esophagitis is characterized by epithelial barrier defects and

eosinophil extracellular trap formation. Allergy 2015.

7. Katzka DA, Ravi K, Geno DM, Smyrk TC, Iyer PG, Alexander JA, et al.

Endoscopic Mucosal Impedance Measurements Correlate with Eosinophilia and

72

Dilation of Intercellular Spaces in Patients with Eosinophilic Esophagitis. Clin

Gastroenterol Hepatol 2015.

8. van Rhijn BD, Weijenborg PW, Verheij J, van den Bergh Weerman MA,

Verseijden C, van den Wijngaard RM, et al. Proton pump inhibitors partially

restore mucosal integrity in patients with proton pump inhibitor-responsive

esophageal eosinophilia but not eosinophilic esophagitis. Clin Gastroenterol

Hepatol 2014; 12:1815-23 e2.

9. Capocelli KE, Fernando SD, Menard-Katcher C, Furuta GT, Masterson JC,

Wartchow EP. Ultrastructural features of eosinophilic oesophagitis: impact of

treatment on desmosomes. J Clin Pathol 2015; 68:51-6.

10. Ravelli A, Villanacci V, Cadei M, Fuoti M, Gennati G, Salemme M. Dilated

intercellular spaces in eosinophilic esophagitis. J Pediatr Gastroenterol Nutr

2014; 59:589-93.

11. Blanchard C, Stucke EM, Burwinkel K, Caldwell JM, Collins MH, Ahrens A, et al.

Coordinate interaction between IL-13 and epithelial differentiation cluster genes

in eosinophilic esophagitis. J Immunol 2010; 184:4033-41.

12. Kottyan LC, Davis BP, Sherrill JD, Liu K, Rochman M, Kaufman K, et al.

Genome-wide association analysis of eosinophilic esophagitis provides insight

into the tissue specificity of this allergic disease. Nat Genet 2014; 46:895-900.

13. Sleiman PM, Wang ML, Cianferoni A, Aceves S, Gonsalves N, Nadeau K, et al.

GWAS identifies four novel eosinophilic esophagitis loci. Nat Commun 2014;

5:5593.

73

14. Benjamin P. Davis EMS, M. Eyad Khorki, Vladislav A. Litosh, Jeffrey K. Rymer,

Mark Rochman, Jared Travers, Leah C. Kottyan, and Marc E. Rothenberg.

Eosinophilic esophagitis–linked calpain 14 is an IL-13–induced protease that

mediates esophageal epithelial barrier impairment. JCI insight 2016; 1(4).

15. Caldwell JM, Collins MH, Stucke EM, Putnam PE, Franciosi JP, Kushner JP, et

al. Histologic eosinophilic gastritis is a systemic disorder associated with blood

and extragastric eosinophilia, TH2 immunity, and a unique gastric transcriptome.

J Allergy Clin Immunol 2014; 134:1114-24.

16. Chen J, Bardes EE, Aronow BJ, Jegga AG. ToppGene Suite for gene list

enrichment analysis and candidate gene prioritization. Nucleic Acids Res 2009;

37:W305-11.

17. Rawson R, Yang T, Newbury RO, Aquino M, Doshi A, Bell B, et al. TGF-beta1-

induced PAI-1 contributes to a profibrotic network in patients with eosinophilic

esophagitis. J Allergy Clin Immunol 2016; 138:791-800 e4.

18. Ma Z, Paek D, Oh CK. Plasminogen activator inhibitor-1 and asthma: role in the

pathogenesis and molecular regulation. Clin Exp Allergy 2009; 39:1136-44.

19. Herman ML, Farasat S, Steinbach PJ, Wei MH, Toure O, Fleckman P, et al.

Transglutaminase-1 gene mutations in autosomal recessive congenital

ichthyosis: summary of mutations (including 23 novel) and modeling of TGase-1.

Hum Mutat 2009; 30:537-47.

20. Andl CD, Mizushima T, Nakagawa H, Oyama K, Harada H, Chruma K, et al.

Epidermal growth factor receptor mediates increased cell proliferation, migration,

74

and aggregation in esophageal keratinocytes in vitro and in vivo. J Biol Chem

2003; 278:1824-30.

21. Kiran KC MER, and Joseph D. Sherrill. In vitro model for studying esophageal

epithelial differentiation and allergic inflammatory responses identifies keratin

involvement in eosinophilic esophagitis. Plos One 2015; accepted for publication.

22. Wen T, Stucke EM, Grotjan TM, Kemme KA, Abonia JP, Putnam PE, et al.

Molecular diagnosis of eosinophilic esophagitis by gene expression profiling.

Gastroenterology 2013; 145:1289-99.

23. Fuchs E. Keratins as biochemical markers of epithelial differentiation. Trends

Genet 1988; 4:277-81.

24. Patel ZH, Kottyan LC, Lazaro S, Williams MS, Ledbetter DH, Tromp H, et al. The

struggle to find reliable results in exome sequencing data: filtering out Mendelian

errors. Front Genet 2014; 5:16.

25. Lambrecht BN, Hammad H. The airway epithelium in asthma. Nat Med 2012;

18:684-92.

26. Rael EL, Lockey RF. Interleukin-13 signaling and its role in asthma. World

Allergy Organ J 2011; 4:54-64.

27. Blanchard C, Stucke EM, Rodriguez-Jimenez B, Burwinkel K, Collins MH, Ahrens

A, et al. A striking local esophageal cytokine expression profile in eosinophilic

esophagitis. J Allergy Clin Immunol 2011; 127:208-17, 17 e1-7.

28. Straumann A, Bauer M, Fischer B, Blaser K, Simon HU. Idiopathic eosinophilic

esophagitis is associated with a T(H)2-type allergic inflammatory response. J

Allergy Clin Immunol 2001; 108:954-61.

75

29. Rothenberg ME, Wen T, Greenberg A, Alpan O, Enav B, Hirano I, et al.

Intravenous anti-IL-13 mAb QAX576 for the treatment of eosinophilic esophagitis.

J Allergy Clin Immunol 2015; 135:500-7.

30. Jarzab J, Filipowska B, Zebracka J, Kowalska M, Bozek A, Rachowska R, et al.

Locus 1q21 Gene expression changes in atopic dermatitis skin lesions:

deregulation of small proline-rich region 1A. Int Arch Allergy Immunol 2010;

151:28-37.

31. Lieden A, Ekelund E, Kuo IC, Kockum I, Huang CH, Mallbris L, et al. Cornulin, a

marker of late epidermal differentiation, is down-regulated in eczema. Allergy

2009; 64:304-11.

32. Ober C, Yao TC. The genetics of asthma and allergic disease: a 21st century

perspective. Immunol Rev 2011; 242:10-30.

33. Palmer CN, Irvine AD, Terron-Kwiatkowski A, Zhao Y, Liao H, Lee SP, et al.

Common loss-of-function variants of the epidermal barrier protein filaggrin are a

major predisposing factor for atopic dermatitis. Nat Genet 2006; 38:441-6.

34. Muller S, Marenholz I, Lee YA, Sengler C, Zitnik SE, Griffioen RW, et al.

Association of Filaggrin loss-of-function-mutations with atopic dermatitis and

asthma in the Early Treatment of the Atopic Child (ETAC) population. Pediatr

Allergy Immunol 2009; 20:358-61.

35. D'Mello RJ, Caldwell JM, Azouz NP, Wen T, Sherrill JD, Hogan SP, et al.

LRRC31 is induced by IL-13 and regulates kallikrein expression and barrier

function in the esophageal epithelium. Mucosal Immunol 2015.

76

36. Vigne S, Palmer G, Martin P, Lamacchia C, Strebel D, Rodriguez E, et al. IL-36

signaling amplifies Th1 responses by enhancing proliferation and Th1

polarization of naive CD4+ T cells. Blood 2012; 120:3478-87.

37. Gresnigt MS, van de Veerdonk FL. Biology of IL-36 cytokines and their role in

disease. Semin Immunol 2013; 25:458-65.

38. Abdulnour-Nakhoul SM, Al-Tawil Y, Gyftopoulos AA, Brown KL, Hansen M,

Butcher KF, et al. Alterations in junctional proteins, inflammatory mediators and

extracellular matrix molecules in eosinophilic esophagitis. Clin Immunol 2013;

148:265-78.

39. Lu TX, Munitz A, Rothenberg ME. MicroRNA-21 is up-regulated in allergic airway

inflammation and regulates IL-12p35 expression. J Immunol 2009; 182:4994-

5002.

40. Garlanda C, Dinarello CA, Mantovani A. The interleukin-1 family: back to the

future. Immunity 2013; 39:1003-18.

41. Afonina IS, Tynan GA, Logue SE, Cullen SP, Bots M, Luthi AU, et al. Granzyme

B-dependent proteolysis acts as a switch to enhance the proinflammatory activity

of IL-1alpha. Mol Cell 2011; 44:265-78.

42. Rochman M, Kartashov AV, Caldwell JM, Collins MH, Stucke EM, Kc K, et al.

Neurotrophic tyrosine kinase receptor 1 is a direct transcriptional and epigenetic

target of IL-13 involved in allergic inflammation. Mucosal Immunol 2015; 8:785-

98.

77

Figures and Figure Legends

Figure 2.1: Altered transcription of esophagus-specific genes in EoE. In A, the graph shows percentage of organ-specific genes altered in the EoE transcriptome, as assessed by RNA sequencing.3 In B, a Venn diagram shows the overlap of genes specifically expressed in the esophagus according to the Human Protein Atlas (Esophagus) with genes differentially expressed in the esophageal tissue of patients with active EoE compared to control esophageal tissue (EoE). For each group of genes the number of upregulated (Up) and downregulated (Down) genes is shown (not to scale). A set of 117 overlapping genes is designated as Eso-EoE genes. In C, normalized expression levels for each gene (FPKM) in esophageal biopsy tissue from patients with active EoE (n = 10) and controls (n = 6) for the 12 genes most highly expressed in the homeostatic esophagus are shown. Genes upregulated in active EoE samples are in orange; the p-value for each gene was determined using the Holm-Sidak method. In D, the Venn diagram shows the overlap of genes specifically expressed in the stomach according to the Human Protein Atlas with genes differentially expressed in the biopsies of patients with eosinophilic gastritis.15

78

Figure 2.2: Functional enrichment analysis of esophagus-specific Eso-EoE genes altered in EoE. In A-D, shown are the 10 most significant terms identified for Eso-EoE genes. Genes for the selected terms (grey bars) are listed with the corresponding fold change in expression for active EoE compared to control samples. Upregulated genes are in orange. The x axes represent the negative log (10) p-value.

79

Figure 2.3: Effect of differentiation on expression of esophagus-specific Eso-EoE genes. (A) Schematic diagram for generating ALI cultures from EPC2 esophageal epithelial cells is shown. Cells are initially grown in submerged culture at low and then high calcium concentrations to promote initial differentiation; terminal differentiation is achieved by growing cells in ALI. Note that the presence of IL-13 during the differentiation process weakens esophageal epithelial integrity, as represented by increased spaces between cell layers. (B) Venn diagram shows overlap between Eso-EoE genes and transcriptional changes during initial differentiation. A heat map represents log2 fold change in gene expression when comparing biopsy specimens from patients with EoE with normal biopsy specimens and comparing day 8 with day 3 (Initial Diff). The analysis is focused on the 90 common Eso-EoE genes. Clustering was performed by using Euclidean distance and average linkage parameters; yellow and blue colors correspond to increased or decreased expression, respectively; and black indicates a nonsignificant change compared with control values.

80

Figure 2.4: Effect of IL-13 treatment on expression of esophagus-specific Eso-EoE genes at the ALI. A, Venn diagram shows overlap of Eso-EoE genes with transcriptional signature of

81 terminally differentiated ALI in the absence or presence of IL-13. D0 represents the first day and D6 represents the last day of terminal differentiation at ALI (see Fig 3 for details). Group 1 shows 32 genes altered in biopsy specimens and in the ALI in the presence and absence of IL- 13. Group 2 shows 8 genes with altered expression in biopsy specimens and in the presence of IL-13 during ALI (D6 vs D6 + IL-13). Group 3 shows 28 genes with altered expression in biopsy specimens and by ALI differentiation but was not affected by IL-13 (D6 vs D0). B, Heat map shows a log2 fold change in expression of genes from the IL-1 and SERPINB gene families in biopsy specimens and ALI culture from patients with EoE at the indicated conditions. Orange color represents increased expression of SERPIN genes in biopsy specimens from patients with

EoE. For Fig 4, A and B, the heat maps show the log2 fold change in expression of genes from indicated groups. For all heat maps, the EoE biopsy specimen represents the expression level of the indicated genes in the EoE transcriptome compared with that in unaffected subjects. Clustering was performed by using Euclidean distance and average linkage parameters. Yellow and blue colors correspond to increased or decreased expression, respectively, and black color indicates a nonsignificant change compared with control subjects. C, Expression level of indicated Eso-EoE genes after the EoE diagnostic panel analysis was assessed by using RT- PCR in control subjects (Ctrl), patients with active EoE (EoE), and patients with gastroesophageal reflux disease (GERD). Each dot represents an individual biopsy specimen, and expression was normalized to GAPDH. Error bars represent the SEM of the mean relative expression. *P < .05, **P < .01, and ***P < .001, 1-way ANOVA. ns, Not significant.

82

Figure 2.5: Protein expression, protease activity, and expression of differentiation markers in esophageal biopsy specimens of patients with EoE. A, Representative Western blot for the indicated proteins in control (Ctrl) and active EoE (EoE) samples (n = 3 biopsy specimens). Red and green arrows point to the bands corresponding to analyzed proteins; molecular weight markers are shown on the right. B, Graphs show relative expression of the indicated proteins normalized to the loading control (HSP90) by using Western blotting (n = 5-10 samples for each group). Error bars represent the SEM of the mean relative expression. *P < .05 and **P < .01, Mann-Whitney test. C-F, Immunofluorescence of distal esophageal biopsy specimens from control subjects (Ctrl, top row) or patients with active EoE (EoE Inflamed, middle row) or proximal esophageal biopsy specimens from regions that do not have active inflammation from patients with active EoE (EoE Non-inflamed, bottom row). Staining with the nuclear dye DAPI or with an antibody directed against the indicated differentiation marker (KRT5, KRT14, KRT4, or CRNN) are indicated by blue or green, respectively. White broken line indicates basal membrane. Scale bar = 100 μm.

83

Figure 2.6: Model of esophagus-specific allergic inflammation. Under healthy conditions, the esophagus expresses tissue-specific markers of terminal epithelial differentiation (represented by KRT4 and CRNN). Profound loss of esophageal differentiation, either acquired or genetically inherited (represented by loss of KRT5 and KRT14), predisposes and/or propagates EoE. Loss of differentiation is triggered by imbalanced proteolytic activity in the esophagus, as indicated by the ratio between proteases and protease inhibitors (note that protease and protease inhibitors can be either upregulated or downregulated in patients with EoE).

84

# EoE Ref Alt Reference Alternate Minor allele Mutation Cases Gene Chr Position Amino Amino allele Allele freq (ExaC) type w/ Acid Acid variant 170215647 G A 0.00306 Nonsyn V M 3 GABRP 5 170224475 C T 0.00012 Nonsyn T M 1

61322980 C T 0.00657 Nonsyn E K 2 Frameshift SERPINB3 18 61326645 A - 0.00014 1 Del 61326659 C T 0.00009 Nonsyn G R 1

147475388 C T 0.00652 Nonsyn R C 2

SPINK5 5 147494001 G A 0.00255 Nonsyn G D 1

147498551 A G 0.00314 Nonsyn E G 1

17588689 T A 0.00378 Nonsyn L H 1

PADI3 1 17597423 C T 0.00666 Nonsyn A V 1

17601263 T C 0.00029 Nonsyn I T 1

9000164 C G 0.00005 Nonsyn P R 1

A2ML1 12 9008113 G A 0.00005 Nonsyn A T 1

9010579 C G 0.00002 Nonsyn Q E 1

52841659 G A 0.00007 Nonsyn R W 1

KRT6B 12 52841685 A G 0.00002 Nonsyn L P 1

52844244 C T 0.00002 Nonsyn R H 1

55273652 T G 0.00103 Nonsyn S A 1 C1orf177 1 55307347 T G 0.00131 Stoploss X E 1

150482145 G A 0.00749 Nonsyn A T 1 ECM1 1 150485285 G C 0.00002 Nonsyn E D 1

31414830 G A 0.00513 Nonsyn L F 1 CAPN14 2 31425971 T G 0.00077 Nonsyn K T 1

71058270 C G 0.00329 Nonsyn A P 1 CD207 2 71062864 C T novel Nonsyn D N 1

58852298 G C 0.00318 Nonsyn Q E 1 C3orf67 3 58853564 C T novel Nonsyn E K 1

78146273 C T 0.00039 Nonsyn P L 1 SCEL 13 78146303 C T 0.00231 Nonsyn S L 1

5831608 G A 0.00386 Nonsyn T M 1 FUT6 19 5831768 C T 0.00007 Nonsyn V M 1

6306810 G A 0.00002 Nonsyn T I 1 ACER1 19 Frameshift 6312165 C - 0.00002 1 Del 46733560 T G 0.00536 Nonsyn C G 1 IGFL1 19 46733750 C T 0.00036 Nonsyn T I 1

31534280 G A 0.00549 Nonsyn A V 1 PLA2G3 22 31536135 - T 0.00404 Frameshift 1

85

Ins

CPA4 7 129948221 G A 0.00235 Stopgain W X 1 Frameshift DSC2 18 28648000 - TC 0.00833 1 Ins

Table 2.1: List of Eso-EoE genes with mutations identified by using WES in 33 unrelated patients with EoE. Nonsyn, non-synonymous mutation; Ins, insertion; Del, deletion; Chr, chromosome; freq, frequency.

86

Figure 2.E1: Effect of IL-1α replenishment on ALI. A, Relative transepithelial electrical resistance (TEER) in EPC2 ALI cultures treated with IL-13 and IL-1α either alone or in combination. IL-13 treatment was started when ALI was induced at day 8, and IL-1α was added for the last 2 days of culture. TEER measurements were normalized to average TEER in the untreated (Ctrl) cultures at day 4. B, Expression level of CCL26 and SERPINB4 was assessed by using RT-PCR. Expression was normalized to GAPDH. Combined data for 2 independent experiments are shown.

87

CHAPTER 3: IL-33 IS INDUCED IN UNDIFFERENTIATED, QUIESCENT ESOPHAGEAL EPITHELIAL CELLS IN EOSINOPHILIC ESOPHAGITIS

J. Travers, BS1#, M. Rochman, PhD1#, J. Caldwell PhD1, J. Besse BS1, C.E. Miracle1, and M.E. Rothenberg, MD, PhD1*

1Division of Allergy and Immunology, Cincinnati Children’s Hospital Medical Center, Cincinnati, Ohio 45229-3039, USA

# These authors contributed equally to this work

*Corresponding author: Marc E. Rothenberg Division of Allergy and Immunology Cincinnati Children's Hospital Medical Center 3333 Burnet Avenue, MLC 7028 Cincinnati, OH 45229-3039 Phone: 513-803-0257 Fax: 513-636-3310 E-mail: [email protected]

CONFLICT OF INTEREST: M.E.R. is a consultant for NKT Therapeutics, Pulm One, Spoon Guru, Celgene, Shire, Astra Zeneca, and Novartis and has an equity interest in the first three listed and Immune Pharmaceuticals and royalties from reslizumab (Teva Pharmaceuticals). M.E.R. is an inventor of several patents, owned by Cincinnati Children’s. All of the other authors have no potential conflicts to disclose.

FUNDING SUPPORT: This work was supported by National Institutes of Health R37 AI045898, R01 AI124355, U19 AI070235, T32 GM063483, and F30 DK109573; the Campaign Urging Research for Eosinophilic Disease (CURED); the Buckeye Foundation; and the Sunshine Charitable Foundation and its supporters, Denise A. Bunning and David G. Bunning.

ACKNOWLEDGEMENTS: The authors wish to thank Shawna Hottinger for editorial assistance. We also wish to thank the Cincinnati Digestive Health Center Pathology Core for tissue processing, sectioning, histology, and immunohistochemical staining (NIH P30 DK078392) and Betsy DiPasquale for assistance with immunohistochemical stains.

SHORT TITLE: IL-33 induction in quiescent epithelial cells in EoE

88

ABSTRACT

BACKGROUND: The molecular and cellular etiology of eosinophilic esophagitis (EoE), an emerging tissue-specific allergic disease, involves dysregulated gene expression in esophageal epithelial cells.

Herein, we assessed the esophageal expression of IL-33, an epithelium-derived alarmin cytokine, in patients with EoE.

METHODS: Immunohistochemistry and immunofluorescence were performed on distal esophageal biopsies obtained from patients with EoE or control individuals. Immunofluorescence was performed on human primary esophageal epithelial cells.

RESULTS: IL-33 protein was markedly overexpressed in a subpopulation of basal layer esophageal epithelial cells in patients with active EoE compared to control individuals (63% vs. 3%, p < 0.0001).

IL-33 exhibited dynamic expression as levels normalized upon EoE remission. IL-33 levels did not correlate with esophageal eosinophil levels. IL-33–positive basal epithelial cells expressed E-cadherin and the undifferentiated epithelial cell markers keratin 5 and 14 but not the differentiation marker keratin 4. Moreover, the IL-33–positive epithelial cells expressed the epithelial progenitor markers p75NTR and ΔNp63 and lacked the proliferation markers Ki67 and phospho-histone H3. Additionally, the IL-33–positive cells had low expression of PCNA. IL-33 expression was detected in ex vivo– cultured primary esophageal epithelial cells in a subpopulation of cells that were not actively proliferating.

CONCLUSIONS: Collectively, we report that IL-33 expression within the epithelium is induced in a quiescent progenitor cell population in patients with active EoE.

89

INTRODUCTION

Eosinophilic esophagitis (EoE) is an emerging chronic, food antigen-driven, inflammatory allergic disorder(1). It is notable for the overwhelming T helper cell type 2 (Th2) inflammation associated with structural changes and the immune cell infiltration into the esophageal epithelium (2). There is a critical need to identify which factors initiate and propagate the excessive Th2 immune responses against food antigens in EoE.

The innate cytokine interleukin 33 (IL-33) is a prominent potentiator of Th2 immunity(3). IL-33 is generally expressed by mucosal epithelial cells, fibroblasts, and endothelial cells within the nucleus(4). Classically, it acts as an alarmin, with its extracellular release following cellular necrosis.

IL-33 can activate a wide variety of immune cells that express its receptor, suppressor of tumorigenicity 2 (ST2). Importantly, IL-33 is a very potent activator of eosinophils(5), mast cells(6), basophils(7, 8), and type 2 innate lymphoid cells (ILC2)(9), which all infiltrate the esophagus in patients with EoE(1, 10-12). Furthermore, intraperitoneal injection of recombinant IL-33 induces esophageal responses that mimic EoE, including eosinophil infiltration, increased proliferation of the epithelium, and production of Th2-associated cytokines(13). In addition, mice genetically deficient in

IL33 or IL1RL1 (encodes ST2) do not develop ovalbumin-induced EoE-like disease(14). These findings are likely to be clinically relevant because there is association between genetic variants in the

IL33 locus and EoE disease risk(15). Herein, we report that IL-33 is not expressed within the esophageal epithelium in control individuals. In patients with active EoE, IL-33 is present in the nuclei of esophageal basal layer cells with high levels of E-cadherin, p75NTR, ΔNp63, and keratins (KRT) 5 and 14 and low expression of proliferating cell nuclear antigen (PCNA). These IL-33–positive basal layer cells also lack KRT4, Ki67, and phospho-histone H3. Levels of IL-33 normalize to undetectable

90 levels following disease remission. Collectively, we propose that IL-33 is specifically induced in an esophageal epithelial progenitor population in patients with EoE.

RESULTS

IL-33 expression in eosinophilic esophagitis

We assessed IL-33 protein expression by immunohistochemistry in esophageal biopsies from patients with active or inactive EoE and from normal controls. In the healthy esophagus, IL-33 protein expression was detected in lamina propria cells, including endothelial cells (Figure 1A). In the healthy esophagus there was no detectable expression of IL-33 within the epithelium (Figure 1A,B). However, in esophageal biopsies of patients with active EoE, IL-33 protein was detected within the epithelium

(Figure 1A). IL-33 expression was limited to a subpopulation of basal layer cells (Figure 1A,B). These

IL-33–positive cells were restricted to the interpapillary basal layer. Notably, IL-33 protein was not detected in the papillary basal layer or in any of the suprabasal layers of the epithelium (Figure 1A,B).

We aimed to determine whether esophageal IL-33 levels were constitutively high in patients with EoE regardless of disease status or were dependent on disease activity. To test this, we compared the esophageal IL-33 expression in patients with active and inactive EoE, defined as patients with 0 or 1 eosinophils per high-power field of esophageal biopsy. IL-33 levels normalized with disease remission

(Figure 1A,B). Despite this, no correlation was observed in patients with active EoE between the proportion of IL-33–expressing basal layer cells and disease severity as assessed by the number of eosinophils per high-power field (Pearson r = -0.03, p = 0.89) (Figure 1C). In all cases, IL-33 was restricted to nuclei as no cytoplasmic or extracellular staining was observed. In summary, these results show that IL-33 is specifically increased within the esophageal epithelium in a sub-population of basal layer cells in patients with active EoE.

91

Characterization of IL-33–positive basal layer cells in vivo

We assessed IL-33 expression in esophageal biopsies by immunofluorescence using two different anti–IL-33 antibodies (a monoclonal antibody raised in mouse and a polyclonal antibody raised in goat). No staining was detected using either antibody within the epithelium in control individuals

(Figure 2A). Strong staining with both antibodies was detected in basal layer epithelial cells in patients with active EoE. Notably, only nuclear expression was found as the staining from the anti–IL-

33 antibodies overlapped with the DNA-binding dye DAPI (Figure 2A). Next, we assessed the differentiation status of these cells by co-staining with markers of different epithelial populations. The

IL-33–positive cells had strong expression of E-cadherin (Figure 2A,E). Furthermore, the IL-33– positive basal layer cells expressed the undifferentiation markers keratin 5 (KRT5) and KRT14

(Figure 2B,C,E). The IL-33–positive cells did not express the differentiation marker KRT4 (Figure

2D,E). Collectively, these results demonstrate that IL-33 is induced in a population of undifferentiated epithelial cells in patients with active EoE.

Because esophageal epithelial progenitor cells exist within the basal layer(16), we hypothesized that the IL-33–expressing cells constituted an epithelial progenitor population. The IL-33–positive basal cells expressed the epithelial progenitor markers p75 (Figure 3A,E) and p63 (Figure 3B,E). Next, the cell cycle status was assessed by performing immunofluorescence with a panel of proliferation markers. The IL-33–positive basal cells did not express the proliferation marker Ki-67 (Figure 3C,E), known to be expressed at all cell cycle stages except for the G0 phase, nor the mitosis marker phospho-histone H3 (Figure 3D,E). Additionally, the IL-33–positive basal layer cells had low expression of PCNA (Figure 3C), which is strongly upregulated during S phase(17). These results

92 indicate that these IL-33–positive basal layer cells constitute a quiescent epithelial progenitor cell population.

Characterization of IL-33 expression ex vivo

To determine whether restriction to a quiescent subpopulation of esophageal epithelial cells is an intrinsic feature of IL-33, we assessed IL-33 expression in ex vivo cultures of primary esophageal epithelial cells. Cells were maintained in an undifferentiated state as nearly all of the cells expressed

KRT5 and p63 (Figure 4A). IL-33 expression was detected in approximately 10% of the cells, and almost all of this 10% expressed KRT5 and p63 (Figure 4A,D). Nuclear expression of IL-33 was detected using two independent anti–IL-33 antibodies (Figure 4B). Additionally, no mitotic cells, defined by positive expression of phospho-histone H3, had detected expression of IL-33 using either antibody (Figure 4B,D). Additionally, the vast majority of IL-33–positive cells lacked Ki-67 and had low expression of PCNA (Figure 4C,D). In total, these results illustrate that IL-33 is expressed in a subpopulation of ex vivo–cultured esophageal epithelial cells that is not actively dividing.

DISCUSSION

We have assessed esophageal expression of IL-33 in patients with active EoE. We report that IL-33 protein expression is increased within the epithelium in nearly all patients with active EoE; IL-33 induction is dynamic and is a function of disease activity as its expression normalizes upon disease remission; esophageal expression of IL-33 in patients with EoE does not correlate with disease severity; IL-33 protein has a nuclear compartmentalization; IL-33 is expressed in a subpopulation of basal layer cells; IL-33–positive basal cells strongly express E-cadherin, KRT5, KRT14, p75, and p63; have low expression of PCNA; and do not express KRT4, Ki-67, or phospho-H3; and IL-33

93 expression ex vivo is restricted to a limited population of primary esophageal epithelial cells not currently dividing. On the basis of these findings, we conclude that IL-33 expression is induced in a quiescent, undifferentiated esophageal epithelial cell population in patients with EoE.

IL-33 levels within the esophageal epithelium normalize upon disease remission, indicating that the increased IL-33 expression is an acquired, rather than constitutive or intrinsic, feature of EoE, consistent with prior reports(13, 18). It remains unclear why IL-33 protein expression is restricted to interpapillary basal layer cells in patients with EoE. In light of the high potency of extracellular IL-33– induced activation of immune cells(4, 5), limiting the number of IL-33–expressing cells could serve to prevent excess release. Interestingly, the IL-33 expression profile in the esophageal epithelium is different than those of thymic stromal lymphopoietin (TSLP) and IL-25, which are expressed in superficial layers(18). It is notable that interpapillary basal layer cells do not actively proliferate. IL-33 has been shown to be restricted to non-proliferating cells in other cell types, including stomach surface mucus cells(19) and endothelial cells(20). Quiescent endothelial cells are known to express

IL-33 through the action of the Notch1 ligands Dll4 and Jagged1(21); therefore, it is possible that

Notch1 signaling induces IL-33 in the basal layer esophageal epithelial cells. Interferon  is known to be a strong inducer of IL-33 expression in primary skin keratinocytes(22-24). Currently it is not known what induces IL-33 in esophageal epithelial cells in EoE so future investigation into the regulation of

IL-33 expression is warranted.

Our study characterizes the IL-33–expressing basal layer cells in EoE as a quiescent progenitor population. This supports ex vivo spheroid culture studies demonstrating that the esophageal epithelial cells with the highest stem cell capacity are present in the basal layer(16). EoE is a hyperproliferative disorder(25, 26) with marked loss of esophageal tissue identity and differentiation 94 within the epithelium(27). Because this cell layer purportedly undergoes occasional mitotic divisions in order to maintain the esophageal epithelium(28), future studies should investigate their contributions to disease pathogenesis. IL-33 has long been proposed to act as a transcriptional regulator through its ability to bind chromatin(29, 30). No rigorously tested evidence for an intracellular nuclear function for IL-33 has been identified. However, the effect of nuclear IL-33 expression in these basal layer cells, especially in the context of allergic inflammation, has not been examined and thus warrants further investigation. Taken together, our data identified that IL-33 is induced in a quiescent esophageal epithelial progenitor population in patients with active EoE. We also found that IL-33 was dynamically expressed as a function of disease activity. These findings underscore the potential value of further understanding the regulation and role of IL-33 in EoE and other allergic diseases.

METHODS

Antibodies: Mouse monoclonal antibody (clone Nessy-1) (#ALX-804-840-C100) was purchased from

Enzo. Rabbit polyclonal antibodies against KRT5 (#ab24647) and Ki-67 (#ab15580) were purchased from Abcam (Abcam, Cambridge, MA). Rabbit polyclonal antibody against KRT14 (#PRB-155P) was purchased from Covance (Covance, Princeton, NJ). Rabbit polyclonal antibody against KRT4

(#HPA034881) was purchased from Sigma (Sigma-Aldrich Corp, St. Louis, MO). Rabbit monoclonal antibodies against E-cadherin (#3195), p75 (#8238), and phospho-histone H3 (#3377) were purchased from Cell Signaling (Cell Signaling Technology, MA). Mouse monoclonal antibody against p63 (#sc-8431) was purchased from Santa Cruz (Santa Cruz Biotechnology, TX). Mouse monoclonal antibody against PCNA (#MAB424) was purchased from Millipore (Billerica, MA). Goat polyclonal antibody against IL-33 (#AF3625) was purchased from R&D (R&D Systems, Minneapolis, MN).

Donkey anti-goat Alexa Fluor 488 (A11055 ), anti-rabbit Alexa Fluor 568 (A10042), and anti-mouse

95

Alexa Fluor 647 (A31571) secondary antibodies were purchased from Life Technologies (Carlsbad,

CA).

Esophageal biopsy collection and processing: Esophageal biopsies were obtained and processed as previously described(31). Briefly, this study was approved by the Institutional Review Board fo

Cincinnati Children’s Hospital Medical Center (CCHMC) before the start of the study. Distal esophageal biopsies were obtained after informed consent was received. Active EoE was defined as having a physician-provided EoE diagnosis and ≥ 15 eosinophils per 400x high-power field in distal esophageal biopsies obtained the same day. Inactive EoE was defined as having a previous history of EoE but with 0 or 1 eosinophils per high-power field. Normal controls were defined as patients with no history of EoE or any other eosinophilic gastrointestinal disorder (EGID) and 0 eosinophils per high-power field. Esophageal biopsies were fixed with formalin and then embedded in paraffin

(FFPE).

Immunohistochemistry and immunofluorescence of esophageal biopsies:

Immunohistochemistry of distal esophageal biopsies using mouse anti-IL-33 antibody (Nessy-1) was performed by the Pathology Research Core at CCHMC. Images were obtained using an Apotome widefield microscope (Zeiss, Thornwood, NY). For immunofluorescence studies, slides with 4-µm sections of FFPE esophageal biopsies underwent deparaffinization (serial incubations with xylene,

100% ethanol, 95% ethanol, 70% ethanol, 50% ethanol), antigen retrieval using sodium citrate buffer

(10 mM sodium citrate, 0.05% Tween 20, pH 6.0), blocked with 10% donkey serum/phosphate- buffered saline (PBS), and then incubated with primary antibody diluted in 10% donkey serum/PBS overnight at 4˚C in a humidified chamber. The next day, slides were washed with PBS, incubated with secondary antibodies diluted in 10% donkey serum/PBS for 1 h at room temperature (RT) in a 96 humidified chamber, and then washed in the presence of DAPI (0.5 µg/mL). Finally, a cover slip was added with ProLong Gold mounting reagent (Molecular Probes). The next day, slides were imaged using a Nikon A1R inverted confocal microscope. Analysis was performed with the Nikon Elements program.

Ex vivo culture of primary esophageal epithelial cells: One human distal esophageal biopsy obtained during routine endoscopy was collected for research purposes in 1 mL keratinocyte serum- free media (KSFM) (Invitrogen) containing human epidermal growth factor (EGF) (1 ng/mL), bovine pituitary extract (50 μg/mL), and 1X penicillin/streptomycin (Invitrogen) and subsequently placed in a

60-mm dish in 3 mL of Leibovitz’s L-15 media (Invitrogen) containing 115 U/mL collagenase, 1.2

U/mL dispase, and 1.25 mg/mL BSA that had been filter sterilized (0.2 μm). The biopsy was mechanically dispersed using scissors into pieces less than 1 mm in size and then incubated at 37°C for 1 h. The digested biopsy was collected and washed twice with 5 mL KSFM containing the same supplements as described above. Cells were then incubated in 1 mL of 0.05% trypsin/EDTA

(Invitrogen) (10 min, 37°C, with agitation every 2 min). Soybean trypsin inhibitor (250 mg/L in 1X

DPBS) was added (5 mL). Cells were pelleted and then resuspended in 1 mL KSFM (containing the same supplements as described above) and transferred to a 35-mm dish. Irradiated NIH 3T3 J2 fibroblasts (162,500 cells) were added to the dish. Media were changed at day 5 and every other day thereafter using KSFM containing the same supplements as describe above. After epithelial cells became 60-70% confluent, they were dispersed from the plate using 0.05% trypsin/EDTA, which was inactivated by STI; cells were then cultured in KSFM containing the same supplements as described above.

97

Immunofluorescence of primary esophageal epithelial cells: Primary esophageal epithelial cells were plated on Ibidi 8-well chambers (# 80826). The next day, cells were fixed with 4% paraformaldehyde for 10 min and quenched with 50 mM ammonium chloride. Cells were blocked with

10% donkey serum/PBS for 30 min and incubated with primary antibody diluted in 10% donkey serum/PBS for 1 h at RT. Cells were washed with PBS, incubated with secondary antibodies diluted in 10% donkey serum/PBS for 1 h at RT, and washed in the presence of DAPI (0.5 µ/mL). Finally, cells were placed in fresh PBS and imaged using a Nikon A1R inverted confocal microscope.

Analysis was performed with the Nikon Elements program.

Statistical Analysis: One-way ANOVA with Holm-Sidak correction for multiple testing was performed using GraphPad Prism 7.0 software.

98

REFERENCES

1. Davis BP, Rothenberg ME. Mechanisms of Disease of Eosinophilic Esophagitis. Annu Rev

Pathol 2016;11:365-393.

2. Travers J, Rothenberg ME. Eosinophils in mucosal immune responses. Mucosal Immunol

2015;8(3):464-475.

3. Liew FY, Girard JP, Turnquist HR. Interleukin-33 in health and disease. Nat Rev Immunol

2016;16(11):676-689.

4. Martin NT, Martin MU. Interleukin 33 is a guardian of barriers and a local alarmin. Nat Immunol

2016;17(2):122-131.

5. Bouffi C, Rochman M, Zust CB, Stucke EM, Kartashov A, Fulkerson PC, et al. IL-33 markedly activates murine eosinophils by an NF-kappaB-dependent mechanism differentially dependent upon an IL-4-driven autoinflammatory loop. J. Immunol. 2013;191(8):4317-4325.

6. Joulia R, L'Faqihi FE, Valitutti S, Espinosa E. IL-33 fine tunes mast cell degranulation and chemokine production at the single-cell level. J Allergy Clin Immunol 2016.

7. Rivellese F, Suurmond J, de Paulis A, Marone G, Huizinga TW, Toes RE. IgE and IL-33- mediated triggering of human basophils inhibits TLR4-induced monocyte activation. Eur J Immunol

2014;44(10):3045-3055.

8. MacGlashan D, Jr. Expression profiling of human basophils: modulation by cytokines and secretagogues. PLoS One 2015;10(5):e0126435.

9. Kim BS, Artis D. Group 2 innate lymphoid cells in health and disease. Cold Spring Harb

Perspect Biol 2015;7(5).

10. Noti M, Wojno ED, Kim BS, Siracusa MC, Giacomin PR, Nair MG, et al. Thymic stromal lymphopoietin-elicited basophil responses promote eosinophilic esophagitis. Nat. Med.

2013;19(8):1005-1013.

99

11. Doherty TA, Baum R, Newbury RO, Yang T, Dohil R, Aquino M, et al. Group 2 innate lymphocytes (ILC2) are enriched in active eosinophilic esophagitis. J Allergy Clin Immunol

2015;136(3):792-794.e793.

12. Abonia JP, Blanchard C, Butz BB, Rainey HF, Collins MH, Stringer K, et al. Involvement of mast cells in eosinophilic esophagitis. J Allergy Clin Immunol 2010;126(1):140-149.

13. Judd LM, Heine RG, Menheniott TR, Buzzelli J, O'Brien-Simpson N, Pavlic D, et al. Elevated

IL-33 expression is associated with pediatric eosinophilic esophagitis, and exogenous IL-33 promotes eosinophilic esophagitis development in mice. Am J Physiol Gastrointest Liver Physiol

2016;310(1):G13-25.

14. Venturelli N, Lexmond WS, Ohsaki A, Nurko S, Karasuyama H, Fiebiger E, et al. Allergic skin sensitization promotes eosinophilic esophagitis through the IL-33-basophil axis in mice. J Allergy Clin

Immunol 2016;138(5):1367-1380.e1365.

15. Kottyan LC, Davis BP, Sherrill JD, Liu K, Rochman M, Kaufman K, et al. Genome-wide association analysis of eosinophilic esophagitis provides insight into the tissue specificity of this allergic disease. Nat. Genet. 2014;46(8):895-900.

16. Jeong Y, Rhee H, Martin S, Klass D, Lin Y, Nguyen le XT, et al. Identification and genetic manipulation of human and mouse oesophageal stem cells. Gut 2016;65(7):1077-1086.

17. Kelman Z. PCNA: structure, functions and interactions. Oncogene 1997;14(6):629-640.

18. Simon D, Radonjic-Hosli S, Straumann A, Yousefi S, Simon HU. Active eosinophilic esophagitis is characterized by epithelial barrier defects and eosinophil extracellular trap formation.

Allergy 2015;70(4):443-452.

19. Buzzelli JN, Chalinor HV, Pavlic DI, Sutton P, Menheniott TR, Giraud AS, et al. IL33 Is a

Stomach Alarmin That Initiates a Skewed Th2 Response to Injury and Infection. Cell Mol

Gastroenterol Hepatol 2015;1(2):203-221.e203.

100

20. Kuchler AM, Pollheimer J, Balogh J, Sponheim J, Manley L, Sorensen DR, et al. Nuclear interleukin-33 is generally expressed in resting endothelium but rapidly lost upon angiogenic or proinflammatory activation. Am J Pathol 2008;173(4):1229-1242.

21. Sundlisaeter E, Edelmann RJ, Hol J, Sponheim J, Kuchler AM, Weiss M, et al. The alarmin IL-

33 is a notch target in quiescent endothelial cells. Am J Pathol 2012;181(3):1099-1111.

22. Sundnes O, Pietka W, Loos T, Sponheim J, Rankin AL, Pflanz S, et al. Epidermal Expression and Regulation of Interleukin-33 during Homeostasis and Inflammation: Strong Species Differences. J

Invest Dermatol 2015;135(7):1771-1780.

23. Meephansan J, Tsuda H, Komine M, Tominaga S, Ohtsuki M. Regulation of IL-33 expression by IFN-gamma and tumor necrosis factor-alpha in normal human epidermal keratinocytes. J Invest

Dermatol 2012;132(11):2593-2600.

24. Seltmann J, Werfel T, Wittmann M. Evidence for a regulatory loop between IFN-gamma and

IL-33 in skin inflammation. Exp Dermatol 2013;22(2):102-107.

25. Steiner SJ, Kernek KM, Fitzgerald JF. Severity of basal cell hyperplasia differs in reflux versus eosinophilic esophagitis. J Pediatr Gastroenterol Nutr 2006;42(5):506-509.

26. Denning KL, Al-Subu A, Elitsur Y. Immunoreactivity of p53 and Ki-67 for dysplastic changes in children with eosinophilic esophagitis. Pediatr Dev Pathol 2013;16(5):331-336.

27. Rochman M, Travers J, Miracle CE, Bedard MC, Wen T, Azouz NP, et al. Profound loss of esophageal tissue differentiation in patients with eosinophilic esophagitis. J Allergy Clin Immunol

2017.

28. Seery JP, Watt FM. Asymmetric stem-cell divisions define the architecture of human oesophageal epithelium. Curr Biol 2000;10(22):1447-1450.

29. Carriere V, Roussel L, Ortega N, Lacorre DA, Americh L, Aguilar L, et al. IL-33, the IL-1-like cytokine ligand for ST2 receptor, is a chromatin-associated nuclear factor in vivo. Proc Natl Acad Sci

U S A 2007;104(1):282-287.

101

30. Roussel L, Erard M, Cayrol C, Girard JP. Molecular mimicry between IL-33 and KSHV for attachment to chromatin through the H2A-H2B acidic pocket. EMBO Rep 2008;9(10):1006-1012.

31. Rochman M, Kartashov AV, Caldwell JM, Collins MH, Stucke EM, Kc K, et al. Neurotrophic tyrosine kinase receptor 1 is a direct transcriptional and epigenetic target of IL-13 involved in allergic inflammation. Mucosal Immunol 2015;8(4):785-798.

102

Figures and Figure Legends

Figure 3.1: Subcellular localization of IL-33 in esophageal epithelial cells. (A-C) Immunohistochemistry for IL-33 protein expression in representative esophageal biopsies from healthy control individuals (Control, left panel), patients with active EoE (Active EoE, middle panel), or patients with inactive EoE (Inactive EoE, right panel) using mouse anti–IL-33 antibody. In (A), the bottom row is a high-power view of the area enclosed in the black square. The black dashed line indicates the basement membrane. (A-B) Biopsies from 19 controls, 20 patients with active EoE, and 7 patients with inactive EoE were stained. (B) Quantification of the proportion of basal layer cells in each biopsy with IL-33 expression. Mean ± standard error of the mean is depicted. (C) Pearson correlation of biopsy eosinophil count per high-power field (hpf) and the percentage of basal layer cells with IL-33 expression in esophageal biopsies of patients with active EoE. **** p < 0.0001.

103

Figure 3.2: Epithelium-specific protein expression in esophageal tissue. (A-D) Immunofluorescence of esophageal biopsies from control individuals (top row) or patients with active EoE (bottom row). Nuclei are indicated by DAPI staining (blue). Green and red indicate staining with the indicated antibodies. Images are representative of biopsies from 4 or 5 patients with active EoE and 4 or 5 control individuals. (E) Quantification of the proportion of IL-33–positive basal layer cells from active EoE biopsies with strong expression of the indicated marker from (A-D). Mean ± standard error of the mean is depicted. Scale bar is 20 µm. gIL-33, goat anti–IL-33 antibody; mIL-33, mouse anti–IL-33 antibody; KRT, keratin.

104

Figure 3.3: Cell cycle and differentiation status in vivo in esophageal tissue. (A-D) Immunofluorescence of esophageal biopsies from control individuals (top row) or patients with active EoE (bottom row). Nuclei are indicated by DAPI staining (blue). Green and red indicate staining with the indicated antibodies. Images are representative of biopsies from 3-6 patients with active EoE and 3-6 control individuals. (E) Quantification of the number of IL-33–positive basal layer cells from active EoE biopsies with strong expression of the indicated marker from (A-D). Mean ± standard error of the mean is depicted. Scale bar is 20 µm. pH3, phospho-histone H3; PCNA, proliferating cell nuclear antigen.

105

Figure 3.4: Cell cycle status of IL-33–expressing esophageal cells ex vivo. (A-C) Immunofluorescence of ex vivo–cultured primary esophageal epithelial cells. Nuclei are indicated by DAPI staining (blue). Green and red indicate staining with the indicated antibodies. Images are representative of three independent experiments. (D) Quantification of the percentage of IL-33–positive primary epithelial cells with strong expression of the indicated marker. Mean ± standard error of the mean of cumulative data from three independent experiments is depicted. Scale bar is 20 µm. gIL-33, goat anti–IL-33 antibody; mIL-33, mouse anti–IL-33 antibody; KRT, keratin; pH3, phospho-histone H3; PCNA, proliferating cell nuclear antigen.

106

CHAPTER 4: CHROMATIN REGULATES THE MAGNITUDE AND KINETICS OF IL-

33 RELEASE AND BIOACTIVITY

J. Travers BS1#, M. Rochman PhD1#, C.E. Miracle1, Jeffery Habel1, Michael Brusilovsky1, Jeffrey K Rymer BS1, and M.E. Rothenberg MD PhD1*

1Division of Allergy and Immunology, Cincinnati Children’s Hospital Medical Center, Cincinnati, Ohio 45229-3039, USA

# These author contributed equally to this work

*Corresponding author: Marc E. Rothenberg Division of Allergy and Immunology Cincinnati Children's Hospital Medical Center 3333 Burnet Avenue, MLC 7028 Cincinnati, OH 45229-3039 Phone: 513-803-0257 Fax: 513-636-3310 E-mail: [email protected]

Conflict of interest: M.E.R. is a consultant for NKT Therapeutics, Pulm One, Spoon Guru, Celgene, Shire, Astra Zeneca and Novartis and has an equity interest in the first three listed and Immune Pharmaceuticals, and royalties from reslizumab (Teva Pharmaceuticals). M.E.R. is an inventor of several patents, owned by Cincinnati Children’s. All of the other authors have no potential conflicts to disclose.

Funding support: This work was supported by National Institutes of Health R37 AI045898, R01 AI124355, U19 AI070235, T32 GM063483, and F30 DK109573; the Campaign Urging Research for Eosinophilic Disease (CURED); the Buckeye Foundation; and the Sunshine Charitable Foundation and its supporters, Denise A. Bunning and David G. Bunning.

107

ABSTRACT

We aimed to uncover the characteristics and functions of the nucleus- localized, chromatin-binding cytokine IL-33, an epithelium-derived alarmin thatinitiates a variety of immune responses following extracellular release. IL-33 exhibited only nuclear localization in human allergic tissue. Epithelial overexpression of IL-33did not change gene expression, as assessed by RNA-sequencing. Fluorescence recovery after photobleaching revealed that wild-type (WT) IL-33 had a ~10-fold slower intranuclear mobility compared with the nuclear cytokine IL-1α (p <0.0001), whereas a truncated form of IL-33 engineered not to bind chromatin was freely mobile. Following cellular necrosis, WT IL-33 demonstrated decreased extracellular release as a high molecular weight species with increased intracellular retention compared to truncated IL-33. Time- lapse microscopy revealed retention of WT IL-33, but not truncated IL-33, after membrane dissolution. WT IL-33 displayed a slow, linear release over time compared with non-chromatin binding IL-33 and IL-1α. Direct interaction between released WT IL-

33 and histone H2B was detected by co-immunoprecipitation. Notably, histones and IL-

33 synergistically activated ST2-mediated responses. We propose that IL-33 is finely- regulated by a “molecular chromatin sink” that limits its ability to be released while simultaneously increasing both the duration of its release and ST2-mediated bioactivity.

108

INTRODUCTION

Cytokines mediate cellular communication through activation of surface receptors upon extracellular release. A classic cytokine contains a leader peptide sequence that mediates either immediate extracellular secretion or storage in cytoplasmic secretory granules for release after cellular activation1. However, a subset of cytokines, including interleukin 1 (IL-1) family members and high mobility group box 1 (HMGB1), lack leader peptide sequences and instead are localized to the nucleus where they bind DNA or chromatin2. These cytokines function as alarmins to indicate cellular damage or stress.

Amongst alarmins, much attention has been focused on IL-33, an IL-1 family member expressed by mucosal epithelial cells3,4, because it is a potent initiator of acute inflammation and primes for subsequent type 2 immune responses5,6. Through its receptor, suppression of tumorigenicity 2 (ST2), IL-33 potently activates a plethora of immune cells, including basophils, mast cells, eosinophils, type 2 innate lymphoid cells, and CD4+ T cells7. IL-33 is distinguished from other cytokines by the extensive post- translational modifications that profoundly modulate its ability to activate ST2- expressing cells. Notably, during apoptosis IL-33 is proteolytically cleaved by caspases

3 and 7 into forms incapable of activating surface ST28. Following acute necrosis, extracellular IL-33 is cleaved into mature forms by the serine proteases derived from neutrophils9 and mast cells10 (e.g. elastase and tryptase), generating highly active forms of IL-33. Additionally, cysteine oxidation diminishes the ability of IL-33 to active ST211.

Taken together, a model is emerging wherein IL-33 is uniquely regulated based on post- translational processes. The potency of IL-33 may have necessitated the development of such complex, post-translational regulatory processes to allow fine-tuning.

109

The IL-33-ST2 axis is notably prominent in the pathogeneses of several allergic disorders, including asthma, atopic dermatitis, and eosinophilic esophagitis (EoE)5,12. A strong genetic association exists between allergy and the IL-33-ST2 axis as variants in the IL33 and IL1RL1 (encodes ST2) genes confer risk for several allergic diseases13-17.

As such, the IL-33-ST2 axis has emerged as a primary target for therapeutic modulation in allergy5.

A perplexing, unanswered question concerning IL-33 is the functional significance of its unique nuclear localization and chromatin binding5,18. Other nuclear cytokines, including HMGB119 and IL-1α20, are considered to be dual-function as they can also act as transcription factors through their ability to bind DNA. IL-33 directly binds to the nucleosome acidic patch composed of the tails of histones H2A and H2B21, which have important roles in regulating chromatin structure22. Several other nucleosome acidic patch-binding proteins act as transcriptional regulators23, including high mobility group

N2 (HMGN2) and latency associated nuclear antigen (LANA) of the Kaposi sarcoma herpesvirus. The chromatin binding domain (CBD) of IL-33 has a remarkably high sequence similarity to that of LANA21, and IL-33 promotes chromatin compaction18,21.

Yet, the nuclear of IL-33 has yet to be elucidated. Herein, we aimed to define the functional significance of the nuclear localization and chromatin binding of IL-33 in epithelial cells. We have identified functional roles for the enigmatic nuclear localization

110 and chromatin binding of IL-33 in altering the magnitude and kinetics of its extracellular release and increasing ST2-mediated inflammatory cytokine production.

111

RESULTS

Effect of nuclear IL-33 on gene expression

In order to establish the nuclear function of IL-33, we first confirmed the reported4,18 nuclear localization of IL-33 protein focusing on human allergic inflammation. Using immunofluorescence with two different antibodies directed against IL-33, only nuclear expression was detected in esophageal epithelial cells in patients with EoE

(Supplemental Figure S1A,B). Similarly, only nuclear IL-33 protein was detected in ex vivo-cultured primary esophageal epithelial cells (Supplemental Figure S1C) and in an esophageal epithelial cell line (TE-7) engineered to constitutively overexpress IL-33

(Supplemental Figure S1D). We then overexpressed different IL-33 variants in esophageal epithelial cells lacking both endogenous IL-33 and the IL-33 receptor ST2

(data not shown). In particular, we utilized lentivirus-mediated, stable transduction to engineer doxycycline (Dox)-inducible overexpression of wild-type (WT) IL-33 or a non- chromating binding18, truncated form of IL-33, composed of amino acid residues 112-

270. IL-33 protein expression was efficiently induced with Dox treatment (Figure 1A,B) at similar levels. Genome-wide RNA-sequencing (RNA-seq) after 48 hour treatment with

Dox or vehicle revealed increased IL33 reads were detected following Dox, but not in controls (Figure 1C,D). Differentially expressed genes for each individual clone were identified based on loose criteria (RPKM ≥ 1 in at least one sample, Benjamini-

Hochberg False Discovery Correction [5%], and a fold change ≥ 1.5 upon treatment with

Dox). Levels of 102 genes were found to change in at least one of the three clones overexpressing WT IL-33; however, IL33 was the only gene differentially expressed in all 3 clones (Figure 1E,F). After decreasing the stringency even further by removing the

112 restriction on fold change, MMP13 was found to be differentially expressed in all 3 clones overexpressing WT IL-33. However, MMP13 levels also changed in control cells

(Figure 1F), consistent with its known regulation by doxycycline24. There were seven genes (IL33, ZHX2, SLC16A2, RDX, LOXL4, CD55, and PTHLH) differentially expressed in all three clones overexpressing truncated IL-33 (Figure 1F). Collectively, these results demonstrate that the presence of nuclear IL-33 does not alter gene expression in esophageal epithelial cells under the conditions tested.

Chromatin-binding properties of IL-33

Our unexpected inability to determine an effect of chromatin binding on gene expression led us to characterize its biophysical properties in order to identify if unique features may exist compared with transcriptionally active alarmins, such as IL-1α. We assessed the strength of the physical association of WT IL-33 with chromatin in esophageal epithelial cells by performing biochemical fractionation with serial extractions of increasing stringency (Figure 2A). Of the three other proteins measured by Western blot

(the cytoplasmic protein heat shock protein 90, histone H3, and the nuclear matrix protein lamin B), the distribution of IL-33 among the different fractions was most similar to that of histone H3 with detectable amounts of IL-33 and H3 present in the S3, S4, and P4 fractions (Figure 2B,C). Notably, even 2 M NaCl does not fully extract WT IL-33 from chromatin. Conversely, truncated IL-33 was fully extracted by Triton X-100 (Figure

2D,E). These data show that IL-33 tightly binds chromatin.

113

Next, we assessed localization of IL-33 within chromatin with C-terminal green fluorescent protein (GFP)-fusion proteins of WT or truncated IL-33 (Figure 3A). Live-cell confocal microscopy showed that WT IL-33-GFP was restricted to the nucleus as expression was not detected outside of the regions containing the DNA-binding dye

Hoechst 33342 (Figure 3B). WT IL-33-GFP exhibited a high correlation with Hoechst

33342 within the nucleus (the Pearson coefficient was 0.55 ± 0.02 [mean ± SEM]).

There was not a statistically significant difference between the co-localizaion with

Hoechst 33342 of WT IL-33-GFP vs H2B-GFP (Holm-Sidak multiple comparisons test p

= 0.59) (Figure 3C), indicating that WT IL-33-GFP is enriched in regions of heterochromatin. In contrast to WT IL-33-GFP, truncated IL-33-GFP was in the nucleus and cytoplasm. Additionally, there was not a statistically significant difference in the co- localization with Hoechst 33342 of truncated IL-33-GFP vs GFP alone (p = 0.49) within the nucleus. These results indicate that the restriction to the nucleus and enrichment in heterochromatin of WT IL-33-GFP occur because of chromatin binding. Finally, the

Pearson coefficient within the nucleus with Hoechst 33342 of IL-1α-GFP was drastically lower than that of WT IL-33-GFP (p < 0.0001) but indistinguishable GFP alone (p =

0.59). This indicates that IL-1α-GFP is not enriched in heterochromatin despite the fact that it is in the same cytokine family as IL-333. In summary, these data demonstrate that chromatin binding causes IL-33 to be enriched in regions of heterochromatin in esophageal epithelial cells, a finding that is not observed with other IL-1 family member nuclear alarmins.

114

We next quantitated the kinetics of the interactions of IL-33 with chromatin using fluorescence recovery after photobleaching (FRAP). WT IL-33-GFP exhibited a relatively slow intranuclear mobility (e.g. it took 13 ± 1 and 30 ± 2 seconds [mean ±

SEM] to recover 50% and 70% of the fluorescence within the ROI, respectively) (Figure

4B,C). The intranuclear mobility of truncated IL-33-GFP was indistinguishable from GFP alone, indicating that the slow mobility of WT IL-33-GFP was dependent on chromatin binding. Notably, WT IL-33-GFP was dramatically less mobile than GFP-tagged IL-1 α, which had a shorter 70% recovery (3.5 ± 0.2 vs 29.7 ± 2.3 seconds, p < 0.0001). In fact, the 50% recovery of IL-1α-GFP was too fast to measure. However, WT IL-33-GFP was dramatically less mobile than the core histone H2B-GFP, which had a 50% recovery of

H2B-GFP beyond the time of measurement of the experiment. Furthermore, WT IL-33-

GFP exhibited a smaller immobile fraction (interpreted as the proportion of the protein that does not change location over the course of the experiment) than H2B-GFP.

Overall, these data demonstrate that IL-33 exhibits dynamic binding to chromatin with a much higher average residence time than IL-1α.

Effect of chromatin binding on IL-33 release during necrosis

We compared the extracellular release of IL-33 variants after inducing necrosis with a 4 hour treatment with calcium ionophore A23187 (20 µM). There was minimal baseline release of WT or truncated IL-33 (Figure 5A,B). Calcium ionophore treatment increased the release of truncated IL-33 compared with WT IL-33 (supernatant/pellet ratio of 1.4 ±

0.5 vs 0.2 ± 0.1 [mean ± SEM], representing a 7.8-fold change between truncated and

115

WT IL-33; two-way ANOVA interaction term p = 0.05) (Figure 5A,B). Furthermore, WT

IL-33 was highly cell-associated despite marked necrosis, indicating intracellular retention. Importantly, a similar rate of necrosis was observed between cells overexpressing WT and truncated IL-33 (p > 0.90) (Figure 5C). In independent experiments, we assessed the release of WT and truncated IL-33 in response to another necrotic stimulus, cryoshock. Accordingly, cryoshock induced increased release of truncated IL-33 compared to WT IL-33 (supernatant/pellet ratio of 7.4 ± 0.7 vs

1.1 ± 0.3, representing a 7.0-fold change between truncated and WT IL-33; Student’s t- test p < 0.001) (Figure 5D,E). As a control, there was no statistically significant differences in cellular viability between WT and truncated IL-33-expressing cells as assessed by trypan blue exclusion (p > 0.80) (Figure 5F). These results demonstrate that chromatin binding decreases the extracellular release of IL-33 by mediating intracellular retention after necrosis has occurred.

Chromatin binding regulates the kinetics of IL-33 release

We next examined whether IL-33 chromatin binding mediated slow release over time from necrotic cells using live-cell time-lapse confocal microscopy. With impaired membrane integrity, a cardinal feature of necrosis, WT IL-33-GFP exhibited increased intracellular retention compared to IL-1α-GFP, truncated 112-270 IL-33-GFP, and GFP alone (Figure 6A-C). The retention of WT IL-33-GFP was consistently lower than that of

H2B-GFP. Notably, a slow, steady decrease in intracellular levels over time was observed for WT IL-33-GFP but not H2B-GFP. Collectively, these data indicate that WT

116

IL-33 exhibits a relatively slow, durable extracellular release upon loss of membrane integrity that is not observed for other chromatin-binding proteins.

Characterization of released IL-33 species

We next tested whether IL-33-chromatin complexes were released during necrosis.

Size-exclusion chromatography of necrotic supernatants revealed that approximately

70% of WT IL-33 was detected as high molecular weight (HMW) complexes (e.g. the void volume of the column) (Figure 7B,C). The fractions with the highest concentrations of WT IL-33 were present in the void volume of the column (Figure 7A,B), which contain species with molecular species of at least 100 kDa. Truncated IL-33 was predominantly detected as low molecular weight (LMW) species (Figure 7B,C), indicating that the presence of the HMW species of WT IL-33 is dependent on chromatin binding. Western blot analysis of pooled fractions demonstrated that WT IL-33 is predominantly present as HMW species whereas truncated IL-33 is not. In addition, DNA and histones were present in high levels in HMW fractions and undetectable in LMW fractions (Figure

7D,E). Furthermore, IL-33 co-immunoprecipitated with an anti-H2B antibody in bulk necrotic supernatants (Figure 7F). We next tested if histones plus IL-33 cooperatively signaled using an IL-33 biosensor assay (ST2-expressing HMC-1 cells25,26). Purified histones (Figure 7G) did not exhibit any ST2 bioactivity (Figure 7H). Synergistic cytokine production was observed with co-treatment with histones WT, but not truncated, IL-33

(Figure 7H). In total, these results demonstrate that during necrosis WT IL-33 is

117 released extracellularly in complex with chromatin and that histones synergize full- length IL-33 to to induce ST2 signaling.

DISCUSSION

The cytokine IL-33 is a potent extracellular activator of innate immunity; however, the reasons for its unique nuclear localization and chromatin binding have remained enigmatic. Herein, we have elucidated biophysical properties of IL-33 chromatin binding and uncovered its functional roles. In particular, we have made the observations that (1)

IL-33 retains its predominantly nuclear localization in vivo, even under conditions classically associated with its release, such as human allergic inflammation; (2) despite its nuclear localization, IL-33 does not affect homeostatic gene expression as assessed by genome-wide transcript profiling of esophageal epithelial cells engineered to overexpress IL-33; (3) the chromatin binding dynamics of IL-33 are remarkably slower than IL-1α27, typical DNA-binding transcription factors28, and all previously studied mammalian nucleosome acidic patch binding proteins29,30; (4) IL-33 chromatin binding curtails its extracellular release as demonstrated by relatively high nuclear retention even within necrotic cells as assessed by analysis of IL-33 structural variants; (5) upon disruption of membrane integrity there is a relatively slow release of IL-33 over time; (6)

IL-33 is released extracellularly in complex with histones, and (7) IL-33 synergizes with chromatin components to induce inflammatory cytokine production. On the basis of these findings, we propose that the cytokine activity of IL-33 is regulated by a

“molecular chromatin sink” that decreases its bioavailability for release while

118 simultaneously enhancing both the duration of its release and the ST2-related bioactivity of released IL-33 (see model in Figure 8). We speculate this chromatin binding-mediated, fine-tune regulation likely arose due to the potency of extracellular IL-

33. For example, IL-33 is a more potent activator of eosinophils than the pro-allergic cytokine IL-4, which utilizes a classical secretion mechanism7.

In our study, we find no evidence for a role for nuclear IL-33 in regulating gene expression, consistent with a recent elegant proteomic analysis which found no reproducible effect of nuclear IL-33 on protein expression in primary endothelial cells31.

It is notable that we obtained similar results with an alternate cell type, namely epithelial cells. In addition, we utilized an independent approach by inducing overexpression of IL-

33wjhereas the prior work focused on a gene-silencing approach. The inducible system likely minimizes compensatory mechanisms that could potentially occur with extended exposure to IL-33. Furthermore, the cells used herein did not express ST2 nor responded to extracellular IL-33, unlike the endothelial cells used in the aforementioned proteomic analysis. Nevertheless, we cannot exclude the existence of a context- dependent intracellular functions of nuclear IL-33.

Our finding that loss of IL-33 chromatin binding increases its extracellular release by mediating intracellular retention is strengthened by the use of multiple independent necrotic stimuli. These findings build upon a previous elegant study showing that

119 deletion of chromatin binding domain of IL-33 by an endogenous knock-in approach causes overwhelming lethal eosinophilic inflammation not seen in control mice32.

The ST2 bioassay studies herein suggest that the existence of extracellular IL-33- complexes could serve as a mechanism to ensure that cells that respond to IL-33 also simultaneously respond to released histones. At present the molecular mechanisms underlying the synergistic cytokine release are not clear. The fact that synergistic cytokine production was not observed with truncated IL-33 suggests suggests that complexing IL-33 with histones enhances activation of the ST2 receptor. However, alternative explanation exist, including synergistic downstream signaling secondary to activation of cell surface receptors by histones. Consistent with this idea, histones have been proposed to act as damage-associated molecular patterns to promote inflammation through activation of the Toll-like receptors TLR2, TLR4, and TLR933-35.

Further studies should investigate the downstream effects and molecular mechanisms of simultaneous cell activation by IL-33 and histones. This approach could potentially uncover aspects of IL-33 biology not previously appreciated.

IL-33 is often grouped together with thymic stromal lymphopoietin (TSLP) and IL-25 as innate, epithelium-derived cytokines expressed at mucosal surfaces that promote type 2 immune responses36. However, IL-33 is the only one of the three stored in the nucleus as a pre-formed molecule bound to chromatin. Its chromatin-binding properties distinguish IL-33from TSLP, IL-25, and other prototypical alarmins such as IL-1α and

120

HMGB1. IL-33 has much slower dynamics of chromatin binding than either of the two

37,38. HMGB1 is known not to be released in similar complexes during necrosis39, and no

IL-1α-chromatin complexes have been described under necrotic conditions. In addition, the chromatin-binding properties also separate IL-33 from other nucleosome acidic patch-binding proteins. To the best of our knowledge, no mammalian nucleosome acidic patch-binding proteins have been shown to exhibit as low of an intranuclear mobility as

IL-33 as assessed by FRAP. The fact that other nucleosome acidic patch-binding proteins, such as HMGN129, HMGN230, are intracellular transcriptional regulators with minor extracellular immune-related activities40 fits with our proposed model that the unique properties of IL-33 chromatin binding arose to fine-tune its extracellular activity.

IL-33 is classically associated with promoting allergy through extracellular release during necrosis. However, in assessing the localization of IL-33 in human allergic esophageal tissue we only detected nuclear expression. This is consistent with previous studies that have only detected IL-33 nuclear localization in the bronchial epithelium as detected by immunostaining despite high levels of IL-33 present in the bronchoalveolar lavage fluid41. Additionally, in other tissues42,43 IL-33 is localized to the nucleus. Perhaps only a small fraction of IL-33 is secreted, which remains relatively undetectable. Additionally, extracellular IL-33 may have a very short half-life due to proteolytic degradation, binding to decoy soluble ST2, or internalization upon binding cell surface ST244,45. Finally, EoE patients may not be undergoing sufficient cellular stimulation to induce cellular necrosis and IL-33 release at the time of the biopsy procurement. It is notable that patients are fasting prior to endoscopy and hence not

121 exposed to the causative food antigens immediately prior to the biopsy. Indeed, the biochemical fractionation studies clearly demonstrate that IL-33 has very tight binding to chromatin as high salt extractions did not completely release it.

In summary, we have further defined chromatin binding properties of IL-33 in epithelial cells, elucidated its relatively strong dynamic binding compared with other classic nuclear cytokines (e.g. IL-1α and HMGB1), and demonstrated functional roles for this chromatin binding. We have shown that chromatin binding regulates both the availability of IL-33 for release and its extracellular cytokine activity. While we cannot completely rule out a transcriptional role for nuclear IL-33, we find no evidence to support this expected function. We propose that the nuclear localization of IL-33 is a consequence of its unique regulation by a “molecular chromatin sink”. As such, we have identified additional mechanisms by which the potent alarmin IL-33 is post-transcriptionally regulated.

METHODS

Cell Culture. The esophageal epithelial cell line TE-7 (a kind gift of Dr. Hainault,

France), and HMC-1 mast cell line were cultured in RPMI-1640 medium (Invitrogen,

Carlsbad, CA) supplemented with 5% fetal bovine serum (FBS). Primary esophageal epithelial cells were cultured in keratinocyte serum-free media (KSFM) (Life

Technologies, Grand Island, NY).

122

Generation of plasmids. In order to generate cells with stable, constitutive overexpression of IL-33, the cDNA for the Homo sapiens full-length IL33 mRNA

(NM_001314044.1) was cloned into the pLVX lentiviral vector using the In-fusion method (Clontech, CA). In order to generate plasmids encoding proteins with GFP fused to the C-terminus, the Infusion method (Clontech) was used to insert into the peGFP-N1 vector the cDNA for mus muscularis full-length IL1A (NM_010554.4) or the homo sapiens full-length IL-33 gene (NM_001314044.1) or a truncated form of the gene that only encodes amino acids 112-270, both with C-terminal FLAG tag (encoding the amino acid residues DYKDDDDK). Plasmid #16680 encoding human H2B-GFP was purchased from Addgene. In order to generate TE-7 cells with stable, doxycycline

(Dox)-inducible overexpression of IL-33, the cDNA for the homo sapiens full-length IL-

33 gene or a truncated form of the gene that only encodes amino acids 112-270, both with an additional C-terminal FLAG tag (encoding the amino acid residues

DYKDDDDK), were cloned into the pINDUCER20 vector via the Gateway cloning system.

Transduction. In order to generate TE-7 cells with stable, constitutive overexpression of IL-33, lentivirus was generated by transiently transfecting HEK 293T cells with either pLVX IL-33 plasmid or empty vector, and then transducing TE-7 cells with supernatants in the presence of polybrene (5 µg/mL) with centrifugation at 2000 x g for 1 hour at room temperature. The following day, puromycin selection was applied at 1 μg/mL to

123 generate pools of cells with stable overexpression. Following at least one week of puromycin treatment, single-cell clones were generated using limiting dilution. In order to generate cells with stable, dox-inducible overexpression of IL-33, TE-7 cells were transduced with lentivirus pINDUCER20 Full-length or 112-270 IL-33-Flag, or the empty vector, as described above except that selection was performed with G418.

IL-33 induction and RNA-sequencing. To induce IL-33 expression, dox-inducible TE-

7 cells were treated with 100 ng/mL of doxycycline (Clontech) or control media for 48 hours. Expression of wild-type and truncated IL-33 was verified by Western blot. RNA was isolated using Tripure reagent and subjected to genome-wide RNA-sequencing through the Cincinnati Children’s Hospital Medical Center Gene Expression Core. The

RNA-seq results were analyzed using BioWardrobe46 (http://biowardrobe.cchmc.org).

Briefly, the RNA-seq Fastq files from the Illumina pipeline were aligned by STAR provided with the human RefSeq transcriptome. Differentially expressed genes upon treatment with doxycycline were identified using differential expression analysis for sequence count data (DESeq)47. Venny (http://bioinfogp.cnb.csic.es/tools/venny/) was used to intersect gene lists. For heatmap generation, Cluster 3.0

(http://bonsai.hgc.jp/~mdehoon/software/cluster/software.htm) was used for clustering data using Euclidian distance with average linkage. For visualization, the Java Treeview

(http://jtreeview.sourceforge.net/) was used. RNA-sequencing data files were uploaded to the GEO database under the accession number XXXXXX.

124

Western blot. Supernatants were obtained by centrifugation at 3,000xg for 5 minutes at

4 °C. NuPage LDS loading buffer, beta-mercaptoethanol, and protease inhibitors

(Roche) were then added to the supernatants. Cell lysates were extracted from the pellets by lysing cells in RIPA buffer (50 mM Tris-HCl pH 8; 150 mM NaCl, 1% Igepal,

0.5% sodium deoxycholate; 0.1% SDS,and 1 mM EGTA), supplemented with beta- mercaptoethanol and protease inhibitors (Roche), followed by sonication for three rounds of 10 seconds. Lysates and supernatant samples were boiled for 15 minutes, loaded onto a 4-12% SDS-PAGE gel (Invitrogen), and subjected to Western blot analysis. Membranes were probed with goat anti-IL-33 (AF3625,R&D), rabbit anti-p38

MAPK XP (8690, Cell Signalling), goat anti-Lamin B (sc-6217, Santa Cruz), rabbit anti- total histone H3 (ab1791, Abcam), mouse anti-histone H2B (ab52484 , Abcam) or mouse anti-Hsp90 (TA500494, Origene) primary antibodies. Secondary IRDye- conjugated antibodies were from LI-COR Biosciences (Lincoln, Nebraska).

Quantification of signal was performed with Image Studio Lite software

(http://www.licor.com/bio/products/software/image_studio_lite/).

Induction of necrosis. IL-33 expression was induced in pools of dox-inducible TE-7 cells as above. Cells were treated for four hours with calcium ionophore A23187 (20

μM) or vehicle. Two sets of supernatants and cell pellets were harvested. Whole cell lysis was performed on one set of cell pellets as described above; the second cell of cell pellets were lysed with 0.5% Triton X-100 in RPMI media). Lactate dehydrogenase activity assay (Promega, catalog # G7890) was performed according to manufacturer’s instructions on one set of supernatants and on the 0.5% Triton X-100 cell lysates.

125

Western blot was performed on the second set of supernatants and cell pellets as described above. In independent experiments, the same pools of TE-7 cells were subjected to two rounds of freeze-thaw (30 minutes at -80 degrees Celsius and 1 minute at 37 degrees Celsius). Viability was assessed by counts with Trypan blue exclusion, then Western blot was performed on supernatants and cell pellets.

Biochemical fractionation. Biochemical fractionation was performed on single-cell clones of TE-7 cells with stable, constitutive overexpression of full-length IL-33 (pLVX

IL-33). Triton X-100 (0.5%) was used to solubilize both the plasma and nuclear membranes in order to release cytoplasmic, nucleoplasmic, and weakly DNA-bound proteins. Proteins with moderate-to-high association with chromatin were then extracted with serial treatment of the residual pellet with micrococcal nuclease (MNase), ammonium sulfate [(NH4)2SO4] (0.65 M), and sodium chloride [NaCl] (2 M). To begin the fractionation, cells were washed with 1 mL of cold PBS and aliquoted equally into 1.5 mL Eppendorf tubes.Cells were resuspended in cold Buffer A (10 mM Tris pH 7.5, 100 mM NaCl, 3 mM MgCl2, 1 mM EGTA and 300 mM sucrose) supplemented with protease inhibitors and 0.5% Triton X-100, and incubated at room temperature for 5 minutes in order to permeabilize the cells. The supernatants (S1), representing cytoplasmic and nucleoplasmic proteins, were then collected after centrifugation for 5 minutes at 350xg.

Pellets were then digested with 2000 gel units of micrococcal nuclease (MNase) (NEB) for 20 minutes at 37 °C in the presence of 5 mM CaCl2 and 5 mM Tris pH 7.5.

Supernatant S2, containing proteins with moderate association with chromatin, was harvested after centrifugation at 2000xg for 5 min. Residual pellets were then exposed

126 to cold Buffer A supplemented with 0.65 M ammonium sulfate for 15 minutes at room temperature to extract remaining chromatin. Supernatant S3, containing proteins with tight association with chromatin, was collected by centrifugation at 2000 g for 5 minutes.

Pellets were resuspended 2 M NaCl in water and left at RT for 15 minutes. Supernatant

S4 was then harvested following centrifugation at 2000xg for 5 minutes. All centrifugation steps were performed at 4 °C. Protein analysis of cell-equivalent volumes of the supernatant and pellet from each fraction was then performed by Western blot.

Live-cell imaging. TE-7 esophageal epithelial cells were transiently transfected plasmid encoding GFP-fusion proteins; 24 hours later live-cell imaging experiments were performed. For Hoechst co-localization experiments, transfected cells were pre- treated with Hoechst 33342 DNA dye (16.2 mM) for 30 minutes prior to imaging in PBS supplemented with 10% FBS with a Nikon A1R LUN-V Inverted confocal microscope with a 60X/1.27NA water objective. Pearson correlation coefficients of fluorescence between GFP and Hoechst 33342 within the nucleus of individual cells was determined using Nikon NIS AR elements program.

In separate experiments, GFP fluorescence and DIC were continuously detected. Triton

X-100 (final concentration 0.13%) was added to the cells to induce immediate necrosis, and imaging continued for an additional three minutes. The amount of intracellular fluorescence in individual cells was determined over time with normalization to the initial fluorescence in that cell before cell death.

127

Fluorescence recovery after photobleaching. TE-7 esophageal epithelial cells were transiently transfected with plasmid encoding GFP-fusion proteins. After 24 hours,

FRAP was performed on cells in PBS supplemented with 10% FBS using a Nikon A1R

LUN-V Inverted confocal microscope using a 60X/1.27NA water objective and Nyquist zoom. A 15 pixel circular region of interest (ROI) was used to bleach heterochromatic regions defined by intense green staining for three loops of 1 second with 25% percent bleach using the 488 nm laser. Fluorescence within the ROI was recorded at approximately 1 second intervals for the following 60 seconds. Fluorescence at each timepoint within the ROI was calculated relative to unbleached area with background subtraction and normalized to the prebleach signal. Recovery of fluorescence occurs because of the influx of other GFP-fusion proteins into that region that exchange with the photobleached proteins48. For chromatin-binding proteins, the amount of time it takes to occur is inversely related to their residence time at binding sites49.

Procuring and processing of esophageal biopsies. This study was performed with the approval of the CCHMC Institutional Review Board. Informed consent was obtained from patients or their legal guardians to donate tissue samples for research and to have their clinical information entered into the Cincinnati Center for Eosinophilic Disorders

(CCED) database. Patients with no history of EoE or other eosinophilic gastrointestinal disorders (EGID) and with a current biopsy indicating 0 eosinophils/400x high power field (HPF) in the distal esophagus and taking no form of glucocorticoid treatment at time of biopsy served as normal controls. Patients with active EoE were defined as those having 15 or more eosinophils/HPF at the time of biopsy and not receiving

128 swallowed glucocorticoid or diet treatment at time of endoscopy. Esophageal biopsies were fixed with formalin and embedded in paraffin (FFPE).

Immunofluorescence. Immunofluorescence of esophageal biopsies was performed as previously described50. Briefly, slides with 4 µm sections of FFPE esophageal biopsies underwent deparaffinization by serial incubations with xylene, 100% EtOH, 95% EtOH,

70% EtOH, and 50% EtOH. Antigen retrieval was performed using sodium citrate buffer

(10 mM sodium citrate, 0.05% Tween 20, pH 6.0). Sections were blocked with blocking buffer (10% donkey serum/PBS) and then incubated with primary antibody directed against IL-33 (mouse clone Nessy-1, ALX-804-840, Enzo or goat polyclonal, AF3625,

R&D) or control antibody diluted in blocking buffer overnight at 4 °C in humidified chambers. The next day, slides were washed with PBS, incubated with AlexaFluor 488- conjugate donkey anti-goat IgG or AlexaFluor 647-conjugated donkey anti-mouse IgG secondary antibodies diluted in blocking buffer for one hour at RT in humidified chambers, and washed in the presence of DAPI (0.5 µg/mL). Finally, cover slip was added with ProLong Gold mounting reagent (Molecular Probes). Slides were imaged using Nikon A1R Inverted confocal microscope.

In separate experiments, immunofluorescence was performed on primary esophageal epithelial cells or in single-cell clones of TE-7 cells with stable, constitutive overexpression of full-length IL-33 (pLVX IL-33). Cells were cultured on Ibidi 8-well chambers. Cells were fixed with 4% PFA for 10 minutes, washed with PBS, then permeabilized with 0.1% Triton X-100 in PBS. Cells were blocked with blocking buffer

129 for 30 minutes and incubated with primay antibodies diluted in blocking buffer for 1-2 hour at RT. Cells were washed with PBS, incubated with secondary antibodies diluted in blocking buffer for one hour at RT, and washed in the presence of DAPI (0.5 µ/mL).

Cells were placed in fresh PBS and imaged using Nikon A1R Inverted confocal microscope.

Size-exclusion chromatography. Necrosis was induced by cryoshock in pools of TE-7 cells with Dox-inducible overexpression of WT or truncated IL-33 as described above.

Size-exclusion chromatography was run on supernatants using GE Healthcare HiPrep

16/60 Sephacryl S-100 HR column. Protein concentration was continuously monitored by measuring UV absorbance at a wavelength of 280 nm. Fractions of 1 mL were collected. IL-33 expression in fractions was dermined by IL-33 ELISA (DY3625, R&D).

In separate experiments, double-stranded DNA concentration in fractions was quantitated by Qubit (Q32854, Thermo Scientific) following the manufacturer’s instructions, then fractions corresponding to elution volumes of 35-39 and 55-59 mL were pooled, concentrated by acetone precipitation, and subjected to Western blot analysis.

Co-immunoprecipitation. Necrosis was induced by cryoshock in pools of TE-7 cells with Dox-inducible overexpression of WT IL-33 as described above in PBS. Protein concentration was determined by BCA assay (23227, Thermo Scientific). Equal amounts of proteins were precleared for two hours at 4 °C with Protein A/G beads (sc-

130

2003, Santa Cruz) in the presence of 250 mM NaCl, 1% NP-40, 1 mM EDTA, and protease inhibitors. Samples were incubated overnight at 4 °C with 2 µg of antibody.

Samples were incubated for one hour at 4 °C with Protein A/G beads before elution with glycine (pH 2.8). Protein expression in eluates was determined by Western blot analysis.

Acid-extraction of histones. Histones were isolated from TE-7 cells as described previously51. All steps were performed at 4 °C. Briefly, cells were washed with PBS, incubated in hypotonic lysis buffer (10 mM Tris-HCl [pH 8.0], 1 mM KCl, 1.5 mM MgCl2,

0.2% Triton-X 10, 1 mM DTT, protease inhibitors) for 30 minutes, spun at 10,000xg for

10 minutes, and incubated overnight in 0.4 N H2SO4. Supernatants were collected after centrifugation at 16,000xg for 10 minutes and incubated for 30 minutes with trichloroacetic acid (final concentration 33%). Proteins were washed with ice-cold acetone and resuspended in MilliQ water. Protein concentration was determined by

BCA assay (23227, Thermo Scientific). Purity of the preparation was assessed by SDS-

PAGE and Coommassie blue staining.

ST2 bioactivity assay. HMC-1 mast cells blocked for one hour on ice with 20 µg/mL of an ST2-blocking antibody or IgG control, plated in 96-well plates at 50,000 cells per well, and stimulated overnight in the presence of 4 µg/mL of antibody with recombinant human truncated IL-33 (200-33, Peprotech), recombinant human WT IL-33

131

(H00090865-P01, Abnova), or acid-extracted histones prepared as described above. IL-

8 levels in supernatants was determined by IL-8 ELISA (431504, BioLegend).

Statistical Analysis. GraphPad Prism software was used for indicated statistical analyses. A p value < 0.05 was considered to be statistically significant.

132

REFERENCES

1 Stanley, A. C. & Lacy, P. Pathways for cytokine secretion. Physiology (Bethesda,

Md.) 25, 218-229, doi:10.1152/physiol.00017.2010 (2010).

2 Bertheloot, D. & Latz, E. HMGB1, IL-1alpha, IL-33 and S100 proteins: dual-

function alarmins. Cellular & molecular immunology 14, 43-64,

doi:10.1038/cmi.2016.34 (2017).

3 Moussion, C., Ortega, N. & Girard, J. P. The IL-1-like cytokine IL-33 is

constitutively expressed in the nucleus of endothelial cells and epithelial cells in

vivo: a novel 'alarmin'? PloS one 3, e3331, doi:10.1371/journal.pone.0003331

(2008).

4 Pichery, M. et al. Endogenous IL-33 is highly expressed in mouse epithelial

barrier tissues, lymphoid organs, brain, embryos, and inflamed tissues: in situ

analysis using a novel Il-33-LacZ gene trap reporter strain. J Immunol 188, 3488-

3495, doi:10.4049/jimmunol.1101977 (2012).

5 Liew, F. Y., Girard, J. P. & Turnquist, H. R. Interleukin-33 in health and disease.

Nat Rev Immunol 16, 676-689, doi:10.1038/nri.2016.95 (2016).

6 Martin, N. T. & Martin, M. U. Interleukin 33 is a guardian of barriers and a local

alarmin. Nat Immunol 17, 122-131, doi:10.1038/ni.3370 (2016).

7 Bouffi, C. et al. IL-33 markedly activates murine eosinophils by an NF-kappaB-

dependent mechanism differentially dependent upon an IL-4-driven

autoinflammatory loop. Journal of immunology (Baltimore, Md. : 1950) 191, 4317-

4325, doi:10.4049/jimmunol.1301465 (2013).

133

8 Luthi, A. U. et al. Suppression of interleukin-33 bioactivity through proteolysis by

apoptotic caspases. Immunity 31, 84-98, doi:10.1016/j.immuni.2009.05.007

(2009).

9 Lefrancais, E. et al. IL-33 is processed into mature bioactive forms by neutrophil

elastase and cathepsin G. Proc Natl Acad Sci U S A 109, 1673-1678,

doi:10.1073/pnas.1115884109 (2012).

10 Lefrancais, E. et al. Central domain of IL-33 is cleaved by mast cell proteases for

potent activation of group-2 innate lymphoid cells. Proc Natl Acad Sci U S A 111,

15502-15507, doi:10.1073/pnas.1410700111 (2014).

11 Cohen, E. S. et al. Oxidation of the alarmin IL-33 regulates ST2-dependent

inflammation. Nature communications 6, 8327, doi:10.1038/ncomms9327 (2015).

12 Venturelli, N. et al. Allergic skin sensitization promotes eosinophilic esophagitis

through the IL-33-basophil axis in mice. The Journal of allergy and clinical

immunology, doi:10.1016/j.jaci.2016.02.034 (2016).

13 Grotenboer, N. S., Ketelaar, M. E., Koppelman, G. H. & Nawijn, M. C. Decoding

asthma: translating genetic variation in IL33 and IL1RL1 into disease

pathophysiology. The Journal of allergy and clinical immunology 131, 856-865,

doi:10.1016/j.jaci.2012.11.028 (2013).

14 Shimizu, M. et al. Functional SNPs in the distal promoter of the ST2 gene are

associated with atopic dermatitis. Human molecular genetics 14, 2919-2927,

doi:10.1093/hmg/ddi323 (2005).

15 Castano, R., Bosse, Y., Endam, L. M. & Desrosiers, M. Evidence of association

of interleukin-1 receptor-like 1 gene polymorphisms with chronic rhinosinusitis.

134

American journal of rhinology & allergy 23, 377-384,

doi:10.2500/ajra.2009.23.3303 (2009).

16 Nieuwenhuis, M. A. et al. Combining genomewide association study and lung

eQTL analysis provides evidence for novel genes associated with asthma.

Allergy 71, 1712-1720, doi:10.1111/all.12990 (2016).

17 Savenije, O. E. et al. Association of IL33-IL-1 receptor-like 1 (IL1RL1) pathway

polymorphisms with wheezing phenotypes and asthma in childhood. The Journal

of allergy and clinical immunology 134, 170-177, doi:10.1016/j.jaci.2013.12.1080

(2014).

18 Carriere, V. et al. IL-33, the IL-1-like cytokine ligand for ST2 receptor, is a

chromatin-associated nuclear factor in vivo. Proc Natl Acad Sci U S A 104, 282-

287, doi:10.1073/pnas.0606854104 (2007).

19 Lotze, M. T. & Tracey, K. J. High-mobility group box 1 protein (HMGB1): nuclear

weapon in the immune arsenal. Nature reviews. Immunology 5, 331-342,

doi:10.1038/nri1594 (2005).

20 Werman, A. et al. The precursor form of IL-1alpha is an intracrine

proinflammatory activator of transcription. Proc Natl Acad Sci U S A 101, 2434-

2439 (2004).

21 Roussel, L., Erard, M., Cayrol, C. & Girard, J. P. Molecular mimicry between IL-

33 and KSHV for attachment to chromatin through the H2A-H2B acidic pocket.

EMBO reports 9, 1006-1012, doi:10.1038/embor.2008.145 (2008).

22 Zhou, J., Fan, J. Y., Rangasamy, D. & Tremethick, D. J. The nucleosome surface

regulates chromatin compaction and couples it with transcriptional repression.

135

Nature structural & molecular biology 14, 1070-1076, doi:10.1038/nsmb1323

(2007).

23 Kalashnikova, A. A., Porter-Goff, M. E., Muthurajan, U. M., Luger, K. & Hansen,

J. C. The role of the nucleosome acidic patch in modulating higher order

chromatin structure. Journal of the Royal Society, Interface 10, 20121022,

doi:10.1098/rsif.2012.1022 (2013).

24 Shlopov, B. V., Smith, G. N., Jr., Cole, A. A. & Hasty, K. A. Differential patterns of

response to doxycycline and transforming growth factor beta1 in the down-

regulation of collagenases in osteoarthritic and normal human chondrocytes.

Arthritis and rheumatism 42, 719-727, doi:10.1002/1529-

0131(199904)42:4<719::aid-anr15>3.0.co;2-t (1999).

25 Gordon, E. D. et al. Alternative splicing of interleukin-33 and type 2 inflammation

in asthma. Proc Natl Acad Sci U S A 113, 8765-8770,

doi:10.1073/pnas.1601914113 (2016).

26 Bae, S. et al. Contradictory functions (activation/termination) of neutrophil

proteinase 3 enzyme (PR3) in interleukin-33 biological activity. J Biol Chem 287,

8205-8213, doi:10.1074/jbc.M111.295055 (2012).

27 Cohen, I. et al. Differential release of chromatin-bound IL-1alpha discriminates

between necrotic and apoptotic cell death by the ability to induce sterile

inflammation. Proceedings of the National Academy of Sciences of the United

States of America 107, 2574-2579, doi:10.1073/pnas.0915018107 (2010).

136

28 Karpova, T. S., Chen, T. Y., Sprague, B. L. & McNally, J. G. Dynamic interactions

of a transcription factor with DNA are accelerated by a chromatin remodeller.

EMBO reports 5, 1064-1070, doi:10.1038/sj.embor.7400281 (2004).

29 Prymakowska-Bosak, M. et al. Mitotic phosphorylation of chromosomal protein

HMGN1 inhibits nuclear import and promotes interaction with 14.3.3 proteins.

Molecular and cellular biology 22, 6809-6819 (2002).

30 Wu, J. et al. High mobility group nucleosomal binding domain 2 (HMGN2)

SUMOylation by the SUMO E3 ligase PIAS1 decreases the binding affinity to

nucleosome core particles. The Journal of biological chemistry 289, 20000-

20011, doi:10.1074/jbc.M114.555425 (2014).

31 Gautier, V. et al. Extracellular IL-33 cytokine, but not endogenous nuclear IL-33,

regulates protein expression in endothelial cells. Scientific reports 6, 34255,

doi:10.1038/srep34255 (2016).

32 Bessa, J. et al. Altered subcellular localization of IL-33 leads to non-resolving

lethal inflammation. Journal of autoimmunity 55, 33-41,

doi:10.1016/j.jaut.2014.02.012 (2014).

33 Chen, R., Kang, R., Fan, X. G. & Tang, D. Release and activity of histone in

diseases. Cell death & disease 5, e1370, doi:10.1038/cddis.2014.337 (2014).

34 Huang, H. et al. Endogenous histones function as alarmins in sterile

inflammatory liver injury through Toll-like receptor 9 in mice. Hepatology

(Baltimore, Md.) 54, 999-1008, doi:10.1002/hep.24501 (2011).

35 Xu, J., Zhang, X., Monestier, M., Esmon, N. L. & Esmon, C. T. Extracellular

histones are mediators of death through TLR2 and TLR4 in mouse fatal liver

137

injury. Journal of immunology (Baltimore, Md. : 1950) 187, 2626-2631,

doi:10.4049/jimmunol.1003930 (2011).

36 (!!! INVALID CITATION !!! {}).

37 Luheshi, N. M., McColl, B. W. & Brough, D. Nuclear retention of IL-1 alpha by

necrotic cells: a mechanism to dampen sterile inflammation. European journal of

immunology 39, 2973-2980, doi:10.1002/eji.200939712 (2009).

38 Mollica, L. et al. Glycyrrhizin binds to high-mobility group box 1 protein and

inhibits its cytokine activities. Chemistry & biology 14, 431-441,

doi:10.1016/j.chembiol.2007.03.007 (2007).

39 Urbonaviciute, V. et al. Induction of inflammatory and immune responses by

HMGB1-nucleosome complexes: implications for the pathogenesis of SLE. The

Journal of experimental medicine 205, 3007-3018, doi:10.1084/jem.20081165

(2008).

40 Feng, Y., Huang, N., Wu, Q. & Wang, B. HMGN2: a novel antimicrobial effector

molecule of human mononuclear leukocytes? Journal of leukocyte biology 78,

1136-1141, doi:10.1189/jlb.0505280 (2005).

41 Christianson, C. A. et al. Persistence of asthma requires multiple feedback

circuits involving type 2 innate lymphoid cells and IL-33. The Journal of allergy

and clinical immunology 136, 59-68.e14, doi:10.1016/j.jaci.2014.11.037 (2015).

42 Meephansan, J. et al. Expression of IL-33 in the epidermis: The mechanism of

induction by IL-17. Journal of dermatological science 71, 107-114,

doi:10.1016/j.jdermsci.2013.04.014 (2013).

138

43 Sundnes, O. et al. Epidermal Expression and Regulation of Interleukin-33 during

Homeostasis and Inflammation: Strong Species Differences. The Journal of

investigative dermatology 135, 1771-1780, doi:10.1038/jid.2015.85 (2015).

44 Zhao, J. et al. Focal adhesion kinase-mediated activation of glycogen synthase

kinase 3beta regulates IL-33 receptor internalization and IL-33 signaling. Journal

of immunology (Baltimore, Md. : 1950) 194, 795-802,

doi:10.4049/jimmunol.1401414 (2015).

45 Zhao, J. et al. F-box protein FBXL19-mediated ubiquitination and degradation of

the receptor for IL-33 limits pulmonary inflammation. Nature immunology 13, 651-

658, doi:10.1038/ni.2341 (2012).

46 Kartashov, A. V. & Barski, A. BioWardrobe: an integrated platform for analysis of

epigenomics and transcriptomics data. Genome biology 16, 158,

doi:10.1186/s13059-015-0720-3 (2015).

47 Anders, S. & Huber, W. Differential expression analysis for sequence count data.

Genome biology 11, R106, doi:10.1186/gb-2010-11-10-r106 (2010).

48 Vivante, A., Brozgol, E., Bronshtein, I. & Garini, Y. Genome organization in the

nucleus: From dynamic measurements to a functional model. Methods (San

Diego, Calif.), doi:10.1016/j.ymeth.2017.01.008 (2017).

49 Catez, F., Ueda, T. & Bustin, M. Determinants of histone H1 mobility and

chromatin binding in living cells. Nature structural & molecular biology 13, 305-

310, doi:10.1038/nsmb1077 (2006).

139

50 Rochman, M. et al. Profound loss of esophageal tissue differentiation in patients

with eosinophilic esophagitis. J Allergy Clin Immunol,

doi:10.1016/j.jaci.2016.11.042 (2017).

51 Shechter, D., Dormann, H. L., Allis, C. D. & Hake, S. B. Extraction, purification

and analysis of histones. Nature protocols 2, 1445-1457,

doi:10.1038/nprot.2007.202 (2007).

140

Figures and Figure Legends

Figure 4.1: Effect of nuclear IL-33 on gene expression. (A) Western blot analysis of IL-33 protein expression, with quantification in (B), in indicated single-cell clones of TE-7 cells with

141 stable, doxycycline (Dox)-inducible expression of wild-type (WT) IL-33, truncated IL-33, or the empty vector with or without treatment for 48 hours with Dox. Total p38 was used as a loading control. (C) TE-7 single-cell clones from (A) were treated with Dox for 48 hours, then RNA was isolated and subjected to genome-wide RNA sequencing (RNA-seq) analysis. (D) IL33 RPKM levels in samples subjected to RNA-seq in (C). (E) Heatmap of fold change in each single-cell clone overexpressing WT IL-33 of genes with statistically significant differential expression in at least one clone. Differentially expressed genes were identified by filtering on RPKM ≥ 1, Benjamini-Hochberg False Discovery Correction of 5%, and fold change ≥ 1.5. Depicted is the log2 ratio of gene expression in cells treated with Dox compared to cells without Dox. (F) Table of genes differentially expressed in all three clones overexpressing WT IL-33, truncated IL-33, or the empty vector, respectively.

142

Figure 4.2: IL-33 association with chromatin. (A) Schematic of biochemical fractionation of IL-33-overexpressing TE-7 cells with serial extraction with indicated treatments. (B) show representative Western blot with quantitation of two independent experiments in (C). S refers to proteins extracted with each treatment, and P represents unextracted proteins remaining in the residual pellet. MNase, micrococcal nuclease. Graph in (C) depicts the percentage of protein levels in indicated fraction compared to total (S1+S2+S3+S4+P4). (D,E) TE-7 cells with Dox- inducible overexpression of wild-type or truncated IL-33 were treated with Triton X-100 (0.5%). (D) is a representative Western blot, and (E) shows quantification of the ratio of IL-33 detected in the supernatant to pellet (S/P) after Triton X-100 treatment from three independent experiments. **, p < 0.01 using parametric two-tailed Student’s t-test. Trunc, truncated; WT, wild-type.

143

Figure 4.3: Heterochromatic localization of IL-33. (A) Schematic of wild-type and truncated IL-33-GFP fusion proteins. (B) Representative live-cell images from one experiment from (C). Top row indicates Hoechst 33342 DNA dye (red), middle row indicates presence of GFP-fusion protein (green), and bottom row contains merged images. (C) Quantification of the Pearson correlation coefficient within the nucleus between fluorescence from Hoechst 33342 DNA dye and GFP after transient transfection of plasmid encoding indicated GFP-fusion protein into TE-7 cells. Graph shows average and standard deviation of combined data from 3-5 independent experiments (between 13-29 cells each). One-way ANOVA p value was < 0.0001, and depicted are p values from Holm-Sidak multiple comparisons test. Trunc, truncated; WT, wild-type.

144

Figure 4.4: Dynamics of IL-33 binding to chromatin. (A) Fluorescence recovery after photobleaching (FRAP) was performed on TE-7 cells after transient transfection with plasmid encoding indicated GFP-fusion proteins. A region of interest (ROI) [white dashed lines] was bleached, and fluorescence within that ROI was determined continuously over the following 60 seconds. Representative images of GFP fluorescence (green) from before bleach (top row), immediately after bleach (middle row), and 60 seconds post-bleach (bottom row) are shown for indicated GFP-fusion protein. For GFP-fusion proteins with significant cytoplasmic localization, the nucleus is outlined with yellow dashed lines. (B) Representative FRAP experiment showing fluorescence within ROI during bleach and for 60 seconds post-bleach for each GFP-fusion protein. Dashed gray lines indicate 50% recovery for wild-type IL-33-GFP. (C) Mean and standard error of the mean of 50% and 70% recoveries of indicated GFP-fusion proteins from 3- 5 independent experiments. N.D. (I.B.), not determined due to insufficient bleach; N.D. (I.R.), not determined due to insufficient recovery.

145

Figure 4.5: Effect of chromatin binding on IL-33 release during necrosis. (A-C) TE-7 pools with stable, Dox-inducible overexpression of WT or truncated IL-33 were treated for 4 with calcium ionophore A23187 (20 μM) [Iono] or vehicle [Veh]. (A) shows a representative Western blot. (B) is quantification of the ratio of IL-33 detected in the supernatant vs the pellet (S/P) by Western blot. (C) is quantification of the percent release of lactate dehydrogenase (LDH) using enzymatic activity assay, defined as LDH activity in supernatant divided by the total amount in the supernatant plus pellet [S/(S+P)]. (B,C) depict mean and standard error of the mean of three independent experiments. Two-way ANOVA interaction terms for IL-33 and LDH % release were p = 0.05 and p > 0.90, respectively. (D-F) The same pools of IL-33-overexpressing TE-7 cells were subjected to cryoshock. (D) shows a representative Western blot. (E,F) are quantification of the ratio of IL-33 detected in supernatant to pellet (E) or cell viability as assessed by Trypan blue exclusion (F); presented is mean and standard error of the mean of three independent experiments; ***, p < 0.001 and n.s., not significant by Student’s t-test. S, supernatants; P, cell pellets.

146

Figure 4.6: Chromatin binding dynamics regulate the kinetics of release of IL-33. (A) TE-7 cells were transiently transfected with plasmid encoding indicate GFP-fusion protein and then treated with Triton X-100 (0.13%) to induce necrosis. Depicted are images from a representative experiment showing 5 seconds prior to cell killing by the Triton X-100 (“Pre-cell death”; top row) or 15, 60, or 120 seconds after death (“Post-cell death”; second, third and fourth rows, respectively) from a representative experiment. (B) Quantification of intracellular fluorescence intensity from three combined independent experiments. Data are normalized to time zero, which is when Triton X-100 was added. Photobleach curve was generated in the absence of Triton X-100. For each depicted curve, dark line indicates mean and light lines indicate standard error of the mean. (C) Mean and standard error of the mean of the percentage of intracellular fluorescence intensity at indicate time points from (B) relative to time zero.

147

Figure 4.7: Characterization of released IL-33 species. TE-7 pools with stable, Dox-inducible overexpression of WT or truncated IL-33 were subjected to cryoshock, and then the presence of high molecular weight IL-33 species in the supernatants was determined by size-exclusion

148 chromatography. (A) Protein levels in fractions from the column as determined by UV absorbance at 280 nm. (B) IL-33 protein levels in different fractions as determined by ELISA, with normalization to the fraction with the highest amount of IL-33. Blue arrows indicates where WT IL-33 is predicted to elute based off of a previously-generated protein standard curve. (C) Proportion of IL-33 present in indicated fractions from (B). (D) DNA concentration in indicated fractions as determined by Qubit. (E) High molecular weight (corresponding to 35-39 mL retention volume, H) and low molecular weight (corresponding to 55-59 mL retention volumes, L) were pooled, concentrated by acetone precipitation, and subjected to Western blot analysis for IL-33 and histone H2B. (A,B,E) depict a representative example of three independent expeirments. (C,D) depict mean and standard error of the mean of cumulative data from three independent experiments. (F) TE-7 pools with stable, Dox-inducible overexpression of WT were subjected to cryoshock, then co-immunoprecipitation with anti-histone H2B antibody, isotype control, or Protein A/G beads alone was performed. Protein expression in eluates of IL-33 and histone H2B was assessed by Western blot analysis. Black dashed arrows indicate bands corresponding to IL-33 and H2B. (G) Acid-extracted histones were run on an SDS-PAGE gel and subjected to Coommassie blue staining; right lane are indicated molecular weight markers. (H) HMC-1 mast cells were blocked for one hour with an an ST2-block antibody or IgG control and treated overnight with indicated amounts (in ng) of acid-purified histones and recombinant wild-type or truncated IL-33, then IL-8 levels in the supernatants were determined by ELISA. Depicted is representative data of three independent experiments. **, p < 0.01 by Student’s t- test

149

Figure 4.8: Proposed model of chromatin binding-mediated regulation of IL-33 extracellular activity. In live cells, IL-33 remains within the nuclei of epithelial cells (rectangular cells) due to binding to histones. When necrosis occurs, the integrity of the plasma and nuclear membranes is lost with partial retention of IL-33. Over time, there is a slow release

150 of IL-33 from the necrotic cell. IL-33 is released either as a monomer or complexes with chromatin. Extracellular IL-33 alone activates ST2 to induce downstream proinflammatory gene expression. IL-33-chromatin complexes induce synergistic ST2 signaling.

151

Supplemental Figure 4.S1: Nuclear localization of IL-33 in esophageal epithelial cells. (A) (A,B) Immunofluorescence (IF) of esophageal biopsies from control individuals (top row) or active EoE patients (bottom row). (C,D) IF of primary esophageal epithelial cells (C) or TE-7 cells with stable, constitutive overexpression of wild-type IL-33 (D). For (A-D), left column is DNA as indicated by DAPI staining (red). Middle column is staining (green) with mouse anti-IL- 33 antibody (A,C,D), goat anti-IL-33 antibody (B), or antibody controls (A-D) as indicated. Right column shows merged images.

152

CHAPTER 5: GENERAL DISCUSSION AND SUMMARY

Introduction

The development of EoE involves complex cross-talk between the epithelium and immune cells. The work contained in this dissertation originated from a desire to further understand the pathogenesis of this disorder, with a particular focus on the epithelium.

We identified in patients with active EoE a profound loss of esophageal epithelial differentiation and esophagus-specific IL-1-related activity (Chapter Two). Furthermore, we identified in patients with EoE a striking induction of expression of the IL-1 family member IL-33 in the nuclei of an undifferentiated esophageal epithelial cell population

(Chapter Three). Most significantly, we have answered the long-standing question in the field of why IL-33, a key cytokine that signals through a plasma membrane receptor, is normally restricted to the nucleus (Chapter Four).

More specifically, in Chapter Two we identified that 39% of esophagus-specific genes have altered expression in EoE (termed Eso-EoE genes), with approximately 90% of them being down-regulated. These transcriptional changes were reproduced with IL-13 treatment of differentiated esophageal epithelial cells. Notably, Eso-EoE genes are enriched for multiple IL-1 family members, proteases, and protease inhibitors.

Additionally, we found dysregulated esophageal epithelial differentiation in patients with

EoE. Collectively, these findings demonstrate a profound loss of esophageal tissue identify in EoE and suggest that esophagus-specific, imbalanced activity of proteases and IL-1 family members is integral to the pathogenesis of EoE.

153

In Chapter Three, we identified that protein expression of the IL-1 family member IL-33 is increased within the esophageal epithelium in nearly all patients with active EoE. IL-

33 induction is a function of disease activity as IL-33 epithelial expression normalizes upon disease remission. IL-33 induction is limited to a sub-population of basal layer cells. These IL-33-positive basal cells strongly express E-cadherin, KRT5, KRT14, p75 and p63, have low expression of PCNA, and do not express KRT4, Ki-67 or phospho-

H3. Finally, IL-33 is expressed in a sub-population of primary esophageal epithelial cells not currently dividing. These results indicate that IL-33 is induced in a quiescent, undifferentiated esophageal epithelial cell population in patients with EoE.

In Chapter Four, we show that intracellular IL-33 does not affect homeostatic gene expression as assessed by genome-wide transcript profiling of esophageal epithelial cells engineered to overexpress IL-33. Rather, we identified that IL-33 chromatin binding regulates its extracellular cytokine activity in several ways. First, it curtails its extracellular release as demonstrated by relatively high nuclear retention even within necrotic cells. Second, following disruption of membrane integrity, the chromatin binding mediates a relatively slow release of IL-33 over time that is unique compared with IL-1α.

Finally, IL-33 is released in complex with chromatin and synergizes with histones to induce receptor signaling via its cognate receptor ST2. We propose that the chromatin binding of IL-33 mediates complex, fine-tune regulation of its extracellular cytokine activity.

154

The implications of our collective findings are summarized and discussed further below.

Dysregulation of Epithelial Differentiation in EoE

Several lines of evidence have implicated interactions between IL-13 and the esophageal epithelium in the pathogenesis of EoE. IL-13 treatment of primary esophageal epithelial cells reproduces a high proportion of the EoE transcriptome1. IL-

13 treatment also induces release of chemokines that cause pathologic immune cell recruitment; a prominent example is the eosinophil-chemoattractant eotaxin-31,2.

Additionally, IL-13 induces disease-relevant structural and functional alterations, including dilated intercellular spaces impaired barrier function, in three dimensional cultures of esophageal epithelial cells3,4. Finally, variants in IL-13-regulated, epithelium- specific genes, such as CAPN14, have genetic association with disease risk for EoE5.

Treatment with an IL-13-neutralizing antibody normalizes the transcriptome of patients with active EoE, including expression of the basal keratin KRT146. In line with these previous findings, we have identified that IL-13 reproduces the dysregulation of Eso-

EoE genes in esophageal epithelial cell cultures. The importance of Eso-EoE genes to the pathogenesis of EoE is demonstrated by the fact that rare variants in Eso-EoE genes are enriched in EoE patients as compared to control individuals as analyzed by whole-exome sequencing. Thus, our findings further support the importance of IL-13- epithelium interactions in the pathogenesis of EoE. Significant study is still needed in order to elucidate the underlying molecular mechanisms underlying the dysregulation of

155 epithelial differentiation and Eso-EoE genes by IL-13. Gene-silencing approaches can be employed to assess the requirement of different components of IL-13 signaling, such as signal transducer and activator of transcription 6 (STAT6). Additionally, the readily- available transcriptomic data from IL-13-stimulated differentiated esophageal epithelial cell can be used to determine whether IL-13 dysregulates p63 or members of the Notch and BMP signaling pathways, which have previously been shown to profoundly regulate esophageal epithelial differentiation7-9.

We have identified a large number of candidate genes with potential contributions to the pathogenesis of EoE, most notably IL-1 family members and regulators of protease activity. The expression of multiple IL-1 family members is dysregulated within the esophagus in patients with EoE as IL-33 is up-regulated whereas IL-36α, IL-1α and IL-

1Rα exhibit decreased expression. Notably, IL-36α promotes T helper type 1 (Th1) responses through direct activation of naïve CD4+ T cells10 whereas IL-33 promotes type 2 immunity11. Therefore, these different IL-1 family members could contribute to the dysregulated balance of Th1 and Th2 immune responses observed in the disorder.

Furthermore, extracellular IL-33 is directly cleaved by immunocyte-derived serine proteases into processed forms with dramatically elevated ST2-related bioactivity12,13 while the cysteine protease calpain processes pro-IL-1α into a highly reactive mature form14. Therefore, altered activity of both serine and cysteine proteases within the esophagus could promote disease by inducing excess cytokine activity of IL-1α and IL-

33. Independently of any effects on IL-1 family members, altered serine and cysteine protease activity can also directly promote EoE by inducing impaired barrier function

156

(IBF) of the esophageal epithelium. IBF promotes disease progression by increasing the exposure of causative food allergens to immune cells present in the underlying lamina propria. IBF has been detected in esophageal epithelial cells with overexpression of the cysteine protease calpain 144. Additionally, overexpression of leucine-rich repeat- containing protein 31 (LRRC31) impairs esophageal IBF with altered activity of multiple kallikrein serine proteases15. These collective results suggest two main series of experiments in the future. The first is to use genetic approaches in differentiated esophageal epithelial cells to assess whether other regulators of serine or cysteine protease activity identified within Eso-EoE alter the barrier function. The second is to determine whether the serine or cysteine protease activity within epithelial cells alters the cytokine activity of IL-33. Genetic or pharmacologic approaches can be used to dysregulate serine or cysteine protease activity in supernatants of differentiated esophageal epithelial cells, and effects on the cytokine activity of IL-33 can be determined using ST2 reporter cell-lines. Further mechanistic studies can then be employed to identify the proteases responsible, such as kallikreins for example. This line of experiments could potentially lead to the novel discovery that epithelial-derived serine proteases can alter the bioactivity of IL-33.

At present it is unclear whether our finding of dramatic dysregulation of esophagus- specific genes in patients with EoE is generalizable to other immune-mediated disorders. We did not identify dysregulation of esophagus-specific genes in clinical specimens of patients with gastro-esophageal reflux disease nor of stomach-specific genes in samples from patients with eosinophilic gastritis. These findings demonstrate

157 that dysregulation of tissue-specific genes is not a universal feature of allergy or inflammation in general. Moving forward, the expression of tissue-specific genes should be assessed in clinical specimens of patients with allergic asthma and atopic dermatitis, which are both notable for elevated IL-13 signaling16, in order to determine the generalizability of this phenomenon to allergic diseases. Discovery that this phenomenon is detected in clinical samples of patients with atopic dermatitis but not asthma would suggest that the findings are specific to allergic diseases of squamous epithelia. A very important case in testing the generalizability of these findings to non- allergic inflammation is psoriasis, a skin condition that is hyperproliferative17 similar to

EoE but with no known allergic component18. Regardless of whether the dysregulation of tissue-specific genes occurs in other immune-mediated disorders, however, it is still clear that this phenomenon is present in patients with EoE and likely contributes to the pathogenesis of the disease (Figure 5.1).

Functions of IL-33-expressing Basal Layer Esophageal Epithelial Cells in EoE

We have identified that the IL-1 family member IL-33 is not present at detectable levels within the homeostatic esophagus but is induced within the esophageal epithelium in patients with active EoE. Through immunofluorescence, we identified that this IL-33- expressing basal layer cell population express markers consistent with being undifferentiated, quiescent esophageal epithelial progenitor cells. Our assertion is supported by a study that suggested that the human esophageal epithelial cells with the highest stem cell function, as determined by ex vivo-sphere-forming capacity, are

158 present in the basal layer contact with the basement membrane7. However, their study did not distinguish between papillary basal layer cells, which are negative for IL-33, and the IL-33-expressing interpapillary basal layer cells. Additional experiment evidence is required to prove our assertion that the IL-33-positive basal layer cells are a quiescent epithelial progenitor population. Further characterization of the markers of these cells could be performed by immunofluorescence would not lead to definitive proof.

Progenitor populations have been identified in other tissues using lineage tracing techniques possible in mice but not humans19. However, these techniques are not applicable in this situation as there are major structural and functional differences between the human and murine esophagus. While both the human and mouse esophagi contain stratified squamous epithelia, only the murine esophagus is highly keratinized. Additionally, the murine esophageal epithelium does not have papillae, so there cannot be a division of the basal layer into papillary and interpapillary basal layer.

Furthermore, there is no Ki-67-negative basal layer present in the murine esophagus20.

Therefore, direct isolation and ex vivo-culture of the IL-33-positive basal layer cells will be required. Finding that the IL-33-positive basal layer cells have higher sphere-forming capacity ex vivo than other esophageal epithelial cell populations would strongly support our assertion. In order to perform these experiments, flow cytometry can be used to generate pure populations of these cells using our described markers, such as p75NTR

(Figure 5.2).

The contributions of this IL-33-expressing basal layer cell population to the pathogenesis of EoE remain unclear. The current literature suggests that this cell

159 population undergo rare mitotic divisions in order to maintain the epithelium7,21,22.

Because basal cell hyperplasia is notable in EoE, this could be due in part to a higher rate of proliferation of these basal cells. While we found that these IL-33-positive basal layer cells generally do not proliferate, we cannot exclude this possibility. This hypothesis can be compared by testing the ex vivo sphere-forming capacity of esophageal interpapillary basal layer cells from patients with EoE or control individuals.

Furthermore, the interpapillary basal layer cells could indirectly promote the pathogenesis of EoE impacting the function of other cells present in the esophagus.

This could be achieved through dysregulation of cell-contact pathways, such as Notch, or secretion of soluble mediators, including cytokines, growth factors, and enzymes. A first step towards answering this question would be to determine the transcriptome of the interpapillary basal layer cells in the homeostatic and allergic esophagus using genome-wide RNA-sequencing. While laser capture microdissection would be the ideal technique to obtain the RNA from these cells, sorting the basal layer cells from esophageal biopsies using cell surface markers might be more feasible. Once transcriptome profiles of these cells have been generated, more specific hypotheses can be generated and tested.

Regulation of IL-33 Expression in EoE

Little is known about the regulation of IL-33 expression within the esophageal epithelium in patients with EoE. It is not known which signals induce IL-33 in the interpapillary basal layer cells, why expression within the esophagus is restricted to this epithelial cell

160 population, or why IL-33 is only expressed within the esophageal epithelium at appreciable levels in patients with EoE, even though IL-33 is homeostatically expressed in endothelia and other epithelia. One cannot help but notice that IL-33-expressing basal layer cells are non-dividing, similar to other IL-33-positive cell populations. For instance, IL-33 was first described as a marker of endothelial cell quiescence as it was only present in non-dividing endothelial cells and was lost when they started proliferating23. Additionally, human umbilical vein endothelial cells IL-33 is only present in Ki-67-negative cells24. Furthermore, in the murine stomach IL-33 is only expressed in

Ki-67-negative surface mucus cells25. In characterizing IL-33 expression in human primary esophageal epithelial cells we identified that IL-33-positive cells did not express

Ki-67 or phospho-H3 and only had low expression of nuclear PCNA. These findings support the notion that the restriction to non-dividing esophageal epithelial cells is an intrinsic feature of IL-33.

At least two possibilities explain why IL-33 expression would be restricted to non- dividing cells. One possibility is that IL-33 interferes with DNA damage response, which would then prevent the repair of typical double-stranded breaks that are induced as a normal part of cell proliferation26. Therefore, expression of IL-33 in highly proliferative cells could promote malignant transformation. In support of this hypothesis, Latency- associated nuclear antigen (LANA), a protein of the Kaposi sarcoma herpes virus family that has a very similar chromatin binding domain to IL-33, has been shown to impair the

DNA damage response. Expression of LANA does not affect the amount of DNA damage induced in response to ionizing radiation as assessed by total levels of

161

γH2Ax27, a marker of DNA double-stranded breaks. However, the chromatin binding domain of LANA impairs the recruitment of repair enzymes, such as 53BP1 and

BRCA1, to sites of DNA damage27. Due to the high similarities between the chromatin binding domains of IL-33 and LANA, it is likely that overexpression of IL-33 can mediate similar effects. This hypothesis can be easily tested by assessing formation of foci of

γH2Ax, 53BP1, and BRCA1 after irradiation or etoposide treatment. While an impact of

IL-33 on DNA damage response is unlikely to be relevant to EoE as there is no evidence of increased cancer risk in patients with EoE nor of DNA damage in esophageal basal layer cells, this could apply to cancers in which IL-33 is known to be up-regulated28-30. Another possibility for the restriction of IL-33 to non-dividing cells is that the nucleus undergoes dramatic alterations during mitosis. Such chromatin changes could include dissolution of the nuclear membrane, condensation of chromatin, and histone modifications31. If these alterations were to decrease the affinity of IL-33 for chromatin, then there could be excess IL-33 release from cells that undergo necrosis during the act of mitosis. Live-cell confocal microscopy with GFP-fusion proteins has demonstrated that IL-33 can bind mitotic chromatin32. However, this study did not examine strength of the interaction with chromatin nor has it been replicated using unlabeled IL-33. A simple test would be to synchronize IL-33-overexpressing cells in different stages of the cell cycle using techniques such as double-thymidine block and assess both IL-33 affinity for chromatin by biochemical fraction and extracellular release after treatment with necrotic stimuli.

162

Even if the studies suggested in the preceding paragraph identify a detrimental effect of

IL-33 expression in mitotic cells, it is still not yet clear why IL-33 is only present in non- dividing cells in the basal layer when there are non-dividing cells throughout the esophageal epithelium. One possibility is that wide expression of IL-33 throughout the epithelium would induce excessive release. In further support of this idea, basal layer cells are likely more protected from environmental insults and acid reflux than the suprabasal layer cells. Additionally, the suprabasal layers of the epithelium express the pro-allergic innate cytokines IL-25 and TSLP33,34, which are also released during cellular damage. Therefore, IL-33 may be restricted to basal layer cells to ensure that type 2 immune responses occur in response to cellular damage within this layer. Infiltration of immune cells, such as eosinophils and mast cells, into the esophageal epithelium requires transit through the lamina propria. This means that infiltrating immune cells must come in close contact to basal layer epithelial cells to the epithelium. This close proximity would allow IL-33 to activate these incoming immune cells than if IL-33 expression were limited to the suprabasal region of the epithelium (Figure 5.3).

It is still unclear which signal(s) induce IL-33 expression in esophageal basal layer epithelial cells. In endothelial cells, it is known that IL-33 is expressed in non-dividing cells due to the action of Notch1 signaling, which both induces IL-33 expression and maintains endothelial cell quiescence24. In fibroblasts, it is known that IL-33 is induced by pro-inflammatory signals such as TNF-alpha35. The effect of these stimuli on IL-33 expression in human primary esophageal epithelial cells should be examined. The two signals most likely to induce IL-33 expression in EoE are interferon-gamma and

163 oncostatin M. Interferon-gamma is known to induce IL-33 protein expression in basal skin and esophageal keratinocytes36,37. It typically is not considered important in the pathogenesis of EoE as it is a Th1 cytokine, but its expression is up-regulated in esophageal biopsies from EoE patients38. Oncostatin M is a proallergic cytokine that is a member of the IL-6 family of cytokines39, is up-regulated in EoE40, and is known to induce IL33 mRNA in liver endothelial cells41 and protein expression in mouse type 2 alveolar cells42. Therefore, the effects of treatment with oncostatin M, as well as the other stimuli mentioned above, should be determined in primary esophageal epithelial cells.

Potential Roles of IL-33 in the Pathogenesis of EoE

Our finding that IL-33 expression is increased within the esophageal epithelium in patients with active EoE and normalizes upon disease remission builds upon previous work suggesting an important role for IL-33 in the pathogenesis of the disorder. There is some association between genetic variants in the IL33 locus and disease risk5. Two other studies have demonstrated that esophageal biopsies from patients with EoE express higher levels of IL-33 protein than those from control individuals33,43.

Additionally, intraperitoneal injection into mice of recombinant IL-33 induces esophageal eosinophilia, epithelial hyperplasia, and production of Th2-associated cytokines43.

Finally, the IL-33-ST2 axis was shown to be required in an experimental murine model of EoE-like disease that employed repeated intranasal challenge with ovalbumin after epicutaneous sensitization44. While concerns do remain with regard to translating these

164 findings to the human disease because of significant differences between this experimental murine model and EoE disease in humans, the above findings collectively demonstrate an importance for IL-33 in the pathogenesis of EoE.

Extracellular IL-33 can potently robust a wide variety of immune cells, including type 2 innate lymphoid cells (ILC2s)45,46, CD4+ T cells, eosinophils47, mast cells48, and basophils44. Importantly, all of these cells populations have been identified to be present in the esophageal epithelium in patients with EoE and have been implicated in the pathogenesis of the disorder2,44,45,49,50. It is tempting to hypothesize that extracellular IL-

33 is critical to the pathogenesis of EoE through activation of some combination of the immune cells above. However, before doing so one must prove that extracellular release of IL-33 actually occurs within the esophagus in patients with EoE. The immunohistochemistry and immunofluorescence experiments presented in this thesis dissertation have not yielded any evidence for the extracellular release of IL-33 nor necrosis in IL-33-expressing cells in patients with EoE. There are several possible explanations for this result. First, patients are resting and avoiding food for the day before biopsies are taken, so it is possible that the necrosis only occurs in the presence of the aggravating food allergens. Also, it is possible that extracellular IL-33 was washed out when the biopsies were collected as significant time can elapse between removal of biopsy specimen from the esophagus to fixation in formalin. IL-33 could be released but below the detection limits of immunohistochemistry and immunofluorescence; this possibility could be assessed using a more sensitive technique, such as electron microscopy with immunogold labeling of IL-33. Moreover, it

165 was recently shown that IL-33 undergoes cysteine oxidation after extracellular release that causes a major conformation shift and changes its detection by ELISA51; it is possible that this change also decreases its recognition by the antibodies utilized in immunofluorescence and immunohistochemistry. Additionally, in vivo there is clearance of necrotic cells by phagocytes include macrophages, which could result in elimination of necrotic cells and released IL-33. Finally, IL-33 is internalized along with ST252,53.

Because of all of these reasons, it seems like it will be difficult to definitively identify extracellular IL-33 in esophageal biopsy specimens from patients with EoE. In line with this, previous studies that have only detected IL-33 nuclear localization in the bronchial epithelium as detected by immunostaining despite high levels of IL-33 present in the bronchoalveolar lavage fluid54. An approach circumvent the difficulty in detecting extracellular IL-33 would be to develop a biomarker for ST2 activity to demonstrate that

ST2-dependent signaling is occurring. A known step in ST2 signaling is internalization of the ST2 receptor and shuttling to early endosomes with degradation in proteasomes but not lysosomes53. Therefore, the presence of the ST2 receptor in early endosomes of immune cells should be assessed in esophageal biopsy specimens. Upon definitive proof of either extracellular IL-33 or of active ST2 signaling, further studies should examine the contributions of IL-33-mediated activation of the different ST2-expressing immune cell populations mentioned above.

While it is uncertain whether extracellular IL-33 is present in the esophagus of patients with EoE, there is very strong evidence that that nuclear IL-33 is present. Therefore, potential contributions of nuclear IL-33 to the pathogenesis of EoE should continue to

166 be studied. While our inability to identify an effect of intracellular IL-33 is consistent with previous work55,this was a surprising finding that warrants further investigation. Despite its strengths, our study was limited in that it employed non-physiologic overexpression of IL-33 in a cancer cell line and did not include stimulation conditions. Importantly, no studies to date have been performed in esophageal basal layer epithelial cells. A good starting point would be to assess the effect of IL-33 gene-silencing on three-dimensional cultures of primary esophageal epithelial cells both at baseline and upon stimulation with IL-13. The importance to the pathogenesis of EoE of any identified effects could then be assessed in subsequent studies.

Comparison of IL-33 to other Nuclear Alarmins

A major theme of our findings in Chapter Four of this dissertation is that the chromatin binding properties of IL-33 impart unique characteristics compared to the prototypic nuclear alarmins IL-1α and HMGB1. These unique characteristics exist because IL-33 directly binds core histone proteins whereas IL-1α and HMGB1 bind DNA nucleotides56.

Fluorescence recovery after photobleaching analysis revealed that IL-33 exhibits a dramatically slower intranuclear mobility than IL-1α or than what has been previously reported for HMGB157. Furthermore, IL-33 exhibited partial intracellular retention in necrotic cells with a relatively slow release over time and was released in complex with chromatin. All of these features of IL-33 were not present in a truncated form engineered not to bind chromatin. Additionally, intracellular retention in necrotic cells or slow release over time was not detected for IL-1α, and previous work has shown that

167

HMGB1 is not released bound to DNA during necrosis58. These collective results demonstrate major differences in the release of these alarmins that exist due to different chromatin binding properties. Furthermore, the existence of our newly-discovered synergistic histone-IL-33 signaling may also explain why a reproducible, non-necrotic active secretion has not been identified despite that fact that such processes are well- known for both IL-1α and HMGB1. HMGB1 is known to have post-translational modifications, including hyper-acetylation of certain residues, that decreases its affinity for DNA59,60 before extracellular secretion. IL-1α is cleaved by calpain into a mature form that lacks a nuclear localization sequence and is subsequently secreted61.

Importantly, both of these processes involve removing the alarmin from the DNA (Table

5.1). It is likely that IL-33 does not undergo similar active secretion to ensure that it is in complex with histones when present extracellularly.

Chromatin-mediated Regulation of IL-33-ST2 bioactivity

Our data clearly demonstrate that the chromatin binding of IL-33 mediates fine-tune regulation of its extracellular activity. The partial intracellular retention decreases the total amount of released IL-33, which should decrease ST2 activation. However, chromatin binding simultaneously allows for a relatively slow release over time, which increases the overall duration of IL-33 release. Additionally, IL-33 is released in chromatin-complexes that cause synergistic ST2-mediated signaling. It is interesting to consider why such complex regulation of IL-33’s ST2 activity evolved; this likely underscores the importance IL-33 in immunity. Furthermore, it is interesting to think

168 about the other ways in which chromatin binding could potentially alter the ST2 bioactivity of IL-33. Released IL-33 undergoes cysteine oxidation over time which induces a conformational change to prevent binding to the ST2 receptor51, and full- length IL-33 can be cleaved by enzymes derived from mast cells and neutrophils into mature forms with higher ST2 bioactivity12,13. Whether either of these two regulatory processes are undermined when IL-33 is present in complex with chromatin should be examined (Figure 5.4).

While we have identified the existence of IL-33-chromatin complexes in necrotic supernatants, it is not yet clear whether these complexes exist in vivo. This can be tested in bronchoalveolar lavage fluid from both asthmatic patients and from mice with allergic airway inflammation induced by repeated intranasal challenges with ovalbumin or Aspergillus fumigatus. While there have been numerous studies on the effects of IL-

33 on immune cells, to date there have not been any studies involving co-treatment of

IL-33 and histones. Upon identification that IL-33-chromatin complexes exist in vivo, the responses of isolated populations of ST2-expressing cells to IL-33 plus histones versus

IL-33 treatment alone should be assessed. In particular, there should be a focus on proliferation, cytokine production, and changes in function, including degranulation and migratory potential. While IL-33 is already known to be a potent activator of immune cells, synergy between histones and IL-33 could allow IL-33 to act as an even more potent inducer.

169

The ST2 bioassay experiments presented in this dissertation demonstrate that purified histones and IL-33 synergize to induce ST2-mediated signaling. Should the existence of released IL-33-chromatin complexes be proven in vivo, it would then be worthwhile to perform further mechanistic studies. While the most likely explanation is that there is altered binding to of IL-33-chromatin complexes to ST2 compared to IL-33 by itself, other alternative hypotheses must first be disproven. Recombinant proteins can be used determine which histones reproduce the synergy. Because IL-33 is known to bind H2A and H2B but not H3, a failure to see synergy with H2A and H2B or observing strong synergy with H3 would contradict the hypothesis. Additionally, the exact ST2-binding dynamics of IL-33-histone complexes can be determined using surface plasmon resonance. If the presence of histones does not affect the binding of IL-33 to ST2, then the most likely mechanism will involve co-activation by the histones of receptors other than ST2. The most likely candidates would be Toll-like receptors (TLRs) 2, 4, and 9, which are proposed receptors for extracellular histones62-64. The requirement of different

TLRs can be assessed using inhibitory antibodies, pharmacologic inhibitors, and/or gene-silencing approaches. As an independent approach, it can be determined which

TLR agonists, such as LPS or CpG, can replace histones in inducing synergistic ST2- mediated signaling. If these studies fail to identify a required TLR, then co- immunoprecipitation of IL-33-chromatin complexes coupled with mass spectrometry can be used to generate a list of candidate receptors for further experimentation.

170

Reflections and Conclusions

In total, this dissertation has combined a diverse set of experimental strategies, including analysis of genome-wide datasets, characterization of patient-derived clinical specimens, and molecular and biochemical techniques, to improve our understanding of both the pathogenesis of EoE and the unique regulation of the extracellular activity of

IL-1 cytokines, especially the prominent pro-allergic cytokine IL-33. We identified that there is a loss of tissue identity, with dysregulation of epithelial differentiation and of the expression of IL-1 family members, within the esophageal epithelium in patients with

EoE with likely contributions to the pathogenesis of the disease. We subsequently identified that the IL-1 family member IL-33, an alarmin with prominent contributions to allergic inflammation, is up-regulated in patients with EoE in an undifferentiated epithelial progenitor population. Because we only observed striking nuclear expression of IL-33 in biopsy tissue, we assessed the effect of nuclear IL-33 on gene expression using genome-wide RNA-sequencing analysis of esophageal epithelial cells but failed to identify an effect. Further characterization of the properties of nuclear IL-33 led us to identify that the chromatin binding regulates both the magnitude and kinetics of extracellular release of IL-33 during necrosis. Additionally, we identified that IL-33 is released in complexes with chromatin and that histones synergize with IL-33 to induce

ST2-dependent signaling. Through these studies, we have identified that the chromatin binding properties of IL-33 impart unique characteristics as compared to other nuclear alarmins such as IL-1α and HMGB1. Our results (Figure 5.5) prompt future investigation into whether co-treatment of IL-33 and histones has differential effects on ST2-

171 expressing cells as compared to IL-33 alone, which has the potential to uncover previously unappreciated downstream effects of extracellular IL-33.

Without a doubt, the work contained herein was a team effort as it would not have been possible without the support and guidance of my mentor Dr. Marc Rothenberg, members of my thesis dissertation committee, Dr. Mark Rochman, Cora Miracle, other members of the Rothenberg laboratory, and countless others. I have learned invaluable skills such as asking important questions, postulating and rigorously testing hypotheses, and reaching conclusions only after fully analyzing the data. This work has undoubtedly proven to be a superb training vehicle.

172

References

1 Blanchard, C. et al. IL-13 involvement in eosinophilic esophagitis: transcriptome

analysis and reversibility with glucocorticoids. The Journal of allergy and clinical

immunology 120, 1292-1300, doi:10.1016/j.jaci.2007.10.024 (2007).

2 Travers, J. & Rothenberg, M. E. Eosinophils in mucosal immune responses.

Mucosal Immunol 8, 464-475, doi:10.1038/mi.2015.2 (2015).

3 Sherrill, J. D. et al. Desmoglein-1 regulates esophageal epithelial barrier function

and immune responses in eosinophilic esophagitis. Mucosal Immunol. 7, 718-

729, doi:10.1038/mi.2013.90 (2014).

4 Davis, B. P. et al. Eosinophilic esophagitis-linked calpain 14 is an IL-13-induced

protease that mediates esophageal epithelial barrier impairment. JCI insight 1,

e86355, doi:10.1172/jci.insight.86355 (2016).

5 Kottyan, L. C. et al. Genome-wide association analysis of eosinophilic

esophagitis provides insight into the tissue specificity of this allergic disease.

Nature genetics 46, 895-900, doi:10.1038/ng.3033 (2014).

6 Rothenberg, M. E. et al. Intravenous anti-IL-13 mAb QAX576 for the treatment of

eosinophilic esophagitis. The Journal of allergy and clinical immunology,

doi:10.1016/j.jaci.2014.07.049 (2014).

7 Jeong, Y. et al. Identification and genetic manipulation of human and mouse

oesophageal stem cells. Gut 65, 1077-1086, doi:10.1136/gutjnl-2014-308491

(2016).

173

8 Ohashi, S. et al. NOTCH1 and NOTCH3 coordinate esophageal squamous

differentiation through a CSL-dependent transcriptional network.

Gastroenterology 139, 2113-2123, doi:10.1053/j.gastro.2010.08.040 (2010).

9 Jiang, M. et al. BMP-driven NRF2 activation in esophageal basal cell

differentiation and eosinophilic esophagitis. J Clin Invest 125, 1557-1568,

doi:10.1172/jci78850 (2015).

10 Vigne, S. et al. IL-36 signaling amplifies Th1 responses by enhancing

proliferation and Th1 polarization of naive CD4+ T cells. Blood 120, 3478-3487,

doi:10.1182/blood-2012-06-439026 (2012).

11 De la Fuente, M., MacDonald, T. T. & Hermoso, M. A. The IL-33/ST2 axis: Role

in health and disease. Cytokine & growth factor reviews 26, 615-623,

doi:10.1016/j.cytogfr.2015.07.017 (2015).

12 Lefrancais, E. et al. Central domain of IL-33 is cleaved by mast cell proteases for

potent activation of group-2 innate lymphoid cells. Proceedings of the National

Academy of Sciences of the United States of America 111, 15502-15507,

doi:10.1073/pnas.1410700111 (2014).

13 Lefrancais, E. et al. IL-33 is processed into mature bioactive forms by neutrophil

elastase and cathepsin G. Proceedings of the National Academy of Sciences of

the United States of America 109, 1673-1678, doi:10.1073/pnas.1115884109

(2012).

14 Kavita, U. & Mizel, S. B. Differential sensitivity of interleukin-1 alpha and -beta

precursor proteins to cleavage by calpain, a calcium-dependent protease. J Biol

Chem 270, 27758-27765 (1995).

174

15 D'Mello, R. J. et al. LRRC31 is induced by IL-13 and regulates kallikrein

expression and barrier function in the esophageal epithelium. Mucosal Immunol

9, 744-756, doi:10.1038/mi.2015.98 (2016).

16 Gandhi, N. A., Pirozzi, G. & Graham, N. M. H. Commonality of the IL-4/IL-13

pathway in atopic diseases. Expert review of clinical immunology 13, 425-437,

doi:10.1080/1744666x.2017.1298443 (2017).

17 Wraight, C. J. et al. Reversal of epidermal hyperproliferation in psoriasis by

insulin-like growth factor I receptor antisense oligonucleotides. Nature

biotechnology 18, 521-526, doi:10.1038/75382 (2000).

18 Zeichner, J. A. & Armstrong, A. The Role of IL-17 in the Pathogenesis and

Treatment of Psoriasis. The Journal of clinical and aesthetic dermatology 9, S3-

s6 (2016).

19 Blanpain, C., Horsley, V. & Fuchs, E. Epithelial stem cells: turning over new

leaves. Cell 128, 445-458, doi:10.1016/j.cell.2007.01.014 (2007).

20 Kalabis, J. et al. A subpopulation of mouse esophageal basal cells has properties

of stem cells with the capacity for self-renewal and lineage specification. J Clin

Invest 118, 3860-3869, doi:10.1172/jci35012 (2008).

21 Seery, J. P. & Watt, F. M. Asymmetric stem-cell divisions define the architecture

of human oesophageal epithelium. Current biology : CB 10, 1447-1450 (2000).

22 Seery, J. P. Stem cells of the oesophageal epithelium. Journal of cell science

115, 1783-1789 (2002).

175

23 Kuchler, A. M. et al. Nuclear interleukin-33 is generally expressed in resting

endothelium but rapidly lost upon angiogenic or proinflammatory activation. Am J

Pathol 173, 1229-1242, doi:10.2353/ajpath.2008.080014 (2008).

24 Sundlisaeter, E. et al. The alarmin IL-33 is a notch target in quiescent endothelial

cells. Am J Pathol 181, 1099-1111, doi:10.1016/j.ajpath.2012.06.003 (2012).

25 Buzzelli, J. N. et al. IL33 Is a Stomach Alarmin That Initiates a Skewed Th2

Response to Injury and Infection. Cellular and molecular gastroenterology and

hepatology 1, 203-221.e203, doi:10.1016/j.jcmgh.2014.12.003 (2015).

26 Jeggo, P. A. & Lobrich, M. DNA double-strand breaks: their cellular and clinical

impact? Oncogene 26, 7717-7719, doi:10.1038/sj.onc.1210868 (2007).

27 Leung, J. W. et al. Nucleosome acidic patch promotes RNF168- and

RING1B/BMI1-dependent H2AX and H2A ubiquitination and DNA damage

signaling. PLoS genetics 10, e1004178, doi:10.1371/journal.pgen.1004178

(2014).

28 Wasmer, M. H. & Krebs, P. The Role of IL-33-Dependent Inflammation in the

Tumor Microenvironment. Frontiers in immunology 7, 682,

doi:10.3389/fimmu.2016.00682 (2016).

29 Schwartz, C., O'Grady, K., Lavelle, E. C. & Fallon, P. G. Interleukin 33: an innate

alarm for adaptive responses beyond Th2 immunity-emerging roles in obesity,

intestinal inflammation, and cancer. Eur J Immunol 46, 1091-1100,

doi:10.1002/eji.201545780 (2016).

176

30 Lu, B., Yang, M. & Wang, Q. Interleukin-33 in tumorigenesis, tumor immune

evasion, and cancer immunotherapy. Journal of molecular medicine (Berlin,

Germany) 94, 535-543, doi:10.1007/s00109-016-1397-0 (2016).

31 Steffen, P. A. & Ringrose, L. What are memories made of? How Polycomb and

Trithorax proteins mediate epigenetic memory. Nature reviews. Molecular cell

biology 15, 340-356, doi:10.1038/nrm3789 (2014).

32 Roussel, L., Erard, M., Cayrol, C. & Girard, J. P. Molecular mimicry between IL-

33 and KSHV for attachment to chromatin through the H2A-H2B acidic pocket.

EMBO reports 9, 1006-1012, doi:10.1038/embor.2008.145 (2008).

33 Simon, D., Radonjic-Hosli, S., Straumann, A., Yousefi, S. & Simon, H. U. Active

eosinophilic esophagitis is characterized by epithelial barrier defects and

eosinophil extracellular trap formation. Allergy 70, 443-452, doi:10.1111/all.12570

(2015).

34 Chandramouleeswaran, P. M. et al. Preferential Secretion of Thymic Stromal

Lymphopoietin (TSLP) by Terminally Differentiated Esophageal Epithelial Cells:

Relevance to Eosinophilic Esophagitis (EoE). PloS one 11, e0150968,

doi:10.1371/journal.pone.0150968 (2016).

35 Martin, N. T. & Martin, M. U. Interleukin 33 is a guardian of barriers and a local

alarmin. Nature immunology 17, 122-131, doi:10.1038/ni.3370 (2016).

36 Sundnes, O. et al. Epidermal Expression and Regulation of Interleukin-33 during

Homeostasis and Inflammation: Strong Species Differences. The Journal of

investigative dermatology 135, 1771-1780, doi:10.1038/jid.2015.85 (2015).

177

37 Shan, J. et al. Interferon gamma-Induced Nuclear Interleukin-33 Potentiates the

Release of Esophageal Epithelial Derived Cytokines. PloS one 11, e0151701,

doi:10.1371/journal.pone.0151701 (2016).

38 Mulder, D. J. et al. Antigen presentation and MHC class II expression by human

esophageal epithelial cells: role in eosinophilic esophagitis. Am J Pathol 178,

744-753, doi:10.1016/j.ajpath.2010.10.027 (2011).

39 Hermanns, H. M. Oncostatin M and interleukin-31: Cytokines, receptors, signal

transduction and physiology. Cytokine & growth factor reviews 26, 545-558,

doi:10.1016/j.cytogfr.2015.07.006 (2015).

40 Pothoven, K. L. et al. Oncostatin M promotes mucosal epithelial barrier

dysfunction, and its expression is increased in patients with eosinophilic mucosal

disease. J Allergy Clin Immunol 136, 737-746.e734,

doi:10.1016/j.jaci.2015.01.043 (2015).

41 Arshad, M. I. et al. Oncostatin M induces IL-33 expression in liver endothelial

cells in mice and expands ST2+CD4+ lymphocytes. Am J Physiol Gastrointest

Liver Physiol 309, G542-553, doi:10.1152/ajpgi.00398.2014 (2015).

42 Richards, C. D. et al. Regulation of IL-33 by Oncostatin M in Mouse Lung

Epithelial Cells. Mediators of inflammation 2016, 9858374,

doi:10.1155/2016/9858374 (2016).

43 Judd, L. M. et al. Elevated IL-33 expression is associated with pediatric

eosinophilic esophagitis, and exogenous IL-33 promotes eosinophilic esophagitis

development in mice. Am J Physiol Gastrointest Liver Physiol 310, G13-25,

doi:10.1152/ajpgi.00290.2015 (2016).

178

44 Venturelli, N. et al. Allergic skin sensitization promotes eosinophilic esophagitis

through the IL-33-basophil axis in mice. J Allergy Clin Immunol 138, 1367-

1380.e1365, doi:10.1016/j.jaci.2016.02.034 (2016).

45 Doherty, T. A. et al. Group 2 innate lymphocytes (ILC2) are enriched in active

eosinophilic esophagitis. J Allergy Clin Immunol 136, 792-794.e793,

doi:10.1016/j.jaci.2015.05.048 (2015).

46 Kim, B. S. & Artis, D. Group 2 innate lymphoid cells in health and disease. Cold

Spring Harbor perspectives in biology 7, doi:10.1101/cshperspect.a016337

(2015).

47 Bouffi, C. et al. IL-33 markedly activates murine eosinophils by an NF-kappaB-

dependent mechanism differentially dependent upon an IL-4-driven

autoinflammatory loop. Journal of immunology (Baltimore, Md. : 1950) 191, 4317-

4325, doi:10.4049/jimmunol.1301465 (2013).

48 Joulia, R., L'Faqihi, F. E., Valitutti, S. & Espinosa, E. IL-33 fine tunes mast cell

degranulation and chemokine production at the single-cell level. J Allergy Clin

Immunol, doi:10.1016/j.jaci.2016.09.049 (2016).

49 Noti, M. et al. Thymic stromal lymphopoietin-elicited basophil responses promote

eosinophilic esophagitis. Nature medicine 19, 1005-1013, doi:10.1038/nm.3281

(2013).

50 Abonia, J. P. et al. Involvement of mast cells in eosinophilic esophagitis. J Allergy

Clin Immunol 126, 140-149, doi:10.1016/j.jaci.2010.04.009 (2010).

51 Cohen, E. S. et al. Oxidation of the alarmin IL-33 regulates ST2-dependent

inflammation. Nature communications 6, 8327, doi:10.1038/ncomms9327 (2015).

179

52 Zhao, J. et al. Focal adhesion kinase-mediated activation of glycogen synthase

kinase 3beta regulates IL-33 receptor internalization and IL-33 signaling. Journal

of immunology (Baltimore, Md. : 1950) 194, 795-802,

doi:10.4049/jimmunol.1401414 (2015).

53 Zhao, J. et al. F-box protein FBXL19-mediated ubiquitination and degradation of

the receptor for IL-33 limits pulmonary inflammation. Nature immunology 13, 651-

658, doi:10.1038/ni.2341 (2012).

54 Christianson, C. A. et al. Persistence of asthma requires multiple feedback

circuits involving type 2 innate lymphoid cells and IL-33. The Journal of allergy

and clinical immunology 136, 59-68.e14, doi:10.1016/j.jaci.2014.11.037 (2015).

55 Gautier, V. et al. Extracellular IL-33 cytokine, but not endogenous nuclear IL-33,

regulates protein expression in endothelial cells. Scientific reports 6, 34255,

doi:10.1038/srep34255 (2016).

56 Bertheloot, D. & Latz, E. HMGB1, IL-1alpha, IL-33 and S100 proteins: dual-

function alarmins. Cellular & molecular immunology 14, 43-64,

doi:10.1038/cmi.2016.34 (2017).

57 Mollica, L. et al. Glycyrrhizin binds to high-mobility group box 1 protein and

inhibits its cytokine activities. Chemistry & biology 14, 431-441,

doi:10.1016/j.chembiol.2007.03.007 (2007).

58 Urbonaviciute, V. et al. Induction of inflammatory and immune responses by

HMGB1-nucleosome complexes: implications for the pathogenesis of SLE. The

Journal of experimental medicine 205, 3007-3018, doi:10.1084/jem.20081165

(2008).

180

59 Bonaldi, T. et al. Monocytic cells hyperacetylate chromatin protein HMGB1 to

redirect it towards secretion. The EMBO journal 22, 5551-5560,

doi:10.1093/emboj/cdg516 (2003).

60 Ito, I., Fukazawa, J. & Yoshida, M. Post-translational methylation of high mobility

group box 1 (HMGB1) causes its cytoplasmic localization in neutrophils. J Biol

Chem 282, 16336-16344, doi:10.1074/jbc.M608467200 (2007).

61 Zheng, Y., Humphry, M., Maguire, J. J., Bennett, M. R. & Clarke, M. C.

Intracellular interleukin-1 receptor 2 binding prevents cleavage and activity of

interleukin-1alpha, controlling necrosis-induced sterile inflammation. Immunity

38, 285-295, doi:10.1016/j.immuni.2013.01.008 (2013).

62 Huang, H. et al. Endogenous histones function as alarmins in sterile

inflammatory liver injury through Toll-like receptor 9 in mice. Hepatology

(Baltimore, Md.) 54, 999-1008, doi:10.1002/hep.24501 (2011).

63 Xu, J., Zhang, X., Monestier, M., Esmon, N. L. & Esmon, C. T. Extracellular

histones are mediators of death through TLR2 and TLR4 in mouse fatal liver

injury. Journal of immunology (Baltimore, Md. : 1950) 187, 2626-2631,

doi:10.4049/jimmunol.1003930 (2011).

64 Chen, R., Kang, R., Fan, X. G. & Tang, D. Release and activity of histone in

diseases. Cell death & disease 5, e1370, doi:10.1038/cddis.2014.337 (2014).

181

Figures and Figure Legends

Figure 5.1: Loss of esophageal epithelial differentiation in EoE. In the homeostatic esophagus, epithelial cells in the basal zone express undifferentiated markers, including KRT5 and KRT14. Epithelial cells in the upper layers express differentiated markers, including KRT4 and CRNN. In EoE, there is a dramatic expansion of the basal zone and KRT5+/KRT14+ epithelial cells (indicated by gray vertical bar). There is a corresponding decrease in epithelial cells expressing KRT4 and CRNN (indicated by black vertical bar). CRNN, cornulin; EoE, eosinophilic esophagitis; KRT, keratin.

182

Figure 5.2: Markers of IL-33-positive esophageal epithelial cells in EoE. In the esophagus of patients with active EoE, IL-33 (yellow circle) is expressed in the nuclei of basal layer cells that also express undifferentiated keratins (KRT5, KRT14) and epithelial progenitor markers (p75, p63). IL-33-positive basal cells do not express the proliferation markers Ki67, pH3, or PCNA; these markers are instead highly expressed in higher layers of the basal zone. Gray vertical bar indicates epithelial cell layers expressing KRT5 and KRT14, and black vertical bar indicates those layers expressing KRT4 and CRNN. CRNN, cornulin; EoE, eosinophilic esophagitis; IL, interleukin; PCNA, proliferating cell nuclear antigen; pH3, phosphorylated histone H3; KRT, keratin.

183

Figure 5.3: Route of immune cell infiltration into the esophagus in EoE. In the esophagus of patients with active EoE, IL-33 (yellow circle) is expressed in basal layer cells that do not express proliferation markers (Ki67, phospho-H3, PCNA), which are instead highly expressed within other layers of the basal zone. The upper layers of the epithelium, which express the differentiated markers KRT4 and CRNN (black vertical bar), also do not express proliferation markers. ST2-expressing immune cells (blue circle), such as eosinophils, travel through the vasculature to the esophagus, traverse through the lamina propria, and then spread throughout the epithelium. Over the course of entering the epithelium, immune cells must come in close physical proximity to IL-33-positive basal layer cells and would be potently activated should any IL-33 be released. Gray vertical bar indicates epithelial cell layers expressing KRT5 and KRT14. CRNN, cornulin; EoE, eosinophilic esophagitis; IL, interleukin; KRT, keratin; PCNA, proliferating cell nuclear antigen; phospho-H3, phosphorylated histone H3.

184

Figure 5.4: Regulation of the ST2-mediated bioactivity of IL-33 through post-translational modification. During necrosis, IL-33 is released as a full-length form that is capable of activating the ST2 receptor. Full-length IL-33 can be processed by serine proteases derived from mast cells or neutrophils into mature forms with increased ST2-mediated bioactivity. Over time, both full-length and processed mature forms of IL-33 are inactivated through the formation of disulfide bonds (blue curve). IL, interleukin; ST2, suppressor of tumorigenicity 2.

185

Figure 5.5: Changes in epithelial differentiation in EoE and regulation of IL-33 cytokine activity by chromatin binding. In patients with EoE, there is a dramatic dysregulation of epithelial differentiation as indicated by the expansion of the basal zone and KRT5+/KRT14+ epithelial cell layers (gray vertical bar) with a corresponding decrease in epithelial cells expressing KRT4 and CRNN (black vertical bar). Additionally, immune cells, including eosinophils, infiltrate into the esophagus, and IL33 is expressed in basal layer epithelial cells. IL- 33 is restricted to the nucleus due to chromatin binding. In response to necrotic stimuli, the integrity of the plasma and nuclear membranes of basal layer cells is lost with partial retention of IL-33. Over time, there is slow release of IL-33 from the necrotic cells, either as a monomer or complexes with chromatin. Released IL-33 and histones then synergize to induce ST2 signaling and downstream inflammatory gene expression from ST2-expressing immune cells. CRNN, cornulin; DNA, deoxyribonucleic acid; EoE, eosinophilic esophagitis; IL, interleukin; KRT, keratin; PCNA, proliferating cell nuclear antigen; pH3, phosphorylated histone H3; ST2, suppressor of tumorigenicity 2.

186

HMGB1 IL-1α IL-33

Release during Yes Yes Yes necrosis?

Released in chromatin complexes during No Not Examined Yes necrosis?

Active (non-necrotic) Yes Yes No secretion?

Mechanism of removal PTM that decrease Cleavage by calpain into mature None from chromatin during affinity for DNA form that lacks NLS described active secretion?

Mode of release during Noncanonical vesicular Noncanonical vesicular secretion None active secretion? secretion pathway pathway described

Table 5.1: Comparison of active secretion of nuclear alarmins. Table lists properties of passive release and active, non-necrotic secretion of the nuclear alarmins HMGB1, IL-1α, and IL-33. Known PTM of HMGB1 that decrease its affinity for DNA include acetylation, methylation, and phosphorylation. DNA, deoxyribonucleic acid; HMGB1, high mobility group B1; IL, interleukin; NLS, nuclear localization sequence; PTM, post-translational modifications.

187