Functional characterization of the new 8q21 Asthma risk locus

Cristina M T Vicente

B.Sc, M.Sc

A thesis submitted for the degree of Doctor of Philosophy at

The University of Queensland in 2017

Faculty of Medicine

Abstract

Genome wide association studies (GWAS) provide a powerful tool to identify genetic variants associated with asthma risk. However, the target genes for many allergy risk variants discovered to date are unknown. In a recent GWAS, Ferreira et al. identified a new association between asthma risk and common variants located on chromosome 8q21. The overarching aim of this thesis was to elucidate the biological mechanisms underlying this association. Specifically, the goals of this study were to identify the gene(s) underlying the observed association and to study their contribution to asthma pathophysiology.

Using genetic data from the 1000 Genomes Project, we first identified 118 variants in linkage disequilibrium (LD; r2>0.6) with the sentinel allergy risk SNP (rs7009110) on chromosome 8q21. Of these, 35 were found to overlap one of four Putative Regulatory Elements (PREs) identified in this region in a lymphoblastoid cell line (LCL), based on epigenetic marks measured by the ENCODE project. Results from analysis of gene expression data generated for LCLs (n=373) by the Geuvadis consortium indicated that rs7009110 is associated with the expression of only one nearby gene: PAG1 - located 732 kb away. PAG1 encodes a transmembrane adaptor protein localized to lipid rafts, which is highly expressed in immune cells.

Results from chromosome conformation capture (3C) experiments showed that PREs in the region of association physically interacted with the promoter of PAG1. Furthermore, results from luciferase reporter assays demonstrated that one of these PREs (PRE 3) acted as a transcriptional enhancer on PAG1 exclusively when it carried the rs2370615:C asthma predisposing allele. This variant, which is in complete LD with rs7009110, was found to disrupt the binding of the Foxo3a transcription factor to PRE3. As such, rs2370615 represents a putative functional variant underlying the association between rs7009110 and asthma.

In addition to PAG1, other genes in the 8q21 region could be targets of asthma risk variants, despite the lack of evidence for an association between rs7009110 and gene expression. The closest gene is ZBTB10 (75 kb away), a repressor of the specificity proteins (Sp1, Sp3 and SP4), which are transcription factors known to regulate several immune-related genes. Therefore, based on distance to rs7009110 and its known function, we prioritised ZBTB10 to investigate if it could represent an additional target gene of 8q21 asthma risk variants. Results from 3C assays showed that the same PREs that interacted with the PAG1 promoter also interacted with the ZBTB10 promoter. Results from luciferase

ii assays for ZBTB10 were inconclusive: PRE 2 enhanced the activity of the ZBTB10 promoter, and this effect was inhibited by an asthma risk allele but these differences were not statistically significant. Together, these results indicate that ZBTB10 is also likely to be a target of 8q21 asthma risk variants.

To study the contribution of PAG1 and ZBTB10 to asthma pathophysiology, we performed in vivo experiments using wild type (WT), Pag1-/- and Zbtb10het mice, which included challenging mice with an allergen or virus, both able to induce airway inflammation. Results from these experiments indicate that Pag1 expression might regulate granulocyte, lymphocyte and/or airway epithelial cell function. On the other hand, the expression of Zbtb10 was critical for embryonic development and could play a role in immune cell development as well as activation during the allergen sensitization phase.

In conclusion we have identified PAG1 and ZBTB10 as target genes of 8q21 asthma risk variants and showed that both genes play important roles in the immune system and development of airway inflammation.

iii Declaration by author

This thesis is composed of my original work, and contains no material previously published or written by another person except where due reference has been made in the text. I have clearly stated the contribution by others to jointly-authored works that I have included in my thesis.

I have clearly stated the contribution of others to my thesis as a whole, including statistical assistance, survey design, data analysis, significant technical procedures, professional editorial advice, financial support and any other original research work used or reported in my thesis. The content of my thesis is the result of work I have carried out since the commencement of my higher degree by research candidature and does not include a substantial part of work that has been submitted to qualify for the award of any other degree or diploma in any university or other tertiary institution. I have clearly stated which parts of my thesis, if any, have been submitted to qualify for another award.

I acknowledge that an electronic copy of my thesis must be lodged with the University Library and, subject to the policy and procedures of The University of Queensland, the thesis be made available for research and study in accordance with the Copyright Act 1968 unless a period of embargo has been approved by the Dean of the Graduate School.

I acknowledge that copyright of all material contained in my thesis resides with the copyright holder(s) of that material. Where appropriate I have obtained copyright permission from the copyright holder to reproduce material in this thesis and have sought permission from co-authors for any jointly authored works included in the thesis.

iv Publications during candidature

Peer-reviewed papers:

Vicente CT, Edwards SL, Hillman KM, Kaufmann S, Mitchell H, Bain L, Glubb DM, Lee JS, French JD, Ferreira MAR. (2015) “Long-range modulation of PAG1 expression by 8q21 allergy risk variants”. American Journal of Human Genetics 97: 1-8

Ferreira MA, Jansen R, Willemsen G, Penninx B, Bain LM, Vicente CT, Revez JA, Matheson MC, Hui J, Tung JY, Baltic S, Le Souëf P, Montgomery GW, Martin NG, Robertson CF, James A, Thompson PJ, Boomsma DI, Hopper JL, Hinds DA, Werder RB, Phipps S; Australian Asthma Genetics Consortium Collaborators. (2017) “Gene-based analysis of regulatory variants identifies 4 putative novel asthma risk genes related to nucleotide synthesis and signalling”. Journal of Allergy and Clinical Immunology 139(4):1148-1157.

Conference abstracts:

Vicente C, Edwards S, Hillman K, Kaufmann S, Mitchell H, Bain L, Glubb D, Lee J, French J, Ferreira M. “Long-range modulation of PAG1 expression by 8q21 allergy risk variants”. Annals of Translational Medicine 2015;3(S2):AB013

v Publications included in this thesis

Vicente CT, Edwards SL, Hillman KM, Kaufmann S, Mitchell H, Bain L, Glubb DM, Lee JS, French JD, Ferreira MAR. (2015) “Long-range modulation of PAG1 expression by 8q21 allergy risk variants”. American Journal of Human Genetics 97: 1-8 – incorporated as Chapter 2.

Contributor Statement of contribution

Concept and design: 10% Cristina T. Vicente Analysis and interpretation: 30% Drafting and production: 30% Concept and design: 25% Stacey L. Edwards Analysis and interpretation: 15% Drafting and production: 10% Concept and design: 0% Kristine M. Hillman Analysis and interpretation: 10% Drafting and production: 0% Concept and design: 0% Susanne Kaufmann Analysis and interpretation: 5% Drafting and production: 0% Concept and design: 0% Hayley Mitchell Analysis and interpretation: 3% Drafting and production: 0% Concept and design: 0% Lisa Bain Analysis and interpretation: 3% Drafting and production: 0% Concept and design: 0% Dylan M. Glubb Analysis and interpretation: 3% Drafting and production: 0% Concept and design: 0% Jason S. Lee Analysis and interpretation: 6% Drafting and production: 0% Concept and design: 25% Juliet D. French Analysis and interpretation: 10% Drafting and production: 10% Concept and design: 40% Manuel A.R. Ferreira Analysis and interpretation: 15% Drafting and production: 50%

vi Contributions by others to the thesis

Overall Thesis The project was conceived and designed with the help of my main supervisor Dr. Manuel Ferreira and co-supervisors Dr. Juliet French and A/Prof. Simon Phipps, who also provided feedback on the interpretation and presentation of the results.

Chapter 1: Introduction Review article submitted for publication together with Joana Revez and Dr. Manuel Ferreira. This publication is currently under review.

Chapter 2: Identification of PAG1 and ZBTB10 as target genes of 8q21 asthma-risk variants Individual contributions as stated in the “Publication included in this thesis” section. Routine reagents used to perform experiments were prepared by several laboratories, depending on the technique. Contributing laboratories include the following from QIMR Berghofer: Asthma Genetics (Dr. Manuel Ferreira); Functional Genetics (Dr. Juliet French); Functional Cancer Genomics (Dr. Stacey Edwards); and Control of Gene Expression (Professor Frank Gannon).

Chapter 3: Mechanisms of regulation of PAG1 and ZBTB10 expression Western blot and hypoxia experiments were done in collaboration with Dr. Jason Lee from the Control of Gene Expression laboratory. The correlation analyses were done by Dr. Manuel Ferreira.

Chapter 4: PAG1 and ZBTB10 expression in human and mouse tissues This study makes use of data generated by the Blueprint Consortium. A full list of the investigators who contributed to the generation of the data is available from www.blueprint- epigenome.eu. Funding for the project was provided by the European Union's Seventh Framework Programme (FP7/2007-2013) under grant agreement no 282510 – BLUEPRINT. The Genotype-Tissue Expression (GTEx) Project was supported by the Common Fund of the Office of the Director of the National Institutes of Health, and by NCI, NHGRI, NHLBI, NIDA, NIMH, and NINDS. The data used for the analyses described in this manuscript were obtained from the GTEx Portal on September 2017. Dr. Vivian Zhang assisted with mouse tissue collection, lung section slide preparation and staining. Dr. Jason Lynch performed mediastinal lymph node collection from experimental

vii animals. Ms Rhiannon Werder collected airway epithelial cells from experimental animals. Routine reagents used to perform experiments were prepared by the Laboratory for Respiratory Mucosal Immunity at the University of Queensland (A/Prof. Simon Phipps) and the Asthma Genetics Laboratory at QIMR Berghofer.

Chapter 5: The contribution of Pag1 to allergen-induced airway inflammation in mice Dr. Ashik Ullah assisted with mouse models, tissue collection, flow cytometry and endpoint measurements (experiments 3, 4 and 5). Dr. Vivian Zhang and Dr. Jason Lynch contributions as described for Chapter 4. Routine reagents used to perform experiments were prepared by the Laboratory for Respiratory Mucosal Immunity at the University of Queensland.

Chapter 6: The contribution of Pag1 to viral-induced airway inflammation in mice Contributions by Dr. Ashik Ullah and Dr. Vivian Zhang as described for Chapter 5. Routine reagents used to perform experiments were prepared by the Laboratory for Respiratory Mucosal Immunity at the University of Queensland.

Chapter 7: The contribution of Zbtb10 to allergen-induced airway inflammation in mice Contributions by Dr. Ashik Ullah, Dr. Vivian Zhang and Dr. Jason Lynch, as described for Chapter 5. Routine reagents used to perform experiments were prepared by the Laboratory for Respiratory Mucosal Immunity at the University of Queensland.

Chapter 8: Conclusions and future directions No contributions by others.

Statement of parts of the thesis submitted to qualify for the award of another degree

None.

viii Research Involving Human or Animal Subjects

All animal experiments were approved and performed in accordance with the Animal Care and Ethics Committees of the University of Queensland (Brisbane, Australia) under the following ethics approvals:

SBMS/350/13/NHMRC Role of different PRRs and its ligands in the development of different endotype of asthma

SBMS/209/13/NHMRC Pneumovirus infections in early life and development of allergic asthma

SBMS/393/16/NHMRC Role of different PRRs and its ligands in the development of different endotype of asthma

SBMS/194/16/NHMRC Pneumovirus infections in early life and development of allergic asthma

Copies of these ethics approval letters are included in the Appendix.

The experiments performed with lymphoblastoid cell lines were approved by the Ethics Committee of QIMR Berghofer (project P710).

ix Acknowledgements

First and foremost, my deep gratitude to my supervisor Manuel Ferreira, who trusted me with this project and gave me the opportunity to come to Australia. Your advice and encouragement throughout these four years have been invaluable and it was a great pleasure to have been your student.

A special thank you to Dr. Juliet French, Dr. Stacey Edwards and A/Prof. Simon Phipps who co-supervised and provided critical feedback on this project, while also welcoming me into their groups, to learn most of the experimental techniques described in this thesis. This thank you note is extended to their respective group members especially, Kristine Hillman and Susanne Kaufmann for their unwearying help during my first year of PhD; and also, Dr. Ashik Ullah and Dr. Vivian Zhang who have introduced me to the world of animal research and helped me countless times during the endless take-down days. Also, to Dr. Jason Lee who dedicated his time to help me with the ChIP and Western blot assays.

To my fellow group members Lisa Bain and Joana Revez, thank you for the support, friendship and regular coffee and lunch breaks throughout the years.

On a more personal note, I extend my gratitude to my friends, old and new, for all the fun times and adventures. You have all made this PhD journey and my time in Brisbane so much more pleasurable.

To my brother António and my cousin Carolina, thank you for your encouragement and support from afar. I have missed you dearly during these four years abroad.

Finally, to my parents who provided me with every opportunity. I wouldn’t be where I am today without you.

x Financial support

This research was supported by an Equity Trustees Queensland Medical Research PhD Scholarship – QPhD2015025.

xi Keywords asthma, gwas, genetics, genes, allergy, bronchiolitis, inflammation, pag1, zbtb10, 8q21.

Australian and New Zealand Standard Research Classifications (ANZSRC)

ANZSRC code: 060406, Genetic Immunology, 40% ANZSRC code: 060407, Genome Structure and Regulation, 20% ANZSRC code: 110311, Medical Genetics, 40%

Fields of Research (FoR) Classification

FoR code: 0604, Genetics, 60% FoR code: 1103, Clinical Sciences, 40%

xii Table of Contents

Abstract ...... ii

List of Figures and Tables ...... xix

List of Abbreviations ...... xxii

CHAPTER 1: Introduction ...... 26

CHAPTER 2: Identification of PAG1 and ZBTB10 as target genes of 8q21 asthma risk variants ...... 94

2.1 INTRODUCTION ...... 95

2.1.1 Background ...... 95

2.1.2 Hypothesis ...... 95

2.1.3 Aims ...... 96

2.2 METHODS ...... 97

2.2.1 Defining the core region of association ...... 97

2.2.2 Identifying genes with expression levels associated with rs7009110 ...... 97

2.2.3 Identifying Putative Regulatory Elements (PREs) in the core region of association ...... 98

2.2.4 Lymphoblastoid Cell Culture ...... 99

2.2.5 Identifying physical interactions between the core region of association and nearby gene promoters using Chromosome Conformation Capture (3C) ...... 100

2.2.6 Identifying candidate causal variants and their impact on PRE function using luciferase reporter assays ...... 103

2.2.7 Identifying transcription factors (TFs) that bind to interacting PREs and determine the impact of the candidate causal variants ...... 106

2.3 RESULTS ...... 108

2.3.1 The 8q21 core region of association spans 69 kb and harbours 118 variants in LD with the index SNP rs7009110 ...... 108

2.3.2 PAG1 expression is significantly associated with 8q21 allergy risk variants 110

2.3.3 There are four putative regulatory elements (PREs) in the core region of association ...... 115

xiii 2.3.4 PRE2 and PRE3 interact with the promoter of PAG1 in lymphoblastoid cell lines (LCLs) ...... 116

2.3.5 PRE3 harbouring the rs2370615 risk allele acts as an enhancer on the PAG1 promoter ...... 119

2.3.6 The rs2370615:C asthma risk allele disrupts the binding of Foxo3a to PRE3 ...... 121

2.3.7 FOXO3A expression is correlated with PAG1 expression in LCLs ...... 123

2.3.8 PRE2 and PRE3 also interact with ZBTB10 in LCLs ...... 124

2.3.9 PRE2 enhancer activity on the ZBTB10 promoter is disrupted by the rs11783496 risk allele ...... 126

2.4 DISCUSSION ...... 128

CHAPTER 3: Mechanisms of regulation of PAG1 and ZBTB10 expression ...... 134

3.1 INTRODUCTION ...... 135

3.1.1 Background ...... 135

3.1.2 Hypothesis ...... 135

3.1.3 Aims ...... 135

3.2 METHODS ...... 136

3.2.1 Assessing the contribution of NF-κB signalling to PAG1 expression ...... 136

3.2.1.1 TNF-α and IFN-α as NF-κB inducers ...... 136

3.2.1.2 Hypoxia as the NF-κB inducer ...... 139

3.2.2 Assessing the effect of allergen exposure on PAG1 and ZBTB10 expression ...... 140

3.2.3 Assessing the effect of phenol on PAG1 and ZBTB10 expression ...... 140

3.2.4 Assessing the contribution of asthma risk SNPs to phenol-induced gene expression ...... 142

3.2.5 Assessing broad immune gene expression profiles in response to allergen and phenol exposure ...... 142

3.3 RESULTS ...... 144

3.3.1 Contribution of NF-κB signalling to PAG1 expression ...... 144

3.3.1.1 TNF-α and IFN-α as NF-κB inducers ...... 144

xiv 3.3.1.2 Hypoxia as the NF-κB inducer ...... 149

3.3.2 Effect of allergen exposure on PAG1 and ZBTB10 expression ...... 153

3.3.3 Effect of phenol on PAG1 and ZBTB10 expression ...... 155

3.3.4 Association between asthma risk SNPs and phenol-induced gene expression ...... 158

3.3.5 Expression profiles of immune-related genes after phenol exposure...... 159

3.4 DISCUSSION ...... 163

CHAPTER 4: PAG1 and ZBTB10 expression in human and mouse tissues ...... 170

4.1 INTRODUCTION ...... 171

4.1.1 Background ...... 171

4.1.2 Aims ...... 171

4.2 METHODS ...... 172

4.2.1 Characterisation of PAG1 and ZBTB10 expression in human and mouse tissues based on information from public databases ...... 172

4.2.2 Quantification of Pag1 and Zbtb10 gene expression in mouse tissues of interest ...... 173

4.2.3 Pag1 protein expression in mouse lung tissue ...... 174

4.3 RESULTS ...... 176

4.3.1 Assessing PAG1 and ZBTB10 expression in human and mouse tissues and cell types based on publicly available databases ...... 176

4.3.2 Quantification of Pag1 and Zbtb10 gene expression in mouse tissues of interest ...... 185

4.3.3 Expression of Pag1 protein in mouse lung tissue ...... 187

4.4 DISCUSSION ...... 188

CHAPTER 5: The contribution of Pag1 to allergen-induced airway inflammation in mice...... 190

5.1 INTRODUCTION ...... 191

5.1.1 Background ...... 191

5.1.2 Hypothesis ...... 191

xv 5.1.3 Aim ...... 191

5.2 METHODS ...... 192

5.2.1 Mouse strains ...... 192

5.2.2 Genotyping ...... 192

5.2.3 Sample collection ...... 193

5.2.4 Statistical analysis ...... 193

5.2.5 Validating knockout of gene expression in the Pag1-/- mouse strain ...... 193

5.2.6 Assessing the contribution of Pag1 to airway inflammation triggered in sensitised mice by acute exposure to a high dose (10μg) of allergen ...... 194

5.2.7 Assessing the contribution of Pag1 to airway inflammation triggered in naïve mice upon first exposure to allergen ...... 197

5.2.8 Assessing the contribution of Pag1 to granulocyte recruitment into the airways ...... 198

5.2.9 Validating findings from the model of acute experimental asthma (section 5.2.6) using an allergen batch from a different source ...... 199

5.2.10 Assessing the contribution of Pag1 to airway inflammation triggered by exposure to a low dose of allergen ...... 200

5.3 RESULTS ...... 202

5.3.1 Validating knockout of gene expression in the Pag1-/- mouse strain ...... 202

5.3.2 Experiment 1: assessing the contribution of Pag1 to airway inflammation triggered in sensitised mice by acute exposure to a high dose of allergen ...... 204

5.3.3 Experiment 2: assessing the contribution of Pag1 to airway inflammation triggered in naïve mice upon first exposure to allergen ...... 207

5.3.4 Experiment 3: assessing the contribution of Pag1 to granulocyte recruitment into the airways ...... 210

5.3.5 Experiment 4: Validating results from experiment 1 (sensitised mice re- challenged with a high dose of allergen) ...... 212

5.3.6 Experiment 5: assessing the contribution of Pag1 to airway inflammation triggered in sensitised mice by acute exposure to a low dose of allergen ...... 219

5.4 DISCUSSION ...... 223

xvi CHAPTER 6: The contribution of Pag1 to viral-induced airway inflammation in mice ...... 230

6.1 INTRODUCTION ...... 231

6.1.1 Background ...... 231

6.1.2 Hypothesis ...... 232

6.1.3 Aim ...... 232

6.2 METHODS ...... 233

6.2.1 Mouse strains ...... 233

6.2.2 Study design ...... 233

6.2.3 Sample collection ...... 233

6.2.4 Assessment of airway inflammation ...... 233

6.2.5 Cytokine measurement ...... 234

6.2.6 Histologic analysis of viral load and mucus production ...... 234

6.2.7 Statistical analysis ...... 235

6.3 RESULTS ...... 236

6.4 DISCUSSION ...... 242

CHAPTER 7: The contribution of Zbtb10 to allergen-induced airway inflammation ...... 244

7.1 INTRODUCTION ...... 245

7.1.1 Background ...... 245

7.1.2 Hypothesis ...... 246

7.1.3 Aims ...... 246

7.2 METHODS ...... 247

7.2.1 Mouse strains ...... 247

7.2.2 Genotyping ...... 247

7.2.3 Sample collection ...... 248

7.2.4 Quantification of Zbtb10 expression and immune cell numbers in the new Zbtb10het mouse strain ...... 248

xvii 7.2.5 Assessing the contribution of Zbtb10 to airway inflammation triggered by acute exposure to a high dose of allergen ...... 249

7.2.6 Measurement of Cytokines ...... 250

7.2.7 Statistical analysis ...... 250

7.3 RESULTS ...... 251

7.3.1 Quantification of Zbtb10 expression and immune cell numbers in the new Zbtb10het mouse strain ...... 251

7.3.2 Assessing the contribution of Zbtb10 to airway inflammation triggered by acute exposure to a high dose of allergen ...... 261

7.4 DISCUSSION ...... 268

CHAPTER 8: Conclusions and future directions ...... 272

8.1 Main findings ...... 273

8.2 Future directions ...... 274

References ...... 276

Appendix ...... 287

Vicente et al. original publication Solutions and reagents used in Chapter 3 List of genes included in the gene expression arrays used in Chapter 3 Zbtb10het mouse strain details Animal Ethics certificates

xviii List of Figures and Tables

Figure 2.1 Location of the PRE2 and PRE3 fragments used in the luciferase assay experiments ...... 105 Figure 2.2 Variants overlapped by PRE2 and PRE3 fragments used in luciferase assay experiments ...... 105 Figure 2.3 The 8q21 region of association ...... 109 Figure 2.4 Results from multivariate association analysis...... 113 Figure 2.5 rs7009110 genotype effect on PAG1 expression ...... 114 Figure 2.6 PRE location in the core region of association...... 115 Figure 2.7 Results from PAG1 3C experiments ...... 117 Figure 2.8 Independent allele-specific 3C biological replicates ...... 117 Figure 2.9 Independent PAG1 3C biological replicates ...... 118 Figure 2.10 PRE3 acts as a transcriptional enhancer on the promoter of PAG1 in the presence of the rs2370615 risk allele ...... 120 Figure 2.11 Binding of Foxo3a TF to PRE3 is disrupted by the rs2370615:C risk allele .. 122 Figure 2.12 rs2370615 genotype modulates the association between FOXO3A and PAG1 expression ...... 123 Figure 2.13 Results from ZBTB10 3C experiments ...... 125 Figure 2.14 Independent ZBTB10 3C biological replicates ...... 125 Figure 2.15 PRE2 transcriptional enhancer activity on the promoter of ZBTB10 is disrupted in the presence of the rs11783496 risk allele ...... 127 Figure 2.16 RelA binding overlapping PRE3 ...... 131 Figure 2.17 Proposed model of PAG1 regulation ...... 131 Figure 3.1 Nuclear RelA levels following...... 144 Figure 3.2 Nuclear levels of the five NF-κB subunits following TNF-α stimulation ...... 145 Figure 3.3 Western blot results for cytoplasmic PAG1 following TNF-α stimulation...... 146 Figure 3.4 Cell membrane PAG1 levels following TNF-α and IFN-α stimulation ...... 147 Figure 3.5 Gene expression results for PAG1 following TNF-α and IFN-α stimulation .... 148 Figure 3.6 PAG1 expression in LCLs under hypoxic conditions ...... 150 Figure 3.7 PAG1 expression in LCLs under severe hypoxic conditions ...... 152 Figure 3.8 PAG1 and ZBTB10 expression in a LCL exposed to HDM and LPS ...... 154 Figure 3.9 TLR4 signalling effect on PAG1 and ZBTB10 expression ...... 154 Figure 3.10 PAG1 and ZBTB10 expression following phenol exposure ...... 156 Figure 3.11 PAG1 and ZBTB10 expression following phenol exposure in mast cells ...... 157

xix Figure 3.12 rs2370615 genotype contribution to phenol-mediated gene expression ...... 158 Figure 3.13 Histogram of gene distribution based on expression ratio in LCLs following phenol exposure ...... 159 Figure 3.14 Expression ratio comparison in LCLs and mast cells following phenol exposure for the six genes differentially expressed ...... 160 Figure 3.15 Correlation analyses between ZBTB10 expression and the expression of genes up-regulated by phenol ...... 162 Figure 4.1 PAG1 expression profile in human tissues ...... 178 Figure 4.2 ZBTB10 expression profile in human tissues ...... 179 Figure 4.3 GAPDH expression profile in human tissues ...... 180 Figure 4.4 PAG1, ZBTB10 and GAPDH expression profiles in human immune cells ...... 181 Figure 4.5 Pag1, Zbtb10 and Gapdh expression profiles in mouse tissues ...... 182 Figure 4.6 Zbtb10 expression profile in mouse tissues and immune cells ...... 183 Figure 4.7 Pag1 and Zbtb10 expression profiles in mouse immune cells ...... 184 Figure 4.8 Pag1 and Zbtb10 expression in mouse tissues of interest ...... 185 Figure 4.9 Pag1 and Zbtb10 expression in mouse airway epithelial cells ...... 186 Figure 4.10 Pag1 immunofluorescence staining in mouse lung tissue ...... 187 Figure 5.1 High dose allergen-induced airway inflammation experimental study design . 194 Figure 5.2 Allergen sensitization experimental study design ...... 197 Figure 5.3 LPS-induced airway inflammation experimental study design ...... 198 Figure 5.4 Low dose allergen-induced airway inflammation experimental study design .. 200 Figure 5.5 Pag1-/- strain validation...... 203 Figure 5.6 Experiment 1: assessment of airway inflammation in previously sensitised mice re-exposed to a high dose of HDM allergen ...... 206 Figure 5.7 Experiment 2: assessment of airway inflammation in naïve mice exposed to HDM allergen for the first time ...... 209 Figure 5.8 Experiment 3: assessment of LPS-induced airway inflammation...... 211 Figure 5.9 Airway inflammation assessment and comparison between different batches of HDM ...... 213 Figure 5.10 Experiment 4: validation of results from experiment 1 ...... 218 Figure 5.11 Experiment 5: assessment of airway inflammation triggered by re-exposure to a low dose of HDM ...... 222 Figure 6.1 Allergen-induced acute asthma model study design ...... 233 Figure 6.2 Assessment of viral clearance and innate immune response to PVM infection ...... 237

xx Figure 6.3 Assessment of dendritic cell recruitment in response to PVM infection ...... 238 Figure 6.4 Assessment of granulocyte recruitment in response to PVM infection ...... 241 Figure 7.1 Zbtb10het strain validation ...... 251 Figure 7.2 Immune cell population screen in relevant tissues ...... 260 Figure 7.3 Assessment of airway inflammation in a HDM-induced model of acute experimental asthma ...... 263 Figure 7.4 Assessment of the contribution of Zbtb10 to adaptive cytokine production ex vivo ...... 265 Figure 7.5 Assessment of the contribution of Zbtb10 to innate cytokine production ex vivo ...... 267

Table 2.1 Primers and coordinates of fragments tested in 3C experiments ...... 102 Table 2.2 Primers used to generate luciferase assay constructs ...... 104 Table 2.3 Primers used for ChIP experiments ...... 107 Table 2.4 List of SNPs located within 1 Mb of rs7009110 and associated with the expression of nearby genes in published GWAS of gene expression...... 112 Table 5.1 List of antibodies used for flow cytometry ...... 195 Table 5.2 Histological scoring method for epithelial airway mucus secreting cell analysis ...... 201 Table 7.1 Summary of genotype results from Zbtb10het crossings ...... 253 Table 7.2 Gender breakdown of mice born from Zbtb10het crossings ...... 253 Table 7.3 Summary of genotype results from WT x Zbtb10het crossings ...... 255 Table 7.4 Gender breakdown of mice born from WT x Zbtb10het crossings ...... 255

xxi List of Abbreviations

3C Chromosome conformation capture A.U. Allergy units AEC Airway epithelial cell AGRF Australian genome research facility APC Antigen presenting cell ASM Airway smooth muscle ATP Adenosine triphosphate BAC Bacterial artificial chromosomes BALF Bronchoalveolar lavage fluid BCR B cell receptor bp Base pairs BSA Bovine serum albumin cDC Conventional dendritic cell ChIP Chromatin immunoprecipitation Chr Chromosome CI Confidence interval Csk C-terminal Src kinase CXCL Chemokine (C-X-C motif) ligand DC Dendritic cell DMEM Dulbecco’s modified eagle medium DMSO Dimethyl sulfoxide DNA Deoxyribonucleic acid dNTPs Deoxyribonucleotide triphosphates dpi Days post infection DTT Dithiothreitol e.g. exempli gratia EBV Epstein–Barr virus EDTA Ethylene diamine tetra acetate EGTA Ethylene glycol-bis(β-aminoethyl ether)-N,N,N',N'-tetraacetic acid eQTL Expression quantitative trait loci FACS Fluorescent-activated cell sorting FANTOM5 Functional annotation of the mammalian genome project FBS Fetal bovine serum gDNA Genomic DNA GTEx Genotype-tissue expression project GWAS Genome-wide association study h Hour HDM House dust mite HET Heterozygous HMGB1 High mobility group box 1 protein HRP Horseradish peroxidase HSC Hematopoietic stem cell i.e. id est

xxii i.n. Intranasal i.p. Intraperitoneal ID Identifier IFN Interferon Ig Immunoglobulin IHC Immunohistochemistry IL Interleukin ILC Innate lymphoid cell JAK Janus-activated kinase KO Knockout LCL Lymphoblastoid cell line LD Linkage disequilibrium log Logarithmic LPS Lipopolysaccharide LRTI Lower respiratory tract infection MAF Minor allele frequency Mb Mega base MHC Major histocompatibility complex Min Minutes miRNA Micro RNA mL Millilitre MLN Mediastinal lymph nodes mM Milimolar MQ Milli-Q mRNA Messenger RNA ng Nanogram NK Natural killer ºC Degrees Celsius OR Odds ratio OVA Ovalbumin PAMP Pathogen-associated molecular patterns PAS Periodic-acid Schiff PBMC Peripheral blood mononuclear cell PBS Phosphate-buffered saline PCR Polymerase chain reaction pDC Plasmacytoid dendritic cell PFU Plaque forming units pg Picogram PMID PubMed identifier PRE Putative regulatory element PVM Pneumovirus of mice QC Quality control RBC Red blood cell RIPA Radio-immunoprecipitation assay RNA Ribonucleic acid ROS Reactive oxygen species

xxiii rpm Rotations per minute RSV Respiratory syncytial virus RT Room temperature SCZ Schizophrenia SD Standard deviation SDS Sodium dodecyl sulfate SDS-PAGE Sodium dodecyl sulfate polyacrylamide gel electrophoresis SEM Standard error of mean SFK Src family kinases SNP Single nucleotide polymorphism SPF Specific pathogen free STAT Signal transducer and activator of transcription TCR T cell receptor TCZ Tocilizumab TF Transcription factor Th T helper cell TLR Toll-like receptor TNF Tumor necrosis factor TPM Transcripts per million Treg Regulatory T cell UNG Uracil-DNA glycosylase UQ The University of Queensland vs. Versus WBC White blood cell WT Wild-type μg Microgram μL Microlitre μM Micromolar

xxiv

“It always seems impossible, until it is done.”

- Nelson Mandela

xxv

1

Introduction

Vicente CT, Revez JA, Ferreira MAR. “Summary and interpretation of findings from the first ten years of genome-wide association studies in asthma”. Clinical and Translational Immunology, 2017 (Manuscript under review)

Summary and interpretation of findings from the first ten years of genome-wide association studies of asthma

Cristina T Vicente, Joana A Revez and Manuel A R Ferreira

QIMR Berghofer Medical Research Institute, Brisbane, Australia

Corresponding author:

Manuel A R Ferreira, PhD

Asthma Genetics lab

QIMR Berghofer Medical Research Institute

Locked Bag 2000,

Royal Brisbane Hospital,

Herston QLD 4029

Australia

Phone: +61 7 3845 3552

Fax: +61 7 3362 0101

Email: [email protected]

Functional characterization of the new 8q21 Asthma risk locus | 27

ABSTRACT

Twenty-five genome-wide association studies (GWAS) of asthma were published between

2007 and 2016, the largest with a sample size of 157,242 individuals. Across these studies,

39 genetic variants in low linkage disequilibrium (LD) with each other were reported to associate with disease risk at a significance threshold of P<5x10-8, including 31 in populations of European ancestry. Results from analyses of the UK Biobank data (n=380,503) indicate that at least 28 of the 31 associations reported in Europeans represent true-positive findings, collectively explaining 2.5% of the variation in disease liability (median of 0.06% per variant).

We identified 49 transcripts as likely target genes of the published asthma risk variants, mostly based on LD with expression quantitative trait loci (eQTL). Of these genes, 16 were previously implicated in disease pathophysiology by functional studies, including TSLP,

TNFSF4, ADORA1, CHIT1 and USF1. In contrast, at present, there is limited or no functional evidence directly implicating the remaining 33 likely target genes in asthma pathophysiology.

Some of these genes have a known function that is relevant to allergic disease, including

F11R, CD247, PGAP3, AAGAB, CAMK4, and PEX14, and so could be prioritized for functional follow-up. We conclude by highlighting three areas of research that are essential to help translate GWAS findings into clinical research or practice, namely validation of target gene predictions, understanding target gene function and their role in disease pathophysiology and genomics-guided prioritization of targets for drug development.

Functional characterization of the new 8q21 Asthma risk locus | 28

INTRODUCTION

Asthma is a common and chronic inflammatory disease of the airways, specifically affecting the bronchi and bronchioli. Bronchial inflammation, which results in airway narrowing and shortness of breath symptoms, is generally caused by innate and adaptive immune responses to inhaled viruses and/or allergens 1. These and other environmental exposures are strong risk factors for disease onset and exacerbations but, based on twin studies, account for less than half of the overall disease liability 2-4. The remaining disease risk is largely explained by inherited genetic factors, with gene-by-environment interaction effects also thought to play a role. Given this high heritability, there has been a long-standing interest in identifying specific genetic risk factors for asthma, initially through linkage analysis (1989 [ref. 5] to 2010 [ref. 6]) and subsequently through candidate-gene association studies (since 1995 [ref. 7]). Linkage studies were largely (if not entirely) unsuccessful because this approach is only adequately powered with realistic sample sizes to identify very large genetic effects 8, which we now know do not exist for asthma. On the other hand, most published candidate-gene studies suffered from a number of methodological limitations (e.g. small number of samples and genetic markers tested) 9, and so the reported candidate gene associations have been largely discounted.

In 2004, it became feasible to genotype hundreds of thousands of genetic variants in a single experiment 10. The development of genotyping arrays enabled the design of genome- wide association studies (GWAS), the first published soon after in 2005 (ref. 11). In the years that followed, not only the cost of genotyping arrays decreased substantially, but also methods for the analysis of GWAS data were developed and refined; notably, these included statistical approaches to account for population structure and to infer individual genotypes for genetic variants not present in the genotyping arrays 12. As a result, GWAS including data from thousands of individuals genotyped for millions of genetic variants became a reality, at last providing a powerful tool to identify genetic associations with disease risk. For asthma, the first GWAS was published in 2007 by Moffatt et al. 13. Since then, and until the end of 2016, 24 additional GWAS of asthma were published. In this review, we summarise and interpret the key genetic findings from these studies, specifically addressing the following questions: are any published risk variants likely to be false-positive associations? How much

Functional characterization of the new 8q21 Asthma risk locus | 29 variation in disease liability do they explain? What are the likely target genes of those risk variants? Do those genes point to potential new mechanisms underlying disease pathophysiology? Lastly, we conclude by highlighting areas of research that are essential to help translate genetic findings into clinical research or practice.

Summary of genetic associations reported in asthma GWAS performed between 2007 and 2016

We searched the NHGRI-EBI catalog of published GWAS 14 to identify studies that tested the association between genetic variants and asthma risk, between 2007 and 2016. We used the search term “Asthma” and applied no filters. The search was performed on the 2nd of August 2017, and returned 73 unique studies, which were individually reviewed for inclusion in our analysis. Of these, 48 (66%) were excluded (Supplementary Table 1), most (34 studies) because the phenotype tested in the GWAS was not asthma, but instead an asthma-related trait (e.g. lung function). Studies were also commonly excluded because they were based on DNA pooling (five studies) or reported genetic interactions (e.g. gene-by-environment; four studies) rather than main effects. We extracted data from the GWAS Catalog for the remaining 25 studies 6, 13, 15-37, which were included for analysis (Supplementary Table 2).

The definition of asthma was not always the same across the 25 published GWAS. For example, six GWAS ascertained asthma cases with disease onset in childhood, whereas in other GWAS cases had more severe symptoms or co-morbid allergies (Supplementary Table 2). The smallest GWAS included 66 cases and 42 controls 31, and the largest 28,399 cases and 128,843 controls 15. For most studies (18 of 25), the primary GWAS included exclusively individuals of European descent; two studies were based on populations of Asian ancestry 27, 29, two of Latino ancestry 18, 36, two of African ancestry 16, 35 and one included multiple ancestries 28. Across these 25 studies, 73 unique genetic variants were reported to associate with disease risk at a genome-wide significance threshold of P<5x10-8 (listed per study in Supplementary Table 3). Some single nucleotide polymorphisms (SNPs) were located in close proximity, and so we used the --clump procedure in PLINK 38 to determine which were likely to represent independent associations and which were simply correlated with previously published variants. Specifically, we assigned each SNP into groups of Functional characterization of the new 8q21 Asthma risk locus | 30 correlated variants based on pairwise linkage disequilibrium (LD). LD was estimated using data from the 1000 Genomes Project (release 20130502_v5a), separately for individuals of European, Asian and African ancestry, as appropriate. Using a conservative LD threshold of r2>0.05, there were 31 groups of correlated risk variants reported in GWAS of European ancestry (Table 1 and Supplementary Table 4), with seven, one and three variants reported in GWAS of Asian, African and Latino ancestry, respectively (Supplementary Table 5). One variant reported in Asians (rs1837253) and two (rs9272346 and rs907092) in Latinos were in strong LD (r2>0.8) with risk variants reported in Europeans, and so do not represent independent associations. Therefore, in total, 39 genetic variants (31 in Europeans and 8 additional in other ancestries) in low LD with each other were reported to associate with asthma risk in GWAS published between 2007 and 2016. In the sections below, we focus on the 31 associations reported in Europeans.

Are any of the published risk variants for asthma likely to represent false-positive associations?

GWAS, even when conducted using strict quality control procedures 39 and an appropriate genome-wide significance threshold 40, are not completely protected from false-positive associations. That is, there is always a small chance that a new genome-wide significant association with asthma risk might not be a true-positive association and instead arise by chance or because of unaccounted methodological biases. As such, whenever possible, it is important to perform an independent and adequately powered replication study to confirm novel associations. This is not always feasible, particularly as GWAS become larger, because there might not be sufficiently large studies available for replication that were not included in the discovery stage.

In the last few years, ~500,000 individuals from the UK have been deeply phenotyped and genotyped as part of the UK Biobank study 41. The genotype data for the full dataset has just been made publicly available 42, providing a unique and timely opportunity to test if the 31 associations with asthma risk reported in Europeans between 2007 and 2016 are reproducible. Briefly, we analysed data for 380,503 unrelated individuals (kinship coefficient indicating <3rd degree relatedness) with (a) European ancestry, confirmed based on analysis Functional characterization of the new 8q21 Asthma risk locus | 31 of allele sharing with individuals from the 1000 Genomes Project; and (b) non-missing information for field 6152 of the touchscreen questionnaire: “Has a doctor ever told you that you have had any of the following conditions?”. A total of 44,003 individuals selected “Asthma” when answering that question, and so were considered as cases. On the other hand, 336,500 individuals did not select “Asthma” and so were considered as controls. Mean age was 56.7 (range 38 to 72), with 54% of participants being female. We tested the association between individual SNPs and case-control status using SNPTEST, including age, sex and SNP chip as fixed effects in the model. We adjusted the association results for an LD Score intercept 43 of 1.073, estimated using 1.2 million HapMap3 SNPs. In this analysis, we were able to test all 31 reported SNPs (all with imputation information >0.98), either directly (30 SNPs) or through a proxy SNP (rs166079 instead of rs200634877, r2=0.75). Of these, 28 (90%) had a statistically significant (P<0.05/31 SNPs = 0.0016) and directionally consistent (same predisposing allele as originally reported) association with disease risk (Table 2 and Figure 1), thereby confirming the original findings as true-positive associations.

For the remaining three SNPs, results from this UK Biobank analysis of self-reported doctor-diagnosed asthma did not support an association with disease risk, and so it is possible that they represent false-positive associations. These associations are located in/near DENND1B/CRB1 34, PDE4D 37 and CDHR3 (ref. 20). Another explanation for the lack of association with these three SNPs is that the case-control definition we used in the UK Biobank analysis is not a good proxy for that used in the original studies. For example, the original association with rs6967330 in the CDHR3 gene 20, which was subsequently supported by results from Pickrell et al. 15 (rs6959584, P=2x10-8, r2=0.72 with rs6967330), was found when studying asthma cases with childhood onset and severe exacerbations. If such an association is specific to that subgroup of asthmatics, and if these only represent a small fraction of the UK Biobank asthma cases, then the power to replicate the original association might have been low. In this respect, it is noteworthy that the direction of effect in the UK Biobank for rs6967330 was the same as originally reported. In conclusion, at least 28 of the 31 SNPs previously reported in GWAS of European ancestry have a significant and consistent association with self-reported doctor-diagnosed asthma in the UK Biobank study and so represent bona fide asthma risk variants. . It was beyond the scope of this review to report associations found in the UK Biobank study that were not located in previously reported

Functional characterization of the new 8q21 Asthma risk locus | 32 asthma risk loci. Studies that report the full results from the UK Biobank study will be reported elsewhere in the near future.

How much variation in disease liability is explained by the published asthma risk variants and by others yet to be discovered?

Twin studies have estimated the heritability of asthma to be between 55% and 74% in adults 2, 3, with even larger estimates reported in young children 4, 44. The aim of GWAS is to identify variants that contribute to this heritability. So it is important to understand to what extent the heritability of asthma is explained by the asthma risk variants discovered to date, and how much heritability remains to be discovered. To answer the first question, we estimated the total variance in disease liability explained in the UK Biobank study by each of the 31 published risk variants, using the formula var(g) / (var(g) * (π2)/3) described by Pawitan et al. 45. This is often referred to as the SNP heritability. In this formula, var(g) for each SNP is given by 2p*(1-p)*(log(OR))2, where p and OR are respectively the frequency and odds ratio for the effect allele, while π is the mathematical constant pi. Using this formula, we found that the median SNP heritability was 0.06% (range 0% to 0.41%; Table 2), while the sum of the 31 SNP heritabilities was 2.5%. That is, in the UK Biobank study, 2.5% of the variation in asthma liability is explained by the 31 asthma risk variants discovered to date.

The second question of interest is how much heritability is likely to be explained by risk variants that remain to be discovered. One approach to address this question might be to simply subtract the SNP heritability explained by the 31 published associations (2.5%) from the overall asthma heritability estimates reported in twin studies (e.g. 55%). Therefore, potentially, asthma risk variants yet to be discovered could account for at least ~52% (55% - 2.5%) of disease liability. However, heritability estimates from twin studies can be inflated, for example, because of violations of study design assumptions 46. Thus, such estimates should be taken as the upper boundary of the total variation in disease liability that is explained by genetic variants.

A more conservative approach to address the same question involves first estimating the disease heritability that is explained collectively by all genetic variants studied in a GWAS,

Functional characterization of the new 8q21 Asthma risk locus | 33 not just those with a strong association with disease risk. This is referred to as the SNP-based disease heritability, which can be estimated using for example GCTA 47, BOLT-REML 48 or LD Score regression 43. In theory, the SNP-based heritability can be lower than the twin-based heritability if genetic variants not tested (or not well tagged) in GWAS contribute to disease risk, which could be the case for uncommon variants (e.g. with minor allele frequency [MAF] < 1%). Differences in heritability estimates could also arise if the twin-based heritability estimate is inflated, as discussed above. When we applied the LD Score regression approach to genome-wide results from the UK Biobank GWAS analysis described above (n=380,503), we found that the overall SNP-based heritability for asthma was 14%, with a standard error (SE) of 1%. This estimate was obtained based on results for 1.2 million common, well imputed HapMap3 SNPs; therefore, it can be considered as the lower boundary of the total variation in disease liability that is explained by common SNPs. If we consider this estimate, then genetic risk variants yet to be discovered, and that could be identified in larger asthma GWAS of common variants, are likely to account for about 11.5% (14% - 2.5%) of the variation in disease liability. This conclusion was supported by the observation that the overall asthma SNP-based heritability obtained after removing from the UK Biobank GWAS the 31 published SNPs (and all variants in LD with them, r2>0.05) was 12% (SE = 1%).

Have we found fewer asthma risk variants than expected based on GWAS sample size?

The largest asthma GWAS published between 2007 and 2016 included 157,242 individuals and identified 27 independent associations with disease risk. This figure is smaller than reported by GWAS of some complex diseases using similar or smaller sample sizes (Table 3). For example, with a similar sample size (n=150,064), Ripke et al. 49 found 128 independent associations with schizophrenia (SCZ), which is 4.7-fold greater than those found for asthma. What underlies this difference in GWAS yield? First, the power to detect an association with a SNP depends on the proportion of variance in disease liability it explains. As discussed above, this can be calculated from the risk allele frequency and the odds ratio. For example, power is about the same to detect an association with two SNPs, one with a risk allele of frequency 0.5 and odds ratio of 1.2, and the other with a risk allele of frequency 0.01 and odds ratio of 2.5 – both SNPs explain about 0.5% of variation in disease liability (see Functional characterization of the new 8q21 Asthma risk locus | 34 formula in section above). When comparing two diseases, one needs to consider that the power to detect an association with a SNP that explains the same proportion of variance in disease liability depends on the disease prevalence 50, 51. Using the formula derived by Yang et al. 51, if we consider a SNP that explains 0.05% of the variation in disease liability, then the expected non-centrality parameter (NCP; which reflects power, but is linearly related to sample size) for the SCZ GWAS listed in Table 3 is 102 (assuming a disease prevalence of 1%). This is 2.6-fold greater than obtained for the similar-sized asthma GWAS (NCP=40, assuming a prevalence of 15%). In other words, although the overall sample size is the same, the SCZ GWAS has substantially greater power to detect an association with such a SNP when compared to the asthma GWAS. Another consideration is the genetic architecture (number of risk loci, their frequency and effect size), which is unknown and may differ across different diseases 52. For example, asthma might have a smaller number of common risk variants than SCZ, or SNP effects might be smaller. So to adequately compare the number of associations reported for different diseases (and quantitative traits), one needs to consider not just the sample size but also disease lifetime risk and the likely genetic architecture, as highlighted previously 50, 53.

Another factor that influences power, and that could have a more severe effect for some diseases than others, is disease heterogeneity and misclassification. If the genetic architecture of asthma is not homogeneous, and instead has components that are specific to clinically distinct subtypes, then lumping all individuals who reported ever having asthma in the same case group could lead to underestimation of SNP heritabilities, which decreases power. As examples of this, the risk allele for susceptibility variants near ORMDL3 is significantly more common in cases with childhood onset asthma 6, 54, while the association near CDHR3 might be specific to children with early onset and severe exacerbations 20, as discussed above. Similarly, including in the control group individuals who do not have asthma but suffer from other common allergic diseases (e.g. hay fever) can significantly decrease power to detect associations with SNPs that affect allergies in general, not just asthma 55. Shared genetic effects are expected to be widespread amongst common allergic diseases, with pairwise genetic correlations estimated to be >50% in twin studies 2, 4, 56, 57. One approach to capitalise on this high genetic correlation is to define cases as those suffering from two or more allergic diseases, and controls as those who do not suffer from any allergic

Functional characterization of the new 8q21 Asthma risk locus | 35 disease 55. We have shown empirically that the estimated SNP effects are indeed larger with such approach, and so power is increased 19. Another approach that also increases power to detect shared genetic effects is to define cases as those who suffer from any allergic disease 55. We have recently performed a large GWAS using this approach (n=360,000) and identified 136 independent associations for allergic disease (Ferreira et al., under review). Of note, the expected NCP in this study design for a SNP with a 0.05% heritability and assuming a 30% disease prevalence was 123, which is comparable to the SCZ GWAS that identified 128 independent associations.

Lastly, if gene-by-environment interactions have a substantial contribution to asthma risk, then ignoring the relevant environmental exposures in GWAS could also result in underestimated SNP effects. For example, gene-by-environment interactions have been reported for variants in the ORMDL3 locus, with larger SNP effects found in children with rhinovirus wheezing illness in early life 58 or exposed to tobacco smoke in early life 59. The latter interaction, however, was not replicated in a large independent study 60. A small number of genome-wide interaction analyses between environmental exposures and asthma risk have been published 61-64, but are not discussed in this review. For reviews of this topic, see for example 65, 66. Thus, in summary, fewer genetic associations have been reported for asthma as compared to some complex diseases with GWAS of similar or smaller size. This likely reflects lower statistical power arising from the relatively high disease prevalence, but also study design limitations, such as not accounting for disease subtypes, information from genetically-correlated diseases and gene-by-environment interaction effects.

What are the likely target genes of the published asthma risk variants?

The aim of asthma GWAS per se is to identify genetic variants associated with disease risk. A genetic variant is associated with asthma risk most likely because that same variant, or another in strong LD with it (e.g. r2>0.8), affects the protein sequence or the transcription patterns of a gene (i.e. the 'target gene') that plays a role in disease pathophysiology. Therefore, knowing which variants are associated with disease risk might highlight specific genes and molecular pathways dysregulated in asthma, and ultimately help better understand why asthma develops in the first place. How do we identify the likely target genes of risk Functional characterization of the new 8q21 Asthma risk locus | 36 variants? First, we can determine if the associated variant, or another in strong LD with it, is a non-synonymous coding variant, using for example ANNOVAR 67. If we focus on the 28 published risk variants that had a reproducible association with asthma in the UK Biobank study (Table 2), and also include the additional correlated risk variants reported in other asthma GWAS (Supplementary Table 4), then this approach identifies eight likely target genes: GSDMA, GSDMB, HLA-DQA1, HLA-DQB1, IL1RL1, IL6R, TLR1 and ZPBP2 (Supplementary Table 6).

A second approach that can be used to identify likely target genes is to determine if the associated SNPs are in strong LD with variants that have been reported to associate with variation in gene expression levels, known as expression quantitative trait loci (eQTL). A plethora of eQTL studies have been reported in recent years, including 39 conducted using tissues relevant to asthma pathophysiology (Supplementary Table 7), for example, whole- blood, lung, skin and individual immune cell types, such as CD4+ T cells. For each of these studies, we extracted eQTL results (i.e. SNP, gene and P-value) from the original publication, keeping only associations in cis (i.e. SNP within 1 Mb of gene) that were significant at a conservative threshold of P<2.3x10-9, which corresponds to a Bonferroni correction for testing each of 21,472 genes 68 for association with 1,000 independent SNPs 69. In each study and for each gene, we then used the --clump procedure in PLINK to reduce the published list of eQTLs (which typically includes many correlated variants) to the subset of strongest eQTLs that were in low LD with each other (r2<0.05), which we refer to as sentinel eQTLs. Finally, we asked if any of the 28 variants with a reproducible association with asthma risk in the UK Biobank study (Table 2), or the additional correlated risk variants reported in other asthma GWAS (Supplementary Table 4), were in strong LD (r2>0.8) with a sentinel eQTL. Using this approach, we found 48 likely target genes (Table 4 and Supplementary Table 8). Of note, these include all eight genes with non-synonymous variants listed above, indicating that for these variation in both protein sequence and transcript levels might be important determinants of cellular function and disease risk.

For many asthma risk variants (12 of 28, or 43%), the two approaches described above failed to identify any likely target gene. This could arise, for example, if the effect of those variants on gene expression is (a) specific to tissues, cell types and/or cellular

Functional characterization of the new 8q21 Asthma risk locus | 37 conditions (e.g. hypoxia) for which eQTL information is not available at present; or (b) too weak to be detected with the sample sizes included in published eQTL studies and at the conservative significance level that we used. In this case, the likely target genes could potentially be identified using functional studies; these can include, for example, experiments to determine the effect of risk variants on promoter-enhancer chromatin interactions, promoter activity or transcription factor binding. We recently used such approaches to identify PAG1 as a likely target gene of the asthma risk variants located on chromosome 8q21 (ref. 70), for which no eQTL support was available at the strict significance threshold used above. In summary, at least 49 genes are likely targets of published asthma risk variants, most (90%) identified based on the LD between risk variants and eQTLs.

Do any of the likely target genes represent potential new players in the pathophysiology of asthma and allergic disease more generally?

To address this question, we performed a PubMed query using the HGNC-approved gene symbols listed in Table 4, as well as all known aliases (Supplementary Table 9), and the allergy-related terms “asthma OR rhinitis OR eczema OR atopic OR dermatitis OR allergy OR allergi* OR hayfever OR 'hay fever'”. We downloaded results from the PubMed query in XML format and then counted the number of unique articles in that file citing both the allergy- related terms and the gene name or aliases in the title, abstract or keyword fields. Based on results from this query, we classified the 49 genes into three groups (Table 5). Group one consisted of nine genes co-mentioned frequently (5 or more studies) with those allergy- related terms prior to 2007, the year the first GWAS of any allergy-related trait was published. The genes were TSLP, IL1RL1, TNFSF4, TLR1, HLA-DQB1, HLA-DQB2, HLA-DQA1, ADORA1 and TAP2. For these genes, GWAS findings did not provide the first clue for a key role in disease pathophysiology.

Group two consisted of 13 genes co-mentioned frequently with allergy-related terms since, but not before, 2007: ORMDL3, GSDMB, ZPBP2, IKZF3, GSDMA, IL6R, CHIT1, FCER1G, SLC22A5, WDR36, IL18RAP, HLA-DQA2 and NDFIP1. The first five are located in the same asthma risk locus, and were first suspected to contribute to asthma pathophysiology because of GWAS findings. Of the remaining eight genes, for five it can be argued that Functional characterization of the new 8q21 Asthma risk locus | 38 functional studies provided the first suggestion of a key role in asthma/allergies, namely IL6R 71, CHIT1 72, FCER1G 73, IL18RAP 74 and NDFIP1 (ref. 75). But that is unlikely to be the case for the other three genes, SLC22A5 (organic cation transporter involved in pulmonary absorption of asthma-related drugs 76, 77), WDR36 (nucleolar protein involved in processing of 18S rRNA 78) and HLA-DQA2 (HLA class II molecule expressed in epidermal Langerhans cells 79).

Lastly, group three was composed of 27 genes co-mentioned infrequently with allergy- related terms, both before and after 2007. To our knowledge, only two of these genes have been suggested to play a role in allergic disease through functional studies: USF1(ref. 80) and STARD3 (ref. 81). However, many have a known function (Supplementary Table 10) that is directly relevant to allergic disease pathophysiology, such as F11R (ref. 82, 83), MICB 84, CD247 (ref. 85, 86), PGAP3 (ref. 87), AAGAB 88, CAMK4 (ref. 89) and PEX14 (ref. 90). In summary, of the 49 likely target genes of published asthma risk variants, only 16 were previously implicated by functional studies in disease pathophysiology.

Are there examples of genetic findings subsequently translated into clinical research or practice?

Based on the Thomson Reuters CortellisTM Drug database, drugs against five of the 49 likely target genes of asthma risk variants are being considered for clinical development (Table 6). Six of these drugs are being (or have been) tested in clinical trials of asthma. To our knowledge, only one of these clinical studies was motivated directly by results from a GWAS, our clinical trial of tocilizumab (TCZ) in participants with mild to moderate asthma. In 2011, we reported the association between a variant in the IL6R gene (rs4129267) and asthma risk 26. A consistent association with this variant was later reported also for eczema 91, 92 and asthma severity 93. We and others noted that the disease protective allele (rs4129267:C) was strongly associated with decreased protein levels of the soluble form of the receptor (sIL-6R)94, but increased mRNA levels of the full length IL6R transcript 95-97, which encodes for the membrane bound form of the receptor (mIL-6R). That is, decreased asthma risk is associated with decreased sIL-6R but increased mIL-6R. By extension, decreased disease risk is likely

Functional characterization of the new 8q21 Asthma risk locus | 39 associated with decreased IL-6 trans-signaling (which requires sIL-6R and is mainly pro- inflammatory) but increased IL-6 classic signaling (which requires mIL-6R and is thought to be mainly regenerative and protective) 98. Based on these genetic findings, it was not immediately obvious what to expect from a drug such as TCZ (approved therapeutic for rheumatoid arthritis and other auto-immune diseases), which blocks both sIL-6R and mIL-6R. A small case study published in 2011 reported decreased clinical activity of atopic dermatitis in three patients treated with TCZ for up to 12 months 99. This was the first suggestion of a protective effect of TCZ in allergic conditions. To characterise the effect of TCZ in asthma, in 2013 we performed pre-clinical studies using mouse models of acute allergic asthma100. In these studies, we found that TCZ had a protective effect on allergen-induced airway inflammation only when the experimental model used resulted in increased levels of sIL-6R in the airways, and so that was likely to involve activation of the IL-6 trans-signaling pathway. When that was not the case, dual receptor blockade resulted in worse airway inflammation when compared to control mice. Based on the genetic findings and the mouse studies, in 2014 we initiated a clinical trial of tocilizumab in participants with mild asthma, specifically those with CT or TT genotype for rs4129267, as these have markedly increased levels of sIL- 6R levels 94. Results from this trial are expected to be published in 2018. On the other hand, if our prediction from the genetic studies and findings from the mouse studies are correct, then a drug that blocks sIL-6R but not mIL-6R might be more desirable. There is one such drug (sgp130Fc, also known as FE301 or olamkicept), which was shown to be safe in phase 1 clinical trials 101 and is now in phase 2 trials for Crohn's disease (Stefan Rose-John, personal communication). Future studies that test the safety and efficacy of this drug in asthma patients are warranted.

Opportunities and challenges

In this section, we highlight three broad research areas that should be addressed in the short- term to help translate findings from asthma GWAS into clinical research or practice.

Functional characterization of the new 8q21 Asthma risk locus | 40

Improving and confirming target gene predictions. As discussed above, eQTL information provides a valuable tool to identify a set of genes for which variation in SNP genotype for an asthma risk SNP is likely to directly cause variation in transcription levels. However, this approach has some caveats, of which we highlight two. First, SNP effects on gene expression can be tissue-specific (e.g. 102), context-dependent (e.g. 103) and/or relatively small. For these reasons, some genes might not be predicted to be target genes because appropriate (right tissue, context and sample size for that gene) eQTL studies have not been performed. For example, to our knowledge, the largest eQTL study conducted using airway epithelial cells was based on just 105 individuals 104 (compared to 5,311 for whole-blood 105), and so our understanding of the genetic control of gene expression in this relevant cell type is still very poor. We thus consider essential to continue expanding eQTL studies to include more relevant tissues, diverse cellular stimuli and larger sample sizes. The second caveat of this approach is that a genetic overlap between risk variants and eQTLs can often arise by chance, because eQTLs are widespread106. We therefore argue that eQTL information should be used to generate predictions that can be subsequently tested by functional studies. However, such studies are laborious and time consuming, often deployed one locus at a time (e.g. 70, 107, 108). As the number of known asthma risk variants increases in the near future, efficient validation of target gene predictions will require high-throughput approaches, such as capture Hi-C 109, multiplexed reporter assays 110 and genome editing 111. Ideally, these studies should be performed in relevant primary cell types, which at present is still technically challenging.

Understanding target gene function and their role in disease pathophysiology. As highlighted above, for many genes it is unclear how variation in gene expression or protein sequence ultimately leads to variation in asthma risk. Therefore, for these, there is an opportunity to make new discoveries regarding the molecular mechanisms underlying the regulation of gene expression, and the impact that this might have on cellular function and disease pathophysiology. The best examples of this to date are ORMDL3 (ref. 112-118) and GSDMB 119, for which recent functional and animal studies have provided the first insights into their contribution to disease pathophysiology. Similarly, we have recently been interested in understanding how variation in PAG1 expression might contribute to asthma pathophysiology.

Functional characterization of the new 8q21 Asthma risk locus | 41

Our functional studies suggested that decreased PAG1 expression is associated with decreased disease risk, contrary to what was expected based on the widely accepted contribution of this transmembrane adaptor protein to the development of immune responses 70. We are currently testing this possibility using mouse models of experimental asthma. Various models have been described 120, including those that mimic mechanisms underlying acute allergic asthma 121, 122; chronic asthma 123; and viral-induced asthma 124 in humans. It is not always straightforward to decide which model is most appropriate to use, and different models can result in different findings 100, presumably because of the different underlying pathophysiology. In this respect, understanding the contribution of gene expression to cellular function might provide the best clue as to which models to begin with. An additional caveat of these models might be encountered when comparing germline knock-out mice against wild- type mice, because of possible genetic compensation mechanisms in the former, that is, the up-regulation of functionally-related genes 125. Potential solutions include temporally controlled gene deletion 126 or using therapeutic agents that block gene activity. Mouse models are also laborious and so are invariably applied on a gene-by-gene basis, which constitutes a significant bottleneck when dealing with the rapidly increasing number of asthma risk genes. Nonetheless, despite the potential limitations of these models, they will continue to represent essential tools to understand the contribution of target genes of risk variants to asthma pathophysiology.

Genomics-guided prioritization of targets for drug development. GWAS findings, in combination with eQTL information, can provide important clues into the predicted directional effect of gene expression on disease risk. This, in turn, can be used to inform drug repositioning or the development of novel drugs: specifically, if the disease protective allele of an asthma risk variant is associated with decreased (increased) gene expression, then the prediction is that a drug that also decreases (increases) gene expression should similarly have a protective effect on asthma symptoms. For example, the disease-protective allele rs1438673:T 19 on chromosome 5q22.1 is strongly associated with decreased expression of TSLP in skin127 (beta=-0.48, P=2x10-15) and whole-blood 103 (beta=-8.8, P=10-18). The same direction of effect on gene expression is observed in at least eight different tissues studied by the GTEx project 127, including adipose, colon and esophagus. Thus, based on these genetic

Functional characterization of the new 8q21 Asthma risk locus | 42 findings, we can postulate that TSLP antagonists (but not agonists) might improve asthma symptoms. This is consistent with the known contribution of TSLP to disease pathophysiology 128 and with results from a recent clinical trial of an anti-TSLP antibody 129. This is not always so straightforward, as the IL6R example discussed above illustrates. Other examples that are harder to interpret occur when the directional effect of a risk variant on gene expression is not the same, or is poorly characterised, in different relevant tissues. For example, the asthma protective allele rs6683383:A is strongly associated with increased expression of ADORA1 in whole-blood (e.g. beta=0.40, P=10-44 [ref. 130]), suggesting that agonists and not antagonists (such as PBF-680, which is currently being tested in a phase 2 trial in asthma: NCT02635945) might be beneficial in asthma. But it is possible that rs6683383:A has the opposite effect on ADORA1 expression in other tissues relevant to asthma, such as airway epithelial cells. This example illustrates the importance of characterising the genetic control of gene expression in multiple relevant tissues and cell types. In this regard, projects such as GTEx 127 but with a focus on asthma-related tissues and cell types could prove very useful. Tissues and cell types could be selected based on results from tissue-specific expression 131 or SNP heritability 132 enrichment analyses. For example, the latter suggests that SNPs associated with asthma are enriched amongst enhancers in Th17 cells 133; yet, to our knowledge, no eQTL study has been performed on this cell type.

Concluding remarks

In the last ten years, GWAS have identified the first reproducible associations between common SNPs and asthma risk. In Europeans, these SNPs account for ~2.5% of the variation in disease liability. The risk SNPs identified provide the starting point for functional studies that can help understand how genetic variation at these loci affects gene expression, cellular function and disease pathophysiology. The target genes of published risk SNPs are likely to include many previously unsuspected players in disease pathophysiology, and so might point to new therapeutic opportunities. Larger GWAS of broadly defined asthma have the potential to identify additional common risk variants that explain at least 12% of disease liability. This

Functional characterization of the new 8q21 Asthma risk locus | 43 could translate into many hundreds or thousands of risk variants, if we consider that the mean SNP heritability is likely to be low (e.g. 0.01% to 0.05%). Larger GWAS will also have increased power to detect individual associations with uncommon variants (e.g. MAF < 1%), and to quantify their overall contribution to disease liability, which has not been addressed to date. Other areas of research that remain largely unexplored in the field of asthma genetics include understanding if different inflammatory subtypes (e.g. neutrophilic and eosinophilic asthma) have distinct genetic components and whether disease remission is a heritable trait. The next decade of research might provide new insights into these important questions.

Functional characterization of the new 8q21 Asthma risk locus | 44

Acknowledgments

This research has been conducted using the UK Biobank Resource. CTV was supported by a PhD scholarship (QPhD2015025) from Equity Trustees (Australia); JAR was supported by a PhD scholarship (SFRH/BD/92907/2013) from Fundacao para a Ciencia e Tecnologia (Portugal); MARF was supported by a Senior Research Fellowship (APP1124501) from the National Health and Medical Research Council (Australia). We also thank Naomi Wray for providing comments on the original manuscript.

Functional characterization of the new 8q21 Asthma risk locus | 45

References

1. Murray CS, Simpson A, Custovic A. Allergens, viruses, and asthma exacerbations. Proc Am Thorac Soc 2004;1(2):99-104. 2. Duffy DL, Martin NG, Battistutta D, Hopper JL, Mathews JD. Genetics of asthma and hay fever in Australian twins. The American review of respiratory disease 1990;142(6 Pt 1):1351-8. 3. Thomsen SF, Ulrik CS, Kyvik KO, Ferreira MA, Backer V. Multivariate genetic analysis of atopy phenotypes in a selected sample of twins. Clin Exp Allergy 2006;36(11):1382-90. 4. van Beijsterveldt CE, Boomsma DI. Genetics of parentally reported asthma, eczema and rhinitis in 5-yr-old twins. Eur Respir J 2007;29(3):516-21. 5. Cookson WO, Sharp PA, Faux JA, Hopkin JM. Linkage between immunoglobulin E responses underlying asthma and rhinitis and chromosome 11q. Lancet 1989;1(8650):1292- 5. 6. Moffatt MF, Gut IG, Demenais F, Strachan DP, Bouzigon E, Heath S, et al. A large- scale, consortium-based genomewide association study of asthma. N Engl J Med 2010;363(13):1211-21. 7. Ohe M, Munakata M, Hizawa N, Itoh A, Doi I, Yamaguchi E, et al. Beta 2 adrenergic receptor gene restriction fragment length polymorphism and bronchial asthma. Thorax 1995;50(4):353-9. 8. Ferreira MA. Linkage analysis: principles and methods for the analysis of human quantitative traits. Twin Res 2004;7(5):513-30. 9. Hoffjan S, Nicolae D, Ober C. Association studies for asthma and atopic diseases: a comprehensive review of the literature. Respir Res 2003;4(1):14. 10. Matsuzaki H, Dong S, Loi H, Di X, Liu G, Hubbell E, et al. Genotyping over 100,000 SNPs on a pair of oligonucleotide arrays. Nat Methods 2004;1(2):109-11. 11. Klein RJ, Zeiss C, Chew EY, Tsai JY, Sackler RS, Haynes C, et al. Complement factor H polymorphism in age-related macular degeneration. Science 2005;308(5720):385-9. 12. Marchini J, Howie B, Myers S, McVean G, Donnelly P. A new multipoint method for genome-wide association studies by imputation of genotypes. Nat Genet 2007;39(7):906-13. 13. Moffatt MF, Kabesch M, Liang L, Dixon AL, Strachan D, Heath S, et al. Genetic variants regulating ORMDL3 expression contribute to the risk of childhood asthma. Nature 2007;448(7152):470-3. 14. MacArthur J, Bowler E, Cerezo M, Gil L, Hall P, Hastings E, et al. The new NHGRI-EBI Catalog of published genome-wide association studies (GWAS Catalog). Nucleic Acids Res 2017;45(D1):D896-D901. 15. Pickrell JK, Berisa T, Liu JZ, Segurel L, Tung JY, Hinds DA. Detection and interpretation of shared genetic influences on 42 human traits. Nat Genet 2016;48(7):709-17. 16. White MJ, Risse-Adams O, Goddard P, Contreras MG, Adams J, Hu D, et al. Novel genetic risk factors for asthma in African American children: Precision Medicine and the SAGE II Study. Immunogenetics 2016;68(6-7):391-400. 17. Barreto-Luis A, Pino-Yanes M, Corrales A, Campo P, Callero A, Acosta-Herrera M, et al. Genome-wide association study in Spanish identifies ADAM metallopeptidase with thrombospondin type 1 motif, 9 (ADAMTS9), as a novel asthma susceptibility gene. J Allergy Clin Immunol 2016;137(3):964-6.

Functional characterization of the new 8q21 Asthma risk locus | 46

18. Galanter JM, Gignoux CR, Torgerson DG, Roth LA, Eng C, Oh SS, et al. Genome- wide association study and admixture mapping identify different asthma-associated loci in Latinos: the Genes-environments & Admixture in Latino Americans study. J Allergy Clin Immunol 2014;134(2):295-305. 19. Ferreira MA, Matheson MC, Tang CS, Granell R, Ang W, Hui J, et al. Genome-wide association analysis identifies 11 risk variants associated with the asthma with hay fever phenotype. J Allergy Clin Immunol 2014;133(6):1564-71. 20. Bonnelykke K, Sleiman P, Nielsen K, Kreiner-Moller E, Mercader JM, Belgrave D, et al. A genome-wide association study identifies CDHR3 as a susceptibility locus for early childhood asthma with severe exacerbations. Nat Genet 2014;46(1):51-5. 21. Lasky-Su J, Himes BE, Raby BA, Klanderman BJ, Sylvia JS, Lange C, et al. HLA-DQ strikes again: genome-wide association study further confirms HLA-DQ in the diagnosis of asthma among adults. Clin Exp Allergy 2012;42(12):1724-33. 22. Ramasamy A, Kuokkanen M, Vedantam S, Gajdos ZK, Couto Alves A, Lyon HN, et al. Genome-wide association studies of asthma in population-based cohorts confirm known and suggested loci and identify an additional association near HLA. PLoS One 2012;7(9):e44008. 23. Li X, Ampleford EJ, Howard TD, Moore WC, Torgerson DG, Li H, et al. Genome-wide association studies of asthma indicate opposite immunopathogenesis direction from autoimmune diseases. J Allergy Clin Immunol 2012;130(4):861-8 e7. 24. Wan YI, Shrine NR, Soler Artigas M, Wain LV, Blakey JD, Moffatt MF, et al. Genome- wide association study to identify genetic determinants of severe asthma. Thorax 2012;67(9):762-8. 25. Karunas AS, Iunusbaev BB, Fedorova I, Gimalova GF, Ramazanova NN, Gur'eva LL, et al. [Genome-wide association study of bronchial asthma in the Volga-Ural region of Russia]. Mol Biol (Mosk) 2011;45(6):992-1003. 26. Ferreira MA, Matheson MC, Duffy DL, Marks GB, Hui J, Le Souef P, et al. Identification of IL6R and chromosome 11q13.5 as risk loci for asthma. Lancet 2011;378(9795):1006-14. 27. Hirota T, Takahashi A, Kubo M, Tsunoda T, Tomita K, Doi S, et al. Genome-wide association study identifies three new susceptibility loci for adult asthma in the Japanese population. Nat Genet 2011;43(9):893-6. 28. Torgerson DG, Ampleford EJ, Chiu GY, Gauderman WJ, Gignoux CR, Graves PE, et al. Meta-analysis of genome-wide association studies of asthma in ethnically diverse North American populations. Nat Genet 2011;43(9):887-92. 29. Noguchi E, Sakamoto H, Hirota T, Ochiai K, Imoto Y, Sakashita M, et al. Genome-wide association study identifies HLA-DP as a susceptibility gene for pediatric asthma in Asian populations. PLoS Genet 2011;7(7):e1002170. 30. Ferreira MA, McRae AF, Medland SE, Nyholt DR, Gordon SD, Wright MJ, et al. Association between ORMDL3, IL1RL1 and a deletion on chromosome 17q21 with asthma risk in Australia. Eur J Hum Genet. 31. DeWan AT, Triche EW, Xu X, Hsu LI, Zhao C, Belanger K, et al. PDE11A associations with asthma: results of a genome-wide association scan. J Allergy Clin Immunol 2010;126(4):871-3 e9. 32. Himes BE, Lasky-Su J, Wu AC, Wilk JB, Hunninghake GM, Klanderman B, et al. Asthma-susceptibility variants identified using probands in case-control and family-based analyses. BMC Med Genet 2010;11:122.

Functional characterization of the new 8q21 Asthma risk locus | 47

33. Li X, Howard TD, Zheng SL, Haselkorn T, Peters SP, Meyers DA, et al. Genome-wide association study of asthma identifies RAD50-IL13 and HLA-DR/DQ regions. J Allergy Clin Immunol;125(2):328-35 e11. 34. Sleiman PM, Flory J, Imielinski M, Bradfield JP, Annaiah K, Willis-Owen SA, et al. Variants of DENND1B associated with asthma in children. N Engl J Med 2010;362(1):36-44. 35. Mathias RA, Grant AV, Rafaels N, Hand T, Gao L, Vergara C, et al. A genome-wide association study on African-ancestry populations for asthma. J Allergy Clin Immunol 2010;125(2):336-46 e4. 36. Hancock DB, Romieu I, Shi M, Sienra-Monge JJ, Wu H, Chiu GY, et al. Genome-wide association study implicates chromosome 9q21.31 as a susceptibility locus for asthma in mexican children. PLoS Genet 2009;5(8):e1000623. 37. Himes BE, Hunninghake GM, Baurley JW, Rafaels NM, Sleiman P, Strachan DP, et al. Genome-wide association analysis identifies PDE4D as an asthma-susceptibility gene. Am J Hum Genet 2009;84(5):581-93. 38. Purcell S, Neale B, Todd-Brown K, Thomas L, Ferreira MA, Bender D, et al. PLINK: a tool set for whole-genome association and population-based linkage analyses. Am J Hum Genet 2007;81(3):559-75. 39. Winkler TW, Day FR, Croteau-Chonka DC, Wood AR, Locke AE, Magi R, et al. Quality control and conduct of genome-wide association meta-analyses. Nature protocols 2014;9(5):1192-212. 40. Fadista J, Manning AK, Florez JC, Groop L. The (in)famous GWAS P-value threshold revisited and updated for low-frequency variants. Eur J Hum Genet 2016;24(8):1202-5. 41. Allen NE, Sudlow C, Peakman T, Collins R, Biobank UK. UK biobank data: come and get it. Science translational medicine 2014;6(224):224ed4. 42. Bycroft C, Freeman C, Petkova D, Band G, Elliott LT, Sharp K, et al. Genome-wide genetic data on ~500,000 UK Biobank participants. bioRxiv 2017. 43. Bulik-Sullivan BK, Loh PR, Finucane HK, Ripke S, Yang J, Schizophrenia Working Group of the Psychiatric Genomics C, et al. LD Score regression distinguishes confounding from polygenicity in genome-wide association studies. Nat Genet 2015. 44. Ullemar V, Magnusson PK, Lundholm C, Zettergren A, Melen E, Lichtenstein P, et al. Heritability and confirmation of genetic association studies for childhood asthma in twins. Allergy 2016;71(2):230-8. 45. Pawitan Y, Seng KC, Magnusson PK. How many genetic variants remain to be discovered? PLoS One 2009;4(12):e7969. 46. Vinkhuyzen AA, Wray NR, Yang J, Goddard ME, Visscher PM. Estimation and partition of heritability in human populations using whole-genome analysis methods. Annu Rev Genet 2013;47:75-95. 47. Yang J, Lee SH, Goddard ME, Visscher PM. GCTA: a tool for genome-wide complex trait analysis. Am J Hum Genet 2011;88(1):76-82. 48. Loh PR, Bhatia G, Gusev A, Finucane HK, Bulik-Sullivan BK, Pollack SJ, et al. Contrasting genetic architectures of schizophrenia and other complex diseases using fast variance-components analysis. Nat Genet 2015;47(12):1385-92. 49. Schizophrenia Working Group of the Psychiatric Genomics C. Biological insights from 108 schizophrenia-associated genetic loci. Nature 2014;511(7510):421-7. 50. Wray NR, Pergadia ML, Blackwood DH, Penninx BW, Gordon SD, Nyholt DR, et al. Genome-wide association study of major depressive disorder: new results, meta-analysis, and lessons learned. Mol Psychiatry 2012;17(1):36-48. Functional characterization of the new 8q21 Asthma risk locus | 48

51. Yang J, Wray NR, Visscher PM. Comparing apples and oranges: equating the power of case-control and quantitative trait association studies. Genet Epidemiol 2010;34(3):254-7. 52. Shi H, Kichaev G, Pasaniuc B. Contrasting the Genetic Architecture of 30 Complex Traits from Summary Association Data. Am J Hum Genet 2016;99(1):139-53. 53. Levinson DF, Mostafavi S, Milaneschi Y, Rivera M, Ripke S, Wray NR, et al. Genetic studies of major depressive disorder: why are there no genome-wide association study findings and what can we do about it? Biol Psychiatry 2014;76(7):510-2. 54. Kreiner-Moller E, Strachan DP, Linneberg A, Husemoen LL, Bisgaard H, Bonnelykke K. 17q21 gene variation is not associated with asthma in adulthood. Allergy 2015;70(1):107- 14. 55. Ferreira MA. Improving the power to detect risk variants for allergic disease by defining case-control status based on both asthma and hay fever. Twin Res Hum Genet 2014;17(6):505-11. 56. Fagnani C, Annesi-Maesano I, Brescianini S, D'Ippolito C, Medda E, Nistico L, et al. Heritability and shared genetic effects of asthma and hay fever: an Italian study of young twins. Twin Res Hum Genet 2008;11(2):121-31. 57. Thomsen SF, Ulrik CS, Kyvik KO, Skadhauge LR, Steffensen I, Backer V. Findings on the atopic triad from a Danish twin registry. The international journal of tuberculosis and lung disease : the official journal of the International Union against Tuberculosis and Lung Disease 2006;10(11):1268-72. 58. Caliskan M, Bochkov YA, Kreiner-Moller E, Bonnelykke K, Stein MM, Du G, et al. Rhinovirus wheezing illness and genetic risk of childhood-onset asthma. N Engl J Med 2013;368(15):1398-407. 59. Bouzigon E, Corda E, Aschard H, Dizier MH, Boland A, Bousquet J, et al. Effect of 17q21 variants and smoking exposure in early-onset asthma. N Engl J Med 2008;359(19):1985-94. 60. Brauner EV, Loft S, Raaschou-Nielsen O, Vogel U, Andersen PS, Sorensen M. Effects of a 17q21 chromosome gene variant, tobacco smoke and furred pets on infant wheeze. Genes Immun 2012;13(1):94-7. 61. Gref A, Merid SK, Gruzieva O, Ballereau S, Becker A, Bellander T, et al. Genome- Wide Interaction Analysis of Air Pollution Exposure and Childhood Asthma with Functional Follow-up. Am J Respir Crit Care Med 2017;195(10):1373-83. 62. Du R, Litonjua AA, Tantisira KG, Lasky-Su J, Sunyaev SR, Klanderman BJ, et al. Genome-wide association study reveals class I MHC-restricted T cell-associated molecule gene (CRTAM) variants interact with vitamin D levels to affect asthma exacerbations. J Allergy Clin Immunol 2012;129(2):368-73, 73 e1-5. 63. Ege MJ, Strachan DP, Cookson WO, Moffatt MF, Gut I, Lathrop M, et al. Gene- environment interaction for childhood asthma and exposure to farming in Central Europe. J Allergy Clin Immunol 2011;127(1):138-44, 44 e1-4. 64. Vonk JM, Scholtens S, Postma DS, Moffatt MF, Jarvis D, Ramasamy A, et al. Adult onset asthma and interaction between genes and active tobacco smoking: The GABRIEL consortium. PLoS One 2017;12(3):e0172716. 65. Bonnelykke K, Ober C. Leveraging gene-environment interactions and endotypes for asthma gene discovery. J Allergy Clin Immunol 2016;137(3):667-79. 66. Kauffmann F, Demenais F. Gene-environment interactions in asthma and allergic diseases: challenges and perspectives. J Allergy Clin Immunol 2012;130(6):1229-40; quiz 41- 2. Functional characterization of the new 8q21 Asthma risk locus | 49

67. Yang H, Wang K. Genomic variant annotation and prioritization with ANNOVAR and wANNOVAR. Nature protocols 2015;10(10):1556-66. 68. Rebhan M, Chalifa-Caspi V, Prilusky J, Lancet D. GeneCards: integrating information about genes, proteins and diseases. Trends in genetics : TIG 1997;13(4):163. 69. Davis JR, Fresard L, Knowles DA, Pala M, Bustamante CD, Battle A, et al. An Efficient Multiple-Testing Adjustment for eQTL Studies that Accounts for Linkage Disequilibrium between Variants. Am J Hum Genet 2016;98(1):216-24. 70. Vicente CT, Edwards SL, Hillman KM, Kaufmann S, Mitchell H, Bain L, et al. Long- Range Modulation of PAG1 Expression by 8q21 Allergy Risk Variants. Am J Hum Genet 2015. 71. Doganci A, Eigenbrod T, Krug N, De Sanctis GT, Hausding M, Erpenbeck VJ, et al. The IL-6R alpha chain controls lung CD4+CD25+ Treg development and function during allergic airway inflammation in vivo. J Clin Invest 2005;115(2):313-25. 72. Seibold MA, Donnelly S, Solon M, Innes A, Woodruff PG, Boot RG, et al. Chitotriosidase is the primary active chitinase in the human lung and is modulated by genotype and smoking habit. J Allergy Clin Immunol 2008;122(5):944-50 e3. 73. Yu M, Tsai M, Tam SY, Jones C, Zehnder J, Galli SJ. Mast cells can promote the development of multiple features of chronic asthma in mice. J Clin Invest 2006;116(6):1633- 41. 74. Lewkowich IP, Rempel JD, HayGlass KT. Prevention of allergen-specific, Th2-biased immune responses in vivo: role of increased IL-12 and IL-18 responsiveness. J Immunol 2005;175(8):4956-62. 75. Oliver PM, Cao X, Worthen GS, Shi P, Briones N, MacLeod M, et al. Ndfip1 protein promotes the function of itch ubiquitin ligase to prevent T cell activation and T helper 2 cell- mediated inflammation. Immunity 2006;25(6):929-40. 76. Nakamura T, Nakanishi T, Haruta T, Shirasaka Y, Keogh JP, Tamai I. Transport of ipratropium, an anti-chronic obstructive pulmonary disease drug, is mediated by organic cation/carnitine transporters in human bronchial epithelial cells: implications for carrier- mediated pulmonary absorption. Mol Pharm 2010;7(1):187-95. 77. Mukherjee M, Cingolani E, Pritchard DI, Bosquillon C. Enhanced expression of Organic Cation Transporters in bronchial epithelial cell layers following insults associated with asthma - Impact on salbutamol transport. Eur J Pharm Sci 2017;106:62-70. 78. Skarie JM, Link BA. The primary open-angle glaucoma gene WDR36 functions in ribosomal RNA processing and interacts with the p53 stress-response pathway. Hum Mol Genet 2008;17(16):2474-85. 79. Lenormand C, Bausinger H, Gross F, Signorino-Gelo F, Koch S, Peressin M, et al. HLA-DQA2 and HLA-DQB2 genes are specifically expressed in human Langerhans cells and encode a new HLA class II molecule. J Immunol 2012;188(8):3903-11. 80. Takahashi K, Nishiyama C, Nishiyama M, Okumura K, Ra C, Ohtake Y, et al. A complex composed of USF1 and USF2 activates the human FcepsilonRI alpha chain expression via a CAGCTG element in the first intron. Eur J Immunol 2001;31(2):590-9. 81. Hannen RF, Michael AE, Jaulim A, Bhogal R, Burrin JM, Philpott MP. Steroid synthesis by primary human keratinocytes; implications for skin disease. Biochem Biophys Res Commun 2011;404(1):62-7. 82. Cera MR, Del Prete A, Vecchi A, Corada M, Martin-Padura I, Motoike T, et al. Increased DC trafficking to lymph nodes and contact hypersensitivity in junctional adhesion molecule-A-deficient mice. J Clin Invest 2004;114(5):729-38. Functional characterization of the new 8q21 Asthma risk locus | 50

83. Mitchell LA, Ward C, Kwon M, Mitchell PO, Quintero DA, Nusrat A, et al. Junctional adhesion molecule A promotes epithelial tight junction assembly to augment lung barrier function. Am J Pathol 2015;185(2):372-86. 84. Maasho K, Opoku-Anane J, Marusina AI, Coligan JE, Borrego F. NKG2D is a costimulatory receptor for human naive CD8+ T cells. J Immunol 2005;174(8):4480-4. 85. Lundholm M, Mayans S, Motta V, Lofgren-Burstrom A, Danska J, Holmberg D. Variation in the Cd3 zeta (Cd247) gene correlates with altered T cell activation and is associated with autoimmune diabetes. J Immunol 2010;184(10):5537-44. 86. Vales-Gomez M, Esteso G, Aydogmus C, Blazquez-Moreno A, Marin AV, Briones AC, et al. Natural killer cell hyporesponsiveness and impaired development in a CD247-deficient patient. J Allergy Clin Immunol 2016;137(3):942-5 e4. 87. Murakami H, Wang Y, Hasuwa H, Maeda Y, Kinoshita T, Murakami Y. Enhanced response of T lymphocytes from Pgap3 knockout mouse: Insight into roles of fatty acid remodeling of GPI anchored proteins. Biochem Biophys Res Commun 2012;417(4):1235-41. 88. Pohler E, Mamai O, Hirst J, Zamiri M, Horn H, Nomura T, et al. Haploinsufficiency for AAGAB causes clinically heterogeneous forms of punctate palmoplantar keratoderma. Nat Genet 2012;44(11):1272-6. 89. Koga T, Ichinose K, Mizui M, Crispin JC, Tsokos GC. Calcium/calmodulin-dependent protein kinase IV suppresses IL-2 production and regulatory T cell activity in lupus. J Immunol 2012;189(7):3490-6. 90. Vijayan V, Srinu T, Karnati S, Garikapati V, Linke M, Kamalyan L, et al. A New Immunomodulatory Role for Peroxisomes in Macrophages Activated by the TLR4 Ligand Lipopolysaccharide. J Immunol 2017;198(6):2414-25. 91. Esparza-Gordillo J, Schaarschmidt H, Liang L, Cookson W, Bauerfeind A, Lee-Kirsch MA, et al. A functional IL-6 receptor (IL6R) variant is a risk factor for persistent atopic dermatitis. J Allergy Clin Immunol 2013;132(2):371-7. 92. Paternoster L, Standl M, Waage J, Baurecht H, Hotze M, Strachan DP, et al. Multi- ancestry genome-wide association study of 21,000 cases and 95,000 controls identifies new risk loci for atopic dermatitis. Nat Genet 2015;47(12):1449-56. 93. Hawkins GA, Robinson MB, Hastie AT, Li X, Li H, Moore WC, et al. The IL6R variation Asp(358)Ala is a potential modifier of lung function in subjects with asthma. J Allergy Clin Immunol 2012;130(2):510-5 e1. 94. Melzer D, Perry JR, Hernandez D, Corsi AM, Stevens K, Rafferty I, et al. A genome- wide association study identifies protein quantitative trait loci (pQTLs). PLoS Genet 2008;4(5):e1000072. 95. van Dongen J, Jansen R, Smit D, Hottenga JJ, Mbarek H, Willemsen G, et al. The contribution of the functional IL6R polymorphism rs2228145, eQTLs and other genome-wide SNPs to the heritability of plasma sIL-6R levels. Behav Genet 2014;44(4):368-82. 96. Revez JA, Bain L, Chapman B, Powell JE, Jansen R, Duffy DL, et al. A new regulatory variant in the interleukin-6 receptor gene associates with asthma risk. Genes Immun 2013;14(7):441-6. 97. Ferreira RC, Freitag DF, Cutler AJ, Howson JM, Rainbow DB, Smyth DJ, et al. Functional IL6R 358Ala Allele Impairs Classical IL-6 Receptor Signaling and Influences Risk of Diverse Inflammatory Diseases. PLoS Genet 2013;9(4):e1003444. 98. Schaper F, Rose-John S. Interleukin-6: Biology, signaling and strategies of blockade. Cytokine & growth factor reviews 2015;26(5):475-87.

Functional characterization of the new 8q21 Asthma risk locus | 51

99. Navarini AA, French LE, Hofbauer GF. Interrupting IL-6-receptor signaling improves atopic dermatitis but associates with bacterial superinfection. J Allergy Clin Immunol 2011;128(5):1128-30. 100. Ullah MA, Revez JA, Loh Z, Simpson J, Zhang V, Bain L, et al. Allergen-induced IL-6 trans-signaling activates gammadelta T cells to promote type 2 and type 17 airway inflammation. J Allergy Clin Immunol 2015;136(4):1065-73. 101. Cottingham I, PETRI NA, inventorsAdministration of a selective il-6-trans-signalling inhibitor2016. 102. Kasela S, Kisand K, Tserel L, Kaleviste E, Remm A, Fischer K, et al. Pathogenic implications for autoimmune mechanisms derived by comparative eQTL analysis of CD4+ versus CD8+ T cells. PLoS Genet 2017;13(3):e1006643. 103. Zhernakova DV, Deelen P, Vermaat M, van Iterson M, van Galen M, Arindrarto W, et al. Identification of context-dependent expression quantitative trait loci in whole blood. Nat Genet 2017;49(1):139-45. 104. Luo W, Obeidat M, Di Narzo AF, Chen R, Sin DD, Pare PD, et al. Airway Epithelial Expression Quantitative Trait Loci Reveal Genes Underlying Asthma and Other Airway Diseases. Am J Respir Cell Mol Biol 2016;54(2):177-87. 105. Westra HJ, Peters MJ, Esko T, Yaghootkar H, Schurmann C, Kettunen J, et al. Systematic identification of trans eQTLs as putative drivers of known disease associations. Nat Genet 2013;45(10):1238-43. 106. Chun S, Casparino A, Patsopoulos NA, Croteau-Chonka DC, Raby BA, De Jager PL, et al. Limited statistical evidence for shared genetic effects of eQTLs and autoimmune- disease-associated loci in three major immune-cell types. Nat Genet 2017;49(4):600-5. 107. Berlivet S, Moussette S, Ouimet M, Verlaan DJ, Koka V, Al Tuwaijri A, et al. Interaction between genetic and epigenetic variation defines gene expression patterns at the asthma- associated locus 17q12-q21 in lymphoblastoid cell lines. Hum Genet 2012;131(7):1161-71. 108. Verlaan DJ, Berlivet S, Hunninghake GM, Madore AM, Lariviere M, Moussette S, et al. Allele-specific chromatin remodeling in the ZPBP2/GSDMB/ORMDL3 locus associated with the risk of asthma and autoimmune disease. Am J Hum Genet 2009;85(3):377-93. 109. Javierre BM, Burren OS, Wilder SP, Kreuzhuber R, Hill SM, Sewitz S, et al. Lineage- Specific Genome Architecture Links Enhancers and Non-coding Disease Variants to Target Gene Promoters. Cell 2016;167(5):1369-84 e19. 110. Tewhey R, Kotliar D, Park DS, Liu B, Winnicki S, Reilly SK, et al. Direct Identification of Hundreds of Expression-Modulating Variants using a Multiplexed Reporter Assay. Cell 2016;165(6):1519-29. 111. Pashos EE, Park Y, Wang X, Raghavan A, Yang W, Abbey D, et al. Large, Diverse Population Cohorts of hiPSCs and Derived Hepatocyte-like Cells Reveal Functional Genetic Variation at Blood Lipid-Associated Loci. Cell Stem Cell 2017;20(4):558-70 e10. 112. Miller M, Tam AB, Mueller JL, Rosenthal P, Beppu A, Gordillo R, et al. Cutting Edge: Targeting Epithelial ORMDL3 Increases, Rather than Reduces, Airway Responsiveness and Is Associated with Increased Sphingosine-1-Phosphate. J Immunol 2017;198(8):3017-22. 113. Miller M, Rosenthal P, Beppu A, Gordillo R, Broide DH. Oroscomucoid like protein 3 (ORMDL3) transgenic mice have reduced levels of sphingolipids including sphingosine-1- phosphate and ceramide. J Allergy Clin Immunol 2017;139(4):1373-6 e4. 114. Loser S, Gregory LG, Zhang Y, Schaefer K, Walker SA, Buckley J, et al. Pulmonary ORMDL3 is critical for induction of Alternaria-induced allergic airways disease. J Allergy Clin Immunol 2017;139(5):1496-507 e3. Functional characterization of the new 8q21 Asthma risk locus | 52

115. Oyeniran C, Sturgill JL, Hait NC, Huang WC, Avni D, Maceyka M, et al. Aberrant ORM (yeast)-like protein isoform 3 (ORMDL3) expression dysregulates ceramide homeostasis in cells and ceramide exacerbates allergic asthma in mice. J Allergy Clin Immunol 2015;136(4):1035-46 e6. 116. Miller M, Rosenthal P, Beppu A, Mueller JL, Hoffman HM, Tam AB, et al. ORMDL3 transgenic mice have increased airway remodeling and airway responsiveness characteristic of asthma. J Immunol 2014;192(8):3475-87. 117. Ha SG, Ge XN, Bahaie NS, Kang BN, Rao A, Rao SP, et al. ORMDL3 promotes eosinophil trafficking and activation via regulation of integrins and CD48. Nature communications 2013;4:2479. 118. Miller M, Tam AB, Cho JY, Doherty TA, Pham A, Khorram N, et al. ORMDL3 is an inducible lung epithelial gene regulating metalloproteases, chemokines, OAS, and ATF6. Proc Natl Acad Sci U S A 2012;109(41):16648-53. 119. Das S, Miller M, Beppu AK, Mueller J, McGeough MD, Vuong C, et al. GSDMB induces an asthma phenotype characterized by increased airway responsiveness and remodeling without lung inflammation. Proc Natl Acad Sci U S A 2016;113(46):13132-7. 120. Lloyd CM. Building better mouse models of asthma. Curr Allergy Asthma Rep 2007;7(3):231-6. 121. Barry J, Loh Z, Collison A, Mazzone S, Lalwani A, Zhang V, et al. Absence of Toll-IL-1 receptor 8/single immunoglobulin IL-1 receptor-related molecule reduces house dust mite- induced allergic airway inflammation in mice. Am J Respir Cell Mol Biol 2013;49(3):481-90. 122. Dullaers M, Schuijs MJ, Willart M, Fierens K, Van Moorleghem J, Hammad H, et al. House dust mite-driven asthma and allergen-specific T cells depend on B cells when the amount of inhaled allergen is limiting. J Allergy Clin Immunol 2017;140(1):76-88 e7. 123. McMillan SJ, Lloyd CM. Prolonged allergen challenge in mice leads to persistent airway remodelling. Clin Exp Allergy 2004;34(3):497-507. 124. Arikkatt J, Ullah MA, Short KR, Zhang V, Gan WJ, Loh Z, et al. RAGE deficiency predisposes mice to virus-induced paucigranulocytic asthma. Elife 2017;6. 125. El-Brolosy MA, Stainier DYR. Genetic compensation: A phenomenon in search of mechanisms. PLoS Genet 2017;13(7):e1006780. 126. Rohlmann A, Gotthardt M, Willnow TE, Hammer RE, Herz J. Sustained somatic gene inactivation by viral transfer of Cre recombinase. Nature biotechnology 1996;14(11):1562-5. 127. Consortium GT. Human genomics. The Genotype-Tissue Expression (GTEx) pilot analysis: multitissue gene regulation in humans. Science 2015;348(6235):648-60. 128. Watson B, Gauvreau GM. Thymic stromal lymphopoietin: a central regulator of allergic asthma. Expert Opin Ther Targets 2014;18(7):771-85. 129. Gauvreau GM, O'Byrne PM, Boulet LP, Wang Y, Cockcroft D, Bigler J, et al. Effects of an anti-TSLP antibody on allergen-induced asthmatic responses. N Engl J Med 2014;370(22):2102-10. 130. Lloyd-Jones LR, Holloway A, McRae A, Yang J, Small K, Zhao J, et al. The Genetic Architecture of Gene Expression in Peripheral Blood. Am J Hum Genet 2017;100(2):228-37. 131. Wells A, Kopp N, Xu X, O'Brien DR, Yang W, Nehorai A, et al. The anatomical distribution of genetic associations. Nucleic Acids Res 2015;43(22):10804-20. 132. Finucane HK, Bulik-Sullivan B, Gusev A, Trynka G, Reshef Y, Loh PR, et al. Partitioning heritability by functional annotation using genome-wide association summary statistics. Nat Genet 2015;47(11):1228-35.

Functional characterization of the new 8q21 Asthma risk locus | 53

133. Farh KK, Marson A, Zhu J, Kleinewietfeld M, Housley WJ, Beik S, et al. Genetic and epigenetic fine mapping of causal autoimmune disease variants. Nature 2015;518(7539):337- 43.

Functional characterization of the new 8q21 Asthma risk locus | 54

Conflict of Interest Statement

The authors have no conflicts of interest.

Functional characterization of the new 8q21 Asthma risk locus | 55

Figure 1. Effect size (odds ratio) for 31 previously reported asthma risk SNPs, comparing results reported in the original GWAS describing each association with those obtained in the analysis of the UK Biobank study. Each SNP is represented by a circle, with its size and color indicating total GWAS sample size (as per Supplementary Table 2) and PubMed identifier (PMID), respectively. The red diagonal line represents equality of odds ratio between published GWAS and UK Biobank (i.e. x = y). The genomic location of the four SNPs that deviate markedly from the diagonal are shown next to the respective circle. For three of these (SNPs in PDE4D, CRB1 and CDHR3), the association in the UK Biobank study had a P-value > 0.05 (see Table 2).

Functional characterization of the new 8q21 Asthma risk locus | 56

Table 1. Variants in low LD with each other (r2<0.05) reported to associate with asthma risk in GWAS conducted between 2007 and 2016 in populations of European ancestry.

-8 First association reported with P<5x10 Correlated SNPs Index Chr Bp Context* reported in other Effect Top SNP OR P-value PMID Year asthma GWAS# allele 1 1 10557251 [PEX14] rs662064 T 0.94 3.2E-08 27182965 2016 No 2 1 154426264 [IL6R] rs4129267 T 1.09 2.4E-08 21907864 2011 No 3 1 161159147 PPOX-[]-ADAMTS4 rs4233366 T 1.09 4.8E-15 27182965 2016 No 4 1 167433420 [CD247] rs1723018 G 0.95 1.4E-08 27182965 2016 No 5 1 173152036 TNFSF18--[]-TNFSF4 rs6691738 T 0.94 2.9E-08 27182965 2016 No 6 1 197325908 [CRB1] rs2786098 A 0.63 8.6E-09 20032318 2009 No 7 1 203100504 [ADORA1] rs6683383 T 1.06 1.1E-08 27182965 2016 No 8 2 8458080 [LINC00299] rs13412757 G 1.06 1.3E-08 27182965 2016 No 9 2 102986222 [IL18R1] rs3771166 A 0.87 3.4E-09 20860503 2010 Yes 10 2 242698640 [D2HGDH] rs34290285 G 1.11 1.8E-15 27182965 2016 No 11 3 188402471 [LPP] rs73196739 T 0.92 6.5E-09 27182965 2016 No 12 4 38799710 [TLR1] rs4833095 T 1.20 5.0E-12 24388013 2013 Yes 13 5 59369794 [PDE4D] rs1588265 G 0.60 2.5E-08 19426955 2009 No 14 5 110401872 SLC25A46--[]-TSLP rs1837253 C 1.19 7.3E-10 21804549 2011 Yes 15 5 131969874 [RAD50] rs6871536 C 1.14 2.4E-09 21907864 2011 Yes 16 5 141529762 [NDFIP1] rs200634877 I 0.94 2.5E-08 27182965 2016 No 17 6 31322197 [HLA-B] rs2428494 T 0.92 1.4E-16 27182965 2016 No 18 6 32728261 HLA-DQA1-[]-HLA-DQB1 rs17843604 T 1.16 1.7E-10 20860503 2010 Yes 19 6 90985198 [BACH2] rs58521088 T 0.93 7.1E-11 27182965 2016 No 20 7 105658451 [CDHR3] rs6967330 A 1.45 1.4E-08 24241537 2013 Yes 21 8 81291879 MIR5708--[]--ZBTB10 rs7009110 T 1.14 4.0E-09 24388013 2013 Yes 22 9 6190076 RANBP6--[]-IL33 rs1342326 C 1.20 9.2E-10 20860503 2010 Yes 23 10 9049253 GATA3---[]---SFTA1P rs12413578 T 0.89 8.1E-12 27182965 2016 No 24 11 76270683 WNT11--[]-LRRC32 rs7130588 G 1.09 1.8E-08 21907864 2011 Yes 25 12 57509055 [STAT6] rs3001426 T 0.94 1.4E-10 27182965 2016 No 26 14 68749927 [RAD51B] rs3784099 G 0.94 1.6E-08 27182965 2016 No 27 15 61069988 [RORA] rs11071559 T 0.85 3.8E-09 21907864 2011 Yes 28 15 67446785 [SMAD3] rs744910 A 0.89 3.9E-09 20860503 2010 Yes 29 16 11228712 [CLEC16A] rs62026376 C 1.17 1.0E-08 24388013 2013 Yes 30 17 38069949 [GSDMB] rs7216389 T 1.45 9.0E-11 17611496 2007 Yes Functional characterization of the new 8q21 Asthma risk locus | 57

31 22 37534034 [IL2RB] rs2284033 A 0.89 1.2E-08 20860503 2010 No Abbreviations: OR, odds ratio; PMID, PubMed identifier.

* If the top SNP is located within the boundaries of a gene, then the gene name is shown inside square brackets. Otherwise, the two nearest genes (upstream and downstream) are listed, with the distance to each represented by the number of '-' between the square bracket and the gene name.

# SNPs reported in other asthma GWAS and with (1) r2>0.05 with top SNP and (2) P<5x10-8 are listed in Supplementary Table 4.

Functional characterization of the new 8q21 Asthma risk locus | 58

Table 2. Association between 31 variants reported in GWAS of European ancestry and self-reported doctor-diagnosed asthma in the UK Biobank study (44,003 cases and 336,500 controls). Index SNP Chr Bp Context Effect allele MAF OR P-value Asthma h2 explained (%) 1 rs662064 1 10557251 [PEX14] C 0.31 1.04 9.7E-006 0.02 2 rs4129267 1 154426264 [IL6R] T 0.41 1.03 5.8E-006 0.02 3 rs4233366 1 161159147 PPOX-[]-ADAMTS4 T 0.27 1.04 4.7E-007 0.02 4 rs1723018 1 167433420 [CD247] G 0.41 0.95 5.7E-012 0.04 5 rs6691738 1 173152036 TNFSF18--[]-TNFSF4 G 0.29 1.04 5.8E-007 0.02 6 rs2786098 1 197325908 [CRB1] G 0.22 1.00 0.5827 0.00 7 rs6683383 1 203100504 [ADORA1] A 0.33 0.95 5.9E-011 0.04 8 rs13412757 2 8458080 [LINC00299] A 0.34 0.94 1.8E-013 0.05 9 rs3771166 2 102986222 [IL18R1] A 0.38 0.90 7.4E-040 0.15 10 rs34290285 2 242698640 [D2HGDH] A 0.26 0.89 1.5E-039 0.15 11 rs73196739 3 188402471 [LPP] T 0.17 0.94 4.1E-010 0.03 12 rs4833095 4 38799710 [TLR1] C 0.21 0.92 1.3E-017 0.06 13 rs1588265 5 59369794 [PDE4D] G 0.31 1.01 0.2859 0.00 14 rs1837253 5 110401872 SLC25A46--[]-TSLP C 0.26 1.12 3.4E-041 0.16 15 rs6871536 5 131969874 [RAD50] C 0.19 1.10 1.0E-024 0.09 16 rs166079 5 141528959 [NDFIP1] T 0.38 1.04 1.8E-008 0.03 17 rs2428494 6 31322197 [HLA-B] A 0.47 1.12 6.1E-055 0.20 18 rs17843604 6 32620283 HLA-DQA1-[]-HLA-DQB1 T 0.42 1.18 1.9E-105 0.41 19 rs58521088 6 90985198 [BACH2] T 0.35 0.93 2.5E-021 0.08 20 rs6967330 7 105658451 [CDHR3] A 0.17 1.01 0.2233 0.00 21 rs7009110 8 81291879 MIR5708--[]--ZBTB10 C 0.38 0.94 9.6E-018 0.06 22 rs1342326 9 6190076 RANBP6--[]-IL33 C 0.16 1.15 6.4E-048 0.17 23 rs12413578 10 9049253 GATA3---[]---SFTA1P T 0.11 0.86 1.0E-033 0.13 24 rs7130588 11 76270683 WNT11--[]-LRRC32 G 0.36 1.10 8.9E-033 0.12 25 rs3001426 12 57509055 [STAT6] C 0.45 1.07 1.4E-021 0.08 26 rs3784099 14 68749927 [RAD51B] A 0.28 1.06 1.4E-012 0.04 27 rs11071559 15 61069988 [RORA] T 0.13 0.91 7.3E-017 0.06 28 rs744910 15 67446785 [SMAD3] A 0.48 0.95 3.5E-014 0.05 29 rs62026376 16 11228712 [CLEC16A] T 0.25 0.92 7.7E-023 0.08 30 rs7216389 17 38069949 [GSDMB] T 0.48 1.11 1.7E-044 0.16 31 rs2284033 22 37534034 [IL2RB] A 0.43 0.97 4.2E-004 0.01 Abbreviations: MAF, minor allele frequency; OR, odds ratio; h2: heritability. Total 2.52

Functional characterization of the new 8q21 Asthma risk locus | 59

Three SNPs with a P>0.05 are highlighted in grey. Median 0.06

Functional characterization of the new 8q21 Asthma risk locus | 60

Table 3. Number of associations reported in published GWAS of other polygenic diseases, which were based on a similar or smaller sample size than the largest asthma GWAS published.

N of independent Disease NCP for a SNP Disease N cases N controls N Total associations in discovery PMID prevalence with h2=0.05% GWAS (P<5x10-8)

Atopic dermatitis 18900 84166 103066 21 26482879 20% 24 Asthma 28399 128843 157242 27 27182965 15% 40 Type 2 diabetes 26676 132532 159208 42 28566273 5% 52 Schizophrenia 36989 113075 150064 128 25056061 1% 102 Rheumatoid arthritis 29880 73758 103638 101 24390342 1% 77 All studies are based on the analysis of 1000 Genome Project SNPs imputed from GWAS arrays (not including fine-mapping arrays). Abbreviations: PMID, PubMed identifier; NCP: non-centrality parameter (which reflects power); h2: heritability.

Functional characterization of the new 8q21 Asthma risk locus | 61

Supplementary Table 1. Seventy-three studies identifed in the GWAS catalog through a search of the term "Asthma" and reviewed for inclusion in our analysis.

Index PMID First author Year Trait Reported in GWAS catalog Included Exclusion Criteria Notes Selected for analysis 1 27182965 Pickrell JK 2016 Asthma Yes - - 2 27142222 White MJ 2016 Asthma (childhood onset) Yes - - 3 26620591 Barreto-Luis A 2015 Asthma Yes - Published 2016 4 24406073 Galanter JM 2014 Asthma Yes - - 5 24388013 Ferreira MA 2013 Asthma and hay fever Yes - Published 2014 6 24241537 Bonnelykke K 2013 Asthma (childhood onset) Yes - Published 2014 7 23181788 Lasky-Su J 2012 Asthma Yes - - 8 23028483 Ramasamy A 2012 Asthma Yes - - 9 22694930 Li X 2012 Asthma Yes - - 10 22561531 Wan YI 2012 Asthma Yes - - 11 22295569 Karunas AS 2011 Asthma Yes - - 12 21907864 Ferreira MA 2011 Asthma Yes - - 13 21804548 Hirota T 2011 Asthma Yes - - 14 21804549 Torgerson DG 2011 Asthma Yes - - 15 21814517 Noguchi E 2011 Asthma Yes - - 16 21150878 Ferreira MA 2010 Asthma Yes - Published 2011 17 20920776 DeWan AT 2010 Asthma Yes - - 18 20860503 Moffatt MF 2010 Asthma Yes - - 19 20698975 Himes BE 2010 Asthma Yes - - 20 20159242 Li X 2010 Asthma Yes - - 21 20032318 Sleiman PM 2009 Asthma Yes - Published 2010 22 19910028 Mathias RA 2009 Asthma Yes - Published 2010 23 19714205 Hancock DB 2009 Asthma (childhood onset) Yes - - 24 19426955 Himes BE 2009 Asthma Yes - - 25 17611496 Moffatt MF 2007 Asthma Yes - - Excluded from analysis 1 27901618 Gref A 2016 Childhood onset asthma (traffic air No Published 2017; Interaction effects - pollution exposure interaction) 2 27611488 Almoguera B 2016 Asthma (childhood onset), Asthma, Adult No Published 2017 - asthma 3 27439200 Nieuwenhuis MA 2016 Bronchial hyperresponsiveness in asthma No Phenotype other than asthma - 4 27387956 Murk W 2016 Asthma (SNP x SNP interaction) No Interaction effects - 5 27523435 Mosteller M 2016 Response to inhaled corticosteroid No Published 2017; Phenotype other - Functional characterization of the new 8q21 Asthma risk locus | 62

treatment in asthma (change in FEV1) than asthma 6 27445529 Bérubé JC 2016 Asthma No GWAS based on DNA pooling - 7 26031901 Dahlin A 2015 Response to zileuton treatment in asthma No Phenotype other than asthma; Published (FEV1 change interaction) interaction effects 2016 8 26542096 Marenholz I 2015 Atopic march No Phenotype other than asthma - 9 26325155 Brehm JM 2015 Post-bronchodilator lung function in No Phenotype other than asthma - asthma (FEV1), Post-bronchodilator lung function in asthma (FEV1/FVC) 10 26083242 Dahlin A 2015 Response to montelukast in asthma (change No Phenotype other than asthma - in FEV1) 11 26073756 McGeachie MJ 2015 Asthma exacerbations No Phenotype other than asthma - 12 26025128 Park HW 2015 Bone mineral accretion in asthma (oral No Phenotype other than asthma; - corticosteroid dose interaction) interaction effects 13 25918132 Yucesoy B 2015 Diisocyanate-induced asthma No Occupational asthma - 14 25562107 Israel E 2015 Bronchodilator response in asthma No Phenotype other than asthma - 15 25601762 Wang Y 2015 Response to inhaled glucocorticoid No Phenotype other than asthma - treatment in asthma (change in FEV1) 16 25221879 Wu K 2014 Pulmonary function in asthmatics No Phenotype other than asthma - 17 25085501 Bunyavanich S 2014 Allergic rhinitis in asthma No Phenotype other than asthma - 18 24993907 Smolonska J 2014 Asthma or chronic obstructive pulmonary No Not all cases had asthma - disease 19 24824216 Myers RA 2014 Asthma (gender interaction) No Interaction effects - 20 24792382 Park TJ 2014 Response to inhaled corticosteroid No Phenotype other than asthma - treatment in asthma (change in FEV1) 21 24486069 Park HW 2014 Asthma (corticosteroid response) No Phenotype other than asthma - 22 24280104 Wu AC 2013 Bronchodilator response in asthma (inhaled No Phenotype other than asthma; Published corticosteroid treatment interaction) interaction effects 2014 23 23992748 Drake KA 2013 Asthma (bronchodilator response) No Phenotype other than asthma Published 2014 24 23508266 Duan QL 2013 Asthma (bronchodilator response) No Phenotype other than asthma Published 2014 25 23984888 Himes BE 2013 Airway hyperresponsiveness No Phenotype other than asthma - 26 23967269 Kim JH 2013 IgE levels in asthmatics (D.f. specific), IgE No Phenotype other than asthma - levels in asthmatics (D.p. specific); IgE levels in asthmatics 27 23886662 Weidinger S 2013 Atopic dermatitis No Phenotype other than asthma - 28 23829686 Ding L 2013 Asthma (childhood onset) No Rank-based analysis - 29 23517042 Melen E 2013 Body mass index in non-asthmatics, Body No Phenotype other than asthma - mass index in asthmatics 30 23180272 Park BL 2012 Aspirin exacerbated respiratory disease in No Phenotype other than asthma Published Functional characterization of the new 8q21 Asthma risk locus | 63

asthmatics 2013 31 22837378 Wilk JB 2012 Airflow obstruction No Phenotype other than asthma - 32 22792082 Himes BE 2012 Asthma (bronchodilator response) No Phenotype other than asthma - 33 22673963 Lasky-Su J 2012 Vitamin D levels No Phenotype other than asthma - 34 22560479 Forno E 2012 Asthma (childhood onset) No Phenotype other than asthma - 35 22538805 Tantisira KG 2012 Asthma No Phenotype other than asthma - 36 22424883 Imboden M 2012 Pulmonary function decline No Phenotype other than asthma - 37 22286170 Cusanovich DA 2012 Lymphocyte counts No Phenotype other than asthma - 38 22051697 Du R 2011 Asthma No Interaction effects Published 2012 39 21696813 Schauberger EM 2011 Asthma (childhood onset) No GWAS based on DNA pooling - 40 21359210 Ricci G 2011 Asthma (childhood onset) No GWAS based on DNA pooling - 41 22188591 Anantharaman R 2011 Asthma No GWAS based on DNA pooling - 42 21991891 Tantisira KG 2011 Asthma treatment response No Phenotype other than asthma - 43 21211648 Ege MJ 2011 Asthma or atopy (interaction) No Interaction effects - 44 21072201 Kim JH 2010 Aspirin intolerance in asthmatics No Phenotype other than asthma - 45 19961619 Castro-Giner F 2009 Atopy No Phenotype other than asthma - 46 19198610 Gudbjartsson DF 2009 Eosinophil counts No Phenotype other than asthma - 47 19187332 Kim SH 2009 Asthma (toluene diisocyanate-induced) No Occupational asthma - 48 18403759 Ober C 2008 YKL-40 levels No Phenotype other than asthma - GWAS Catalog search terms: 'Asthma' Filters: None Date search performed: 2 August 2017 Total studies retrieved and reviewed for inclusion: 73 Studies remaining after exclusion criteria: 25

Functional characterization of the new 8q21 Asthma risk locus | 64

Supplementary Table 2. Details for the twenty-five GWAS of asthma published between 2007 and 2016 that were included in this review. N N GWAS Inde First associatio PMID Date Journal Title Trait Ancestry N cases control N SNPs catalog x author ns with s URL P<5x10-8

Detection and interpretation of

1 Pickrell JK 27182965 5/16/2016 Nat Genet shared genetic Asthma European 28399 128843 13747813 27 Pickrell JK influences on 42 human traits.

Novel genetic risk factors for asthma in Immunogeneti African American Childhood

2 White MJ 27142222 5/3/2016 African American 812 415 797128 0 White MJ cs children: Precision onset asthma Medicine and the SAGE II Study.

Genome-wide association study in Spanish identifies ADAM Barreto- J Allergy Clin metallopeptidase Barreto- 3 26620591 4/24/2015 Asthma European 380 552 6467565 0

Luis A Immunol with thrombospondin Luis A type 1 motif, 9 (ADAMTS9), as a novel asthma susceptibility gene.

Genome-wide association study and admixture mapping identify different Galanter J Allergy Clin asthma-associated Galanter 4 24406073 7/3/2014 Asthma Latino 1893 1881 818154 1

JM Immunol loci in Latinos: the JM Genes-environments & Admixture in Latino Americans study.

Functional characterization of the new 8q21 Asthma risk locus | 65

Genome-wide association analysis Ferreira J Allergy Clin identifies 11 risk Asthma with Ferreira 5 24388013 12/30/2013 European 6685 14091 4972397 11

MA Immunol variants associated hay fever MA with the asthma with hay fever phenotype.

A genome-wide association study Childhood identifies CDHR3 as Bonnelykke onset asthma Bonnelykke 6 24241537 11/17/2013 Nat Genet a susceptibility locus European 1173 2522 124514 5

K with severe for early childhood K exacerbations asthma with severe exacerbations.

HLA-DQ strikes again: genome-wide 1238 association study Clin Exp cases;

7 Lasky-Su J 23181788 12/4/2012 further confirms Asthma European 1348 455089 1 Allergy 1269 Lasky-Su J HLA-DQ in the parents diagnosis of asthma among adults.

Genome-wide association studies of asthma in population-based Ramasamy cohorts confirm Ramasamy 8 23028483 9/28/2012 PLoS One Asthma European 1716 16888 ~2500000 2

A known and suggested A loci and identify an additional association near HLA.

Genome-wide association studies of asthma indicate J Allergy Clin opposite

9 Li X 22694930 6/11/2012 Asthma European 813 1564 474271 0 Immunol immunopathogenesis Li X direction from autoimmune diseases.

Functional characterization of the new 8q21 Asthma risk locus | 66

Genome-wide association study to Severe

10 Wan Yi 22561531 5/5/2012 Thorax identify genetic European 933 3346 480889 1 asthma Wan YI determinants of severe asthma.

Genome-wide association study of Mol Biol Karunas 11 Karunas AS 22295569 11/1/2011 bronchial asthma in Asthma European 330 348 550915 0

(Mosk) the Volga-Ural AS region of Russia.

Identification of IL6R and Ferreira Ferreira 12 21907864 9/10/2011 Lancet chromosome Asthma European 12475 19967 421334 11

MA 11q13.5 as risk loci MA for asthma.

Genome-wide association study identifies three new

13 Hirota T 21804548 7/31/2011 Nat Genet Asthma Asian (Japanese) 1532 3304 458847 5 susceptibility loci for Hirota T adult asthma in the Japanese population.

3246 Meta-analysis of 3385 unrelated genome-wide unrelated cases; association studies of European; African controls; Torgerson 1702 trios; Torgerson 14 21804549 7/31/2011 Nat Genet asthma in ethnically Asthma American/Caribbean; 468 >2000000 6

DG 355 diverse North Latino family- DG family- American based based populations. controls cases Genome-wide association study identifies HLA-DP Childhood

15 Noguchi E 21814517 7/21/2011 PLoS Genet as a susceptibility Asian (Japanese) 938 2376 450326 2 onset asthma Noguchi E gene for pediatric asthma in Asian populations.

Functional characterization of the new 8q21 Asthma risk locus | 67

Association between ORMDL3, IL1RL1 Ferreira Eur J Hum and a deletion on Ferreira 16 21150878 12/8/2010 Asthma European 986 1846 ~2380000 0

MA Genet chromosome 17q21 MA with asthma risk in Australia.

PDE11A associations with Childhood J Allergy Clin

17 DeWan AT 20920776 10/1/2010 asthma: results of a onset asthma European 66 42 396207 0 Immunol DeWan AT genome-wide with allergies association scan. A large-scale, consortium-based

18 Moffatt MF 20860503 9/23/2010 N Engl J Med genomewide Asthma European 10365 16110 582892 6 Moffatt MF association study of asthma. Asthma- susceptibility 403 trios variants identified BMC Med (359

19 Himes BE 20698975 8/10/2010 using probands in Asthma European 846 516617 0 Genet unrelated Himes BE case-control and cases) family-based analyses.

Genome-wide association study of J Allergy Clin asthma identifies Severe

20 Li X 20159242 2/1/2010 European 473 1892 292443 0 Immunol RAD50-IL13 and asthma Li X HLA-DR/DQ regions.

Variants of Persistent Sleiman DENND1B Sleiman 21 20032318 12/23/2009 N Engl J Med asthma with European 793 1988 ~2200000 1

PM associated with daily ICS PM asthma in children.

464 A genome-wide unrelated association study on J Allergy Clin African cases; 163 Mathias 22 Mathias RA 19910028 11/10/2009 African-ancestry Asthma 471 644709 0

Immunol American/Caribbean families populations for RA with 929 asthma. individuals

Functional characterization of the new 8q21 Asthma risk locus | 68

Genome-wide association study implicates Hancock chromosome Childhood Hancock 23 19714205 8/28/2009 PLoS Genet Latino 492 trios NA 520767 0

DB 9q21.31 as a onset asthma DB susceptibility locus for asthma in mexican children.

Genome-wide association analysis Am J Hum

24 Himes BE 19426955 7/5/2009 identifies PDE4D as Asthma European 359 846 518230 1 Genet Himes BE an asthma- susceptibility gene.

Genetic variants regulating ORMDL3 Childhood

25 Moffatt MF 17611496 7/26/2007 Nature expression contribute European 994 1243 307328 1 onset asthma Moffatt MF to the risk of childhood asthma.

Functional characterization of the new 8q21 Asthma risk locus | 69

Supplementary Table 3. Genetic variants reported to associate with disease risk (P<5x10-8) in published asthma GWAS. P-value Chr Effect PMID First author Year Index bp (hg18) SNP OR GWAS+replication Notes band allele GWAS (if GWAS P>5x10-8) 27182965 Pickrell JK 2016 1 17q12 chr17:38026169 rs11655198 T 0.85 1.0E-63 NA NA 2 6p21.32 chr6:32603487 rs3104367 T 0.87 1.0E-40 NA NA

3 9p24.1 chr9:6208030 rs144829310 T 1.17 1.3E-31 NA NA

4 5q22.1 chr5:110401872 rs1837253 T 0.88 3.3E-31 NA NA

5 2q12.1 chr2:102913643 rs202011557 I 0.84 5.1E-31 NA NA

6 15q22.33 chr15:67448363 rs56375023 G 0.90 2.4E-21 NA NA

7 11q13.5 chr11:76293758 rs7936323 G 0.92 1.4E-16 NA NA

8 6p21.33 chr6:31322197 rs2428494 T 0.92 1.4E-16 NA NA

9 5q31.1 chr5:131901225 rs2244012 G 1.10 2.1E-16 NA NA

10 2q37.3 chr2:242698640 rs34290285 G 1.11 1.8E-15 NA NA

11 16p13.13 chr16:11230703 rs7203459 T 1.09 3.5E-15 NA NA

12 1q23.3 chr1:161159147 rs4233366 T 1.09 4.8E-15 NA NA

13 10p14 chr10:9049253 rs12413578 T 0.89 8.1E-12 NA NA

14 8q21.13 chr8:81285139 rs10957978 T 0.93 1.1E-11 NA NA

15 15q22.2 chr15:61068704 rs10519068 G 1.10 3.8E-11 NA NA

16 4p14 chr4:38798648 rs5743618 C 1.08 3.9E-11 NA NA

17 6q15 chr6:90985198 rs58521088 T 0.93 7.1E-11 NA NA

18 12q13.3 chr12:57509055 rs3001426 T 0.94 1.4E-10 NA NA

19 3q28 chr3:188402471 rs73196739 T 0.92 6.5E-09 NA NA

20 1q32.1 chr1:203100504 rs6683383 T 1.06 1.1E-08 NA NA

21 2p25.1 chr2:8458080 rs13412757 G 1.06 1.3E-08 NA NA

22 1q24.2 chr1:167433420 rs1723018 G 0.95 1.4E-08 NA NA

23 14q24.1 chr14:68749927 rs3784099 G 0.94 1.6E-08 NA NA

24 7q22.3 chr7:105676505 rs6959584 T 1.09 2.0E-08 NA NA

25 5q31.3 chr5:141529762 rs200634877 I 0.94 2.5E-08 NA NA

26 1q25.1 chr1:173152036 rs6691738 T 0.94 2.9E-08 NA NA

27 1p36.22 chr1:10557251 rs662064 T 0.94 3.2E-08 NA NA

24406073 Galanter JM 2014 1 17q21 chr17:37922259 rs907092 A 0.67 5.7E-13 NA NA 24388013 Ferreira MA 2013 1 6p21.32 chr6:32626601 rs9273373 G 1.24 4.0E-14 NA NA 2 4p14 chr4:38799710 rs4833095 T 1.20 5.0E-12 NA NA

3 5q22.1 chr5:110467499 rs1438673 C 1.16 3.0E-11 NA NA

Functional characterization of the new 8q21 Asthma risk locus | 70

4 2q12.1 chr2:102966549 rs10197862 A 1.24 4.0E-11 NA NA

5 11q13.5 chr11:76299194 rs2155219 T 1.16 4.0E-11 NA NA

6 17q21.1 chr17:38122680 rs7212938 G 1.16 4.0E-10 NA NA

7 5q22.1 chr5:110401872 rs1837253 C 1.17 1.0E-09 NA NA

8 9p24.1 chr9:6175855 rs72699186 T 1.26 2.0E-09 NA NA

9 8q21.13 chr8:81291879 rs7009110 T 1.14 4.0E-09 NA NA

10 15q22.33 chr15:67468285 rs17294280 G 1.18 4.0E-09 NA NA

11 16p13.13 chr16:11228712 rs62026376 C 1.17 1.0E-08 NA NA

Bonnelykke 24241537 2013 1 17q12 chr17:38062196 rs2305480 G 2.28 1.3E-48 NA NA K 2 9p24.1 chr9:6213387 rs928413 G 1.50 4.2E-13 NA NA 3 5q31.1 chr5:131969874 rs6871536 C 1.44 1.8E-09 NA NA 4 2q12.1 chr2:102765865 rs1558641 G 1.56 6.6E-09 NA NA 5 7q22.3 chr7:105658451 rs6967330 A 1.45 1.4E-08 NA NA 23181788 Lasky-Su J 2012 1 6p21.32 chr6:32604372 rs9272346 G 0.93 1.2E-07 2.2E-08 NA 23028483 Ramasamy A 2012 1 2q12.1 chr2:102955082 rs13408661 G 1.23 3.9E-06 1.1E-09 NA 2 6p21.32 chr6:32379489 rs9268516 T 1.15 1.2E-07 1.1E-08 NA

22561531 Wan Yi 2012 1 17q12.21 chr17:38089344 rs4794820 A 0.75 1.5E-06 1.0E-08 NA 21907864 Ferreira MA 2011 1 17q12 chr17:38092713 rs8079416 C 1.19 2.4E-22 NA NA 2 2q12.1 chr2:102986222 rs3771166 A 0.86 7.9E-15 NA NA 3 9p24.1 chr9:6190076 rs1342326 C 1.20 3.5E-14 NA NA 4 17q12 chr17:37817482 rs2271308 T 1.14 1.7E-10 NA NA 5 22q12.3 chr22:37534034 rs2284033 A 0.89 5.0E-10 NA NA 6 5q31.1 chr5:131969874 rs6871536 C 1.14 2.4E-09 NA NA 7 15q22.33 chr15:67446785 rs744910 G 1.11 2.7E-09 NA NA 8 15q22.2 chr15:61069988 rs11071559 T 0.85 3.8E-09 NA NA 9 5q22.1 chr5:110464008 rs1043828 C 1.11 1.1E-08 NA NA 10 11q13.5 chr11:76270683 rs7130588 G 1.09 1.2E-07 1.8E-08 NA 11 1q21.3 chr1:154426264 rs4129267 T 1.09 2.0E-06 2.4E-08 NA 21804548 Hirota T 2011 1 4q31.21 chr4:144003159 rs7686660 T 1.16 1.3E-04 1.8E-12 NA 2 5q22.1 chr5:110401872 rs1837253 C 1.17 3.5E-05 1.2E-16 NA

3 6p21.32 chr6:32184345 rs404860 A 1.21 2.3E-06 4.1E-23 NA

4 10p14 chr10:8972018 rs10508372 C 1.16 3.5E-05 1.8E-15 NA

5 12q13.2 chr12:56412487 rs1701704 G 1.19 2.2E-06 2.3E-13 NA

Torgerson Multi-ancestry 21804549 2011 1 17q12 chr17:38064405 rs11078927 C 1.27 1.2E-14 NA DG analysis

Functional characterization of the new 8q21 Asthma risk locus | 71

Multi-ancestry 2 5q22.1 chr5:110401872 rs1837253 C 1.19 7.3E-10 NA analysis Multi-ancestry 3 2q12.1 chr2:102957348 rs10173081 C 1.20 1.4E-08 NA analysis Latino only analysis 4 3q27.3 chr3:187417968 rs2017908 A 1.63 4.4E-09 NA (P=0.003 in GWAS+replication) Multi-ancestry 5 9p24.1 chr9:6193455 rs2381416 C 1.18 2.5E-07 1.7E-12 analysis African 6 1q23.1 chr1:158932555 rs1101999 T 1.34 3.6E-07 3.9E-09 American/Caribbean only analysis 21814517 Noguchi E 2011 1 8q24.11 chr8:118025645 rs3019885 G 1.55 1.3E-14 NA NA A third SNP (rs2281389) was 2 6p21.32 chr6:33042880 rs987870 C 1.51 7.5E-09 NA reported but is in LD

(r2=0.32) with rs987870 The strongest SNP in this locus (rs9273349) was not available in the 1000G data, so we 20860503 Moffatt MF 2010 1 6p21.32 chr6:32728261 rs17843604 T 1.16 1.7E-10 NA used the second strongest SNP reported, rs17843604

2 9p24.1 chr9:6190076 rs1342326 C 1.20 9.2E-10 NA NA 3 2q12.1 chr2:102986222 rs3771166 A 0.87 3.4E-09 NA NA 4 15q22.33 chr15:67446785 rs744910 A 0.89 3.9E-09 NA NA 5 17q21.1 chr17:38121993 rs3894194 A 1.17 4.6E-09 NA NA 6 22q12.3 chr22:37534034 rs2284033 A 0.89 1.2E-08 NA NA 20032318 Sleiman PM 2009 1 1q31.3 chr1:197325908 rs2786098 A 0.63 8.6E-09 NA NA 19426955 Himes BE 2009 1 5q12.1 chr5:59369794 rs1588265 G 0.60 4.3E-07 2.5E-08 NA 17611496 Moffatt MF 2007 1 17q21 chr17:38069949 rs7216389 T 1.45 9.0E-11 NA NA

Functional characterization of the new 8q21 Asthma risk locus | 72

Supplementary Table 4. Variants in LD with top SNP listed in Table 1 that were reported to associate with asthma risk in other published asthma GWAS.

Index Chr Bp Context Top SNP SNPs in LD with top SNP reported in other asthma GWAS

9 2 102913643 IL1RL2-[]-IL18R1 rs3771166 rs1558641,rs13408661,rs10173081,rs10197862,rs202011557 12 4 38799710 [TLR1] rs4833095 rs5743618 14 5 110401872 SLC25A46--[]-TSLP rs1837253 rs1043828,rs1438673 15 5 131901225 [RAD50] rs6871536 rs2244012 18 6 32603487 HLA-DRB1-[]-HLA-DQA1 rs17843604 rs9268516,rs9272346,rs9273373,rs3104367 20 7 105658451 [CDHR3] rs6967330 rs6959584 21 8 81285139 MIR5708--[]--ZBTB10 rs7009110 rs10957978 22 9 6208030 RANBP6--[]-IL33 rs1342326 rs72699186,rs2381416,rs928413,rs144829310 24 11 76293758 WNT11--[]-LRRC32 rs7130588 rs2155219,rs7936323 27 15 61068704 [RORA] rs11071559 rs10519068 28 15 67448363 [SMAD3] rs744910 rs17294280,rs56375023 29 16 11230703 [CLEC16A] rs62026376 rs7203459 30 17 38026169 [ZPBP2] rs7216389 rs2271308,rs2305480,rs11078927,rs4794820,rs8079416,rs3894194,rs11655198 Abbreviations: OR, odds ratio; PMID, PubMed identifier.

Functional characterization of the new 8q21 Asthma risk locus | 73

Supplementary Table 5. Variants in low LD with each other (r2<0.05) reported to associate with asthma risk in GWAS conducted between 2007 and 2016 in populations of Asian, African or Latino ancestry. Effect Index Chr Bp SNP Context OR P-value Year PMID allele Asians 1 4 144003159 rs7686660 INPP4B--[]--USP38 T 1.16 1.9E-12 2011 21804548 2 5 110401872 rs1837253 SLC25A46--[]-TSLP C 1.17 1.2E-16 2011 21804548 3 6 32184345 rs404860 [NOTCH4] A 1.21 4.1E-23 2011 21804548 4 6 33042880 rs987870 [HLA-DPA1] C 1.51 7.5E-09 2011 21814517 5 8 118025645 rs3019885 [SLC30A8] G 1.55 1.3E-14 2011 21814517 6 10 8972018 rs10508372 GATA3---[]---SFTA1P C 1.16 1.8E-15 2011 21804548 7 12 56412487 rs1701704 [IKZF4] G 1.19 2.3E-13 2011 21804548 Africans 1 1 158932555 rs1101999 [PYHIN1] T 1.34 3.9E-09 2011 21804549 Latinos 1 3 187417968 rs2017908 [RTP2] A 1.63 4.4E-09 2011 21804549 2 6 32604372 rs9272346 HLA-DRB1-[]-HLA-DQA1 G 0.93 2.2E-08 2012 23181788 3 17 37922259 rs907092 [IKZF3] A 0.67 5.7E-13 2014 24406073 Abbreviations: OR, odds ratio; PMID, PubMed identifier.

Functional characterization of the new 8q21 Asthma risk locus | 74

Supplementary Table 6. Likely target genes of published asthma risk variants identified based on LD with non-synonymous SNPs. GWAS Coding LD Index Gene Amino acid change ClinVar ID SNP SNP (r2) 2 rs4129267 rs2228145 1.00 IL6R IL6R:NM_000565:exon9:c.A1073C:p.D358A RCV000015767.2|RCV0 0 00029243.2 2 rs4129267 rs2228145 1.00 IL6R IL6R:NM_000565:exon9:c.A1073G:p.D358G NA 0 2 rs4129267 rs2228145 1.00 IL6R IL6R:NM_000565:exon9:c.A1073T:p.D358V RCV000015767.2|RCV0 0 00029243.2 9 rs3771166 rs4988956 1.00 IL1RL1 IL1RL1:NM_016232:exon11:c.G1297A:p.A433T IL1RL1:NM_016232:exo 0 n11:c.C1646T:p.T549I 9 rs3771166 rs10192036 1.00 IL1RL1 IL1RL1:NM_016232:exon11:c.C1501A:p.Q501K IL1RL1:NM_016232:exo 0 n11:c.T1652C:p.L551S 9 rs3771166 rs10204137 1.00 IL1RL1 IL1RL1:NM_016232:exon11:c.A1502G:p.Q501R NA

9 rs3771166 rs10192157 1.00 IL1RL1 IL1RL1:NM_016232:exon11:c.C1646T:p.T549I NA

9 rs3771166 rs10206753 1.00 IL1RL1 IL1RL1:NM_016232:exon11:c.T1652C:p.L551S NA

12 rs4833095 rs4833095 1.00 TLR1 TLR1:NM_003263:exon4:c.A743G:p.N248S;TLR1:NM_003263:ex TLR1:NM_003263:exon RC on4:c.A743G:p.N248S 4:c.A743G:p.N248S;TL V00 R1:NM_003263:exon4:c 000 .A743G:p.N248S 886 6.2 12 rs4833095 rs5743618 0.83 TLR1 TLR1:NM_003263:exon4:c.G1805T:p.S602I;TLR1:NM_003263:ex TLR1:NM_003263:exon RC on4:c.G1805T:p.S602I 4:c.G1805T:p.S602I;TL V00 R1:NM_003263:exon4:c 000 .G1805T:p.S602I 886 5.5 12 rs5743618 rs5743618 1.00 TLR1 TLR1:NM_003263:exon4:c.G1805T:p.S602I;TLR1:NM_003263:ex TLR1:NM_003263:exon RC on4:c.G1805T:p.S602I 4:c.G1805T:p.S602I;TL V00 R1:NM_003263:exon4:c 000 .G1805T:p.S602I 886 5.5

Functional characterization of the new 8q21 Asthma risk locus | 75

12 rs5743618 rs4833095 0.83 TLR1 TLR1:NM_003263:exon4:c.A743G:p.N248S;TLR1:NM_003263:ex TLR1:NM_003263:exon RC on4:c.A743G:p.N248S 4:c.A743G:p.N248S;TL V00 R1:NM_003263:exon4:c 000 .A743G:p.N248S 886 6.2 18 rs17843604 rs1047989 0.99 HLA- HLA-DQA1:NM_002122:exon1:c.C22A:p.L8M NA DQA1 18 rs3104367 rs1071630 0.93 HLA- HLA-DQA1:NM_002122:exon2:c.T122C:p.F41S;HLA- NA DQA1 DQA1:NM_002122:exon2:c.T122C:p.F41S;HLA- DQA1:NM_002122:exon2:c.T122C:p.F41S 18 rs3104367 rs1049092 0.92 HLA- HLA-DQB1:NM_001243961:exon3:c.T603G:p.D201E,HLA- NA DQB1 DQB1:NM_002123:exon3:c.T603G:p.D201E;HLA- DQB1:NM_001243961:exon3:c.T603G:p.D201E,HLA- DQB1:NM_002123:exon3:c.T603G:p.D201E;HLA- DQB1:NM_001243961:exon3:c.T603G:p.D201E,HLA- DQB1:NM_002123:exon3:c.T603G:p.D201E 18 rs3104367 rs9272793 0.92 HLA- HLA-DQA1:NM_002122:exon4:c.A722G:p.Q241R;HLA- NA DQA1 DQA1:NM_002122:exon4:c.A722G:p.Q241R;HLA- DQA1:NM_002122:exon4:c.A722G:p.Q241R 18 rs3104367 rs1049057 0.85 HLA- HLA-DQB1:NM_001243961:exon1:c.A35G:p.D12G,HLA- NA DQB1 DQB1:NM_002123:exon1:c.A35G:p.D12G;HLA- DQB1:NM_001243961:exon1:c.A35G:p.D12G,HLA- DQB1:NM_002123:exon1:c.A35G:p.D12G;HLA- DQB1:NM_001243961:exon1:c.A35G:p.D12G,HLA- DQB1:NM_002123:exon1:c.A35G:p.D12G 18 rs9272346 rs1071630 0.86 HLA- HLA-DQA1:NM_002122:exon2:c.T122C:p.F41S;HLA- NA DQA1 DQA1:NM_002122:exon2:c.T122C:p.F41S;HLA- DQA1:NM_002122:exon2:c.T122C:p.F41S 18 rs9272346 rs1049092 0.85 HLA- HLA-DQB1:NM_001243961:exon3:c.T603G:p.D201E,HLA- NA DQB1 DQB1:NM_002123:exon3:c.T603G:p.D201E;HLA- DQB1:NM_001243961:exon3:c.T603G:p.D201E,HLA- DQB1:NM_002123:exon3:c.T603G:p.D201E;HLA- DQB1:NM_001243961:exon3:c.T603G:p.D201E,HLA- DQB1:NM_002123:exon3:c.T603G:p.D201E

18 rs9272346 rs9272793 0.85 HLA- HLA-DQA1:NM_002122:exon4:c.A722G:p.Q241R;HLA- NA DQA1 DQA1:NM_002122:exon4:c.A722G:p.Q241R;HLA- DQA1:NM_002122:exon4:c.A722G:p.Q241R

Functional characterization of the new 8q21 Asthma risk locus | 76

18 rs9272346 rs1049057 0.81 HLA- HLA-DQB1:NM_001243961:exon1:c.A35G:p.D12G,HLA- NA DQB1 DQB1:NM_002123:exon1:c.A35G:p.D12G;HLA- DQB1:NM_001243961:exon1:c.A35G:p.D12G,HLA- DQB1:NM_002123:exon1:c.A35G:p.D12G;HLA- DQB1:NM_001243961:exon1:c.A35G:p.D12G,HLA- DQB1:NM_002123:exon1:c.A35G:p.D12G 18 rs9273373 rs1071630 0.99 HLA- HLA-DQA1:NM_002122:exon2:c.T122C:p.F41S;HLA- NA DQA1 DQA1:NM_002122:exon2:c.T122C:p.F41S;HLA- DQA1:NM_002122:exon2:c.T122C:p.F41S 18 rs9273373 rs9272793 0.98 HLA- HLA-DQA1:NM_002122:exon4:c.A722G:p.Q241R;HLA- NA DQA1 DQA1:NM_002122:exon4:c.A722G:p.Q241R;HLA- DQA1:NM_002122:exon4:c.A722G:p.Q241R 18 rs9273373 rs1049092 0.98 HLA- HLA-DQB1:NM_001243961:exon3:c.T603G:p.D201E,HLA- NA DQB1 DQB1:NM_002123:exon3:c.T603G:p.D201E;HLA- DQB1:NM_001243961:exon3:c.T603G:p.D201E,HLA- DQB1:NM_002123:exon3:c.T603G:p.D201E;HLA- DQB1:NM_001243961:exon3:c.T603G:p.D201E,HLA- DQB1:NM_002123:exon3:c.T603G:p.D201E 18 rs9273373 rs1049057 0.90 HLA- HLA-DQB1:NM_001243961:exon1:c.A35G:p.D12G,HLA- NA DQB1 DQB1:NM_002123:exon1:c.A35G:p.D12G;HLA- DQB1:NM_001243961:exon1:c.A35G:p.D12G,HLA- DQB1:NM_002123:exon1:c.A35G:p.D12G;HLA- DQB1:NM_001243961:exon1:c.A35G:p.D12G,HLA- DQB1:NM_002123:exon1:c.A35G:p.D12G 18 rs9273373 rs1129740 0.85 HLA- HLA-DQA1:NM_002122:exon2:c.G101A:p.C34Y NA DQA1

18 rs9273373 rs1142328 0.81 HLA- HLA-DQA1:NM_002122:exon2:c.G212T:p.W71L NA DQA1 30 rs11078927 rs2305480 0.99 GSDMB GSDMB:NM_018530:exon6:c.C865T:p.P289S,GSDMB:NM_00104 NA 2471:exon8:c.C892T:p.P298S,GSDMB:NM_001165959:exon8:c.C9 04T:p.P302S,GSDMB:NM_001165958:exon9:c.C931T:p.P311S;GS DMB:NM_018530:exon6:c.C865T:p.P289S,GSDMB:NM_00104247 1:exon8:c.C892T:p.P298S,GSDMB:NM_001165959:exon8:c.C904T :p.P302S,GSDMB:NM_001165958:exon9:c.C931T:p.P311S;GSDM B:NM_018530:exon6:c.C865T:p.P289S,GSDMB:NM_001042471:e xon8:c.C892T:p.P298S,GSDMB:NM_001165959:exon8:c.C904T:p. P302S,GSDMB:NM_001165958:exon9:c.C931T:p.P311S

Functional characterization of the new 8q21 Asthma risk locus | 77

30 rs11078927 rs2305479 0.84 GSDMB GSDMB:NM_018530:exon6:c.G844A:p.G282R,GSDMB:NM_0010 NA 42471:exon8:c.G871A:p.G291R,GSDMB:NM_001165959:exon8:c. G883A:p.G295R,GSDMB:NM_001165958:exon9:c.G910A:p.G304 R;GSDMB:NM_018530:exon6:c.G844A:p.G282R,GSDMB:NM_00 1042471:exon8:c.G871A:p.G291R,GSDMB:NM_001165959:exon8: c.G883A:p.G295R,GSDMB:NM_001165958:exon9:c.G910A:p.G30 4R;GSDMB:NM_018530:exon6:c.G844A:p.G282R,GSDMB:NM_0 01042471:exon8:c.G871A:p.G291R,GSDMB:NM_001165959:exon8 :c.G883A:p.G295R,GSDMB:NM_001165958:exon9:c.G910A:p.G30 4R;GSDMB:NM_018530:exon6:c.G844A:p.G282R,GSDMB:NM_0 01042471:exon8:c.G871A:p.G291R,GSDMB:NM_001165959:exon8 :c.G883A:p.G295R,GSDMB:NM_001165958:exon9:c.G910A:p.G30 4R 30 rs11078927 rs11557467 0.82 ZPBP2 ZPBP2:NM_198844:exon4:c.G452T:p.S151I,ZPBP2:NM_199321:ex NA on5:c.G518T:p.S173I;ZPBP2:NM_198844:exon4:c.G452T:p.S151I, ZPBP2:NM_199321:exon5:c.G518T:p.S173I;ZPBP2:NM_198844:e xon4:c.G452T:p.S151I,ZPBP2:NM_199321:exon5:c.G518T:p.S173I; ZPBP2:NM_198844:exon4:c.G452T:p.S151I,ZPBP2:NM_199321:ex on5:c.G518T:p.S173I 30 rs11655198 rs11557467 1.00 ZPBP2 ZPBP2:NM_198844:exon4:c.G452T:p.S151I,ZPBP2:NM_199321:ex NA on5:c.G518T:p.S173I;ZPBP2:NM_198844:exon4:c.G452T:p.S151I, ZPBP2:NM_199321:exon5:c.G518T:p.S173I;ZPBP2:NM_198844:e xon4:c.G452T:p.S151I,ZPBP2:NM_199321:exon5:c.G518T:p.S173I; ZPBP2:NM_198844:exon4:c.G452T:p.S151I,ZPBP2:NM_199321:ex on5:c.G518T:p.S173I 30 rs11655198 rs2305479 0.97 GSDMB GSDMB:NM_018530:exon6:c.G844A:p.G282R,GSDMB:NM_0010 NA 42471:exon8:c.G871A:p.G291R,GSDMB:NM_001165959:exon8:c. G883A:p.G295R,GSDMB:NM_001165958:exon9:c.G910A:p.G304 R;GSDMB:NM_018530:exon6:c.G844A:p.G282R,GSDMB:NM_00 1042471:exon8:c.G871A:p.G291R,GSDMB:NM_001165959:exon8: c.G883A:p.G295R,GSDMB:NM_001165958:exon9:c.G910A:p.G30 4R;GSDMB:NM_018530:exon6:c.G844A:p.G282R,GSDMB:NM_0 01042471:exon8:c.G871A:p.G291R,GSDMB:NM_001165959:exon8 :c.G883A:p.G295R,GSDMB:NM_001165958:exon9:c.G910A:p.G30 4R;GSDMB:NM_018530:exon6:c.G844A:p.G282R,GSDMB:NM_0 01042471:exon8:c.G871A:p.G291R,GSDMB:NM_001165959:exon8 :c.G883A:p.G295R,GSDMB:NM_001165958:exon9:c.G910A:p.G30 4R

Functional characterization of the new 8q21 Asthma risk locus | 78

30 rs11655198 rs2305480 0.82 GSDMB GSDMB:NM_018530:exon6:c.C865T:p.P289S,GSDMB:NM_00104 NA 2471:exon8:c.C892T:p.P298S,GSDMB:NM_001165959:exon8:c.C9 04T:p.P302S,GSDMB:NM_001165958:exon9:c.C931T:p.P311S;GS DMB:NM_018530:exon6:c.C865T:p.P289S,GSDMB:NM_00104247 1:exon8:c.C892T:p.P298S,GSDMB:NM_001165959:exon8:c.C904T :p.P302S,GSDMB:NM_001165958:exon9:c.C931T:p.P311S;GSDM B:NM_018530:exon6:c.C865T:p.P289S,GSDMB:NM_001042471:e xon8:c.C892T:p.P298S,GSDMB:NM_001165959:exon8:c.C904T:p. P302S,GSDMB:NM_001165958:exon9:c.C931T:p.P311S 30 rs2305480 rs2305480 1.00 GSDMB GSDMB:NM_018530:exon6:c.C865T:p.P289S,GSDMB:NM_00104 NA 2471:exon8:c.C892T:p.P298S,GSDMB:NM_001165959:exon8:c.C9 04T:p.P302S,GSDMB:NM_001165958:exon9:c.C931T:p.P311S;GS DMB:NM_018530:exon6:c.C865T:p.P289S,GSDMB:NM_00104247 1:exon8:c.C892T:p.P298S,GSDMB:NM_001165959:exon8:c.C904T :p.P302S,GSDMB:NM_001165958:exon9:c.C931T:p.P311S;GSDM B:NM_018530:exon6:c.C865T:p.P289S,GSDMB:NM_001042471:e xon8:c.C892T:p.P298S,GSDMB:NM_001165959:exon8:c.C904T:p. P302S,GSDMB:NM_001165958:exon9:c.C931T:p.P311S 30 rs2305480 rs2305479 0.85 GSDMB GSDMB:NM_018530:exon6:c.G844A:p.G282R,GSDMB:NM_0010 NA 42471:exon8:c.G871A:p.G291R,GSDMB:NM_001165959:exon8:c. G883A:p.G295R,GSDMB:NM_001165958:exon9:c.G910A:p.G304 R;GSDMB:NM_018530:exon6:c.G844A:p.G282R,GSDMB:NM_00 1042471:exon8:c.G871A:p.G291R,GSDMB:NM_001165959:exon8: c.G883A:p.G295R,GSDMB:NM_001165958:exon9:c.G910A:p.G30 4R;GSDMB:NM_018530:exon6:c.G844A:p.G282R,GSDMB:NM_0 01042471:exon8:c.G871A:p.G291R,GSDMB:NM_001165959:exon8 :c.G883A:p.G295R,GSDMB:NM_001165958:exon9:c.G910A:p.G30 4R;GSDMB:NM_018530:exon6:c.G844A:p.G282R,GSDMB:NM_0 01042471:exon8:c.G871A:p.G291R,GSDMB:NM_001165959:exon8 :c.G883A:p.G295R,GSDMB:NM_001165958:exon9:c.G910A:p.G30 4R 30 rs2305480 rs11557467 0.82 ZPBP2 ZPBP2:NM_198844:exon4:c.G452T:p.S151I,ZPBP2:NM_199321:ex NA on5:c.G518T:p.S173I;ZPBP2:NM_198844:exon4:c.G452T:p.S151I, ZPBP2:NM_199321:exon5:c.G518T:p.S173I;ZPBP2:NM_198844:e xon4:c.G452T:p.S151I,ZPBP2:NM_199321:exon5:c.G518T:p.S173I; ZPBP2:NM_198844:exon4:c.G452T:p.S151I,ZPBP2:NM_199321:ex on5:c.G518T:p.S173I

Functional characterization of the new 8q21 Asthma risk locus | 79

30 rs3894194 rs3894194 1.00 GSDMA GSDMA:NM_178171:exon2:c.G53A:p.R18Q;GSDMA:NM_178171 NA :exon2:c.G53A:p.R18Q

30 rs7216389 rs2305479 0.93 GSDMB GSDMB:NM_018530:exon6:c.G844A:p.G282R,GSDMB:NM_0010 NA 42471:exon8:c.G871A:p.G291R,GSDMB:NM_001165959:exon8:c. G883A:p.G295R,GSDMB:NM_001165958:exon9:c.G910A:p.G304 R;GSDMB:NM_018530:exon6:c.G844A:p.G282R,GSDMB:NM_00 1042471:exon8:c.G871A:p.G291R,GSDMB:NM_001165959:exon8: c.G883A:p.G295R,GSDMB:NM_001165958:exon9:c.G910A:p.G30 4R;GSDMB:NM_018530:exon6:c.G844A:p.G282R,GSDMB:NM_0 01042471:exon8:c.G871A:p.G291R,GSDMB:NM_001165959:exon8 :c.G883A:p.G295R,GSDMB:NM_001165958:exon9:c.G910A:p.G30 4R;GSDMB:NM_018530:exon6:c.G844A:p.G282R,GSDMB:NM_0 01042471:exon8:c.G871A:p.G291R,GSDMB:NM_001165959:exon8 :c.G883A:p.G295R,GSDMB:NM_001165958:exon9:c.G910A:p.G30 4R 30 rs7216389 rs11557467 0.89 ZPBP2 ZPBP2:NM_198844:exon4:c.G452T:p.S151I,ZPBP2:NM_199321:ex NA on5:c.G518T:p.S173I;ZPBP2:NM_198844:exon4:c.G452T:p.S151I, ZPBP2:NM_199321:exon5:c.G518T:p.S173I;ZPBP2:NM_198844:e xon4:c.G452T:p.S151I,ZPBP2:NM_199321:exon5:c.G518T:p.S173I; ZPBP2:NM_198844:exon4:c.G452T:p.S151I,ZPBP2:NM_199321:ex on5:c.G518T:p.S173I

30 rs8079416 rs3894194 0.87 GSDMA GSDMA:NM_178171:exon2:c.G53A:p.R18Q;GSDMA:NM_178171 NA :exon2:c.G53A:p.R18Q

Functional characterization of the new 8q21 Asthma risk locus | 80

Supplementary Table 7. Genome-wide association studies of gene expression levels queried to identify cis-acting eQTLs.

cis effects First author PMID Tissue/cell type Sample size N genes N associations P<2.3x10-9 Andiappan 26259071 NEUTROPHILS 114 310 3776 Barreiro 22233810 DENDRITIC-NotInfected 65 36 36 DENDRITIC-TBInfected 65 216 216 Battle 24092820 WHOLE-BLOOD 922 7792 7793 WHOLE-BLOOD-ase 922 310 537 WHOLE-BLOOD-splice 922 532 818 Brumpton 27155841 LCLS 356 27 2139 Caliskan 25874939 PBMCS-baseline 98 173 173 PBMCS-rhinovirus 98 169 169 PBMCS-rhinovirus-reQTL 98 12 12 Davenport 26917434 LEUCOCYTES 265 879 7699 Dimas 19644074 LCLS 75 97 121 TCELLS 75 85 107 FIBROBLASTS 75 102 136 Di Narzo 27336838 WHOLE-BLOOD 149 1195 2220 Ding 21129726 LESIONAL-SKIN 57 15 295 NORMAL-SKIN 53 19 350 UNINVOLVED-SKIN 53 12 160 Dixon 17873877 LCLS 400 634 6735 Fairfax 2012 22446964 BCELLS 283 236 1723 MONOCYTES 283 549 4051 Fairfax 2014 24604202 MONOCYTES-IFN 367 2048 21366 MONOCYTES-LPS2 261 686 4658 MONOCYTES-LPS24 322 1412 10740 MONOCYTES-NAIVE 414 2024 24676 Fehrmann 21829388 WHOLE-BLOOD 1469 2944 23614 Ferraro 24610777 Tconv 65 26 118 Tregs 65 24 105 Franco 23878721 WHOLE-BLOOD-Influenza 247 65 329 Lappalainen 24037378 LCLS 373 3525 1397237 Grundberg 22941192 LCLS 856 1550 68025 SKIN 856 879 31390 Functional characterization of the new 8q21 Asthma risk locus | 81

GTEx 25954001 FIBROBLASTS 272 3910 368252 LCLS 114 1062 88242 LUNG 278 3148 335019 SKIN 302 4026 613093 SPLEEN 89 926 74864 WHOLE-BLOOD 338 3331 343913 Hao 23209423 LUNG 1111 4177 6339 Huang 25951796 LCLs 368 2196 3091 PBMCs 240 1939 2617 SKIN 110 384 453 Jansen Under Review WHOLE-BLOOD 4896 4377 7460 Kasela 28248954 CD4TCELLS 293 1256 121508 CD8TCELLS 293 947 83337 Kim 25327457 MONOCYTES-Baseline 137 681 5669 MONOCYTES-Differential 137 62 377 MONOCYTES-LPS 137 554 4802 Kukurba 27197214 WHOLE-BLOOD 922 90 8333 Lee 24604203 DENDRITIC-Baseline 528 82 82 DENDRITIC-Flu 342 105 105 DENDRITIC-Flu-delta 342 44 44 DENDRITIC-IFNb 284 86 86 DENDRITIC-IFNb-delta 284 23 24 DENDRITIC-LPS 356 96 96 DENDRITIC-LPS-delta 356 31 31 LloydJones 28065468 WHOLE-BLOOD 2765 5669 1373201 Luo 26102239 SMALL-AIRWAYS 105 152 174 Murphy 20833654 CD4-TCELLS 200 294 986 Naranbhai 26151758 NEUTROPHILS 101 493 547 Nedelec 27768889 MACROPHAGES-baseline 95 451 451 MACROPHAGES-baseline-asQTL 95 634 951 MACROPHAGES-listeria 95 422 422 MACROPHAGES-listeria-asQTL 95 557 827 MACROPHAGES-listeria-reQTL 95 93 93 MACROPHAGES-salmonella 95 433 434 MACROPHAGES-salmonella-asQTL 95 487 741 MACROPHAGES-salmonella-reQTL 95 199 199 Peters 27015630 BCELLS 80 66 318 Functional characterization of the new 8q21 Asthma risk locus | 82

CD4-TCELLS 121 396 2187 CD8-TCELLS 108 277 1484 MONOCYTES 124 564 3254 NEUTROPHILS 121 341 1766 Quach 27768888 MONOCYTES-baseline 100 467 469 MONOCYTES-IAV 100 366 366 MONOCYTES-LPS 100 464 465 MONOCYTES-Pam3CSK4 100 536 537 MONOCYTES-R848 100 478 479 Raj 24786080 CD4-TCELLS 407 1898 150019 MONOCYTES 401 2494 196870 Walsh 27140173 WHOLE-BLOOD 377 3777 546913 Westra 24013639 WHOLE-BLOOD 5311 4285 343167 Yao 28285768 WHOLEBLOOD 5257 1681 93834 Ye 25214635 CD4-TCELLS-48h328 348 51 51 CD4-TCELLS-48hTh17 348 45 45 CD4-TCELLS-4h328 348 24 24 CD4-TCELLS-4hIFNb 348 27 27 CD4-TCELLS-UNST 348 19 19 Zeller 20502693 MONOCYTES 1490 2322 31353 Zhernakova 27918533 WHOLEBLOOD-exon-primary 2116 15028 88928 WHOLEBLOOD-exonratio-primary 2116 6064 20222 WHOLEBLOOD-gene-contextspecific 2116 14670 27186 WHOLEBLOOD-gene-primary 2116 15158 3669632 WHOLEBLOOD-polyAratio-primary 2116 1286 1843

Functional characterization of the new 8q21 Asthma risk locus | 83

Supplementary Table 8. eQTLs in LD with GWAS SNP reported in tissues and/or studies other than that listed in Table 4.

Index GWAS SNP Gene eQTL reported in other studies/tissues 1 rs662064 PEX14 Nedelec_MACROPHAGES-salmonella-reQTL:rs585870;Hao_LUNG:rs607941;Fairfax_MONOCYTES-NAIVE:rs636291

2 rs4129267 IL6R Battle_WHOLEBLOOD-splice:rs4537545;LloydJones_WHOLEBLOOD:rs4845372;Raj_CD4TCELLS:rs4133213

3 rs4233366 FCER1G Fairfax_MONOCYTES- NAIVE:rs4233366;Raj_MONOCYTES:rs4233366;Davenport_LEUCOCYTES:rs4233366;Fairfax_MONOCYTES:rs4233366;Zhernakova_W HOLEBLOOD-exon-primary:rs2070901;Zhernakova_WHOLEBLOOD-gene-primary:rs2070901;Zhernakova_WHOLEBLOOD-gene- contextspecific:rs2070901;Zhernakova_WHOLEBLOOD-exon-primary:rs2070901;Zhernakova_WHOLEBLOOD-exon- primary:rs2070901;Zhernakova_WHOLEBLOOD-exon-primary:rs2070901;Zhernakova_WHOLEBLOOD-exon- primary:rs2070901;LloydJones_WHOLEBLOOD:rs2070901;Battle_WHOLEBLOOD:rs2070901;Walsh_WHOLEBLOOD:rs2070901;Walsh_ WHOLEBLOOD:rs2070901;Battle_WHOLEBLOOD-ase:rs2070901;Naranbhai_NEUTROPHILS:rs2070901;Battle_WHOLEBLOOD- ase:rs2070901;GTE_WHOLEBLOOD:rs2070901;Dinarzo_WHOLEBLOOD:rs2070901;GTE_SKIN:rs2070901;Hao_LUNG:rs2070902

3 rs4233366 USF1 Westra_WHOLEBLOOD:rs2070902;LloydJones_WHOLEBLOOD:rs2070902;Westra_WHOLEBLOOD:rs2070902

4 rs1723018 CD247 Raj_CD4TCELLS:rs2056626;Raj_CD4TCELLS:rs2056626;Battle_WHOLEBLOOD:rs2995091;Westra_WHOLEBLOOD:rs1214596;Westra _WHOLEBLOOD:rs1214596;Zhernakova_WHOLEBLOOD-gene-primary:rs34276574;Zhernakova_WHOLEBLOOD-gene- contextspecific:rs34276574;Jansen_WHOLEBLOOD:rs35014055;LloydJones_WHOLEBLOOD:rs2949669;LloydJones_WHOLEBLOOD:rs2 949669

7 rs6683383 MYBPH Fairfax_MONOCYTES-NAIVE:rs7555556

7 rs6683383 ADORA1 Fairfax_MONOCYTES:rs10920570;Battle_WHOLEBLOOD:rs7555556;Zhernakova_WHOLEBLOOD-exon- primary:rs7538512;Fairfax_MONOCYTES- NAIVE:rs7555556;LloydJones_WHOLEBLOOD:rs7538512;Raj_MONOCYTES:rs7555556;Zhernakova_WHOLEBLOOD-gene- contextspecific:rs722915;Zhernakova_WHOLEBLOOD-exonratio-primary:rs903361;Hao_LUNG:rs903361

7 rs6683383 PPFIA4 Zhernakova_WHOLEBLOOD-exon-primary:rs6696205;Zhernakova_WHOLEBLOOD-gene- primary:rs6427990;Zhernakova_WHOLEBLOOD-exon-primary:rs6696205

7 rs6683383 CHIT1 Zhernakova_WHOLEBLOOD-gene-primary:rs6683383;Zhernakova_WHOLEBLOOD-gene-contextspecific:rs6683383

7 rs6683383 RP11-335O13.7 Zhernakova_WHOLEBLOOD-gene-primary:rs722915;Zhernakova_WHOLEBLOOD-gene-contextspecific:rs722915

12 rs4833095 TLR1 Zhernakova_WHOLEBLOOD-gene-primary:rs5743604;Zhernakova_WHOLEBLOOD-gene- contextspecific:rs5743604;Westra_WHOLEBLOOD:rs2101521;Jansen_WHOLEBLOOD:rs2101521

14 rs1438673 TSLP GTE_FIBROBLASTS:rs10073816;Zhernakova_WHOLEBLOOD-exon-primary:rs2289277;Zhernakova_WHOLEBLOOD-gene- contextspecific:rs2289277;Zhernakova_WHOLEBLOOD-gene-primary:rs2289277;GTE_SKIN:rs2289277;Zhernakova_WHOLEBLOOD- gene-primary:rs2289277;Zhernakova_WHOLEBLOOD-gene-contextspecific:rs2289277;GTE_SKIN:rs2289277;Huang_SKIN:rs6594497

14 rs1438673 CTC-551A13.2 Zhernakova_WHOLEBLOOD-gene-contextspecific:rs2289277;Zhernakova_WHOLEBLOOD-gene-primary:rs2289277

Functional characterization of the new 8q21 Asthma risk locus | 84

16 rs166079 NDFIP1 Zhernakova_WHOLEBLOOD-gene-primary:rs12655465;Zhernakova_WHOLEBLOOD-exon- primary:rs2338821;Yao_WHOLEBLOOD:rs6863411;Fairfax_MONOCYTES-NAIVE:rs4912622;Raj_MONOCYTES:rs10875596

18 rs9273373 HLA-DQA1 Geuvadis_LCLS:rs9274660;Geuvadis_LCLS:rs9274660;Geuvadis_LCLS:rs9274660;Geuvadis_LCLS:rs9274660;Geuvadis_LCLS:rs9274660; Geuvadis_LCLS:rs9274660;Geuvadis_LCLS:rs9274660;Geuvadis_LCLS:rs9274660;Geuvadis_LCLS:rs9274660;Geuvadis_LCLS:rs9274660; Geuvadis_LCLS:rs9274660;Geuvadis_LCLS:rs9274660;GTE_SKIN:rs9273382;GTE_SKIN:rs9273382;Quach_MONOCYTES- baseline:rs9273479;Quach_MONOCYTES-IAV:rs9273479;Quach_MONOCYTES- R848:rs9273479;GTE_SKIN:rs9273382;GTE_SKIN:rs9273382

18 rs9272346 HLA-DQB1 Luo_SMALLAIRWAYS:rs9272346;Ding_SKIN- NORMAL:rs9272346;Geuvadis_LCLS:rs9274660;Geuvadis_LCLS:rs9274660;Geuvadis_LCLS:rs9274660;Geuvadis_LCLS:rs9274660;Geuv adis_LCLS:rs9274660;Kim_MONOCYTES-Baseline:rs1063355;Kim_MONOCYTES- LPS:rs1063355;Geuvadis_LCLS:rs9274660;Dimas_LCLS:rs3817963;LloydJones_WHOLEBLOOD:rs3828789;LloydJones_WHOLEBLOOD :rs3828789;Geuvadis_LCLS:rs9274660;Geuvadis_LCLS:rs9274660;Geuvadis_LCLS:rs9274660;Geuvadis_LCLS:rs9274660;Geuvadis_LCLS :rs9274660;Kim_MONOCYTES-Baseline:rs1063355;Kim_MONOCYTES- LPS:rs1063355;Geuvadis_LCLS:rs9274660;LloydJones_WHOLEBLOOD:rs3828789;LloydJones_WHOLEBLOOD:rs3828789;Geuvadis_LC LS:rs9274660;Geuvadis_LCLS:rs9274660;Geuvadis_LCLS:rs9274660;Geuvadis_LCLS:rs9274660;Geuvadis_LCLS:rs9274660;Kim_MONO CYTES-Baseline:rs1063355;Kim_MONOCYTES- LPS:rs1063355;Geuvadis_LCLS:rs9274660;Luo_SMALLAIRWAYS:rs9272346;Ding_SKIN- NORMAL:rs9272346;LloydJones_WHOLEBLOOD:rs3828789;Luo_SMALLAIRWAYS:rs9272346;LloydJones_WHOLEBLOOD:rs382878 9;Ding_SKIN-NORMAL:rs9272346;Zhernakova_WHOLEBLOOD-gene-primary:rs9274669;Zhernakova_WHOLEBLOOD-gene- primary:rs9274669

18 rs9273373 HLA-DQA2 Quach_MONOCYTES-LPS:rs9272545;Quach_MONOCYTES-LPS:rs9272545

18 rs9272346 HLA-DRB5 Zeller_MONOCYTES:rs9268429;Yao_WHOLEBLOOD:rs9272320

27 rs10519068 RP11-554D20.1 Zhernakova_WHOLEBLOOD-gene-primary:rs11633029;Zhernakova_WHOLEBLOOD-gene-primary:rs11633029

28 rs56375023 AAGAB Zhernakova_WHOLEBLOOD-gene-contextspecific:rs17293632

30 rs4794820 GSDMB Battle_WHOLEBLOOD:rs12950743;Fehrmann_WHOLEBLOOD:rs11557467;Jansen_WHOLEBLOOD:rs11557467;Fehrmann_WHOLEBL OOD:rs11557467;Geuvadis_LCLS:rs12950743;Geuvadis_LCLS:rs12950743;Geuvadis_LCLS:rs12950743;Geuvadis_LCLS:rs12950743;Geu vadis_LCLS:rs12950743;Geuvadis_LCLS:rs12950743;Geuvadis_LCLS:rs12950743;Zhernakova_WHOLEBLOOD-exon- primary:rs7359623;Battle_WHOLEBLOOD-splice:rs2305479;Zhernakova_WHOLEBLOOD-exonratio- primary:rs2305479;Raj_CD4TCELLS:rs2872507;Kasela_CD8TCELLS:rs2305479;Raj_CD4TCELLS:rs2872507;Kasela_CD8TCELLS:rs230 5479;Peters_CD8TCELLS:rs2305479;Davenport_LEUCOCYTES:rs10852936;LloydJones_WHOLEBLOOD:rs12936231;LloydJones_WHO LEBLOOD:rs12936231;Raj_CD4TCELLS:rs2872507;LloydJones_WHOLEBLOOD:rs12936231;Raj_CD4TCELLS:rs2872507;GTE_WHOL EBLOOD:rs8067378;Huang_PBMCS:rs8067378;Dinarzo_WHOLEBLOOD:rs7224129;Huang_LCLS:rs8067378;GTE_LCLS:rs12936231;Gr undberg_LCLS:rs8067378;Grundberg_LCLS:rs8067378;Davenport_LEUCOCYTES:rs10852936;Walsh_WHOLEBLOOD:rs8074437;Walsh_ WHOLEBLOOD:rs8074437;Kasela_CD4TCELLS:rs4795397;Kasela_CD4TCELLS:rs4795397;Walsh_WHOLEBLOOD:rs8074437;Kasela_ CD4TCELLS:rs4795397;Kasela_CD4TCELLS:rs4795397;GTE_FIBROBLASTS:rs8071789;Battle_WHOLEBLOOD- splice:rs2305479;Zhernakova_WHOLEBLOOD-exonratio- primary:rs2305479;Kasela_CD8TCELLS:rs2305479;Kasela_CD8TCELLS:rs2305479;Peters_CD8TCELLS:rs2305479;Hao_LUNG:rs229040 0;Hao_LUNG:rs2290400;Battle_WHOLEBLOOD:rs12950743;Fehrmann_WHOLEBLOOD:rs11557467;Jansen_WHOLEBLOOD:rs1155746 7;Fehrmann_WHOLEBLOOD:rs11557467;Geuvadis_LCLS:rs12950743;GTE_SPLEEN:rs4065275;Geuvadis_LCLS:rs12950743;Geuvadis_L CLS:rs12950743;Geuvadis_LCLS:rs12950743;Geuvadis_LCLS:rs12950743;Geuvadis_LCLS:rs12950743;Geuvadis_LCLS:rs12950743;Zhern akova_WHOLEBLOOD-exon- primary:rs7359623;Westra_WHOLEBLOOD:rs907091;LloydJones_WHOLEBLOOD:rs12936231;LloydJones_WHOLEBLOOD:rs12936231;

Functional characterization of the new 8q21 Asthma risk locus | 85

LloydJones_WHOLEBLOOD:rs12936231;GTE_WHOLEBLOOD:rs8067378;Huang_PBMCS:rs8067378;Huang_LCLS:rs8067378;GTE_LC LS:rs12936231;Grundberg_LCLS:rs8067378;Grundberg_LCLS:rs8067378;Kasela_CD4TCELLS:rs4795397;Kasela_CD4TCELLS:rs4795397 ;GTE_FIBROBLASTS:rs8071789;Battle_WHOLEBLOOD-splice:rs2305479;Zhernakova_WHOLEBLOOD-exonratio- primary:rs2305479;Raj_CD4TCELLS:rs2872507;Kasela_CD8TCELLS:rs2305479;Raj_CD4TCELLS:rs2872507;Kasela_CD8TCELLS:rs230 5479;Peters_CD8TCELLS:rs2305479;Davenport_LEUCOCYTES:rs10852936;Battle_WHOLEBLOOD- splice:rs2305479;Zhernakova_WHOLEBLOOD-exonratio- primary:rs2305479;Kasela_CD8TCELLS:rs2305479;Kasela_CD8TCELLS:rs2305479;Dinarzo_WHOLEBLOOD:rs7224129;Peters_CD8TCE LLS:rs2305479;Walsh_WHOLEBLOOD:rs8074437;Walsh_WHOLEBLOOD:rs8074437;Walsh_WHOLEBLOOD:rs8074437;Battle_WHOL EBLOOD:rs12950743;Battle_WHOLEBLOOD:rs12950743;Fehrmann_WHOLEBLOOD:rs11557467;Fehrmann_WHOLEBLOOD:rs1155746 7;Jansen_WHOLEBLOOD:rs11557467;Jansen_WHOLEBLOOD:rs11557467;Fehrmann_WHOLEBLOOD:rs11557467;Fehrmann_WHOLEB LOOD:rs11557467;Geuvadis_LCLS:rs12950743;Geuvadis_LCLS:rs12950743;Geuvadis_LCLS:rs12950743;Geuvadis_LCLS:rs12950743;Ge uvadis_LCLS:rs12950743;Geuvadis_LCLS:rs12950743;Geuvadis_LCLS:rs12950743;Geuvadis_LCLS:rs12950743;Geuvadis_LCLS:rs129507 43;Geuvadis_LCLS:rs12950743;Geuvadis_LCLS:rs12950743;Geuvadis_LCLS:rs12950743;Geuvadis_LCLS:rs12950743;Geuvadis_LCLS:rs1 2950743;Zhernakova_WHOLEBLOOD-exon-primary:rs7359623;Zhernakova_WHOLEBLOOD-exon- primary:rs7359623;Westra_WHOLEBLOOD:rs907091;Hao_LUNG:rs2290400

30 rs11655198 ORMDL3 Grundberg_LCLS:rs12950743;Peters_CD8TCELLS:rs11557467;Raj_CD4TCELLS:rs4795398;Jansen_WHOLEBLOOD:rs36038753;Zhernak ova_WHOLEBLOOD-exon- primary:rs8067378;Battle_WHOLEBLOOD:rs8067378;LloydJones_WHOLEBLOOD:rs12936231;Yao_WHOLEBLOOD:rs8067378;Geuvadi s_LCLS:rs8067378;Geuvadis_LCLS:rs8067378;Raj_CD4TCELLS:rs4795398;Geuvadis_LCLS:rs8067378;GTE_WHOLEBLOOD:rs9303281 ;Huang_LCLS:rs8067378;GTE_LCLS:rs12936231;Jansen_WHOLEBLOOD:rs36038753;Huang_PBMCS:rs8067378;Geuvadis_LCLS:rs8067 378;Geuvadis_LCLS:rs8067378;Dixon_LCLS:rs7219923;Huang_PBMCS:rs8067378;Huang_LCLS:rs8067378;GTE_SPLEEN:rs7216558;Dix on_LCLS:rs7219923;Walsh_WHOLEBLOOD:rs1453559;Walsh_WHOLEBLOOD:rs1453559;Kasela_CD4TCELLS:rs9303279;Kasela_CD4 TCELLS:rs9303279;Dimas_LCLS:rs9303277;Grundberg_LCLS:rs12950743;Peters_CD8TCELLS:rs11557467;GTE_FIBROBLASTS:rs1260 3332;Westra_WHOLEBLOOD:rs907091;Zhernakova_WHOLEBLOOD-exon- primary:rs8067378;Battle_WHOLEBLOOD:rs8067378;LloydJones_WHOLEBLOOD:rs12936231;Yao_WHOLEBLOOD:rs8067378;Geuvadi s_LCLS:rs8067378;Geuvadis_LCLS:rs8067378;Geuvadis_LCLS:rs8067378;Huang_LCLS:rs8067378;GTE_LCLS:rs12936231;Huang_PBM CS:rs8067378;Geuvadis_LCLS:rs8067378;Geuvadis_LCLS:rs8067378;Huang_PBMCS:rs8067378;Huang_LCLS:rs8067378;Fehrmann_WH OLEBLOOD:rs4795405;Zhernakova_WHOLEBLOOD-exonratio- primary:rs4795405;Murphy_CD4TCELLS:rs4795405;Raj_CD4TCELLS:rs4795398;Walsh_WHOLEBLOOD:rs1453559;Jansen_WHOLEBL OOD:rs36038753;Walsh_WHOLEBLOOD:rs1453559;GTE_WHOLEBLOOD:rs9303281;Kasela_CD4TCELLS:rs9303279;Dixon_LCLS:rs7 219923;GTE_LUNG:rs3816470;GTE_SPLEEN:rs7216558;Dixon_LCLS:rs7219923;Dimas_LCLS:rs9303277;Fehrmann_WHOLEBLOOD:rs 4795405;Zhernakova_WHOLEBLOOD-exonratio- primary:rs4795405;Murphy_CD4TCELLS:rs4795405;Fehrmann_WHOLEBLOOD:rs4795405;Grundberg_LCLS:rs12950743;Grundberg_LC LS:rs12950743;Zhernakova_WHOLEBLOOD-exonratio- primary:rs4795405;Peters_CD8TCELLS:rs11557467;Peters_CD8TCELLS:rs11557467;Murphy_CD4TCELLS:rs4795405;Fehrmann_WHOL EBLOOD:rs4795405;Zhernakova_WHOLEBLOOD-exonratio- primary:rs4795405;Murphy_CD4TCELLS:rs4795405;Westra_WHOLEBLOOD:rs907091

30 rs11655198 ZPBP2 Hao_LUNG:rs10445308;Hao_LUNG:rs10445308

30 rs2271308 STARD3 Raj_MONOCYTES:rs1053651

30 rs11655198 RP11-94L15.2 Zhernakova_WHOLEBLOOD-gene-contextspecific:rs11658278;Zhernakova_WHOLEBLOOD-exon- primary:rs12936231;Zhernakova_WHOLEBLOOD-gene-contextspecific:rs11658278;Zhernakova_WHOLEBLOOD-exon- primary:rs12936231;Zhernakova_WHOLEBLOOD-gene-contextspecific:rs11658278;Zhernakova_WHOLEBLOOD-gene- contextspecific:rs11658278

Functional characterization of the new 8q21 Asthma risk locus | 86

30 rs11655198 IKZF3 Zhernakova_WHOLEBLOOD-gene-contextspecific:rs2305479;Zhernakova_WHOLEBLOOD-exon- primary:rs7224129;Battle_WHOLEBLOOD:rs8067378;Zhernakova_WHOLEBLOOD-gene- contextspecific:rs2305479;Yao_WHOLEBLOOD:rs9303277;LloydJones_WHOLEBLOOD:rs907091;Jansen_WHOLEBLOOD:rs9909593;Ja nsen_WHOLEBLOOD:rs9909593;Battle_WHOLEBLOOD:rs8067378;Zhernakova_WHOLEBLOOD-gene- contextspecific:rs2305479;Zhernakova_WHOLEBLOOD-exon-primary:rs7224129;Zhernakova_WHOLEBLOOD-gene- contextspecific:rs2305479;Yao_WHOLEBLOOD:rs9303277;LloydJones_WHOLEBLOOD:rs907091

Functional characterization of the new 8q21 Asthma risk locus | 87

Supplementary Table 9. Terms used in PubMed query* for each of the likely target genes. Gene Query terms AAGAB KPPP1 OR "PPKP1" OR "PPKP1A" OR "p34" OR "AAGAB" OR "alpha- and gamma-adaptin binding protein" AC007278.3 AC007278.3

ADAMTS4 ADAMTS-2 OR "ADAMTS-4" OR "ADMP-1" OR "ADAMTS4" OR "ADAM metallopeptidase with thrombospondin type 1 motif 4" ADORA1 RDC7 OR "ADORA1" OR "adenosine A1 receptor" B4GALT3 beta4Gal-T3 OR "B4GALT3" OR "beta-1,4-galactosyltransferase 3" CAMK4 CaMK IV OR "CaMK-GR" OR "caMK" OR "CAMK4" OR "calcium/calmodulin dependent protein kinase IV" CD247 CD3-ZETA OR "CD3H" OR "CD3Q" OR "CD3Z" OR "IMD25" OR "T3Z" OR "TCRZ" OR "CD247" OR "CD247 molecule" CHIT1 CHI3 OR "CHIT" OR "CHITD" OR "CHIT1" OR "chitinase 1" CTC- 551A13.2 CTC-551A13.2 CIDE-A OR "CIDEA" OR "DFF-45" OR "DFF1" OR "ICAD" OR "DFFA" OR "CIDEB" OR "CIDE-3" OR "CIDE3" OR "FPLD5" OR "FSP27" OR "CIDEC" OR "CICE" OR "CIDECP" OR "cell death-inducing DFFA-like effector a" OR "DNA fragmentation factor subunit alpha" OR "cell death-inducing DFFA-like effector b" OR "cell death inducing DFFA like effector c" OR "cell death-inducing DFFA-like effector c DFFA pseudogene" F11R CD321 OR "JAM1" OR "JAM-A" OR "F11R" OR "F11 receptor" FCER1G FCRG OR "FCER1G" OR "Fc fragment of IgE receptor Ig" GSDMA FKSG9 OR "GSDM" OR "GSDM1" OR "GSDMA" OR "gasdermin A" GSDMB GSDML OR "PP4052" OR "PRO2521" OR "GSDMB" OR "gasdermin B" HLA-DQA1 CELIAC1 OR "DQ-A1" OR "HLA-DQA" OR "HLA-DQA1" OR "major histocompatibility complex, class II, DQ alpha 1" HLA-DQA2 DX-ALPHA OR "HLA-DXA" OR "HLA-DQA2" OR "major histocompatibility complex, class II, DQ alpha 2" HLA-DQB1 OR "CELIAC1" OR "HLA-DQB" OR "IDDM1" OR "HLA-DQB1-AS1" OR "G protein subunit alpha o1" OR "major histocompatibility complex, class II, DQ beta 1" OR "major histocompatibility complex, class II, DQ beta 2" OR "HLA-DQB1 antisense RNA 1" HLA-DQB1 OR "GNAO1" OR "HLA-DQB2" HLA-DQB1- AS1 HLA-DQB1-AS1 OR "HLA-DQB1 antisense RNA 1" HLA-DQB2 HLA-DQB1 OR "HLA-DXB" OR "HLA-DQB2" OR "major histocompatibility complex, class II, DQ beta 2" HLA-DRB5 HLA-DRB5 OR "major histocompatibility complex, class II, DR beta 5" HLA-DRB6 HLA-DRB6 OR "major histocompatibility complex, class II, DR beta 6 (pseudogene)" IKZF3 AIO OR "AIOLOS" OR "ZNFN1A3" OR "IKZF3" OR "IKAROS family zinc finger 3" ACPL OR "CD218b" OR "CDw218b" OR "IL-18R-beta" OR "IL-18RAcP" OR "IL-18Rbeta" OR "IL-1R-7" OR "IL-1R7" OR "IL-1RAcPL" OR IL18RAP "IL18RB" OR "IL18RAP" OR "interleukin 18 receptor accessory protein" IL1RL1 DER4 OR "FIT-1" OR "IL33R" OR "ST2" OR "ST2L" OR "ST2V" OR "IL1RL1" OR "interleukin 1 receptor like 1"

IL6R CD126 OR "IL-6R-1" OR "IL-6RA" OR "IL6Q" OR "IL6RA" OR "IL6RQ" OR "gp80" OR "IL6R" OR “IL-6R” OR "interleukin 6 receptor"

Functional characterization of the new 8q21 Asthma risk locus | 88

MFSD9 MFSD9 OR "major facilitator superfamily domain containing 9" MICB PERB11.2 OR "MICB" OR "MHC class I polypeptide-related sequence B" MYBPH MYBPH OR "myosin binding protein H" NDFIP1 N4WBP5 OR "NDFIP1" OR "Nedd4 family interacting protein 1" ORMDL3 ORMDL3 OR "ORMDL sphingolipid biosynthesis regulator 3" PAG1 PAG1 OR "phosphoprotein membrane anchor with glycosphingolipid microdomains 1" PEX14 NAPP2 OR "PBD13A" OR "Pex14p" OR "dJ734G22.2" OR "PEX14" OR "peroxisomal biogenesis factor 14" PGAP3 AGLA546 OR "CAB2" OR "PERLD1" OR "PP1498" OR "hCOS16" OR "PGAP3" OR "post-GPI attachment to proteins 3" PPFIA4 PPFIA4 OR "PTPRF interacting protein alpha 4" PPOX PPOX OR "V290M" OR "hypocretin neuropeptide precursor" OR "protoporphyrinogen oxidase" PPP1R1B DARPP-32 OR "DARPP32" OR "PPP1R1B" OR "protein phosphatase 1 regulatory inhibitor subunit 1B" RP11- 335O13.7 RP11-335O13.7 RP11- 554D20.1 RP11-554D20.1 RP11-94L15.2 RP11-94L15.2 SLC22A5 CDSP OR "OCTN2" OR "SLC22A5" OR "solute carrier family 22 member 5" CAB1 OR "MLN64" OR "es64" OR "STARD3" OR "MENTHO" OR "STARD3NL" OR "StAR related lipid transfer domain containing 3" OR STARD3 "STARD3 N-terminal like" ABC18 OR "ABCB3" OR "APT2" OR "D6S217E" OR "PSF-2" OR "PSF2" OR "RING11" OR "TAP2" OR "transporter 2, ATP binding cassette TAP2 subfamily B member" OR "SEC14 like lipid binding 3" OR "SEC14L3" TLR1 CD281 OR "TIL" OR "TIL. LPRS5" OR "rsc786" OR "TLR1" OR "toll like receptor 1" CD134L OR "CD252" OR "GP34" OR "OX-40L" OR "OX40L" OR "TNLG2B" OR "TXGP1" OR "TNFSF4" OR "tumor necrosis factor TNFSF4 superfamily member 4" TOMM40L TOMM40B OR "TOMM40L" OR "translocase of outer mitochondrial membrane 40 like" TSLP TSLP OR "thymic stromal lymphopoietin" USF1 FCHL OR "FCHL1" OR "HYPLIP1" OR "MLTF" OR "MLTFI" OR "USF1" OR "upstream transcription factor 1" WDR36 GLC1G OR "TA-WDRP" OR "TAWDRP" OR "UTP21" OR "WDR36" OR "WD repeat domain 36" ZPBP2 ZPBPL OR "ZPBP2" OR "zona pellucida binding protein 2" * The PubMed query used was: (asthma OR rhinitis OR eczema OR atopic OR dermatitis OR allergy OR allergi* OR hayfever OR "hay fever") AND (gene_name/aliases listed above)

Functional characterization of the new 8q21 Asthma risk locus | 89

Supplementary Table 10. Summary of known functions of likely target genes of published asthma risk variants.

Gene Description Function References (PMID) AAGAB Cytosolic protein that binds alpha- and gamma- Vesicle transport and epidermal proliferation 23064416 adaptin AC007278.3 Intronic sense RNA Unknown - ADAMTS4 ADAM Metallopeptidase With Thrombospondin Degrades aggrecan, a major proteoglycan of 10811839, 11520168 Type 1 Motif 4 cartilage. Induced by IL-1 and TNF ADORA1 Adenosine A1 Receptor Adenosine receptor that may play an anti- 15630442, 26593641 inflammatory role by regulating the production of proinflammatory mediators in the lung. Adenosine has been shown a key neuromodulators controlling respiration B4GALT3 Beta-1,4-Galactosyltransferase 3 Nucleotide synthesis, activation of β1-integrin 9597550, 11588157, 25659296 CAMK4 Calcium/Calmodulin Dependent Protein Kinase Thymocyte maturation, generation and function of 17909078, 22942433, 8065343, IV regulatory T cells, mediates DC survival, regulates 11274168 the activity of transcriptional activators involved in immune response and inflammation CD247 CD247 Molecule Key component of the T-cell receptor-CD3 12507424, 11390434, 26542031 complex involved in intracellular signal transduction following antigen recognition, involved in NK cell development and function CHIT1 Chitinase 1 Enzyme secreted by activated human macrophages 7592832, 21674664 that plays a role in the degradation of chitin- containing pathogens. CTC-551A13.2 Antisense RNA overlapping TSLP exon 2 and Unknown - WDR36 exon 1 DFFA DNA Fragmentation Factor Subunit Alpha Mediates apoptosis 10894162, 10601318 F11R F11 Receptor Involved in epithelial tight junction formation, 11489913, 10753840, 11812992 platelet adhesion and aggregation, leukocyte transendothelial migration FCER1G Fc Fragment Of IgE Receptor Ig Component of the high affinity IgE receptor that 8611682, 8551243, 8313472 plays a critical role in the receptor assembly, cell surface expression and signal transduction. GSDMA Gasdermin A Autophagy and cell death 17471240, 25825937

Functional characterization of the new 8q21 Asthma risk locus | 90

GSDMB Gasdermin B Essential for caspase-11-dependent pyroptosis and 26375259, 26375003, 27281216, interleukin-1β maturation. Regulates AHR and 27799535 airway remodeling without airway inflammation. HLA-DQA1 Major Histocompatibility Complex, Class II, DQ Forms heterodimer with HLA-DQB1, involved in 16557259 Alpha 1 antigen presentation HLA-DQA2 Major Histocompatibility Complex, Class II, DQ Forms heterodimer with HLA-DQB2, expressed in 22407913 Alpha 2 Langerhans cells HLA-DQB1 Major Histocompatibility Complex, Class II, DQ Forms heterodimer with HLA-DQA1, involved in 16557259 Beta 1 antigen presentation HLA-DQB1-AS1 Antisense RNA Unknown - HLA-DQB2 Major Histocompatibility Complex, Class II, DQ Forms heterodimer with HLA-DQA2, expressed in 22407913 Beta 2 Langerhans cells HLA-DRB5 Major Histocompatibility Complex, Class II, DR Component of the DR heterodimer involved in 18395261, 22826457 Beta 5 antigen presentation. Paralog of HLA-DRB1. HLA-DRB6 Major Histocompatibility Complex, Class II, DR Unknown 8248165, 22826457 Beta 6 IKZF3 IKAROS Family Zinc Finger 3 Key regulator of lymphocyte differentiation, 9806640, 14718515, 10369681 involved in the regulation of BCL2 expression and T-cell apoptosis. IL18RAP Interleukin 18 Receptor Accessory Protein Essential for IL18-dependent signal transduction 9792649 (NF-kappa-B and JNK activation) IL1RL1 Interleukin 1 Receptor Like 1 Negatively regulates type I interleukin 1 receptor 15004556, 9618516, 27699235 (IL-1RI) and Toll-like receptor 4 (TLR4) signaling, key regulator of endotoxin tolerance, critical for Th2 effector function IL6R Interleukin 6 Receptor Receptor for IL-6, a pleiotropic cytokine playing 12829785, 15668741, 25898198, key roles in cell growth, differentiation and immune 18845487 response regulation. MFSD9 Major Facilitator Superfamily Domain Containing Member of the major facilitator superfamily with 22458847 9 members involved in small solute transport accross the cell membrane MICB MHC Class I Polypeptide-Related Sequence B Acts as a stress-induced self-antigen that binds to 8022771, 10746790, 10426993, the KLRK1/NKG2D receptor which is involved in 9497295, 16849432 gamma delta T and NK cell activation MYBPH Myosin Binding Protein H Regulates cell motility and morphology, 23068101, 22085929 transcriptional target of TTF-1 which is a TF necessary for physiological lung functions and has

Functional characterization of the new 8q21 Asthma risk locus | 91

been linked to pathological conditions of the lung

NDFIP1 Nedd4 Family Interacting Protein 1 Represses cell proliferation, regulates metal ion 25801959, 19706893, 25631046, transport, involved in ATM-mediated DNA repair 28051111, 27786273, 24520172, response, restricts Th17 cell potency, regulates IgE- mediated mast cell activation, mediates peripheral tolerance to antigen in T cells ORMDL3 ORMDL Sphingolipid Biosynthesis Regulator 3 Regulates sphingolipid synthesis, endoplasmic 20182505, 23011799, 26969910 reticulum homeostasis, involved in immune cell migration, proinflammatory cytokine production and allergen-induced asthma pathologies. PAG1 Phosphoprotein associated with Transmembrane adaptor protein localized to lipid 24213579 glycosphingolipid-enriched microdomains 1 rafts, involved in immunoreceptor signaling PEX14 Peroxisomal Biogenesis Factor 14 Key component of peroxisomal protein import 9653144, 20154681, 21525035, machinery, membrane anchor for microtubules, 17921697 required for peroxisome motility, required for pexophagy PGAP3 Glycophosphatidylinositol(GPI)-specific Control the association between GPI-anchored 22227195 phospholipase A2 enzyme that is expressed in the proteins and lipid rafts; T cell activation Golgi PPFIA4 PTPRF Interacting Protein Alpha 4 Induced by nickel exposure and hypoxia, cell 20599943, 21829649 junction maintenance PPOX Protoporphyrinogen Oxidase Enzyme involved in heme biosynthesis 3996415 PPP1R1B Protein phosphatase 1, regulatory subunit 1B Signal transduction molecule that controls a 10604473 serine/threonine kinase and phosphatase. RP11-335O13.7 Antisense RNA overlapping ADORA1 exon 1 Unknown - RP11-554D20.1 Antisense RNA overlapping RORA intron 1 Unknown - RP11-94L15.2 lincRNA overlapping IKZF3 exon 3 Unknown - SLC22A5 Solute Carrier Family 22 Member 5 Sodium-dependent high affinity carnitine 9685390, 17884651, 23112839, transporter that plays a critical role in gut 20722056 development and homeostasis STARD3 Start domain-containing protein 3 Endosomal membrane protein invlved in 28377464 cholesterol transport TAP2 Transporter ATP-binding cassette, MHC 2 Translocates peptides from the cytosol to MHC 11381133 class I molecules in the endoplasmic reticulum TLR1 Toll Like Receptor 1 Crucial for pathogen recognition and activation of 12077222, 12697090 innate immunity

Functional characterization of the new 8q21 Asthma risk locus | 92

TNFSF4 TNF Superfamily Member 4 Regulates T-cell proliferation and cytokine 7913952, 16924108, 17975668 production TOMM40L Translocase Of Outer Mitochondrial Membrane 40 Paralog of TOMM40 which is crucial for protein 17437969, 9228044, 25140902 Like precursor import into mitochondria and plays a role in antiviral immunity TSLP Thymic Stromal Lymphopoietin Induces the release of T-cell-attracting chemokines 11418668, 17242164, 2872121, from monocytes, enhances CD11C+ dendritic cell 23433457 maturation, involved in mast cell activation and B- cell development USF1 Upstream Transcription Factor 1 Involved in TGF-β signaling pathway, required for 27649659, 24831529, 15054483 cell fate decisions in response to DNA-damage, regulates lipid and glucose homeostasis WDR36 WD Repeat Domain 36 Critical for nucleolar processing of SSU 18S rRNA, 21051332, 15177553 Involved in T-cell activation ZPBP2 Zona Pellucida Binding Protein 2 Implicated in sperm-oocyte interaction during 17664285 fertilization in mice

Functional characterization of the new 8q21 Asthma risk locus | 93

2

Identification of PAG1 and ZBTB10 as target genes of 8q21 asthma risk variants

Vicente CT, Edwards SL, Hillman KM, Kaufmann S, Mitchell H, Bain L, Glubb DM, Lee JS, French JD, Ferreira MAR. “Long-Range Modulation of PAG1 Expression by 8q21 Allergy Risk Variants”. The American Journal of Human Genetics, 2015. 97(2): 329-36.

Chapter 2 2.1 INTRODUCTION

2.1.1 Background

Having identified thousands of single nucleotide polymorphisms (SNPs) associated with several complex diseases, GWAS have revolutionized how the genetic contribution to human diseases is assessed. However, finding the target genes of newly identified disease risk variants is a complex task and is largely influenced by the genomic context of the associated variants. While coding variants have been the focus of research into disease associated SNPs over the past decades, non-coding variants represent the largest proportion of GWAS findings (>80% of variants detected) [1]. Previously overlooked, intergenic regions harbour a multitude of regulatory elements such as promoters, silencers, enhancers and contain numerous untranslated RNA transcripts likely to have regulatory functions [2]. These elements are directly relevant to human genetics and suggest that a large proportion of variation in disease risk may be caused by variation in non-coding regions. One such region is the 8q21 locus, a region that contains asthma risk variants [3] and is the focus of this Chapter. In Ferreira et al [3], the SNP with the strongest association with asthma risk in this locus was rs7009110 (odds ratio [OR]=1.14, P=4x10-9), an intergenic SNP located 105kb from the nearest gene (ZBTB10). We refer to this SNP as the sentinel SNP in this locus. As found for other diseases [4, 5], the underlying causal variant§ is not necessarily the sentinel SNP, but instead it could be a variant in high linkage disequilibrium§ (LD) with it.

2.1.2 Hypothesis

The asthma-associated sentinel SNP rs7009110, or a variant in LD with it, contributes to variation in the expression of a nearby gene by influencing the activity of a regulatory element.

§ Causal variant is defined as the functional genetic variant that modulates disease risk and explains the observed association. Linkage disequilibrium (LD) is the non-random association of alleles in different loci for a given population. When the frequency of association between alleles is different than what would be expected if the loci were independent and associated randomly, then loci are said to be in LD with each other.

Functional characterization of the new 8q21 Asthma risk locus | 95

Chapter 2 2.1.3 Aims

The overall aim of the experiments reported in this Chapter was to identify the actual causal variant(s) and the likely target gene(s) underlying the reported association between rs7009110 and asthma risk. The specific aims were to:

(i) Define the core region of association based on the LD between rs7009110 and other nearby SNPs; (ii) Identify genes with expression levels associated with rs7009110; (iii) Identify putative regulatory elements (PREs) based on publicly available functional genomic annotations (e.g. histone marks) measured in the core region of association; (iv) Identify physical interactions between PREs and gene promoters using chromosome conformation capture (3C); (v) Identify candidate causal variants overlapping interacting PREs and determine their impact on PRE activity using luciferase reporter assays; (vi) Identify transcription factors (TFs) that bind to interacting PREs and determine the impact of the candidate causal variants on (a) TF binding and (b) the correlation between TF expression levels and target gene expression levels.

The results described in sections 2.3.1 to 2.3.7, which relate to the identification of PAG1 as a target gene of 8q21 variants, were published in 2015 [6]. The original publication is included in this Thesis as part of the Appendix.

Functional characterization of the new 8q21 Asthma risk locus | 96

Chapter 2 2.2 METHODS

2.2.1 Defining the core region of association

To prioritize variants in the 8q21 risk locus for functional analyses, we identified all SNPs in moderate to high LD (r2≥0.6) with rs7009110 in Europeans, using PLINK [7]. LD was calculated using genetic data for 376 individuals of European ancestry sequenced as part of the 1000 Genomes Project [8]. These data were made available by the Geuvadis Project [9], and were downloaded from EBI ArrayExpress (accession numbers E-GEUV-1 and E- GEUV-2; http://www.ebi.ac.uk/Tools/geuvadis-das/). We restricted our analysis to 3,692 SNPs that were located within 1 Mb of rs7009110 and passed quality control (QC) filters, namely a minor allele frequency (MAF) ≥0.05, genotype call rate ≥0.95 and with a Hardy Weinberg equilibrium test P>10-6. The core region of association was defined as the region delimited by the left-most and right-most variants with an LD r2≥0.6 with rs7009110.

2.2.2 Identifying genes with expression levels associated with rs7009110

2.2.2.1 Based on results from published studies

To prioritize genes in the 8q21 risk locus for functional analyses, we first investigated if rs7009110 was in LD (r2≥0.6) with variants reported to associate with the expression of a nearby gene (+/- 1 Mb) in published expression quantitative trait loci (eQTL) studies conducted in six tissues or cell types relevant to asthma: B cells [10], lymphoblastoid cell lines (LCLs) [9, 11], lung [12], monocytes [10], peripheral blood mononuclear cells (PBMCs) [13] and whole-blood [14]. We used the following procedure to identify expression-associated SNPs for genes within 1 Mb of rs7009110. First, we extracted association results from the individual eQTL studies listed above, specifically SNP ID, gene ID and association P- value. Second, for each gene, we used the --clump procedure in PLINK [7] to identify the SNPs most associated with gene expression that were in low LD (r2<0.1) with each other. We refer to these as ‘sentinel eQTLs’ for a given gene. Finally, we estimated the LD between rs7009110 and each sentinel eQTL, as described in section 2.2.1.

2.2.2.2. Based on re-analysis of RNA-seq data from the Geuvadis consortium

The analysis described in the previous section (2.2.2.1) had two major limitations. First, weakly associated eQTLs are often not reported; and second, the majority of published eQTL studies used expression microarrays that have incomplete coverage of gene Functional characterization of the new 8q21 Asthma risk locus | 97

Chapter 2 expression patterns for most genes. To address these limitations, we re-analysed the RNA-seq gene expression dataset generated and made available by the Geuvadis consortium [9]. This dataset consisted of exon-level read counts obtained from lymphoblastoid cell lines (LCLs) of 373 Europeans of the 1000 Genomes Project. The Geuvadis consortium normalized read counts by library depth and removed technical variation by PEER normalization. As in their original association analyses, read counts were further quantile normalised and adjusted for ancestry informative covariates and genotype imputation status. We only considered data for exons expressed in >90% of individuals, namely: ten exons in TPD52; nine in PAG1; seven in ZBTB10; and five exons in each of HEY1, MRPS28 and FABP5. Due to low relative abundance, ZNF704 and STMN2 (expressed in only 56% and 12% of all individuals, respectively) were not included in the analysis. To test the association between rs7009110 and the expression of all exons of each gene, we used a multivariate test that improves power when compared to the alternative strategy of testing each exon individually [15]. Multivariate association analyses were first performed for the sentinel SNP rs7009110 and then extended to all correlated variants within the core region of association.

2.2.3 Identifying Putative Regulatory Elements (PREs) in the core region of association

Nucleosomes are the basic repeating functional and structural units of chromatin [16], consisting of positively charged proteins (histones) that package negatively charged DNA in eukaryotic cell nuclei. Histones are arranged in an octamer conformation containing dimmers of each core histone: H2A, H2B, H3, and H4. Modification of these histones by post-translational events, such as methylation, constitutes an important mechanism that is known to modulate gene transcription. Histone methylation can differ in degree (mono, di or tri-methylation) and can occur in all basic residues: arginine, histidine and lysine. Lysine 4 of histone 3 (H3K4) is one of the most studied methylation sites and has been widely associated with transcriptionally active genes and active/poised enhancers (depending on the methylation status) [17]. Another indicator of transcriptional regulation are DNase I hypersensitivity sites (regions of the chromatin that are cleaved by DNase I enzyme), which are a marker of loose chromatin and, therefore, of exposed and accessible DNA [18]. Both these markers (and several others) have been measured by the ENCODE project to generate genome-wide maps of epigenetic profiles in a multitude of cell lines [19].

Functional characterization of the new 8q21 Asthma risk locus | 98

Chapter 2 The ENCODE project data can be accessed through http://genome.ucsc.edu/encode/; specifically, data tracks on the UCSC Genome browser can be customized by accessing the ENCODE histone modification option under the Regulation settings tab. Using this resource, we studied the Histone Modification ChIP-Seq Signal tracks from ENCODE/Broad institute measured in a lymphoblastoid cell line (GM12878), based on the Human Feb. 2009 (GRCh37/hg19) assembly. In this analysis, the distribution of epigenetics markers was assessed within the core region of association (Chr8: 81,246,659-81,315,490) and included mono-, di- and tri-methylation of lysine 4 of histone 3 (H3K4me1, H3K4me2, H3K4me3), as well as DNase I hypersensitivity sites.

2.2.4 Lymphoblastoid Cell Culture

Lymphoblastoid cell lines (LCLs) are obtained by immortalizing primary B-cell lines through infection with the Epstein-Barr virus (EBV). All LCLs used in this thesis were previously generated by the Molecular Epidemiology Group (Prof. Grant Montgomery) at QIMR

Berghofer. Briefly, the B95-8 marmoset cell-line was cultured at 37°C with 5% CO2, using RPMI 1640 media with 15% Fetal bovine serum (FBS) for 6 days to produce Epstein-Barr infectious virus particles. The virus-containing cell culture supernatant was collected, and used to re-suspend and culture PBMCs, obtained and purified from 10mL of whole blood from each individual volunteer, by the Ficoll-Paque method. PBMCs were incubated at 37°C for 2h to allow EBV infection of B-cells, pelleted and cultured in 96-well plates for 2-5 weeks using RPMI 1640 media with 20% FBS and supplemented with phytohaemagglutinin (PHA-P, Sigma-Aldrich) to prevent the generation of cytotoxic T lymphocytes that kill the infected B cells. Following culture establishment, cells were expanded in 24-well plates, 12-well plates, and later 75cm2 flasks until LCLs were stable and growing continuously (5-6 weeks). At this stage, cells were pelleted, resuspended in freeze mix (20% FBS, 10% DMSO in RPMI 1640) and stored in liquid nitrogen until needed.

For all experiments described in this Chapter, LCL stocks were thawed and cultured at

37°C with 5% CO2, using RPMI 1640 media (made in-house) with 10% FBS (Bovogen).

Functional characterization of the new 8q21 Asthma risk locus | 99

Chapter 2 2.2.5 Identifying physical interactions between the core region of association and nearby gene promoters using Chromosome Conformation Capture (3C)

2.2.5.1 3C library preparation

Libraries were prepared from four LCLs of four different donors, two for each rs7009110 genotype: IDs 8673401 and 8630601, homozygote for rs7009110:C asthma protective allele (referred to as samples 1 and 2, respectively); and IDs 8642701 and 8643801, homozygote for the rs7009110:T risk allele (referred to as samples 3 and 4, respectively). To prepare the lysates, 3x107 cells were pelleted, washed with PBS and fixed with 10ml of 1% formaldehyde for 10 minutes (min) at room temperature (RT) in a roller mixer. The reaction was subsequently quenched by pelleting the cells followed by resuspension in 10ml of ice-cold 0.125M PBS-Glycine. Cells were then centrifuged for 5min at 600g at 4°C and the cell pellets washed with 25ml of cold PBS followed by another centrifugation using the same conditions (Beckman Coulter Allegra X-15R). Afterwards, the cells were resuspended in 10ml of ice-cold Lysis Buffer (10mM Tris pH 7.5, 10mM NaCl, 0.2%

IGEPAL – Sigma-Aldrich, 1x Protease inhibitor cocktail tablet – Roche, in H2O) and incubated on ice for 30mins with occasional mixing. Cell lysis was completed manually with 10 strokes of a Dounce homogenizer and the nuclei collected with a 6min centrifuge at 800g at 4°C. The nuclei pellets were resuspended in 600μL of dH20, 120μL of NEB restriction enzyme Buffer and 24μL of 10% SDS (0.3% final concentration, Sigma-Aldrich), mixed and incubated for 30min at 37°C with shaking. To inactivate the SDS, 200μL of 10% Triton X-100 (2% final concentration, Sigma-Aldrich) was added to the solution, mixed and then incubated for another 30min at 37°C with shaking. A total of 1500U of EcoRI restriction enzyme (New England Biolabs, NEB) was added over a 6 hour period (one aliquot of 5ul every 3 hours) and the samples incubated overnight at 37°C with shaking. Following this incubation, two 20ul aliquots were collected from each sample to assess digestion efficiency – one for agarose gel electrophoresis following crosslink reversal and the other for qPCR of purified digested DNA. The remainder of the sample was used to continue to protocol. To inactivate the EcoRI restriction enzyme, 150μL of 10% SDS (1.6% final concentration) was added to the remainder of each sample and incubated at 65°C for 20min with shaking every 5min. Samples were then transferred to 50ml Falcon tubes and to each one was added, in the following order: 5.1ml of dH20, 800μL 10% Triton, 920μL Ligase Buffer

(500mM Tris pH 7.5, 100mM DTT, 100mM MgCl2), 80μL 10mg/ml BSA (NEB) and 80μL of freshly prepared 100mM ATP (Sigma-Aldrich). Samples were incubated in a water bath at Functional characterization of the new 8q21 Asthma risk locus | 100

Chapter 2 37°C for 1 hour. To ligate the DNA fragments, 2μL of 2000U/μL T4 DNA ligase (NEB) was added to each sample followed by a 4 hour incubation in a 16°C water bath, then 30min at RT on the bench. 30μL of 10mg/ml proteinase K (Astral Scientific) was then added to each tube and incubated overnight in a 65°C water bath to de-crosslink the samples. On the following day, samples were brought to RT and 30μL 10mg/ml RNAseA (Sigma-Aldrich Aldrich) was added followed by a 45min incubation at 37°C. 8ml of phenol-chloroform (Invitrogen) followed by a 15min centrifugation at 8,000g and 4°C (JA-12 rotor on a Beckman Coulter Avanti J-25i High Speed Centrifuge), then 8ml of chloroform and another centrifugation using the same conditions, were used to purify the samples. The DNA was then precipitated by transferring the supernatant to high speed centrifuge tubes and adding to each one, in the following order: 4ml of dH2O, 10μL glycoblue (Ambion), 1ml 3M CH₃COONa and 25ml of 100% ethanol. Samples were incubated at -80°C for 1 hour and then centrifuged at 20,000g for 1 hour at 4°C (JA-17 rotor on a Beckman Coulter Avanti J- 25i High Speed Centrifuge). Pellets were then washed 4-6 times with 70% ethanol to remove precipitated salts and finally, dissolved in 150μL of Tris-HCL (pH 7.5) and incubated at 4°C overnight in a roller mixer. To further purify the dissolved DNA, samples were passed through Amicon-Ultra spin columns (EMD-Millipore). qPCR was performed to quantify each library using diluted gDNA samples as standards, and then stored in aliquots at -20°C.

2.2.5.2 3C control library preparation from Bacterial Artificial Chromosomes (BAC)

To create an artificial ligation product library to normalize for qPCR efficiency, 10ug of DNA from BAC clones CTD-2368L6, CTD-2283C1 and CTD-2210P7 (overlapping the core region of association, ZBTB10 and PAG1 promoters respectively, Invitrogen) were combined, digested with EcoRI (which generated 17 fragments) and re-ligated with T4 DNA ligase (NEB).

2.2.5.3 3C qPCR

Primers were designed to flank each one of the 17 contiguous fragments that cover the core region of interaction as well as the promoters of the genes of interest (ZBTB10 and PAG1; Table 2.1). To detect the presence of ligation products, each fragment primer was used in combination with a promoter primer (bait primer) at 0.25μM in a 20μL qPCR reaction, together with 0.1μL of MyTaq (bioline), 5x MyTaq Buffer, 5μM SYTO9 (Life

Technologies) and dH2O. 300ng of 3C library DNA was added to each reaction as Functional characterization of the new 8q21 Asthma risk locus | 101

Chapter 2 template. qPCR cycling conditions included an initial activation step of 2mins at 95°C; 50x cycles of [20secs at 95°C, 20secs at 66°C and 30secs at 72°C]; and a final extension of 4mins at 72°C (Rotor-Gene Q, Qiagen). Following PCR amplification, all products were run in a 1% agarose gel and the bands corresponding to the interacting fragments of interest were excised, extracted using the Qiagen gel extraction kit and Sanger sequenced (Australian Genome Research Facility, AGRF) to confirm their identity. Chromatin interaction frequencies were calculated by normalizing the qPCR Ct values to the relative primer efficiency of each primer pair (measured on the artificial BAC ligation product library).

Primer 3C Interaction Fragment Fragment Primer Sequence (5' - 3') Coordinates Fragment Size (bp)

1 CTGTCAGCCTATTATTCTGCCCACCACG 81.248.205-81.248.232 81.243.654-81.248.331 4678 2 GATTGTCCTTTGCAATGGAACTGCCTAGC 81.252.072-81.252.100 81.248.338-81.252.262 3925 3 GCACACTCTGGCTTTCCTCCTAGTTAACTGG 81.254.901-81.254.931 81.252.269-81.255.043 2775 4 CCACGACTCTTCCCTCAAATGTAGATTTCTGTC 81.255.113-81.255.145 81.255.050-81.258.454 3405 5 CTCTTTCAAGATCTTGGTAAACATGATGAACATGG 81.263.132-81.263.166 81.258.553-81.263.168 4616 6 AAGAAAGCTTTTCAGGCCAGGCATGG 81.265.288-81.265.313 81.263.175-81.265.494 2320 7 GAGCTGACATTGCACCACTGCAGTCC 81.274.947-81.274.972 81.265.501-81.275.183 9683 8 CCTCTGTGGTAGGATGTCTGGTAATTATTTTGCTC 81.275.375-81.275.409 81.275.190-81.275.897 708 9 CCTCTTGTGAATTGCCTGCTCTACATGCTG 81.276.023-81.276.052 81.275.904-81.277.025 1122 10 GCATTCTCATGCTGTTGAGTGCCTATGG 81.277.126-81.277.153 81.277.032-81.279.569 2538 11 CCAGCCTGTTTCCTTTCCACTGATGC 81.279.715-81.279.740 81.279.576-81.290.220 10645 12 CCTTCTGATTTGGGCAGGACTTACTGACTCA 81.290.335-81.290.365 81.290.227-81.299.133 8907 13 CCATTGGCTGACACCGTTGACCTCTT 81.299.290-81.299.315 81.299.140-81.304.960 5821 14 CCAGTGATGGTCTCTAGTCACCACTGTACGC 81.305.069-81.305.099 81.304.967-81.311.710 6744 15 GAGTTGCACTGAGGTCACCTCCAGTTGC 81.313.861-81.313.888 81.311.717-81.313.898 2182 16 GGATGCAAAGGGTCTCCTGATGATCATG 81.314.008-81.314.035 81.313.905-81.314.467 563 17 GCACCACCCAAGAATCAAGAGAGCCA 81.314.581-81.314.606 81.314.503-81.315.368 866

ZBTB10 GCTACCCCTTTCAGGAAAACGTCCAGC 81.400.473-81.400.499 promoter

PAG1 CCTGGAGCAGAGCTTCTGAAACAGTTGG 82.024.998-82.025.025 promoter Table 2.1 Primers and coordinates of fragments tested in 3C experiments.

2.2.5.4 Allele-specific 3C

To determine whether one rs7009110 allele was preferentially involved in chromatin looping, additional 3C libraries were prepared from two rs7009110:C/T heterozygote LCLs (IDs 8615601 and 8625401), as previously described. The final 3C qPCR was performed in these libraries, using the PAG1 promoter primer and a fragment primer overlapping

Functional characterization of the new 8q21 Asthma risk locus | 102

Chapter 2 PRE2 (Fragment 8) and a risk SNP of interest (rs11783496, in LD with rs7009110, r2=0.74), such that the resulting amplicon contained the SNP as well as DNA from Fragment 8 and the PAG1 promoter. The amplified product was run on a 1% agarose gel, extracted using the Qiagen gel extraction kit and Sanger sequenced (AGRF).

2.2.6 Identifying candidate causal variants and their impact on PRE function using luciferase reporter assays

2.2.6.1 DNA extraction

Genomic DNA (gDNA) was extracted from an LCL using the Salting-Out Method. Briefly, 4x107 cells were pelleted, washed once in PBS and lysed by incubation overnight at 56°C in a Buffer containing 1mL Nuclei lysis solution (Tris 0.1M, EDTA 2mM, NaCl 0.12M), 50μL Proteinase K (10mg/mL, Promega) and 50μL of 20% SDS. DNA was precipitated using 6M NaCl, ice-cold 100% ethanol and 70% ethanol sequentially, followed by a centrifugation step of 20mins at 20,000g to pellet the DNA (Sigma Zentrifugen 1-16). Left to dry at RT overnight, the DNA was then resuspended in 1xTE Buffer and quantified using a NanoDrop Spectrophotometer (Thermo Fisher Scientific).

2.2.6.2 Cloning and luciferase assay

Gene promoters (ZBTB10 and PAG1) and regulatory element fragments (PRE2 and PRE3) were amplified from LCL gDNA with specific primers (Table 2.2) and using the KAPA HiFi system (Kapa Biosystems). Due to size constraints, only a portion of each regulatory element was amplified for cloning. These fragments were chosen based on the location of the 3C interaction fragments as well as the distribution of epigenetic marks that each element overlaps (Figure 2.1). All fragments were first cloned into the pBLUNT plasmid (Thermo Fisher) to be replicated and then into the pGL3-Basic plasmid (Promega) for experiments. The primers used to amplify all fragments contained added restriction enzyme sites to assist cloning into this vector system. For these experiments, using standard DNA cloning techniques, PAG1 or ZBTB10 promoters were inserted in the pGL3- Basic plasmid upstream of the Luciferase gene to drive its expression. Three sets of promoter-driven luciferase constructs were generated for each gene, including: (i) promoter alone; (ii) promoter combined with PRE2; and (iii) promoter combined with PRE3. To determine if PRE activity was affected by asthma risk variants, additional

Functional characterization of the new 8q21 Asthma risk locus | 103

Chapter 2 constructs were generated containing the allergy predisposing allele of each SNP that: (i) was in LD with rs7009110 (r2>0.6); and (ii) overlapped the respective PRE fragment cloned. In total, seven additional constructs were generated using the gBlocks system (Integrated DNA Technologies) to individually incorporate each variant: four for PRE2 (containing the minor allele of rs11783496, rs11786685, rs11786704 or rs13275449) and three for PRE3 (containing the minor allele of rs4739737, rs10957979 or rs2370615) (Figure 2.2). The original cloned fragments contained the allergy protective allele for these SNPs. The DNA sequence of each construct was confirmed through Sanger sequencing (AGRF). Subsequently, 1μg of each plasmid and 50ng of pRLTK (Renilla) were electroporated into 1x106 cells in SF Buffer using the program EH-100 on the Amaxa Nucleofector 96-well Shuttle System (Lonza). Cells were then incubated in 24-well plates at 37°C for 24hrs, after which the media was removed and luciferase/renilla activity measured using the Dual-Glo Luciferase system (Promega) on a Biotek Synergy H4 Multi Mode Plate Reader. Luciferase activity values were normalized to the corresponding renilla activity value to control for variation in transfection efficiency. A two-way ANOVA test with Dunnett’s correction for multiple comparisons was used to analyse the data.

Primer Fragment Cloning Primer Primer Sequence (5' - 3') Coordinates Size (bp)

ZBTB10promoter_fwd GGTACCAAGAGTGGTACAACGGGCAACC 81.396.851-81.396.872 1126 ZBTB10promoter_rev CTCGAGTTGTTCCTTCACAAAGACTGGACC 81.397.953-81.397.976 PAG1promoter_fwd CCACCATTAGTCAACTCATGTCCAGG 82.025.770-82.025.795 1660 PAG1promoter_rev CAAGGGAATCACGGCTCAATTAGG 82.024.136-82.024.159 PRE2_fwd GAGGTACCACCGGTGCAAATTCCTCTGTGGTAGGATGTCTGG 81.275.368-81.275.395 1893 PRE2_rev GATCTAGACCTGCAGGGTAGGGTCCAGTGGGAGAGACACTTGC 81.277.234-81.277.260 PRE3_fwd GAGAAGCTTACCGGTCCTTTGATGGACATGATGTCACC 81.289.408-81.289.433 1560 PRE3_rev GATCTAGACCTGCAGGCACGTGGCTGCCTTCTTATGAAAAGC 81.290.942-81.290.967 Table 2.2 Primers used to generate luciferase assay constructs.

Functional characterization of the new 8q21 Asthma risk locus | 104

Chapter 2

Figure 2.1 Location of the PRE2 and PRE3 fragments used in the luciferase assay experiments (green bars). Also represented are the full length PREs 2 and 3 (blue bars); peak 3C interaction fragments (black/grey bars); and histone modifications and DNase I hypersensitivity tracks from an ENCODE LCL (GM12878).

Figure 2.2 Variants overlapped by PRE2 and PRE3 fragments used in luciferase assay experiments. Location of PRE2 and PRE3 (purple bars); 3C interacting fragments (black bars); the two fragments cloned for luciferase assays (green bars); and the SNPs (in LD with rs7009110; r2 > 0.6) that overlap the cloned fragments and individually incorporated in additional constructs.

Functional characterization of the new 8q21 Asthma risk locus | 105

Chapter 2 2.2.7 Identifying transcription factors (TFs) that bind to interacting PREs and determine the impact of the candidate causal variants

2.2.7.1 TF binding using chromatin immunoprecipitation (ChIP)

LCLs with all three rs7009110 genotypes (CC, CT and TT) were used for these experiments (IDs 8630601, 8615601 and 8642701, respectively). For each cell line, 1x107 cells were crosslinked with 1% formaldehyde at RT for 10 mins followed by two washes with ice-cold PBS. Cells were resuspended in 100mM Tris-HCL (pH 9.4) with 10mM DTT, incubated for 15min at 30°C and pelleted by centrifugation at 2,000g for 5min. Cell pellets were sequentially washed with 1mL of ice-cold PBS, Buffer I (0.25% Triton X-100, 10mM EDTA, 0.5mM EGTA, 10mM HEPES, pH 6.5) and Buffer II (200mM NaCl, 1mM EDTA, 0.5mM EGTA, 10mM HEPES, pH 6.5), and centrifuged for 5min at 2,000g. Cells were resuspended in 300μL of Lysis Buffer (1% SDS, 10mM EDTA, 50mM Tris-HCl pH8.1, 1x Protease inhibitor cocktail tablet – Roche) and incubated at RT for 5min. Samples were then sonicated three times at the maximum setting for 10sec with average size of sheared fragments 100-500bp (Branson SLPt Digital Sonifier). The soluble chromatin fraction was collected, diluted 1:10 with Dilution Buffer (1% Triton X-100, 2mM EDTA, 50mM Tris-HCL pH 8.1, 1x Protease inhibitor cocktail tablet – Roche) and incubated overnight at 4°C with an anti-Foxo3a rabbit mAb (2497S, Cell Signaling Technology) or an IgG control antibody (SC-8334, Santa Cruz Biotechnology). Immunoprecipitation was performed using 45μL Protein A Sepharose 4 Fast Flow Beads (GE Healthcare), 2μg Salmon Sperm DNA and a 1h incubation at 4°C. Sepharose Beads were harvested by centrifugation (10min at 20,000g) and washed sequentially for 10min each with Buffers TSE I (0.1% Triton X-100, 2mM EDTA, 20mM Tris-HCL pH 8.1, 150mM NaCl), TSE II (0.1% SDS, 1% Triton X-100, 2mM EDTA, 20mM Tris-HCL pH 8.1, 500mM NaCl) and Buffer III (0.25M LCl, 1% NP-40, 15 deoxycholate, 1mM EDTA, 10mM Tris-HCl pH 8.1) to precipitate DNA. Protein-bound immunoprecipitated DNA was washed three times with TE Buffer, eluted twice with 150μL

Elution Buffer (1% SDS, 0.1M NaHCO3) and incubated at 65°C overnight to reverse the formaldehyde crosslink. DNA was then extracted using a QIAquick Spin Kit (Qiagen) and eluted in 50μL. 1uL of DNA was used in a 10uL PCR reaction including 5uL 2x PowerUp SYBR Green Master Mix (Applied Biosystems) and 0.2μL of 10pMol specific primers (Table 2.3). PCR cycling conditions included an initial activation step of 2mins at 50°C followed by 10mins at 95°C followed by 40x cycles of [15secs at 95°C and 1min at 66°C] (ViiA 7 Real-Time PCR System, Thermo Fisher Scientific). Percent input was first

Functional characterization of the new 8q21 Asthma risk locus | 106

Chapter 2 determined and Foxo3a values were normalised to an IgG control to calculate the fold enrichment. Statistical analysis was performed using unpaired t test.

Primer Primer Sequence (5' - 3') Region Amplified

PRE3_test_region_fwd CTTCCCTTCTGATTTGGGCAGG 81.290.331-81.290.507 PRE3_test_region_rev TTGGCATCATCAAAAGCAACAGC control_region_fwd GAGGTCGGGAGTTTGAGAACAGC 81.283.900-81.284.074 control_region_rev GCAATCTTGGCTCAGTGCAACC

Table 2.3 Primers used for ChIP experiments.

Functional characterization of the new 8q21 Asthma risk locus | 107

Chapter 2 2.3 RESULTS

2.3.1 The 8q21 core region of association spans 69 kb and harbours 118 variants in LD with the index SNP rs7009110

In the 8q21 locus, SNP rs7009110 had the strongest association with asthma risk in the Ferreira et al. GWAS [20] (odds ratio [OR] = 1.14, P= 4 x 10-9). However, this does not necessarily mean that rs7009110 is the actual causal variant underlying the association with asthma in this locus. Instead, this might be another SNP in LD with rs7009110, which might have less significant association in the GWAS because, for example, it was not imputed§ as accurately as rs7009110.

Therefore, we first used data from the 1000 Genomes Project [21] to identify all the variants in LD (r2 ≥ 0.6) with the index SNP, as described in the methods section. A total of 118 such variants were identified, spanning a 69 kb region that we refer to as the ‘core region of association’ (Chr8: 81,246,659-81,315,490). The core region of association does not overlap any known transcripts, being located between MIR5708 and ZBTB10. However, eight protein coding genes and one miRNA are located nearby (within 1 Mb of rs7009110; Figure 2.3).

§ SNP imputation refers to the statistical inference of unobserved genotypes, based on known haplotypes for a given population.

Functional characterization of the new 8q21 Asthma risk locus | 108

Chapter 2

Figure 2.3 The 8q21 region of association. Regional association results (-log10(P-value), y-axis) for chromosome 8q21. The most-associated SNP is shown in purple, and the colour of the remaining markers reflects the linkage disequilibrium (r2) with the top SNP. Genes within this 1Mb region are also shown.

Functional characterization of the new 8q21 Asthma risk locus | 109

Chapter 2 2.3.2 PAG1 expression is significantly associated with 8q21 allergy risk variants

A useful approach to identify potential target genes is to search for expression quantitative trait loci (eQTLs) that are in LD with the sentinel variants. eQTLs represent genetic variants that determine variation in transcription levels. Results from many eQTL studies are publicly available, including two types of eQTLs – cis and trans. Cis-eQTLs or “local eQTLs” are SNPs that affect the expression of a nearby gene (typically defined as <1 Mb away). Conversely, trans-eQTLs or “distant eQTLs” represent distal effects (>1 Mb), including between different chromosomes [22].

In published eQTL studies, the sentinel SNP was not in high LD with eQTLs for any gene, in cis or trans (Table 2.4) This could potentially represent a false negative finding because, for example, the eQTL association was weak and therefore not reported; and/or PAG1 association was not tested due to incomplete coverage from the expression microarray. To address this possibility, we re-analysed publicly available RNA-seq data (n=373) generated by the Geuvadis consortium, as described in the Methods section. Results from this analysis are shown in Figure 2.4. We were able to test 6 of the 9 nearby transcripts and found that the sentinel SNP (purple square) was significantly associated with the expression of a single gene, PAG1 (P=0.0017). Several SNPs in high LD with the index SNP (r2 ≥ 0.8) were associated with PAG1 expression; the strength of association increased with physical proximity to rs7009110 but did not exceed the significance observed for this SNP. Increased expression of PAG1 was associated with the rs7009110:T risk allele (Figure 2.5).

Therefore, based on results from this eQTL analysis, we conclude that PAG1 is the most likely target gene of asthma risk variants in this region. As such, we prioritized this gene for functional studies.

Functional characterization of the new 8q21 Asthma risk locus | 110

Chapter 2

r2 with eQTL Gene Index SNP Study Ref Tissue Best SNP P-value rs7009110

1 FABP5 rs4739622 [13] PBMCs rs4739622 2.14E-07 0.00

2 FAM164A rs2705496 [10] Monocytes rs2705496 9.60E-04 0.00 3 rs3812459 [10] Monocytes rs3812459 4.38E-06 0.01 FTHL11 4 rs1026830 [10] Monocytes rs1026830 1.53E-05 0.00 5 rs282850 [14] Blood rs282850 5.81E-56 0.01 6 rs12544114 [14] Blood rs12544114 7.27E-43 0.00 7 rs17534643 [14] Blood rs17534643 4.76E-17 0.00 8 rs390691 [14] Blood rs390691 1.38E-15 0.00 9 rs6996298 [14] Blood rs6996298 2.00E-15 0.01 10 rs2956251 [14] Blood rs2956251 8.00E-14 0.01 11 rs2029824 [14] Blood rs2029824 5.44E-06 0.01 12 HEY1 rs2920944 [14] Blood rs2920944 5.75E-06 0.00 13 rs12543376 [14] Blood rs12543376 6.89E-06 0.00 14 rs2467778 [14] Blood rs2467778 1.59E-04 0.01 15 rs2278667 [14] Blood rs2278667 4.51E-04 0.00 16 rs7845273 [14] Blood rs7845273 7.10E-04 0.00 17 rs2979707 [14] Blood rs2979707 1.09E-03 0.00 18 rs13281919 [14] Blood rs13281919 1.19E-03 0.01 19 rs1157639 [14] Blood rs1157639 1.22E-03 0.00 20 IL7 rs1863593 [10] Monocytes rs1863593 4.87E-04 0.00 21 IMPA1 rs17582647 [13] PBMCs rs17582647 7.38E-06 0.01 22 rs3812460 [10] Monocytes rs3812460 7.97E-04 0.00 LOC646374 23 rs10110615 [10] Monocytes rs10110615 2.63E-03 0.00 24 MRPS28 rs903583 [10] Monocytes rs903583 1.53E-04 0.01 25 [11] LCLs rs4500045 5.20E-11 0.00 25 [10] B cells rs12677218 3.37E-04 0.00 25 [10] Monocytes rs16908663 2.85E-17 0.01 25 rs13279056 [9] LCLs rs12677218 5.35E-08 0.00 PAG1 25 [12] Lung rs13266020 7.59E-08 0.00 25 [14] Blood rs13279056 1.55E-97 0.00 25 [13] PBMCs rs6473263 3.54E-38 0.01 26 rs10097731 [11] LCLs rs10504730 5.00E-08 0.00 26 [10] B cells rs10504730 2.54E-04 0.00 26 [9] LCLs rs10504730 1.97E-08 0.00 26 [14] Blood rs10097731 1.64E-88 0.00 27 [10] Monocytes rs1896216 2.51E-03 0.00 27 rs4739791 [14] Blood rs4739791 2.54E-31 0.00 27 [13] PBMCs rs10957999 2.65E-17 0.00 28 rs4237093 [14] Blood rs4237093 7.38E-25 0.00 Functional characterization of the new 8q21 Asthma risk locus | 111

Chapter 2

29 [10] Monocytes rs1445558 5.20E-04 0.00 rs920983 29 [14] Blood rs920983 3.52E-11 0.00 30 [14] Blood rs17494473 8.95E-04 0.00 rs11988289 30 [13] PBMCs rs11988289 7.11E-07 0.00 31 [10] Monocytes rs9650270 1.39E-06 0.00 rs9650270 31 [14] Blood rs7009399 4.39E-06 0.00 32 rs7831986 [13] PBMCs rs7831986 2.51E-06 0.01 33 rs2705499 [14] Blood rs2705499 2.06E-05 0.00 34 rs7017984 [14] Blood rs7017984 5.09E-04 0.00 35 [12] Lung rs10755965 9.02E-07 0.01 rs1863436 35 [14] Blood rs1863436 6.31E-52 0.00 36 TPD52 rs6473202 [14] Blood rs6473202 3.94E-11 0.01 37 rs4440674 [14] Blood rs4440674 8.81E-06 0.01 38 rs4739729 [14] Blood rs4739729 1.58E-05 0.13 39 [12] Lung rs9298338 0.00E+00 0.00 rs9298338 39 [9] LCLs rs201499880 1.90E-11 0.01 40 ZBTB10 [12] Lung rs3863246 9.76E-11 0.00 rs3863246 40 [9] LCLs rs368280 5.39E-09 0.01 41 rs11781854 [9] LCLs rs11781854 9.86E-09 0.02 42 rs3812459 [10] Monocytes rs3812459 1.71E-04 0.01 43 ZFAND1 rs1026830 [10] Monocytes rs1026830 4.19E-04 0.00 44 rs6473199 [10] Monocytes rs6473199 9.32E-04 0.01

Table 2.4 List of SNPs located within 1 Mb of rs7009110 and associated with the expression of nearby genes in published GWAS of gene expression. rs7009110 and index SNPs, are in low LD (r2<0.1) with each other; from each study queried, we selected the SNP correlated with the Index SNP (r2>0.1) that had the most significant association with gene expression. Table adapted from Vicente et al [6].

Functional characterization of the new 8q21 Asthma risk locus | 112

Chapter 2

Figure 2.4 Results from multivariate association analysis. Association was tested in LCL samples from 373 individuals, between allergy risk variants in the core region of association and gene expression levels of six genes within 1 Mb of index SNP rs7009110.

Functional characterization of the new 8q21 Asthma risk locus | 113

Chapter 2

Figure 2.5 rs7009110 genotype effect on PAG1 expression. The rs7009110:T is associated with higher PAG1 expression as well as increased risk of allergic disease. Plotted are the read counts (N=373) for exon2 of PAG1 normalized by library depth, with technical variation removed by PEER normalization and after quantile normalization. The association between rs7009110 (coded additively: 0, 1 or 2 copies of the T minor allele) and variation in gene expression levels was significant (P= 0.0018).

Functional characterization of the new 8q21 Asthma risk locus | 114

Chapter 2 2.3.3 There are four putative regulatory elements (PREs) in the core region of association

PAG1 is located 732 kb away from the core region of association. Therefore, we hypothesised that regulatory elements located in the core region of association physically interact with the PAG1 promoter through long-range chromatin looping.

To identify PREs in the core region of association, we studied markers of transcriptional regulation measured by the ENCODE project in LCLs (GM12878), including histone modifications and DNase I hypersensitivity sites. Based on these markers, we found that at least four PREs were located in the core region of association, which we refer to as PRE1 to PRE4 (Figure 2.6). Of the 118 variants in LD with rs7009110, 35 overlap one of the identified PREs; notably, rs7009110 overlaps PRE3.

Chromosome 8 position, kb

Figure 2.6 PRE location in the core region of association. Based on the distribution of histone marks and DNase I hypersensitivity sites in an LCL analysed by the ENCODE project.

Functional characterization of the new 8q21 Asthma risk locus | 115

Chapter 2 2.3.4 PRE2 and PRE3 interact with the promoter of PAG1 in lymphoblastoid cell lines (LCLs)

To test the hypothesis that a PRE physically interacts with the promoter of PAG1, we divided the core region of association in 17 contiguous fragments using restriction enzymes and quantified the frequency of DNA interactions between each fragment and the PAG1 promoter, using Chromosome Conformation Capture (3C). 3C was performed in LCLs from two individuals homozygous for the rs7009110:T risk allele, and two for the C protective allele.

Results show that for 15 of the 17 fragments tested, the frequency of chromatin interactions was low and consistent with being due to chance (Figure 2.7A). However, for two fragments (F9 and F11) the interaction frequency was significantly above the background level. F9 is 1.1 kb long and overlaps PRE2. Interactions with this fragment and the promoter of PAG1 were observed in LCLs homozygous for the rs7009110:C protective allele but not for the rs7009110:T risk allele. Consistent with this observation, when heterozygous LCLs for rs11783496 (a variant in LD with rs7009110 [r2 = 0.75] and located near fragment F9) were tested in an allele-specific 3C assay, looping of PRE2 to the PAG1 promoter was more strongly associated with the rs11783496:C protective allele (Figures 2.7B and 2.8). F11 is 10.6 kb long and overlaps PRE3. Unlike observed for F9, interactions between F11 and the PAG1 promoter were detected in all LCLs, irrespective of rs7009110 genotype. Two additional independent replicate experiments for each LCL were performed, with the same pattern of results (Figure 2.9).

Functional characterization of the new 8q21 Asthma risk locus | 116

Chapter 2

Figure 2.7 Results from PAG1 3C experiments. (A) Interaction profile between the core region of association and the PAG1 promoter. Frequency of DNA interactions (y-axis, mean ± SD) between 17 fragments (horizontal black lines) located in the 8q21 core region of association (x-axis) and the promoter of PAG1. Triangles represent SNPs in LD with rs7009110 (r2>0.6), with colour gradient representing the association strength with PAG1 expression (least associated in grey to most associated in red). (B) Allele-specific chromatin looping between fragment F9 and the PAG1 promoter. Results show Sanger sequencing chromatograms of the PCR product of a 3C assay performed in a heterozygous LCLs with rs11783496 as the surrogate marker.

Figure 2.8 Independent allele-specific 3C biological replicates. Allele-specific chromatin looping between F9 overlapping PRE2 and the PAG1 promoter is observed in all replicates.

Functional characterization of the new 8q21 Asthma risk locus | 117

Chapter 2

Figure 2.9 Independent PAG1 3C biological replicates. The 3C interaction profile between the 8q21risk region and the promoter of PAG1 is conserved between replicates.

Functional characterization of the new 8q21 Asthma risk locus | 118

Chapter 2 2.3.5 PRE3 harbouring the rs2370615 risk allele acts as an enhancer on the PAG1 promoter

The previous experiments demonstrated that PRE2 and PRE3 physically interact with the PAG1 promoter, 732 kb away from rs7009110. Multiple asthma risk variants overlap these PREs (shown by triangles in Figure 2.7A). Next, we used luciferase reporter assays to determine if asthma risk variants modulate the regulatory ability of these PREs. In these experiments we measured luciferase activity with several constructs, including those containing: (i) the PAG1 promoter alone; (ii) the PAG1 promoter combined with PRE2; and (iii) the PAG1 promoter combined with PRE3. These constructs contained the asthma protective haplotype, that is, the haplotype containing the asthma protective allele for all overlapping asthma risk variants. To specifically test if PRE activity was affected by asthma risk variants, we tested additional constructs containing the asthma predisposing allele for individual SNPs, as specified in the Methods section.

Results from these experiments showed that, PRE2 had no detectable effect on the activity of the PAG1 promoter, either with the asthma protective haplotype or the predisposing alleles for individuals SNPs (Figure 2.10A). On the other hand, for PRE3 we found that the construct containing the minor allele (C) of rs2370615 – a variant in complete LD (r2=1) with rs7009110 – significantly increased PAG1 promoter activity by 1.8-fold when compared to the promoter-only construct (P=0.0051; Figure 2.10B). This difference was slightly greater when compared to the construct with both the promoter and PRE3 containing the asthma protective haplotype (2.3-fold; P=0.0005). PRE3 constructs containing the predisposing allele for other variants in LD rs7009110 (rs4739737 and rs10957979) did not affect the activity of the PAG1 promoter. Based on these experiments, we conclude that rs2370615 genotype determines the ability of PRE3 to drive gene expression through the PAG1 promoter.

Functional characterization of the new 8q21 Asthma risk locus | 119

Chapter 2

Figure 2.10 PRE3 acts as a transcriptional enhancer on the promoter of PAG1 in the presence of the rs2370615 risk allele. PRE2 (A) or PRE3 (B) was cloned under the control of a PAG1 promoter luciferase reporter with and without the putative causal SNPs. Graphs represent three independent experiments assayed in duplicate. Error bars correspond to 95% CI and P values were calculated using a two-way ANOVA followed by Dunnett’s multiple comparisons test (*p<0.01, **p<0.001).

Functional characterization of the new 8q21 Asthma risk locus | 120

Chapter 2 2.3.6 The rs2370615:C asthma risk allele disrupts the binding of Foxo3a to PRE3

To understand how rs2370615 might affect PRE3 activity, we used HaploReg [23] and RegulomeDB [24] to search for transcription factors (TFs) for which binding to PRE3 was predicted to be disrupted by rs2370615 genotype. We found that rs2370615 was predicted to disrupt the binding motif (TTGTTTAC) for five Forkhead box (Fox) TFs: Foxf1, Foxj2, Foxo1, Foxo3a and Foxq1. Of these, Foxo3a was noteworthy because it is an NF-κB antagonist that plays a role in the inhibition of lymphocyte activation and proliferation [25], a function that is directly relevant to asthma pathophysiology. Other family members have been implicated in: mesenchymal cell migration (Foxf1) [26]; oncogenic functions (Foxj2 and Foxq1) [27-30]; and insulin signalling and adipogenesis (Foxo1) [31, 32].

Therefore, to confirm the prediction that rs2370615 disrupted Foxo3a binding to PRE3, we performed Chromatin Immunoprecipitation (ChIP) assays in DNA extracted from LCLs with the rs2370615:TT, CT or CC genotype. We found a significant enrichment in Foxo3a binding to PRE3 for the TT (3.4-fold, P=0.001) and TC (1.3-fold, P=0.006) but not CC (0.94-fold, P=0.820) genotype, when compared to an IgG binding control (Figure 2.11A). This is consistent with the predicted disrupting effect of rs2370615:C on Foxo3a binding. TF binding was weak and not genotype-dependent in a control region 4 kb upstream of PRE3 which does not include the canonical Fox binding motif (Figure 2.11B), further confirming that Foxo3a binds to PRE3 with high affinity.

Functional characterization of the new 8q21 Asthma risk locus | 121

Chapter 2

Figure 2.11 Binding of Foxo3a TF to PRE3 is disrupted by the rs2370615:C risk allele. ChIP was assayed in LCLs with the indicated genotypes using either an anti-Foxo3a antibody or a control IgG antibody. Foxo3a binding to PRE3 (A) and control region (B) is shown as fold enrichment over IgG (horizontal line). Graphs represent three independent experiments performed in triplicate. Error bars represent 95% CI and P values were calculated by unpaired t test.

Functional characterization of the new 8q21 Asthma risk locus | 122

Chapter 2 2.3.7 FOXO3A expression is correlated with PAG1 expression in LCLs

Two key findings from the experiments above were that (i) PRE3 enhanced the PAG1 promoter activity; and (ii) Foxo3a binding to PRE3 was higher in individuals with the rs2370615:TT genotype, because the TF binding motif was not disrupted. Based on these findings, we hypothesised that in individuals with the TT genotype, FOXO3A expression levels should be correlated with PAG1 expression levels, whereas this should not be the case in individuals with the CC genotype, because the TF binding motif is disrupted. To test this hypothesis, we again used gene expression data measured by the Geuvadis Project in LCLs from 373 individuals (as described in the Methods section 2.2.2.2). In this analysis, we found a significant positive correlation between FOXO3A and PAG1 expression in TT (r=0.31, 95% CI 0.15-0.45, P=0.0003) but not CC (r=-0.12, 95% CI -0.34- 0.15, P=0.3960) individuals, as predicted (Figure 2.12). The magnitude of the correlation in heterozygote individuals was intermediate between that of the homozygotes (r=0.25, 95% CI 0.06-0.34, P=0.0055). Collectively, our results are consistent with a causal effect of rs2370615 on PAG1 expression through disruption of Foxo3a binding to PRE3.

Figure 2.12 rs2370615 genotype modulates the association between FOXO3A and PAG1 expression. Association between FOXO3a and PAG1 expression in rs2370615:TT, CT and CC LCLs, tested by linear regression. Horizontal coloured solid lines represent the mean expression of PAG1 for each genotype class (as shown in Figure 2.5).

Functional characterization of the new 8q21 Asthma risk locus | 123

Chapter 2 2.3.8 PRE2 and PRE3 also interact with ZBTB10 in LCLs

In the previous experiments, we focused on PAG1 because it was the only gene for which there was some evidence from eQTL studies that it was likely to be a target gene of asthma risk variants. But the lack of an eQTL signal for other nearby genes does not strictly imply that they are not target genes of asthma risk variants. The lack of an eQTL signal might arise, for example, if asthma risk variants only influence variation in gene expression in specific cell types or physiological conditions (e.g. hypoxia) that were not tested in published eQTL studies. Because ZBTB10 was the closest gene to the sentinel asthma risk variant, and given its predicted role in immune cell function [33, 34], we performed initial experiments to test if it was also likely to be a target gene of asthma risk variants.

First, 3C assays were performed as described above to test if PREs in the core region of association physically interacted with the promoter of ZBTB10. We found that fragments overlapping PRE2 (F8, spanning 708 bp) or PRE3 (F12, spanning 8,907 bp and including the sentinel SNP rs7009110) physically interacted with the promoter of ZBTB10. Both interactions were observed consistently in all four LCLs tested, irrespective of rs7009110 genotype (Figure 2.13). Two additional independent replicate experiments for each LCL were also performed and reproduced these results (Figure 2.14).

These results indicate that the PREs that were found to interact with the PAG1 promoter also interact with the ZBTB10 promoter.

Functional characterization of the new 8q21 Asthma risk locus | 124

Chapter 2

Figure 2.13 Results from ZBTB10 3C experiments. Interaction profile between the core region of association and the ZBTB10 promoter. Frequency of DNA interactions (y-axis, mean ± SD) between 17 fragments (horizontal black lines) located in the 8q21 core region of association (x-axis) and the promoter of ZBTB10. Triangles represent SNPs in LD with rs7009110 (r2>0.6).

Figure 2.14 Independent ZBTB10 3C biological replicates. The 3C interaction profile between the 8q21risk region and the promoter of ZBTB10 is conserved between replicates.

Functional characterization of the new 8q21 Asthma risk locus | 125

Chapter 2 2.3.9 PRE2 enhancer activity on the ZBTB10 promoter is disrupted by the rs11783496 risk allele

To determine if asthma risk variants modulate the regulatory ability of the interacting PREs, we used again luciferase reporter assays. In these experiments, we measured luciferase activity with several constructs, including those containing: (i) the ZBTB10 promoter alone; (ii) the ZBTB10 promoter combined with PRE2; and (iii) the ZBTB10 promoter combined with PRE3. Additional constructs that incorporated the risk allele for individual SNPs were also tested. No significant effects on ZBTB10 promoter activity were observed with any construct containing PRE3. On the other hand, the PRE2 construct containing the asthma protective haplotype resulted in a 2-fold increase in ZBTB10 expression, which was borderline statistically significant (P=0.05). Amongst the four PRE2 constructs containing the risk allele for individual SNPs, only for one did we observe a significant difference in luciferase activity when compared to the PRE2 construct with the asthma protective haplotype: the construct containing rs11783496:T (Figure 2.15). The differences were again only borderline significant (P=0.05), but suggest that PRE2 acts as a transcriptional enhancer on the promoter of ZBTB10 and that this activity is disrupted by the rs11783496:T asthma risk allele.

A query of public databases (HaploReg and RegulomeDB) revealed that rs11783496 is predicted to alter the binding motif (AAAGTCCAG) of a histone deacetylase (HDAC2), associated with gene silencing [35]. Individuals carrying the rs11783496:T risk allele would therefore harbour the intact binding domain allowing HDAC2 to bind to PRE2 and repress ZBTB10 expression through chromatin looping

Functional characterization of the new 8q21 Asthma risk locus | 126

Chapter 2

Figure 2.15 PRE2 transcriptional enhancer activity on the promoter of ZBTB10 is disrupted in the presence of the rs11783496 risk allele. PRE2 (A) or PRE3 (B) was cloned under the control of a ZBTB10 promoter luciferase reporter with and without the putative causal SNPs. Graphs represent three independent experiments assayed in duplicate. Error bars correspond to 95% CI and P values were calculated using a two-way ANOVA followed by Dunnett’s multiple comparisons test.

Functional characterization of the new 8q21 Asthma risk locus | 127

Chapter 2 2.4 DISCUSSION

The aim of this Chapter was to provide new insights into the molecular mechanisms that explain the association between 8q21 variant and asthma risk.

We first defined a 69 kb core region of association based on the location of variants in LD with our index SNP rs7009110. We used a relatively liberal LD cut-off (r2 >0.6) to ensure that the putative causal variant was captured. Next, based on multivariate eQTL analyses, we identified only one gene for which expression levels were associated with rs7009110 genotype: PAG1, which encodes for a transmembrane receptor that regulates Csk activity [36], with direct effects on immunoreceptor signalling [37]. The rs7009110:T asthma risk allele was associated with increased PAG1 expression suggesting that increased transcription of PAG1 has a pro-inflammatory effect in the context of asthma. As we discuss below, this finding is not consistent with the widely accepted role of PAG1 in immune cell function.

We then performed a series of functional experiments to confirm our hypothesis that PAG1 was a target gene of 8q21 asthma risk variants. We first studied epigenetic marks in the core region of association and identified four putative regulatory elements (PREs) in LCLs. Supporting these results, in a genome-wide study of H3K4me2 sites (which are associated with both active and poised enhancers), a significant gain of this methylation marker was detected in two sites overlapping PRE2, when comparing primary human CD4+ memory T cells with naive CD4+ T cells [38]. Furthermore, in another study, the distribution of lysine 27 acetylation on histone 3 (H3K27ac, also associated with enhancer activity) was assessed, and both PRE2 and PRE3 were predicted to be enhancers in a broad range of human cell types (e.g. H3K27ac marker overlapped PRE2 in lung tissue and PRE3 in thymus and spleen tissue) [39]. Of the 118 variants spaning the core region of association, we found that 35 overlapped one of the four PREs identified. We thus hypothesized that rs7009110, or a correlated variant, disrupts the function of a PRE that controls the transcription of PAG1.

To test this hypothesis, we used 3C assays to identify physical interactions between the promoter of PAG1 and the core region of association. This analysis identified an interaction between the PAG1 promoter and PREs 2 and 3. The interaction with PRE2 was observed in an allele-specific fashion, suggesting that an asthma protective allele overlapping this PRE regulates the establishment of long-range chromatin interactions between PRE2 and the promoter of PAG1. It is not clear what the genotype implications Functional characterization of the new 8q21 Asthma risk locus | 128

Chapter 2 underlying this observation are however, we speculate that a consequence of this observation could be that, in individuals carrying asthma risk alleles, PRE2 would respond to environmental cues (e.g. allergen exposure) and activate PAG1 transcription. In individuals carrying the protective alleles, PRE2 would not activate PAG1 transcription, potentially dampening or not triggering an immune response. The interaction with PRE3 was observed irrespective of rs7009110 genotype, suggesting that asthma risk variants do not influence the establishment of chromatin looping between this PRE and the promoter of PAG1.

Secondly, we used luciferase reporter assays to determine the regulatory ability of these PREs, which revealed that PRE3 modulates PAG1 expression. Furthermore, we also showed that rs2370615 was the functional variant within this PRE. The observation that PRE3 harbouring the rs2370615:C asthma risk allele acted as a transcriptional enhancer of PAG1 expression, corroborated our finding from the eQTL analyses that asthma predisposing alleles were associated with increased PAG1 expression. Together, these results validated PAG1 as a target gene of 8q21 allergy risk variants and indicated that rs2370615 might represent the underlying putative causal variant.

Then, we hypothesized that rs2370615 could alter the binding affinity of transcription factors (TFs) that interact with PRE3 and, consequently, modulate PAG1 expression. To investigate this, we analysed ENCODE ChIP-seq data measured in a LCL and found that this variant overlaps the binding site of two TFs. The first was RelA (or p65) subunit of the nuclear factor κB (NF- κB) TF, a key player in innate and adaptive immune responses [40]. The second was POU2F2 (or Oct-2), the POU domain class 2 transcription factor 2 which regulates B-cell specific genes [41, 42]. The following lines of evidence suggest that binding of these TFs to PRE3 might be required for activation of PAG1 expression. First, both TFs co-occur in LCLs [43] and synergistically regulate enhancer activity [44]. Second, RelA is associated with TNF-α-induced transcriptional activation of target genes when this TF is recruited to enhancers involved in long-range looping interactions [45]. Third, RelA binding to PRE3 in LCLs is highest specifically over rs2370615 (Figure 2.16). Despite the latter observation, rs2370615 was not predicted to disrupt the binding motif for either RelA or POU2F2, suggesting that this variant might modulate the binding affinity of a different TF that recruits RelA to PRE3 and subsequently activates PAG1 expression.

A query of public databases found that the rs2370615 T/C polymorphism is predicted to alter the binding motif of Foxo3a, an NF-κB antagonist that plays a key role in

Functional characterization of the new 8q21 Asthma risk locus | 129

Chapter 2 immunoregulation mainly through lymphocyte activation and proliferation [25]. Results from our ChiP assay and genetic association analyses confirmed our prediction that the rs2370615:C risk allele disrupts the binding motif of Foxo3a and, hence, the ability of this TF to bind to PRE3. These results suggest that in individuals carrying the rs2370615:C asthma risk, Foxo3a is not able to bind to PRE3, thereby allowing other TF to bind to this region and modulate PAG1 expression (e.g. RelA). This hypothesis is yet to be confirmed.

Collectively, our results are consistent with a model of PAG1 regulation (Figure 2.17) where: (i) PRE2 is a dynamic regulatory element with unconfirmed induction/suppression capacity for PAG1. This PRE might require additional triggers to become active – i.e. signalling from PRE3, appropriate cellular activation such as allergen or viral stimulation, or binding of transcription factors that have low/no expression in the LCLs tested and; (ii) PRE3 is a poised regulatory element for PAG1 that promotes its transcription in all individuals, irrespective of genotype. Also (iii) individuals carrying the rs2370615:C allergy risk allele have overall increased PAG1 expression, possibly due to the inability of Foxo3a to bind to PRE3 and thus allowing for another more powerful TF to bind and drive higher levels of PAG1 transcription (e.g. RelA). On the other hand, (iv) individuals carrying the rs2370615:T protective allele have overall lower PAG1 expression, possibly because Foxo3a binds to PRE3 and drives lower levels of transcription. Another explanation might be that PRE2 (shown by 3C to physically interact with the PAG1 promoter in these individuals) suppresses PRE3 activity. We believe the proposed model is plausible, however, additional experiments to further elucidate the molecular mechanisms underlying the ability of PRE2 and PRE3 to modulate PAG1 expression, are warranted.

To our knowledge, PAG1 has not been directly implicated in asthma or allergic disease pathophysiology previously. Nevertheless, several studies have implicated PAG1 in immune-related processes [46-52]. These studies have mainly been carried out in the context of T and B-cell function, two cell types that express high levels of PAG1 [36, 53, 54] and are critical in asthma pathophysiology [55, 56]. These studies have specifically implicated PAG1 in signal transduction from lymphocyte surface receptors [37] which could, therefore, represent the cellular mechanisms underlying the association with disease risk. However, results from these studies have been conflicting [37], with: (i) in vitro studies suggesting that PAG1 might play an anti-inflammatory role by inhibiting immune cell activation [46, 47, 49, 50]; and (ii) in vivo studies suggesting that PAG1 is not required for normal immune cell development and function [51, 52]. Our finding that allergy predisposing alleles increase the transcription of PAG1 in human B cell lines contradicts these previous

Functional characterization of the new 8q21 Asthma risk locus | 130

Chapter 2 studies, instead suggesting that PAG1 overexpression might have a pro-inflammatory role in the context of asthma (e.g by promoting B cell activation). In subsequent Chapter, we report results from a series of experiments that attempt to understand the contribution of PAG1 to asthma pathophysiology.

Figure 2.16 RelA binding overlapping PRE3. The blue bar represents PRE3 location and the green bar represents the PRE3 fragment used in our luciferase assays. Also shown are the location of SNPs in linkage disequilibrium (in green the ones that overlap the green PRE3 fragment) with rs7009110 (red), and transcription factor ChIP-seq from ENCODE (bottom panel) and RelA binding based on ChIP-seq data in LCLs (Top panel). The latter track was downloaded from: ftp://ftp.ncbi.nlm.nih.gov/geo/samples/GSM1329nnn/GSM1329660/suppl/GSM1329660_RelA.bw

Figure 2.17 Proposed model of PAG1 regulation. PRE2 and PRE3 differentially modulate the expression of PAG1 in individuals carrying the (A) rs2370615:C asthma risk allele that disrupts Foxo3a binding to PRE3, allowing other TFs to bind to it and drive higher levels of PAG1 transcription; or (B) rs2370615:T protective allele that allows Foxo3a to bind to PRE3 and drive lower levels of PAG1 transcription. Functional characterization of the new 8q21 Asthma risk locus | 131

Chapter 2 Although we identified PAG1 as a target gene of 8q21, we cannot rule out the possibility that additional genes might also interact with PREs in this locus. Due to its proximity to the index SNP, one such gene was ZBTB10, a repressor of the specificity proteins (Sp1, Sp3 and Sp4) [57]. Results from our 3C assays showed that the promoter of ZBTB10 physically interacted with the same PREs identified when we studied PAG1 (PRE2 and PRE3). However, the specific interacting fragments found to interact with ZBTB10 did not coincide with those observed to interact with PAG1, suggesting that different transcriptional mechanisms are involved in the regulation of both genes. The luciferase reporter assays performed failed to identify a statistically significant effect of asthma risk variants on PRE function and ZBTB10 promoter activity. It is nonetheless possible that the effect of the asthma risk variants on ZBTB10 promoter activity is not very pronounced in the baseline (i.e. unstimulated) state. It is not uncommon for changes in luciferase expression to be detected only when cells are incubated with the appropriate trigger [58] which, for example, might be required to induce the recruitment of appropriate TFs to the PREs involved. Therefore, we suggest that the luciferase reporter assays should be repeated using experimental conditions that up-regulate ZBTB10 expression. It will be interesting to determine if these experiments are able to reproduce the observation that the rs11783496:T asthma risk allele overlapping PRE2 decreased the transcription efficiency of the ZBTB10 promoter (albeit not significantly so). This could be important, given the prediction that rs11783496 disrupts HDAC2 binding to PRE2, and the critical role played by histone deacetylases in the anti-inflammatory effect of glucocorticoids, the most widely used therapy for asthma [59].

ZBTB10, also known as RIN ZF, is a member of the zinc finger gene family [57] and a putative repressor of the Specificity Proteins (Sp1, Sp3 and Sp4), which regulate several immune-related genes [33, 34]. As such, ZBTB10 could play a role in immune responses, although this has not been previously demonstrated. ZBTB10 has been mostly studied in the context of cancer. Its expression is regulated by the oncogenic microRNA-27a that promotes tumour growth and metastasis [60]. Interestingly, it interacts with CEBPB, a bZIP transcription factor critical for normal macrophage function as well as regulation of genes implicated in immune and inflammatory responses [61]. A related member of the zinc finger family is ZBTB20 (34% similarity with ZBTB10), which is expressed in hematopoietic tissues and, among its various roles [62], influences immune responses through activation of TLR-signalling [63]. As for PAG1, subsequent Chapters in this thesis report results from

Functional characterization of the new 8q21 Asthma risk locus | 132

Chapter 2 pilot experiments that aim to provide some insights into the potential contribution of ZBTB10 to asthma pathophysiology.

Finally, we cannot rule out the possibility that other genes in the region might also be targets of the asthma risk variants. For example, a review of published studies suggests that TPD52 and FABP5 are biologically plausible candidates. TPD52 or Tumour protein D52, is a member of a highly conserved family of small coiled-coil motif-bearing polypeptides [64] and is potentially involved in vesicle trafficking [65]. Highly expressed in a variety of cancer types and cell lines, including breast [66], ovarian [67] and prostate [68] cancers, it has also been suggested to be implicated in different molecular processes such as proliferation, tumour dissemination and regulation of apoptosis [69]. A role during B-cell maturation has also been suggested for TPD52 due to selective expression in mature B- cells and plasma cells [70]. On the other hand, FABP5 encodes for the fatty acid binding protein-5, also known as psoriasis-associated-FABP (PA-FABP). It is a member of the intracellular lipid-binding protein (iLBP) family of small and highly conserved cytoplasmic proteins, known to play a key role in the uptake, transportation and storage of fatty acids [71]. FABP5 is expressed in epidermal cells and was first identified as being up-regulated in psoriatic tissue [71]. Up-regulation of this gene in cancer cells has been implicated in increased cell proliferation and invasiveness [72]. A possible relationship between FABP5 and asthma has also been recently hypothesized, given that FABP5 expression is associated with linoleic acid metabolite 13-S-HODE levels [73], which in turn cause airway epithelial injury [74]. Involvement in susceptibility to viral-induced lung inflammation has also been described [75, 76].

In conclusion, this research represents one of the few studies to follow up on GWAS findings in the context of asthma and the first attempt to functionally characterize the 8q21 risk locus. Overall, we identified two target genes underlying the association with allergic disease (PAG1 and ZBTB10) as well as at least one putative causal variant in this locus (rs2370615).

Functional characterization of the new 8q21 Asthma risk locus | 133

3

Mechanisms of regulation of PAG1 and ZBTB10 expression

Chapter 3 3.1 INTRODUCTION

3.1.1 Background

PAG1 and ZBTB10 are two target genes of 8q21asthma risk variants. We have identified the putative causal SNP that regulates PAG1 expression and is in LD with the asthma risk variants in this locus. We demonstrated that in individuals carrying the asthma protective allele for this SNP, the transcription factor Foxo3a binds to the corresponding enhancer and promotes PAG1 expression. On the other hand, in individuals carrying the asthma predisposing allele, we hypothesised (but were unable to demonstrate) that RelA (a subunit of the NF-κB complex) and not Foxo3a binds to the enhancer and promotes greater PAG1 expression. For ZBTB10 even less information is known about the mechanisms controlling gene transcription, with only suggestive clues arising from the initial functional work in this locus. Thus, additional in vitro studies are needed to further elucidate the mechanisms and triggers underlying the activation of both PAG1 and ZBTB10 transcription.

3.1.2 Hypothesis

The expression of PAG1 and ZBTB10 in immune cells is induced after exposure to stimuli that are directly relevant to asthma (e.g. allergen) and this effect is modified by the genotype of asthma risk variants.

3.1.3 Aims

The overall aim of the experiments reported in this Chapter was to understand how PAG1 and ZBTB10 expression might be regulated in the context of allergic disease. The specific aims were to characterise:

(i) The contribution of NF-κB signalling to PAG1 expression; (ii) The effect of allergen exposure on PAG1 and ZBTB10 expression; (iii) The effect of phenol on PAG1 and ZBTB10 expression; (iv) The contribution of asthma risk SNPs to phenol-induced PAG1 and ZBTB10 expression; (v) Expression profiles of immune-related genes after allergen and phenol exposure.

Functional characterization of the new 8q21 Asthma risk locus | 135

Chapter 3 3.2 METHODS

3.2.1 Assessing the contribution of NF-κB signalling to PAG1 expression

3.2.1.1 TNF-α and IFN-α as NF-κB inducers

3.2.1.1.1 Cell culture and conditions used

Lymphoblastoid cell lines (LCLs) were obtained and cultured as described in Chapter 2, section 2.2.4. Extracts were prepared from two different LCLs, one homozygote for each rs7009110 genotype – IDs 8630601 (rs7009110:CC) and 8642701 (rs7009110:TT). To assess the effect of NF-κB on PAG1 expression, cells were stimulated for 30mins at 37ºC, with two different cytokines that strongly activate the NF-κB signalling pathway [77, 78]: TNF- α (Recombinant Human Tumor Necrosis Factor α, R&D systems) or IFN-α2 (Recombinant Human Interferon α2, BioLegend).

3.2.1.1.2 Quantification of PAG1 gene expression

(i) mRNA extraction

5x106 cells were seeded in 24-well plates and incubated with TNF-α or IFN-α at a concentration of 5ng/mL. Following incubation, cells were pelleted, resuspended in 200μL of TRIzol Reagent (Invitrogen) and stored at -80°C. Before mRNA extraction, cell lysates were thawed at RT for 15mins. Chloroform (Chem-Supply) was added to the lysates (20% of initial TRIzol volume), mixed thoroughly for 15secs and incubated at RT for 2-3 mins followed by centrifugation at 13,000rpm for 15mins at 4°C. The colourless aqueous phase was then collected and transferred into a new tube, to which isopropanol (Chem-Supply) was added (50% of initial TRIzol volume) to precipitate the nucleic acids. Samples were vortexed and incubated at RT for 10mins followed by centrifugation at 13,000rpm for 10mins at 4°C. The supernatant was discarded and the pellets washed in 100% ethanol (100% of initial TRIzol volume, Chem-Supply), vortexed and centrifuged at 13,000rpm for 5mins at 4°C. Supernatant was discarded and the pellets briefly air dried at RT before being resuspended in RNAse-free water (30-50μL depending on pellet size). mRNA quality and quantification was measured using a NanoDrop Spectrophotometer (Thermo Fisher Scientific).

Functional characterization of the new 8q21 Asthma risk locus | 136

Chapter 3 (ii) cDNA conversion cDNA synthesis from mRNA was performed using the High-Capacity RNA-to-cDNA Kit (Applied Biosystems). Up to 1μg of RNA was used in a reaction containing 5μL of 2x RT

Buffer, 0.5μL of 20x RT Enzyme Mix and H2O up to 10μL total reaction volume. Samples were incubated at 37°C for 1h followed by a 5mins step at 95°C to stop the reaction, in a T100 Thermal Cycler (BioRad). Samples were stored at -20°C until downstream gene expression analysis.

(iii) Gene expression analysis

Gene expression was measured by Real-Time PCR using the TaqMan Gene Expression assays Hs00179693_m1 (PAG1), Hs02758991_g1 (GAPDH) Hs01060665_g1 (ACTB) and 18S (Hs03003631_g1) from Life Technologies. 0.5μL of each 20x Assay Mix was used in a 10μL reaction containing 2μL of cDNA from each sample (50ng per reaction),

5μL of TaqMan Fast Advanced Master Mix (Applied Biosystems) and 2.5μL of H2O. Amplification was measured using 384-well optical plates in the ABI Viia 7 or ABI Quantstudio 5 systems (Applied Biosystems). RT-PCR cycling conditions included an initial incubation at 50°C for 2mins (to activate the Uracil-N-Glycosylase enzyme (UNG) present in the reaction Master Mix and minimize PCR cross-contamination), followed by a Polymerase activation step of 20secs at 95°C and then 40x cycles of [1sec at 95°C and 20secs at 60°C]. Results were normalized against housekeeping gene controls (Glyceraldehyde 3-phosphate dehydrogenase or GAPDH, Actin Beta or ACTB and 18S ribosomal RNA or 18S), analysed using the Comparative Ct method and shown as fold change values.

3.2.1.1.3 Quantification of PAG1 protein expression

(i) Extraction of nuclear and cytoplasmic protein fractions

1x107 cells were harvested and washed twice in ice-cold PBS. Pellets were thoroughly resuspended in 800μL ice-cold Buffer A (10mM HEPES pH7.9, 10mM KCl, 0.1mM EDTA, 0.1mM EGTA, 1mM DTT and 0.5mM PMSF; all from Sigma-Aldrich) followed by a 15min incubation on ice to promote cell swelling. To lise the cells, 25μL of a 10% Nonidet P-40 solution (Sigma-Aldrich) was added to the cell suspension. Tubes were then vortexed and centrifuged at 4,000rpm for 3min. The supernatants containing cytoplasmic proteins were collected and stored at -80°C until needed. To extract nuclear proteins, nuclear pellets Functional characterization of the new 8q21 Asthma risk locus | 137

Chapter 3 were resuspended in 100μL ice-cold buffer C (20mM HEPES pH7.9, 0.4M NaCl, 1mM EDTA, 1mM DTT and 1mM PMSF; NaCl from Chem-Supply and everything else Sigma- Aldrich), followed by vigorous rocking on a tube shaking platform for 10min at 4°C and a centrifugation step for 5min at 12,000rpm at 4°C. The supernatants containing nuclear proteins were collected and stored at -80°C until needed.

(ii) Cell membrane protein extraction

1x107 cells were harvested and washed once in ice-cold PBS. Cell pellets were resuspended in 300μL 0.25M sucrose (Sigma-Aldrich) and homogenised with 10 strokes of a dounce homogenizer. This suspension was then centrifuged at 14,000rpm for 15mins at 4°C to pellet mitochondria and nuclei. Supernatants containing crude cell membranes were stored at -80°C until needed.

(iii) Extraction of soluble protein fraction

Extractions were performed as described in Schörg et al [79]. Briefly, 1x107 cells were harvested and washed twice in ice-cold PBS. Soluble proteins were extracted in 500μL of buffer containing 10mM Tris-HCL pH8.0, 1mM EDTA, 400mM NaCl, 0.1% Nonidet P-40, and supplemented with 1x protease inhibitor cocktail tablet (Sigma-Aldrich), and centrifuged at 14,000rpm for 10mins at 4°C. Supernatants containing soluble proteins were collected and stored at -80°C until needed.

(iv) Western Blot analysis

Protein concentration was estimated using a Bradford assay (Bio-Rad) or a NanoDrop Spectrophotometer (Thermo Fisher Scientific). 20-40μg total protein were mixed with Loading Buffer* and denatured at 100°C for 7min following a brief incubation on ice for 2- 3min. Samples and were loaded into home-made acrylamide gels* and separated in 1x running buffer* at 150V (volts) for 1h in a Mini-PROTEAN Tetra Vertical Electrophoresis Cell (Bio-Rad). Proteins were then transferred onto a methanol-activated Immobilon-P PVDF Membrane (0.45µm, Merck) in transfer buffer*, using an ice-cold wet system with the Bio-Rad Mini Trans-Blot Cell for 1h at 0.30A (amperes). Membranes were then incubated with blocking buffer* at RT for 1h, and overnight at 4°C using the following antibodies: rabbit polyclonal (pAb) anti-NFκB p50 (sc-114), anti-NFκB p52 (sc-298), anti- NFκB p65 (sc-372), anti-RelB (sc-226) anti-c-Rel (sc-71), all from Santa Cruz

Functional characterization of the new 8q21 Asthma risk locus | 138

Chapter 3 Biotechnology, anti-Lamin A/C (Abcam) and anti-Rab18 (Sigma-Aldrich); rabbit monoclonal (mAb) anti-PAG1 (clone EPR9705, Abcam), anti-αTubulin (Abcam). The following day, membranes were washed in Tris-buffered saline with Tween 20 (TBST) buffer* four times for 10min at RT, incubated with a goat anti-rabbit IgG-HRP antibody (sc- 2004) for 1h at RT and washed in TBST as described. Proteins were then detected using the Chemiluminescence Amersham ECL Prime detection kit (GE Healthcare) and developed using Fuji RX medical X-ray films in the SRX-101A Medical film processor (Konica Minolta). Proteins of interest were identified by molecular weight (Precision Plus Protein Dual Color Standards, Bio-Rad). Western Blot results were analysed and quantified using the ImageJ software (National Institutes of Health, Bethesda, USA).

*All buffers and solutions used in these assays are listed in the Appendix.

3.2.1.2 Hypoxia as the NF-κB inducer

3.2.1.2.1 Cell culture and conditions used

LCLs were obtained and cultured as described in Chapter 2, section 2.2.4. Extracts were prepared from two different LCLs, one homozygote for each rs7009110 genotype – IDs 8630601 (rs7009110:CC) and 8642701 (rs7009110:TT). To evaluate the impact of oxygen depletion on gene and protein expression, cells were incubated in hypoxic (1% O2) or severe hypoxic (0.2% O2) conditions using the InvivO2 1000 hypoxic chamber (Baker Ruskinn). Incubation times vary and are indicated in each figure legend.

3.2.1.2.2 Quantification of PAG1 gene expression

As described in section 3.2.1.1.2 (i) and (ii).

(iii) Gene expression analysis

Gene expression was measured by Real-Time PCR using the TaqMan Gene Expression assays Hs00179693_m1 (PAG1), Hs01060665_g1 (ACTB) and 18S (Hs03003631_g1) from Life Technologies. Results were normalized against housekeeping gene controls ACTB and 18S. All other procedures were performed as described in section 3.2.1.1.2 (iii).

Functional characterization of the new 8q21 Asthma risk locus | 139

Chapter 3 3.2.1.2.3 Quantification of PAG1 protein expression

As described in section 3.2.1.1.3 (iii) and (iv).

3.2.2 Assessing the effect of allergen exposure on PAG1 and ZBTB10 expression

3.2.2.1 Cell culture and conditions used

LCLs were obtained and cultured as described in Chapter 2, section 2.2.4. Extracts were prepared from one rs7009110:TT LCL – ID 8642701. To evaluate the effect of allergen exposure on gene expression, 5x106 cells were seeded in 24 well-plates and incubated at 37ºC for 24h with: different concentrations of Lipopolysaccharides (stated in corresponding figures; LPS from Escherichia coli, Sigma-Aldrich), 30μg LPS-RS (LPS from Rhodobacter sphaeroides; TLR4 antagonist; IvivoGen), 5,000 Allergy Units (A.U.) of Standardized Mites Allergenic Extract (HollisterStier) or a 1:2 dilution of a 50% (V/V) Glycerine buffer (HollisterStier).

3.2.2.2 Quantification of PAG1 and ZBTB10 gene expression

As described in section 3.2.1.1.2 (i) and (ii).

(iii) Gene expression analysis

Gene expression was measured by Real-Time PCR using the TaqMan Gene Expression assays Hs00179693_m1 (PAG1), Hs00397261_m1 (ZBTB10), Hs02758991_g1 (GAPDH) and Hs01060665_g1 (ACTB) from Life Technologies. Results were normalized against housekeeping gene controls GAPDH and ACTB. All other procedures were performed as described in section 3.2.1.1.2 (iii).

3.2.3 Assessing the effect of phenol on PAG1 and ZBTB10 expression

3.2.3.1 Cell culture and conditions used

3.2.3.1.1 Lymphoblastoid cell lines (LCLs)

LCLs were obtained and cultured as described in Chapter 2, section 2.2.4. Extracts were prepared from one rs7009110:TT LCL – ID 8642701. To evaluate the effect of phenol on

Functional characterization of the new 8q21 Asthma risk locus | 140

Chapter 3 gene expression, 5x106 cells were seeded in 24 well-plates and incubated at 37ºC for 24h with: 5,000 Allergy Units (A.U.) of Standardized Mites Allergenic Extract (HollisterStier), a 1:2 dilution of a 50% (V/V) Glycerine buffer (HollisterStier),PBS (phosphate buffered saline), 50μg/mL Dermatophagoides pteronyssinus extract (CiteQ biologics) or pure phenol (0.2%, 0.4% and 0.8% total volume, Sigma-Aldrich).

3.2.3.1.2 LUVA Cell Line (Human Mast Cell Line)

The LUVA mast cell line, which is derived from a donor with aspirin exacerbated respiratory disease, was purchased as a frozen stock in pZerve cryopresevative (Protide Pharmaceuticals) from Kerafast [80, 81]. For all experiments described in this Chapter, LUVA cell stocks were thawed and cultured at 37°C with 5% CO2, using StemPro-34 SFM media (500mL, Gibco) supplemented with L-glutamine (2mM final), 5mL Pen Strep (10,000 U/mL), 1mL Primocin (50mg) and StemPro-34 nutrient supplement. To evaluate the effect of phenol on gene expression, 1x106 cells were seeded in 24 well-plates and incubated at 37ºC for 24h with pure phenol (0.4% total volume, Sigma-Aldrich).

3.2.3.2 Quantification of PAG1 and ZBTB10 gene expression

As described in section 3.2.2.2.

3.2.3.3 Measurement of cell viability

10μL of cell cultures containing: (i) media only; (ii) glycerine buffer; and (iii) two concentrations of a HDM solution (5,000A.U. or 10,000A.U) prepared in the glycerine buffer (phenol content of 0.4% and 0.8% total volume, respectively) after 1h and 24h incubation, were collected and stained with 10μL of trypan blue stain (NanoEnTek). 10μL of stained cells were loaded onto a cell counting slide (EVE cell counting slide, Nano EnTek) and cell viability was measured using an automatic cell counter (EVE Automatic cell counter, Nano EnTek).

Functional characterization of the new 8q21 Asthma risk locus | 141

Chapter 3 3.2.4 Assessing the contribution of asthma risk SNPs to phenol-induced gene expression

3.2.4.1 Cell culture and conditions used

LCLs were obtained and cultured as described in Chapter 2, section 2.2.4. To evaluate the contribution of SNP rs7009110 on gene expression, cell pellets were obtained from thirteen different LCLs, six rs7009110:CC (IDs 8673401, 8630601, 8660601, 8625602, 8639602 and 8634701) and seven rs7009110:TT (IDs: 8693001, 8665002, 8604301, 8604601, 8608602, 8642701 and 8677801). 5x106 cells were seeded in 24 well-plates and incubated at 37ºC for 24h with pure phenol (0.4% or 0.2% total volume, Sigma-Aldrich).

3.2.4.2 Quantification of PAG1 and ZBTB10 gene expression

As described in section 3.2.2.2.

3.2.5 Assessing broad immune gene expression profiles in response to allergen and phenol exposure

3.2.5.1 Cell culture and conditions used

3.2.5.1.1 Lymphoblastoid cell lines (LCLs)

LCLs were obtained and cultured as described in Chapter 2, section 2.2.4. Extracts were prepared from two different LCLs, one homozygote for each rs7009110 genotype – IDs 8630601 (rs7009110:CC) and 8642701 (rs7009110:TT). To evaluate the effect of phenol and allergen exposure on broad gene expression, 5x106 cells were seeded in 24 well- plates and incubated at 37ºC for 24h with pure phenol (0.4% total volume, Sigma-Aldrich).

3.5.1.1.2 LUVA Cell Line (Human Mast Cell Line)

Mast cells were derived and cultured as described in section 3.2.3.1 (ii). To evaluate the effect of phenolic compounds on broad gene expression, 1x106 cells were seeded in 24 well-plates and incubated at 37ºC for 24h with pure phenol (0.4% total volume, Sigma- Aldrich).

Functional characterization of the new 8q21 Asthma risk locus | 142

Chapter 3 3.2.5.2 Quantification of gene expression

As described in section 3.2.1.1.2 (i) and (ii).

(iii) Gene expression analysis

The effect of phenol on broad immune gene expression was determined by RT-PCR, using the Human Immune Panel TaqMan Arrays (Applied Biosystems) which includes 90 immune target genes that fall into 9 categories: cytokines and cytokine receptors, chemokines and chemokine receptors, cell surface receptors, stress response, signal transduction, cell cycle and protein kinases, proteases and oxidoreductases and transcription factors; and 6 endogenous control genes. To measure gene expression, 100μL of a reaction containing 50μL of TaqMan Fast Advanced Master Mix (Applied

Biosystems), 800ng of cDNA and H2O (up to 50μL combined) was loaded into each port of a 384-microfluidic card which was then centrifuged twice at 1,200rpm for 1min using a Heraeus Multifuge X3 centrifuge (Thermo Scientific) and sealed. The ABI Viia 7 system (Applied Biosystems) was used to quantify gene expression using the TaqMan Array Micro Fluidic Card Thermal Cycling Block and the following cycling profile: initial UNG incubation at 50°C for 2mins, polymerase activation at 92°C for 10mins and 40x cycles of [1sec at 97°C and 20secs at 62°C]. Results were normalized against the endogenous control genes (GAPDH, ACTB, 18S, TFRC, GUSB, PGK1), analysed using the Comparative Ct method and shown as fold change values. A complete list of all genes analysed is included in the Appendix.

3.2.5.3 Correlation analyses

Correlation between the expression of ZBTB10 and the expression of genes whose transcription is modulated by phenol exposure, was tested using the RNA-seq gene expression dataset of LCLs of 373 individuals measured by the Geuvadis consortium, as described in Chapter 2.

Functional characterization of the new 8q21 Asthma risk locus | 143

Chapter 3 3.3 RESULTS

3.3.1 Contribution of NF-κB signalling to PAG1 expression

3.3.1.1 TNF-α and IFN-α as NF-κB inducers

TNF-α induces RelA translocation into the nucleus

To investigate if RelA regulates PAG1 expression, we first determined if we could induce its translocation into the nucleus, where this NF-κB subunit is functional. For this, homozygote rs2370615:TT and CC LCLs were incubated with either TNF-α or IFN-α, two stimuli that are known to activate the NF-κB pathway. Nuclear protein extracts were prepared from these cells where RelA protein level was measured. Results are show in Figure 3.1 and suggest that only TNF-α incubation induced RelA translocation into the nucleus in these cells, regardless of rs2370615 genotype. Nuclear RelA levels following IFN-α incubation are similar to those observed in baseline physiologic conditions.

Figure 3.1 Nuclear RelA levels following TNF-α stimulation. (A) 1x107 cells from rs2370615:CC and TT LCLs (n=1 for each genotype) were harvested in resting state (time point 0) or following stimulation. Experimental conditions used were 30 minutes of incubation with 5ng/mL of TNF-α or IFN-α. Nuclear protein extracts were prepared and separated by SDS- PAGE. Membranes were immunoblotted using an anti-RelA antibody or anti-Lamin A/C as loading control. (B) Quantification of the relative intensity of RelA expression.

Functional characterization of the new 8q21 Asthma risk locus | 144

Chapter 3 RelA is the only NF-κB subunit that translocates into the nucleus following TNF-α incubation

The NF-κB family includes five subunits: NF-κB1 (or p105/p50), NF-κB2 (or p100/p52), RelA (or p65), RelB and c-Rel, that exist in cells as homo- and hetero-dimers that play different roles depending on the dimerized subunit(s). To investigate if other NF-κB subunits could potentially contribute to PAG1 regulation by associating with RelA, we assessed the translocation into the nucleus of all NF-κB subunits, following stimulation with TNF-α. Results are shown in Figure 3.2 and indicate that (i) RelA was the only subunit translocated into the nucleus after incubation with TNF-α, regardless of genotype; and (ii) RelB and c-Rel nuclear levels were higher in the LCL carrying the rs2370615:C disease risk allele.

Figure 3.2 Nuclear levels of the five NF-κB subunits following TNF-α stimulation. (A) Nuclear protein extracts were prepared from rs2370615:CC and TT LCLs (n=1 for each genotype) incubated with 5ng/mL TNF-α for 30 mins at 37ºC, and separated by SDS-PAGE. Membranes were immunoblotted using antibodies against each subunit or anti-Lamin A/C as loading control. (B) Quantification of the relative intensity of NF-κB subunits expression.

Functional characterization of the new 8q21 Asthma risk locus | 145

Chapter 3 Stimulation with TNF-α has no effect on PAG1 protein levels

Following confirmation that TNF-α incubation induces the translocation of RelA into the nucleus, we then investigated the effect of this stimulus and RelA nuclear localization on PAG1 cytoplasmic protein levels, the preferential location of this protein. We found that at baseline, PAG1 levels were similar between rs2370615:TT and CC LCLs (Figure 3.3). There was a slight decrease in cytoplasmic PAG1 levels following exposure to TNF-α.

Figure 3.3 Western blot results for cytoplasmic PAG1 following TNF-α stimulation. (A) Cytoplasmic protein extracts were prepared from rs2370615:CC and TT LCLs (n=1 for each genotype) following TNF-α incubation (5ng/mL for 30 minutes) and separated by SDS-PAGE. Membranes were immunoblotted using antibodies against PAG1 or tubulin as loading control. (B) Quantification of the relative intensity of PAG1 expression.

Functional characterization of the new 8q21 Asthma risk locus | 146

Chapter 3 Stimulation with TNF-α or IFN-α has no effect on cell membrane PAG1 content

Although the cytoplasm is the preferential intracellular location of PAG1, its functional site is the cell membrane, where it facilitates downstream signalling from immune cell surface receptors. Therefore, we hypothesised that the slight reduction in PAG1 cytoplasmic levels observed in the experiment above was due to protein trafficking to the cell membrane. To test this hypothesis, we extracted the cell membrane protein fraction from rs2370615:TT and CC LCLs, through a sucrose gradient and assessed PAG1 cell membrane content following TNF-α stimulation (Figure 3.4A). PAG1 levels were also measured following IFN-α stimulation to determine if this cytokine could be affecting PAG1 expression through a different pathway independent of NF-κB (Figure 3.4B). Results from these experiments show no difference in PAG1 levels following TNF-α or IFN-α stimulation, regardless of rs2370615 genotype. Similar results were obtained in LCLs from two additional donors (one per rs2370615 genotype; Figure 3.4C). Therefore, PAG1 protein levels do not seem to be modulated by exposure to TNF-α or IFN-α cytokines.

Figure 3.4 Cell membrane PAG1 levels following TNF-α and IFN-α stimulation. Cell membrane protein extracts were prepared from rs2370615:CC and TT LCLs (n=1 for each genotype) following (A) TNF-α or (B) IFN-α exposure. (C) Independent replicate results from additional LCLs. Experimental conditions used were 30 minutes of incubation with 5ng/mL of TNF-α or IFN-α. Proteins were separated by SDS-PAGE and membranes were immunoblotted using antibodies against PAG1 or Rab18 as loading control. Tables show the quantification of the relative intensity of PAG1 expression for each respective assay. Functional characterization of the new 8q21 Asthma risk locus | 147

Chapter 3 Stimulation with TNF-α or IFN-α has no effect on PAG1 gene expression levels

Despite no obvious effect on PAG1 protein levels, it was possible that TNF-α or IFN-α induced PAG1 transcription. To test this hypothesis LCLs were exposed to these cytokines and PAG1 expression was measured using TaqMan assays. Similarly to the findings from our protein assays, PAG1 mRNA levels also did not change following TNF-α or IFN-α exposure (Figure 3.5).

Figure 3.5 Gene expression results for PAG1 following TNF-α and IFN-α stimulation. Total RNA was extracted from rs2370615:CC and TT LCLs (n=1 for each genotype) following TNF-α or IFN-α exposure (5ng/mL of TNF-α or IFN-α for 30 minutes), using the Trizol method and gene expression was measured using TaqMan assays. (A) PAG1 expression by rs2370615 genotype; and (B) combined genotype analysis Results represent the average of four independent replicate experiments and are calculated as fold change from baseline physiologic conditions (No stimulus) for each rs2370615 genotype. Bars represent mean ± SD and the dotted line indicates baseline expression.

Functional characterization of the new 8q21 Asthma risk locus | 148

Chapter 3 3.3.1.2 Hypoxia as the NF-κB inducer

Results from our experiments using TNF-α and IFN-α showed that these stimuli had no effect on PAG1 expression, suggesting that other stimuli might modulate PAG1 activation. One such stimulus, that is also known to induce the activation of the NF-κB pathway, is hypoxia. Schörg et al [79] identified PAG1 as a target of hypoxia-inducible transcription factor (HIF) and thus, we hypothesized that hypoxic conditions could be the appropriate stimulus to modulate PAG1 expression. As such, LCLs from two individuals (one with TT and one with CC genotype for rs2370615) were incubated in hypoxic conditions (1% O2) and collected at different time points for protein and RNA extraction. Results indicate that cells incubated in hypoxic conditions for 1h had slightly increased PAG1 expression (Figure 3.6A and B) however, this difference was not statistically different (P=0.15, n=2) and also not evident at the protein level (Figure 3.6 C). Longer incubation periods in hypoxic conditions did not have an effect on PAG1 gene or protein expression.

Functional characterization of the new 8q21 Asthma risk locus | 149

Chapter 3

Figure 3.6 PAG1 expression in LCLs under hypoxic conditions. (A) and (B) RNA and (C) cell membrane proteins were extracted from rs2370615:CC and TT LCLs (n=1 for each genotype) harvested in hypoxic conditions (1% O2) at different time points. RNA was extracted using the Trizol method, gene expression was measured using TaqMan assays and results are calculated as fold change from baseline physiologic conditions (Normoxia 20% O2) for each rs2370615 genotype. (A) PAG1 expression by rs2370615 genotype; and (B) combined genotype analysis. Bars represent mean ± SD and the dotted line indicates baseline expression. (C) Cell membrane protein extracts were separated by SDS-PAGE and membranes were immunoblotted using antibodies against PAG1 or Rab18 as loading control. The table shows quantification of the relative intensity of PAG1 expression.

Functional characterization of the new 8q21 Asthma risk locus | 150

Chapter 3 Severe hypoxia has no effect on PAG1 protein or mRNA expression

To assess if more extreme hypoxic conditions would amplify the effects on PAG1 expression observed with standard hypoxic conditions and to replicate the results described by Schörg et al, we measured PAG1 levels in rs2370615:TT and CC LCLs incubated in severe hypoxia (0.2% O2). We found that severe hypoxic conditions did not have an impact on PAG1 mRNA levels (Figure 3.7A and B). Similarly, PAG1 protein levels also remained constant under severe hypoxia compared to normoxic conditions (Figure 3.7C).

Functional characterization of the new 8q21 Asthma risk locus | 151

Chapter 3

Figure 3.7 PAG1 expression in LCLs under severe hypoxic conditions. (A) and (B) RNA and (C) soluble proteins were extracted from rs2370615:CC and TT LCLs (n=1 for each genotype) harvested in severe hypoxic conditions (0.2% O2) at different time points. RNA was extracted using the Trizol method, gene expression was measured using TaqMan assays and results are calculated as fold change from baseline physiologic conditions (Normoxia, 20% O2) for each rs2370615 genotype LCL. (A) PAG1 expression by rs2370615 genotype; and (B) combined genotype analysis. Bars represent mean ± SD and the dotted line indicates baseline expression. (C) Soluble protein fraction was extracted as described by Schörg et al and separated by SDS-PAGE. Membranes were immunoblotted using antibodies against PAG1 or Rab18 as loading control. The table shows quantification of the relative intensity of PAG1 expression.

Functional characterization of the new 8q21 Asthma risk locus | 152

Chapter 3 3.3.2 Effect of allergen exposure on PAG1 and ZBTB10 expression

Given the negative results observed with activators of the NF-κB pathway, we hypothesised that other stimuli relevant to asthma might be required to up-regulate PAG1 transcription. Allergen exposure is a critical driver of asthma development in genetically susceptible individuals and PAG1 plays a role in the signalling cascade following allergen exposure [82]. Therefore, we first measured PAG1 expression in LCLs following allergen exposure. Due to the critical nature of this exposure in disease pathophysiology, we also assessed its impact on the expression of ZBTB10, the other target gene of 8q21 variants.

LCLs were incubated for 24h with a HDM extract prepared in a glycerine buffer. As a negative control we used the glycerine buffer alone. Both solutions are routinely used for skin prick tests in clinical practice. Results from this experiment (Figure 3.8) showed that both the glycerine buffer and the HDM solution, significantly up-regulated both PAG1 and ZBTB10, the latter between 4- to 7-fold when compared to expression levels measured with no exposure.

We hypothesised that the effect seen with the HDM and glycerine solutions could arise due to the presence of high levels of Lipopolysaccharides (LPS). To test this possibility, cells were first incubated with increasing concentrations of LPS, but this did not result in upregulation of either PAG1 or ZBTB10 (Figure 3.8), suggesting that LPS was unlikely to explain the observed effect. To confirm this, the same LCL was incubated with the HDM and glycerine solutions with or without the presence of a Toll-like receptor 4 TLR4 antagonist (LPS-RS); TLR4 is the main receptor that mediates the effect of LPS in immune cells [83]. Results from this experiment are shown in Figure 3.9. In the absence of the TLR4 antagonist, PAG1 and ZBTB10 expression was up-regulated by the HDM and glycerine solutions as observed in the previous experiment (Figure 3.8). Furthermore, the same pattern of results was observed when the TLR4 signalling was inhibited, confirming that gene upregulation was independent of LPS. These results suggest that other components present in both the HDM and glycerine solutions explain the increase in both PAG1 and ZBTB10 expression.

Functional characterization of the new 8q21 Asthma risk locus | 153

Chapter 3

Figure 3.8 PAG1 and ZBTB10 expression in a LCL exposed to HDM and LPS. A rs2370615:TT LCL (n=1) was exposed to the different conditions for 24h at 37ºC and RNA was extracted using the Trizol method. Both glycerine and HDM in glycerine solutions were sourced from the same company and are normally used for human Skin prick reactivity testing. Gene expression was measured using TaqMan assays and results are calculated as fold change from baseline physiologic conditions (No stimulus) for each gene. Bars represent mean ± SD and the dotted line indicates baseline expression.

Figure 3.9 TLR4 signalling effect on PAG1 and ZBTB10 expression. A rs2370615:TT LCL (n=1) was exposed to the different conditions for 24h at 37ºC in the presence or absence of LPS-RS, a TLR4 inhibitor. RNA was extracted using the Trizol method and (A) PAG1 and (B) ZBTB10 expression was measured using TaqMan assays. Results are calculated as fold change from baseline physiologic conditions (No stimulus) for each gene. Bars represent mean ± SD and the dotted line indicates baseline expression.

Functional characterization of the new 8q21 Asthma risk locus | 154

Chapter 3 3.3.3 Effect of phenol on PAG1 and ZBTB10 expression

Of all the ingredients listed as present in both the glycerine buffer and HDM solutions, phenol, a widely used stabilizing agent, was particularly interesting. Phenolic compounds have been extensively implicated in oxidative stress regulation and more recently, a modulatory role in ZBTB10 gene expression was also proposed [84]. As such, we hypothesised that phenol present in both the HDM and glycerine solutions could be driving PAG1 and ZBTB10 expression.

To test this hypothesis, we exposed an LCL to the following stimuli: (a) glycerine buffer; (b) the HDM extract prepared in this same buffer; (c) pure phenol, in equivalent concentrations to the amount present in the glycerine buffer or the HDM solution (0.2% or 0.4% total volume, respectively); or (d) HDM extract prepared in PBS instead of glycerine buffer. Results from this experiment (Figure 3.10A and B) showed that (i) exposure to HDM prepared in glycerine but not PBS up-regulated both PAG1 and ZBTB10, more strikingly so for the latter. These results indicate that gene upregulation following exposure to the glycerine buffer is most likely driven by its phenol content.

We also tested additional concentrations of phenol (~3% and 6%) which proved to be toxic and induced cell death (not shown). We did not measure cell viability specifically in samples incubated with pure phenol, but we did assess this in samples incubated with the glycerine buffer and two concentrations of HDM prepared in this buffer, after 1h and 24h of incubation. Results in Figure 3.10C showed that, compared to unstimulated cells: (i) incubation with the glycerine buffer did not induce cell death; (ii) incubation with the lowest concentration of HDM prepared in the glycerine buffer (containing 0.4% phenol) slightly induced cell death both after 1h (8% decrease in viability from baseline) and 24h (13% decrease in viability from baseline) of incubation; and (iii) incubation with the highest concentration of HDM prepared in the glycerine buffer (containing 0.8% phenol) induced significant cell death (32% decrease in viability at 1h and 24h). Therefore, we decided to proceed with only one concentration of phenol (0.4%), which is equivalent to the amount present in the HDM solution prepared in the glycerine buffer that we used in previous experiments. This amount induces highest PAG1 and ZBTB10 expression without seriously compromising cell viability in culture, suggesting that at this concentration the reactive oxygen species (ROS) content within the cell does not exceed the threshold that LCLs can cope with.

Functional characterization of the new 8q21 Asthma risk locus | 155

Chapter 3

Figure 3.10 PAG1 and ZBTB10 expression following phenol exposure. A rs2370615:TT LCL (n=1) was exposed to each condition for 24h at 37ºC. RNA was extracted using the Trizol method and (A) PAG1 and (B) ZBTB10 expression was measured using TaqMan assays. Results are calculated as fold change from baseline physiologic conditions (No stimulus) for each gene. Bars represent mean ± SD and the dotted line indicates baseline expression. (C) Cell viability measured using trypan blue in cells cultured for 1h and 24h with each stimuli.

Functional characterization of the new 8q21 Asthma risk locus | 156

Chapter 3 Phenol-induced up-regulation of PAG1 and ZBTB10 expression is also observed in mast cells

To investigate if the effect of phenol exposure was specific to LCLs or if it could be recapitulated in other cell lines, we exposed LUVA cells, a human mast cell line, to pure phenol and measured PAG1 and ZBTB10 expression as described above. We found that phenol also up-regulated the expression of both genes in human mast cells, albeit the impact on ZBTB10 expression was not as dramatic as observed in LCLs.

Figure 3.11 PAG1 and ZBTB10 expression following phenol exposure in mast cells. Mast cells (n=1) were exposed to 0.4% phenol for 24h at 37ºC. RNA was extracted using the Trizol method and gene expression was measured using TaqMan assays. Results represent the average of two independent replicate experiments and are calculated as fold change from baseline physiologic conditions (No stimulus) for each gene. Bars represent mean ± SD and the dotted line indicates baseline expression.

Functional characterization of the new 8q21 Asthma risk locus | 157

Chapter 3 3.3.4 Association between asthma risk SNPs and phenol-induced gene expression

To investigate the contribution of rs2370615 genotype to phenol-induced gene expression, LCLs from different individuals homozygous for either genotype (n=6 for TT and n=7 for CC) were exposed to phenol. Individuals with the CC asthma predisposing genotype had lower PAG1 and higher ZBTB10 expression after phenol exposure when compared to those with the TT genotype, but these differences were not statistically significant (Figure 3.12).

Figure 3.12 rs2370615 genotype contribution to phenol-mediated gene expression. Homozygous rs2370615:TT (n=6) and CC (n=7) LCLs were exposed to 0.2 or 0.4% phenol for 24h at 37ºC. RNA was extracted using the Trizol method and (A) PAG1 and (B) ZBTB10 expression was measured using TaqMan assays. Results represent the average of independent LCLs and are calculated as fold change from baseline physiologic conditions (No stimulus) for each gene. Bars represent mean ± SEM and the dotted line indicates baseline expression. *P<0.05 vs. no stimulus, calculated using a Wilcoxon signed-rank test.

Functional characterization of the new 8q21 Asthma risk locus | 158

Chapter 3 3.3.5 Expression profiles of immune-related genes after phenol exposure.

Phenol exposure resulted in a dramatic increase in the expression of ZBTB10, a transcription factor thought to regulate the expression of immune-related genes [34, 85]. We were interested in identifying likely target genes of ZBTB10 in LCLs, which could be studied in greater detail in future functional studies (e.g. using ChIP). To this end, we measured the expression of 96 immune-related genes in LCLs from four individuals (two homozygous LCLs for each rs2370615 genotype) before and after exposure to phenol. Six genes (ACTB, GAPDH, GUSB, PGK1, TFRC and 18S) were used as endogenous controls to normalize gene expression. After restricting the analysis to genes expressed in at least three out of four LCLs, both before and after phenol exposure, 71 genes remained for analysis. Of these, 65 were up-regulated (fold difference >1), 4 were down-regulated (fold- difference <1), and 2 remained unchanged following phenol exposure (Figure 3.13). Statistically significant differences in gene expression were observed for only six genes, all up-regulated by phenol exposure: SKI (8.7-fold, P=0.044), CD40 (9.3, P=0.020), IL5 (9.8- fold, P=0.0002), PTGS2 (10.6-fold, P=0.038), NOS2A (10.6-fold, P=0.047) and CSF1 (10.7-fold, P=0.035).

Figure 3.13 Histogram of gene distribution based on expression ratio in LCLs following phenol exposure. Homozygous rs2370615:TT (n=2) and CC (n=2) LCLs were exposed to 0.4% phenol for 24h at 37ºC or harvested in resting state. RNA was extracted using the Trizol method and gene expression was measured using TaqMan assays. Baseline expression was calculated as independent values for each LCL tested, and expression following phenol exposure was calculated as fold-change from baseline using the ∆CT method. An expression ratio for each gene was 푬풙풑풓풆풔풔풊풐풏 풂풇풕풆풓 풑풉풆풏풐풍 풆풙풑풐풔풖풓풆 calculated as 풓풂풕풊풐 = . Histogram depicts the relative frequency of genes (Y axis) sharing 푬풙풑풓풆풔풔풊풐풏 풃풆풇풐풓풆 풑풉풆풏풐풍 풆풙풑풐풔풖풓풆 the same ratio value (X axis), being 1 the equivalent of no expression change between conditions.

Functional characterization of the new 8q21 Asthma risk locus | 159

Chapter 3 To confirm these results using a different immune cell type, we used the same approach to measure gene expression in mast cells of one individual (in duplicate), before and after phenol exposure. Results from this analysis indicate that five of the six genes up-regulated in LCLs by phenol, were also up-regulated by phenol in mast cells (Figure 3.14). For NOS2A, we did not detect expression at baseline in mast cells and therefore we were unable to calculate an expression ratio for this gene. However, we detected expression following phenol exposure, indicating that NOS2A is also up-regulated with this stimulus. Therefore, we conclude that the transcription factor ZBTB10 might regulate the expression of these six genes in immune cells.

Figure 3.14 Expression ratio comparison in LCLs and mast cells following phenol exposure for the six genes differentially expressed. Expression values for each gene were calculated in LCLs (n=4) and mast cells (n=1) exposed to 0.4% phenol for 24h. PAG1 and ZBTB10 expression was measured in these samples using TaqMan assays as described before and represented as an expression ratio. The dotted line indicates a ratio value of 1.

Functional characterization of the new 8q21 Asthma risk locus | 160

Chapter 3 Finally, if any of these six genes are indeed regulated by ZBTB10, then we would expect a positive correlation between their expression levels. To test this possibility, we used gene expression data generated by the Geuvadis consortium for LCLs of 373 individuals, as described in Chapter 2. Data were available for three (SKI, CD40 and CSF1) of the six genes, in addition to ZBTB10. For two of these, there was a significant positive correlation with ZBTB10 expression: SKI (beta=0.144, P=0.007) and CD40 (beta=0.148, P=0.000015). Moreover, the correlation between ZBTB10 and CD40 was genotype dependent, with a significant positive correlation being observed in CC (beta=0.33, P=0.0008) and CT (beta=0.15, P=0.0006) but not TT (beta=0.05, P=0.434) LCLs (Figure 3.15). The correlation between ZBTB10 and CSF1 was not statistically significant (beta=0.078, P=0.084). Thus, we conclude that SKI and CD40 are likely targets of ZBTB10

Functional characterization of the new 8q21 Asthma risk locus | 161

Chapter 3

Figure 3.15 Correlation analyses between ZBTB10 expression and the expression of genes up-regulated by phenol. Correlation between ZBTB10 expression and the expression of (A) SKI, (B) CD40 and (C) CSF1, was tested by linear regression using gene expression data generated by the Geuvadis consortium for LCLs of 373 individuals.

Functional characterization of the new 8q21 Asthma risk locus | 162

Chapter 3 3.4 DISCUSSION

Results from the previous Chapter showed that (i) individuals carrying the rs2370615:C allergy-predisposing allele (which is in the same haplotype as the rs7009110:T allele), have higher PAG1 expression levels; (ii) a PRE construct carrying the rs2370615:C allele is able to drive greater PAG1 expression; and (iii) Foxo3a binds less frequently to PRE3 in the presence of the rs2370615:C allele. These observations suggest that the rs2370615:C allele promotes greater PAG1 transcription by allowing other TFs to bind to PRE3. We hypothesised that one such TF was RelA, because a binding site for this TF overlaps PRE3 and, specifically, rs2370615. RelA, like all inactive NF-κB subunits, is localized in the cytoplasm and needs to be activated and translocated into the nucleus to be functional. NF-κB plays a key role in regulating the immune response and thus, activation requires a finely tuned cascade of events to take place. When inactive, NF-κB dimers are retained in the cytoplasm by inhibitory IκB proteins that bind to them, only dissociating and allowing NF-κB translocation into the nucleus upon appropriate activation. Two major signalling pathways activate the translocation of NF-κB dimers into the nucleus – the classical (or canonical) and the alternative (or noncanonical) pathway [40]. The classical pathway is crucial for innate immunity, being activated by several inflammatory signals such as TNF-α or pathogen-associated molecular patterns (PAMPs), leading to the transcriptional activation of multiple inflammatory and innate immune genes [86]. On the other hand, the alternative pathway plays a central role in lymphoid organ development and adaptive immunity, being activated by several different cytokines including IFN-α [78].

Thus, we hypothesized that (i) in resting LCLs, RelA is preferentially located in the cytoplasm and therefore does not activate PAG1 transcription; (ii) upon translocation into the nucleus, RelA would induce PAG1 expression in rs2370615:CC LCLs more strongly than in rs2370615:TT LCLs, because in the latter Foxo3a preferentially binds to PRE3; (iii) stimulating LCLs with either TNF-α or IFN-α would allow us to investigate which NF-κB pathway is involved in PAG1 regulation.

To test these hypotheses, we first investigated if we could promote RelA translocation from the cytoplasm into the nucleus. For this, we stimulated rs2370615:CC and TT LCLs with TNF-α or IFN-α and assessed RelA protein levels in nuclear protein extracts by Western Blot. Results from these experiments demonstrated that stimulation with TNF-α induced translocation of RelA into the nucleus in both rs2370615:TT and CC LCLs. Surprisingly, stimulation with IFN-α failed to induce RelA translocation into the nucleus in LCLs. One

Functional characterization of the new 8q21 Asthma risk locus | 163

Chapter 3 possible explanation for this could be the experimental conditions used, specifically IFN-α concentration. NF-κB transactivation has been shown to be IFN-α dose dependent [87] and thus, at the single dose tested in this experimental we might not be able to detect these changes. We did not test additional IFN-α concentrations to test this possibility and instead used only TNF-α stimulation to induce RelA activation in subsequent experiments.

In addition to RelA (or p65), the NF-κB family includes four other subunits: NF-κB1 (or p105/p50), NF-κB2 (or p100/p52), RelB and c-Rel. These proteins exist in cells as homo- and heterodimers with different functions depending on the dimerized subunit(s). Both p50 and p52 lack a transcription activation domain and therefore homodimers of these subunits function as repressors. On the other hand, a transcription activation domain is present in RelA, RelB and c-Rel subunits, and so, dimers containing either RelA or c-Rel function as transcriptional activators. RelB is a more dynamic regulatory subunit that can act as both a repressor or activator [40]. RelA homodimers are the most powerful transcriptional activator NF-κB dimers, however we cannot rule out the hypothesis that RelA might be regulating PAG1 expression together with another NF-κB subunit. As such, a search of available databases (e.g. RegulomeDB [24]) showed evidence that additional NF-κB subunits might bind to PRE3 overlapping rs2370615 (e.g. p50). To assess this possibility, we extended our analysis to all NF-κB subunits, investigating the translocation of each one into the nucleus, by western blot in nuclear protein fractions following TNF-α treatment. Results showed that RelA was the only NF-κB subunit that translocated into the nucleus with TNF- α exposure. Interestingly, when compared to rs2370615:TT, rs2370615:CC LCLs contained higher nuclear levels (even in resting state) of RelB and c-Rel subunits, both known to act as transcriptional activators. We did not further investigate or replicate this result in additional LCLs and thus, it is unclear what the genotype contribution is to such observation. However, previous studies report that RelB and c-Rel are the main subunits that are activated during dendritic cell differentiation and maturation, in both humans and mice [88, 89]; and a role for RelB in B cell antigen presenting function has also been attributed [90]. Therefore, it is possible that individuals carrying the asthma-risk rs2370615:C genotype, have increased levels of RelB and c-Rel subunits, which in turn may increase airway inflammation by modulating the aforementioned events.

Additionally, we measured PAG1 protein levels in the cytoplasmic protein fraction of these cells, which is where PAG1 is preferentially located. Results showed that following stimulation with TNF-α, the levels of PAG1 did not increase as we predicted. In fact, our results suggested that there might be a slight reduction of PAG1 in the cytoplasm following

Functional characterization of the new 8q21 Asthma risk locus | 164

Chapter 3 TNF-α exposure. Given these results, we hypothesized that the observed reduction of PAG1 levels was due to increased protein trafficking to the plasma membrane – the primary established functional location of PAG1. Thus, we measured PAG1 in membrane protein extracts prepared from rs2370615:TT and CC LCLs and the results showed that TNF-α had no effect on cell membrane PAG1 content. We therefore concluded that the apparent decrease in PAG1 levels in the cytoplasm following TNF-α exposure was likely a gel loading or film exposure artefact of the technique.

Although IFN-α did not induce the translocation of RelA into the nucleus, we nonetheless assessed the possibility that it could modulate PAG1 expression through a mechanism independent of NF-κB (e.g. JAK/STAT pathway [91]). However, IFN-α exposure also failed to induce changes in PAG1 levels.

We also investigated the possibility that TNF-α and IFN-α might affect PAG1 transcription, despite no effect on protein levels. However, analysis of PAG1 mRNA levels showed that baseline transcription levels remained unchanged after TNF-α or IFN-α exposure. These results demonstrate that neither cytokine per se is able to strongly induce PAG1 transcription in LCLs. We therefore studied other potential stimuli that could be required to modulate PAG1 transcription.

Because PAG1 was recently shown to be the target of a hypoxia-inducible transcription factor (HIF) [79] and hypoxia has a critical effect on inflammatory processes [92, 93], we first studied the effect of oxygen deprivation on PAG1 expression. We observed only a slight and non-significant increase in PAG1 transcription in LCLs incubated in hypoxic conditions for 1 hour, with no major effect of rs2370615 genotype. PAG1 protein levels on the cell membrane were also unaffected by the same hypoxic conditions, suggesting that they were not sufficient to affect PAG1 expression. We then exposed LCLs to severe hypoxic [79] conditions (0.2% O2) to replicate the observations reported by Schörg et al . However, even in these more extreme conditions, PAG1 remained unchanged. These results indicate that the effect of hypoxia on PAG1 expression reported by Schörg et al. in cancer cell lines (HeLa, Hep3B, U2OS, MCF-7) does not extend to LCLs.

Aeroallergens are known to play a critical role in the development and exacerbation of asthma [94], including house dust mite, pollen, mould and animal dander. The most common of these in the home environment, and to which most asthmatics are allergic to, are house dust mite (HDM) species Dermatophagoides farinae and Dermatophagoides pteronyssinus. The major protein allergens identified in these are Der f 1 and Der p 1

Functional characterization of the new 8q21 Asthma risk locus | 165

Chapter 3 respectively, cysteine proteases found in the mites’ gut and faeces that are not only immunogenic but also increase the permeability of the bronchial mucosa when inhaled. As a consequence, the amount of allergen crossing the epithelium is increased, allowing access to and sensitization of immune cells [95, 96]. Because SNPs in the 8q21 region are associated asthma and other allergic diseases [97], we hypothesised that allergen exposure might trigger a key pathway that involves modulation of PAG1 expression. We extended this hypothesis to ZBTB10, given that this gene was also found to be a target of 8q21 asthma risk variants.

We began by incubating LCLs with a HDM solution that is frequently used for skin prick reactivity testing in humans. The glycerine buffer used in this HDM solution was included as a negative control. Results from this experiment showed that both solutions resulted in significant increase in PAG1 expression (about 2-fold compared to baseline). For ZBTB10, this effect was even stronger, with gene expression increasing 4- to 7-fold when compared to baseline. To explain these results, we first hypothesised that both solutions might be rich in LPS content, a proinflammatory molecule that can activate cells via Toll-like receptor (TLRs), specifically TLR4 [98]. As such, we exposed LCLs to increasing concentrations of LPS, before and after exposure to a TLR4-antagonist, but did not detect any significant effect on PAG1 and ZBTB10 expression. These results suggested that another component of the HDM and glycerine solutions was responsible for up-regulation of PAG1 and ZBTB10.

When we examined the ingredients of the HDM and glycerine solutions, we found that phenol was common to both. Phenol is a common excipient used as a stabilizing agent and preservative in several pharmaceutical formulations including vaccines. Phenolic compounds have been extensively linked to the modulation of reactive oxygen species (ROS) and oxidative stress. ROS are chemically reactive species that contain oxygen and are produced within the cell in a biological context as a natural by-product of mitochondrial functions, playing crucial roles in regulatory pathways [99]. These species are constantly generated and eliminated in a biological system through a finely tuned system to maintain homeostasis, and a shift in the balance of events to either side (excess or scarcity of ROS) has been shown to have deleterious effects in cell structures when the system is not able to counteract this shift and restore the ROS pool back to physiological levels. Several agents used for cancer therapy induce ROS and this plays an important role in their anticancer activities. The use of natural phenols, such as curcumin – the active ingredient of the Indian spice turmeric – has been shown to have promising results in cancer therapy.

Functional characterization of the new 8q21 Asthma risk locus | 166

Chapter 3 The positive outcomes of these compounds have been attributed to their ability to modulate ROS production and down-regulate the expression of the Sp transcription factors, as well as several pro-oncogenic Sp-regulated genes that are important for cancer cell proliferation, survival and metastasis. This regulation has been shown to be due to ROS-mediated destabilization and degradation of a micro RNA – mir-27a, of which ZBTB10 is a target [84]. As mentioned before, ZBTB10 is also a repressor of the Sp proteins that are known to be highly expressed in cancer cells and tumours [100]. Also, overexpression of Sp1 in different cells lines leads to increased production of chemokine CXCL4 and immune cell recruitment both in vivo and in vitro[85]. Thus, we hypothesised that phenol was the component of HDM and glycerine solutions that was driving increased expression of ZBTB10 and PAG1.

To test this hypothesis we exposed LCLs to increasing concentrations of pure phenol. We also tested an additional experimental condition, HDM allergen extract resuspended in PBS instead of the glycerine buffer. Results from this experiment showed a striking increase of ZBTB10 expression following phenol exposure and that this response was dose dependent, with a significant effect at the lowest dose and peaking at the concentration that is equivalent to that present in the HDM solution. Interestingly, we also observed an up-regulation of PAG1 in cells stimulated with phenol, suggesting that both genes play a role in the response to oxidative stress. Incubation with a HDM extract in PBS had no effect on the expression of either gene when compared to baseline. Together, these results indicate that phenol is responsible for the up-regulation of gene expression observed in cells exposed to the glycerine or HDM solutions.

To investigate if the phenol effect was cell type specific, we also tested another human cell line (LUVA), derived from human mast cells. Mast cells are immune cells that play a critical role in asthma pathophysiology by releasing several proinflammatory mediators and cytokines [101]. These cells were also incubated with phenol and expression levels of PAG1 and ZBTB10 were measured. Both PAG1 and ZBTB10 were also up-regulated in these cells, although not as dramatically as observed in LCLs. These results indicate that the phenol effect is not cell type specific and instead extends to other immune cells. Our results also support the hypothesis that both genes are involved in the response to oxidative stress.

To investigate whether the phenol-induced changes were genotype-dependent, PAG1 and ZBTB10 expression were measured in LCLs from individuals homozygous for either the

Functional characterization of the new 8q21 Asthma risk locus | 167

Chapter 3 asthma predisposing rs2370615:T (n=6) or the protective rs2370615:C (n=7) alleles. We found no significant differences in phenol-induced gene expression for both genes. However, because the sample size was small, these results do not rule out the possibility that there are true differences in expression levels between genotypes, detectable only with larger sample sizes. Future studies with increased power to test this hypothesis are warranted.

We did not perform any experiments to dissect the molecular mechanisms underlying the effect of phenol on PAG1 and ZBTB10 expression. For example, we do not know which of the interacting PREs (see Chapter 2, section 2.3.4) and TFs are involved. We speculate that RelA is involved, because phenol induces the NF-κB pathway [102] and therefore nuclear translocation of RelA. Foxo3a might also play a role, because this TF is up- regulated by ROS via a mir-27a-dependent mechanism [103], and has a binding site disrupted by the rs2370615:C allele (see Chapter 2, section 2.3.6). To further elucidate the regulatory mechanisms involved, we suggest that future studies should repeat our experimental conditions using LCLs with disrupted PREs (e.g. CRISPR/Cas9 deletion of the Foxo3a binding site, or of an entire PRE).

To identify likely target genes of ZBTB10, we analysed the expression profile of a panel of 96 immune-related genes in LCLs after exposure to phenol, which strongly up-regulates ZBTB10 expression. We identified six genes significantly up-regulated by phenol and so, that could be potentially modulated by ZBTB10: SKI, a repressor of TGF-β signalling [104]; CD40, a cell surface receptor that induces cytokine and chemokine production upon ligand interaction [105]; IL5, a proinflammatory Th2-type cytokine and key mediator or eosinophil activation [106]; PTGS2, a key player in the production of inflammatory prostaglandins [107]; NOS2A, required for IL12 signalling in innate immunity [108]; and CSF1, critical for differentiation, survival and proliferation of macrophages [109]. All six genes were also up- regulated in a mast cell line, suggesting that the observed effect might be extended to other immune cell types. If these genes are indeed targets of ZBTB10, then we would expect that their expression levels are correlated with that of ZBTB10. We tested this possibility using gene expression measured in LCLs from 373 individuals by the Geuvadis consortium and showed that ZBTB10 is correlated with two of the six genes, SKI and CD40. Furthermore, the correlation between ZBTB10 and CD40 was genotype dependent, being strongest in individuals carrying the rs2370615:C allele. These findings suggest that CD40 transcription is induced by ZBTB10 in individuals that have the asthma predisposing allele, but not (or to a lower extend) otherwise. No previous association with asthma or

Functional characterization of the new 8q21 Asthma risk locus | 168

Chapter 3 allergic disease was found for SKI, however a multitude of studies have been done on CD40 and its role in asthma pathophysiology, specifically in allergen-mediated disease. CD40 is a member of the Tumor Necrosis Factor Receptor superfamily and SNPs in this gene have been implicated in asthma risk, mainly by impacting IgE production [110]. Results from in vivo and in vitro studies showed that CD40: (i) is significantly increased in airway smooth muscle from asthmatics [111, 112], (ii) regulates B cell proliferation and IgE production [113], (iii) is expressed in airway epithelial cells and may modulate airway inflammation [114]; and (iv) modulates the antigen presenting abilities of dendritic cells by regulating the expression of MHC class II and co-stimulatory molecules [115, 116]. Moreover, both SP1 and NF-κB subunit RelB have been implicated in pathways involving CD40 [34, 116]. Therefore, because: (a) ZBTB10 is a SP1 repressor; (b) expression of ZBTB10 and CD40 are correlated; and (c) RelB levels are higher in individuals carrying the asthma- risk rs2370615:C, it is possible that ZBTB10 also plays a role in the mechanisms mentioned above, underlying its association with asthma.

Overall, results from this chapter indicate that: (i) phenol can up-regulate PAG1 and ZBTB10 expression in both LCLs and a mast cell line, suggesting that both genes are involved in oxidative stress response; (iii) SKI and CD40 are likely targets of ZBTB10.

Functional characterization of the new 8q21 Asthma risk locus | 169

4

PAG1 and ZBTB10 expression in human and mouse tissues

Chapter 4 4.1 INTRODUCTION

4.1.1 Background

In the previous Chapters, we identified PAG1 and ZBTB10 as target genes of 8q21 allergy risk variants. The main aim for the remaining Chapters of this thesis is to investigate the role of both genes in asthma pathophysiology. To inform those studies, it is important to determine in which cell types and tissues these genes are expressed. Results from large scale expression studies such as the Functional Annotation of the Mammalian Genome

(FANTOM5) and Genotype-Tissue Expression (GTEx) projects, can be used to address this question, both for human and mouse tissues. However, not all relevant cell types are represented in those studies. To address this limitation, we measured the expression of

Pag1 and Zbtb10 in a small number of mouse tissues and cell types that were of interest but not represented in publicly available databases.

4.1.2 Aims

To identify the main tissues and cell types expressing PAG1 and ZBTB10, in humans and mice, using both publicly available information and experimental approaches.

Functional characterization of the new 8q21 Asthma risk locus | 171

Chapter 4 4.2 METHODS

4.2.1 Characterisation of PAG1 and ZBTB10 expression in human and mouse tissues based on information from public databases

4.2.1.1 PAG1 and ZBTB10 expression in human tissues

The GTEx database [117] was queried to determine gene expression in a range of human broad tissue types. In the database https://www.gtexportal.org we used the search terms “PAG1” and “ZBTB10” and selected “Homo sapiens” as the organism. Results shown were organized by gene and by expression level in descending order.

The Expression Atlas database was queried to determine gene expression in human immune cells. In the database https://www.ebi.ac.uk/gxa/home we used the search terms “PAG1” and “ZBTB10” and selected “Homo sapiens” as the organism. Results shown correspond to “The BLUEPRINT Epigenome project” dataset.

4.2.1.2 Pag1 and Zbtb10 expression in mouse tissues

The Expression Atlas database was also queried to determine gene expression in a spectrum of mouse tissues using the search terms “Pag1” and “Zbtb10”. For Pag1, results were obtained from the FANTOM5 project [118], as this was the most comprehensive dataset available. Data for Zbtb10 was not available in this project, and so we selected results from the dataset with the most comprehensive tissue coverage for this gene: dataset “9” [119]. The selected organism was “Mus musculus” strain C57BL/6J.

The Immunological Genome Project (ImmGen) database [120] was queried to determine gene expression in mouse immune cells. In the database https://www.immgen.org/ we chose the “My GeneSet” browser and used the search term “Pag1,Zbtb10” using all cell types from the “V2 dataset”. Results shown correspond to the interactive heat map and were organized by gene and expression level in descending order. The selected organism was Mus musculus.

Functional characterization of the new 8q21 Asthma risk locus | 172

Chapter 4 4.2.2 Quantification of Pag1 and Zbtb10 gene expression in mouse tissues of interest

4.2.2.1 Mouse strains

All WT mice used were of C57/Bl6 background and specific pathogen free (SPF) purchased from UQ Biological Resources. Experiments were performed in animals between eight to twelve weeks of age. All experiments were approved and performed in accordance with the UQ’s Animal Care and Ethics Committees.

4.2.2.2 Collection of lung tissue, mediastinal lymph nodes and airway epithelial cells

Following euthanasia by pentobarbital overdose, lungs (n=7) and mediastinal lymph nodes (MLN; n=4) were harvested from wild-type (WT) mice. Lung lobes were dissected, snap frozen in dry ice and stored at -80°C until needed. MLN cell suspensions were achieved using a cell strainer (40μm, BD Falcon), pelleted, resuspended in 1mL Gey’s lysis buffer to lyse red blood cells and washed twice in 1% FBS in PBS buffer. Cells were then counted using a Neubauer improved chamber (Hirschman EM Techcolor) and 1x106 cells were transferred to an eppendorf tube, and pelleted by centrifugation at 1,200rpm for 5min. Cells were resuspended in 100μL of TRIzol Reagent (Invitrogen) and stored at -80°C until needed. Airway epithelial cells from naïve WT mice tracheas (n=2) were harvested as described by Rock et al [121]. Cells were resuspended in 100μL of TRIzol and stored at - 80°C until needed.

4.2.2.3 Quantification of Pag1 and Zbtb10 gene expression

4.2.2.3.1 mRNA extraction mRNA from lung tissue was extracted from the right post-caval lobe using the TRIzol method. Briefly, tissue was homogenized in 500μL of TRIzol using a tissue tearor homogenizer (T10 basic Ultra-Turrax, IKA). After homogenization, samples were topped up to 1mL of TRIzol, incubated at RT for 5min and centrifuged at 13,000rpm for 10min at 4°C. Supernatants were then transferred into fresh tubes. 200μL Chloroform was added to the samples, mixed thoroughly for 15secs and incubated at RT for 5mins followed by centrifugation at 13,000rpm for 10mins at 4°C. The colourless aqueous phase was then collected and transferred into a new tube. The solution volume was doubled with isopropanol (Chem-Supply) and samples were vortexed for 10secs, followed by a 5min

Functional characterization of the new 8q21 Asthma risk locus | 173

Chapter 4 incubation at RT, before being centrifuged at 13,000rpm for 10mins at 4°C. The supernatants were discarded and the pellets washed in 75% ethanol, and centrifuged at 10,000rpm for 5mins at 4°C. Supernatants were discarded and the pellets air dried at RT, before being resuspended in 50μL T10E1 buffer (Tris 10 mM, EDTA 1 mM, pH 8). mRNA from MLN and airway epithelial cells was extracted as described in Chapter 3, section 3.2.1.1.2 (i).

4.2.2.3.2 cDNA conversion

See Chapter 3, section 3.2.1.1.2 (ii).

4.2.2.3.3 Gene expression analysis

Measured as described in Chapter 3, section 3.2.1.1.2 (iii), using the Taqman Gene Expression assays Mm00474700_m1 (Pag1) and Mm01281740_m1 (Zbtb10) from Life Technologies). Results were normalized against Gapdh (assay Mm99999915_g1). Each gene was measured in duplicate and results are shown as delta Ct (ΔCt) values, calculated as ΔCt = CtGapdh – Ctgene of interest. A negative ΔCt value indicates that the gene of interest (Pag1 or Zbtb10) was expresser at lower levels (i.e. have a higher Ct value) when compared to Gapdh. A ΔCt value close to 0 indicates similar expression levels between Gapdh and the gene of interest.

4.2.3 Pag1 protein expression in mouse lung tissue

4.2.3.1 Tissue collection

Lungs were harvested from two WT mice previously challenged with HDM (n=1) or PBS (n=1) as follows. At day 0, mice were challenged intranasally with PBS/HDM (100μg, batch 14, Greer Laboratories). At days 14, 15, 16 and 17, mice were re-challenged with PBS/HDM (10μg) and euthanized by pentobarbital overdose 3h after the last challenge. Mice challenged with HDM according to this protocol develop airway inflammation with a characteristic Th2 eosinophilic component [122]. Lung lobes were dissected and the right superior lobe fixed in formalin and used to prepare 5μm paraffin-embedded sections for histologic analysis of Pag1 expression.

Functional characterization of the new 8q21 Asthma risk locus | 174

Chapter 4 4.2.3.2 – Histologic analysis of Pag1 expression (Immunofluorescence)

Lung tissue sections were deparaffinised and rehydrated through sequential incubations in Xylene and absolute or 70% ethanol solutions (Chem-Supply). Antigen retrieval was performed by incubating the slides in citrate buffer (10mM Citric Acid, 0.05% Tween 20, pH 6.0) in a pressure cooker for 10min using the maximum heat setting of a microwave. Slides were then allowed to cool down to RT before being incubated with 0.5% TritonX- 100/PBS for 10min to permeabilize the cells. Samples were rinsed twice with 0.05% Tween-20/TBS (TBS-T) and blocked with 10% normal goat serum (Sigma-Aldrich) for 30min, to avoid unspecific antibody binding. The blocking solution was then discarded and an anti-Pag1 antibody (rabbit polyclonal ab14989; 1:200 dilution in 10% FCS/PBS; Abcam) was added to stain samples overnight at RT in a humidity chamber. The following day, samples were washed with 0.6% Tween-20/PBS (PBS-T) for 10min and then twice with 0.05% TBS-T for 3min, before being incubated with an anti-rabbit Alexa fluor 488- conjugated secondary antibody (AF488-conjugated polyclonal goat anti-rabbit IgG; 1:500 dilution in 10%FCS/PBS; Invitrogen) for 1 hour at RT. Slides were washed with 0.6% PBS- T for 10min, 0.05% Tween-20/PBS (TBS-T), and 0.05% Tween-20/PBS both for 3min.

Nuclei were stained with DAPI (1:100 dilution in MQ H20; Sigma-Aldrich) for 5min at RT followed by three washes of 2min with H20. Finally, slides were mounted using anti-fade mounting media (Dako) and scanned using a fluorescent slide scanner (Aperio ScanScope FL). The acquired images were analysed using the ImageScope software.

Functional characterization of the new 8q21 Asthma risk locus | 175

Chapter 4 4.3 RESULTS

4.3.1 Assessing PAG1 and ZBTB10 expression in human and mouse tissues and cell types based on publicly available databases

4.3.1.1 Human tissues

When considering the 53 broad human tissue types studied by the GTEx consortium [123], medium expression levels (defined as 11-1,000 transcripts per million [TPM]) for PAG1 were observed in seven tissues: lung (median 30 TPM), two brain regions (30 TPM), whole-blood (19 TPM), spleen (16 TPM), small intestine (14TPM) and LCLs (14 TPM; Figure 4.1). All other tissues had low expression levels (defined as 0.5-10 TPM) of PAG1. On the other hand, ZBTB10 is expressed at medium levels in 11 tissues (Figure 4.2), none of which immediately relevant to asthma pathophysiology: for example, artery (16 TPM), uterus (14 TPM), brain (13 TPM) and prostate (12 TPM). Low expression levels were reported in all other tissues, including lung (8 TPM) and spleen (5 TPM). Of note, the tissue with weakest expression of ZBTB10 was whole-blood (0.5 TPM). For comparison, the median expression of the house-keeping gene GAPDH was high (defined as >1,000 TPM) for >50% of the tissues studied by the GTEx consortium (Figure 4.3).

Expression results from 27 human immune cells studies by the BLUEPRINT Epigenome project are shown in Figure 4.4. PAG1 was more highly expressed in CD8-positive (144 TPM) and CD4-positive (136 TPM) alpha-beta thymocytes, neutrophils (112 TPM), a range of alpha-beta T cells, and regulatory T cells (Tregs; 73 TPM). On the other hand, ZBTB10 was more highly expressed in CD8-positive alpha-beta T cells (101 TPM), conventional dendritic cells (cDCs; 81 TPM) and naïve B cells (55 TPM). For comparison, in the immune cells studied by this project, the median expression of GAPDH varied between 81 and 2000 (seven tissues with expression >1,000 TPM; 17 tissues between 200 and 1000 TPM).

4.3.1.2 Mouse tissues

Across 35 broad mouse tissues studied by the FANTOM5 project, Pag1 was most highly expressed in lymph nodes (12 TPM), bone (9 TPM), thymus (9 TPM) and lung (7 TPM) (Figure 4.5). For comparison, expression levels for Gapdh in these tissues varied between 0.5 and 6 TPM. Across the nine broad tissues studied by Huntley et al [119], Zbtb10 was most highly expressed in testis (16 TPM), lung (10 TPM), brain (8 TPM), heart (6 TPM)

Functional characterization of the new 8q21 Asthma risk locus | 176

Chapter 4 and spleen (5 TPM). The expression of Gapdh in these tissues varied between 10 and 421 TPM. In Siggs et al [1] , gene expression datasets from BioGPS [124] were used to analyse the expression of all members of the BTB-ZF family, including Zbtb10 (Figure 4.6). Results from his study corroborate the high expression of Zbtb10 found in testis, lung and heart tissues. Zbtb10 was also highly expressed in retina, pancreas and large intestine.

Expression results for murine immune cells were obtained from the Immunological Genome project and are shown in Figure 4.7. The highest expression of Pag1 was observed in neutrophils and lymphocytes, similar to the pattern observed in human immune cells. In addition, macrophages and basophils also expressed Pag1 at high levels. On the other hand, Zbtb10 expression was highest in natural killer (NK) cells, innate lymphocytes, mast cells and stromal cells. The expression profile described for Zbtb10 by Siggs et al. also included immune cells; highest Zbtb10 expression was observed in NK cells, Tregs, plasmacytoid dendritic cells (pDCs), mast cells and haematopoietic stem cells (Figure 4.6).

Functional characterization of the new 8q21 Asthma risk locus | 177

Chapter 4

Figure 4.1 PAG1 expression profile in human tissues. Data obtained from the GTEx portal using the search term “PAG1”. Gene expression was measured and presented as Transcripts Per Million (TPM) and sorted left to right, from highest to lowest expression.

Functional characterization of the new 8q21 Asthma risk locus | 178

Chapter 4

Figure 4.2 ZBTB10 expression profile in human tissues. Data obtained from the GTEx portal using the search term “ZBTB10”. Gene expression was measured and presented as Transcripts Per Million (TPM) and sorted left to right, from highest to lowest expression.

Functional characterization of the new 8q21 Asthma risk locus | 179

Chapter 4

Figure 4.3 GAPDH expression profile in human tissues. Data obtained from the GTEx portal using the search term “GAPDH”. Gene expression was measured and presented as Transcripts Per Million (TPM) and sorted left to right, from highest to lowest expression.

Functional characterization of the new 8q21 Asthma risk locus | 180

Chapter 4

Figure 4.4 PAG1, ZBTB10 and GAPDH expression profiles in human immune cells. Data obtained from The Expression Atlas database using the search terms “PAG1”, “ZBTB10” and “GAPDH”, using Homo sapiens as the organism. Gene expression was measured and presented as Transcripts Per Million (TPM) with darker shades of blue corresponding to higher expression.

Functional characterization of the new 8q21 Asthma risk locus | 181

Chapter 4

Figure 4.5 Pag1, Zbtb10 and Gapdh expression profiles in mouse tissues. Data obtained from The Expression Atlas database using the search terms “Pag1”, “Zbtb10” and “Gapdh”, using Mus musculus as the organism. Gene expression was measured and presented as Transcripts Per Million (TPM) with darker shades of blue corresponding to higher expression.

Functional characterization of the new 8q21 Asthma risk locus | 182

Chapter 4

Figure 4.6 Zbtb10 expression profile in mouse tissues and immune cells. Hematopoietic and non-hematopoietic tissues are represented in gray and black columns, respectively. Figure adapted from Siggs et al [1].

Functional characterization of the new 8q21 Asthma risk locus | 183

Chapter 4

Figure 4.7 Pag1 and Zbtb10 expression profiles in mouse immune cells. Data obtained from The Immunological Genome Project (ImmGen) database using the search terms “Pag1,Zbtb10”. The dataset presented corresponds to the cell type V2 dataset from Mus musculus. Results are shown as a heat map (one per gene), with cell types sorted from highest (left, in red) to lowest (right, in blue) gene expression levels. Each cell line name (shown above the heat map) and assigned a colour that represents the main cell population it belongs to (see legend on the right) Functional characterization of the new 8q21 Asthma risk locus | 184

Chapter 4 4.3.2 Quantification of Pag1 and Zbtb10 gene expression in mouse tissues of interest

To confirm that both Pag1 and Zbtb10 are expressed in relevant tissues, lung and mediastinal lymph nodes (MLN) were harvested from WT mice. Gene expression was measured using TaqMan assays. We found that Pag1 was expressed at similar levels in lung tissue and MLN, whereas Zbtb10 was more highly expressed in lung tissue (Figure 4.8).

Figure 4.8 Pag1 and Zbtb10 expression in mouse tissues of interest. RNA was extracted from WT lung tissue (n=7) and mediastinal lymph nodes (n=4) using the TRIzol method. Gene expression was measured using TaqMan assays. Pag1 and Zbtb10 gene expression was normalized to Gapdh control and shown as ΔCt values. A ΔCt value of 0 indicates that Gapdh and the gene of interest were expressed at similar levels. A negative ΔCt value indicates that the gene of interest was expressed at lower levels than Gapdh. **P<0.01, *P<0.05, calculated using a Mann- Whitney test.

Functional characterization of the new 8q21 Asthma risk locus | 185

Chapter 4 Following confirmation that Pag1 and Zbtb10 were expressed in lung tissue, we thought to investigate if these genes were expressed specifically in airway epithelial cells (AECs). These cells line the respiratory tract and constitute the first point of contact with allergens and potential pathogens, playing a critical role in asthma pathophysiology [125]. Thus, airway epithelial cells were harvested from the tracheas of WT mice and gene expression measured using TaqMan assays. Results are shown in Figure 4.9 and indicate that both genes were expressed in airway epithelial cells.

Figure 4.9 Pag1 and Zbtb10 expression in mouse airway epithelial cells. RNA was extracted from WT airway epithelial cell samples (n=2) using the TRIzol method. Gene expression was measured using TaqMan assays. Pag1 and Zbtb10 gene expression was normalized to Gapdh control and shown as ΔCt values. A ΔCt value of 0 indicates that Gapdh and the gene of interest were expressed at similar levels. A negative ΔCt value indicates that the gene of interest was expressed at lower levels than Gapdh.

Functional characterization of the new 8q21 Asthma risk locus | 186

Chapter 4 4.3.3 Expression of Pag1 protein in mouse lung tissue

Following the observation that Pag1 mRNA was expressed in airway epithelial cells, we thought to confirm that Pag1 protein was also detectable in these cells. For this, lung tissue sections were prepared from two WT mice (one previously challenged with PBS, the other with HDM) and Pag1 protein stained using an immunofluorescence protocol. We found that AECs from both mice stained positively for Pag1 (Figure 4.10A and B). Pag1 staining in the HDM-challenged mouse was also observed within immune cells in the lung tissue (Figure 4.10C).

Figure 4.10 Pag1 immunofluorescence staining in mouse lung tissue. Lung tissue from two WT mice exposed to (A) PBS (n=1) or (B) HDM (n=1), were used to stain Pag1 protein. Shown are example images from airways and the second image corresponds to a close-up of the white box section in the first picture of each row. (C) Examples of positively stained immune cells for Pag1 in the lung tissue from the HDM-challenged mouse. Pag1 is stained green using an AF488 antibody and nucleic acids are stained blue with DAPI.

Functional characterization of the new 8q21 Asthma risk locus | 187

Chapter 4 4.4 DISCUSSION

Results from this Chapter provide an overview of PAG1 and ZBTB10 expression in human and murine tissues and cell types.

We first used results from published large scale expression studies to investigate the pattern of PAG1 and ZBTB10 expression in human tissues and immune cells. This analysis identified tissues that have very distinct patterns of expression between PAG1 and ZBTB10, for example, whole blood and LCLs both with higher expression of PAG1 and lower expression of ZBTB10; and tissues from the reproductive and cardiovascular systems, where the reverse was observed. However, we also identified tissues for which the relative expression levels where comparable between the two genes, notably, lung and spleen.

The distinct pattern of expression for PAG1 and ZBTB10 was also evident when we considered immune cell subtypes. On the one hand, PAG1 had highest expression in lymphocytes and neutrophils, pointing to a potential role for this gene in both innate and adaptive immune response, where these cells play critical roles [126, 127]. On the other hand, ZBTB10 expression was highest in cDCs and naïve B cells, the former suggesting a role for ZBTB10 in innate immune responses [128].

Results from gene expression in mouse tissues and immune cells, measured by the FANTOM5 and Immunological Genome projects, respectively, showed a similar pattern to that of humans for both Pag1 and Zbtb10. For Pag1, expression was highest in mouse lung tissue, neutrophils and lymphocytes, as was in humans. These observations again suggest that Pag1 might influence both innate and adaptive immune response. For Zbtb10, higher expression in mouse tissues from the cardiovascular and reproductive systems was also consistent with the observations in human tissues. At the level of immune cell types, similarities were not so obvious between the mouse and human studies. Results from Siggs et al. [1] confirmed that Zbtb10 is more highly expressed in mouse dendritic cells, supporting a potential role for this gene in the regulation of innate immune responses.

Results from our own experiments confirmed the expression of both Pag1 and Zbtb10 in murine lung tissue. Moreover, our findings support the observation in the FANTOM5 dataset that the expression of Zbtb10 in lung tissue was higher than that of Pag1.

Functional characterization of the new 8q21 Asthma risk locus | 188

Chapter 4 Finally, we showed that both Pag1 gene and protein were expressed specifically in airway epithelial cells. Few studies have described the patterns of PAG1 expression in different tissues and cell types [36, 54], and findings from these mainly confirm the results that we have described above. To our knowledge, Pag1 expression in murine AECs has not been reported before, representing a novel finding from this Chapter. Pag1 expression needs to be assessed in additional samples to draw definitive conclusions however, our findings suggest that PAG1 might be involved in the response of AECs to allergen challenge For example, PAG1 might modulate the production of cytokines or mucus by AECs and/or directly affect cellular adhesion and consequently the epithelium architecture [125].

Overall, results from this Chapter indicate that both PAG1 and ZBTB10 are expressed in tissues and immune cell types directly relevant to asthma, suggesting that these genes may play important roles in airway inflammation. This possibility is addressed in the following Chapters of this thesis.

Functional characterization of the new 8q21 Asthma risk locus | 189

5

The contribution of Pag1 to allergen- induced airway inflammation in mice

Chapter 5 5.1 INTRODUCTION

5.1.1 Background

PAG1 is mostly expressed in the immune system, notably T and B cells [36, 53, 54] playing an important role in the regulation of Csk activity [36] with direct effects on immunoreceptor signalling [37]. To our knowledge, PAG1 has not been previously implicated in allergic disease pathophysiology; however, its involvement in the development of immune responses has been extensively studied. Results from these studies, that included both in vivo and in vitro experimental systems, have been conflicting [37]. On the one hand, in vitro studies suggest that PAG1 might play an anti-inflammatory role due to its inhibitory activity in immune cell activation when this gene is overexpressed [46, 47, 49, 50]. On the other hand, results from in vivo studies indicate that immune cell development and function is not affected in Pag1-deficient animals, suggesting that PAG1 could have a redundant role [51, 52]. Our finding from Chapter 2 that allergy predisposing alleles increase the transcription of PAG1 in human B cell lines suggests that PAG1 overexpression might instead have a pro- inflammatory role, for example by promoting B cell activation.

5.1.2 Hypothesis

Pag1 knockout (KO, Pag1-/-) mice have attenuated airway inflammation after allergen challenge when compared to wild-type mice (WT).

5.1.3 Aim

To characterise the airway inflammation that develops after allergen challenge in Pag1-/- and WT mice.

Functional characterization of the new 8q21 Asthma risk locus | 191

Chapter 5 5.2 METHODS

5.2.1 Mouse strains

All mice used were of C57/Bl6 background. Pag1-/- mice were kindly donated by Prof Jonathan Lindquist from Otto-von-Guericke University (Magdeburg, Germany) [129]. A Pag1-/- strain was re-derived at the University of Queensland (UQ; Brisbane, Australia) using specific pathogen free (SPF) WT C57Bl/6 mice purchased from UQ Biological Resources. Littermate WT and Pag1-/- mice were obtained through Pag1het crossing and used for experiments between eight to twelve weeks of age. All experiments were approved and performed in accordance with the UQ’s Animal Care and Ethics Committees.

5.2.2 Genotyping

5.2.2.1 DNA extraction

DNA for genotyping was extracted from toe clippings from each animal. Samples were incubated overnight at 56ºC in 80μL of lysis buffer (100mM Tris-HCl pH8.0, 5mM EDTA, 200mM NaCl, 0.2% SDS, 10mg/mL Proteinase K; all Sigma-Aldrich). The following day, samples were centrifuged at 13,000rpm for 10min and the supernatant collected, and transferred into a new tube containing 150μL isopropanol (Chem-Supply) followed by a centrifugation step as before. Pellets were washed once in 500μL 70% ethanol, air dried and resuspended in 50μL TE10-0.1 Buffer (10mM Tris-HCl pH8.0, 0.1mM EDTA). Samples were then incubated at 65ºC for 1h and stored at 4ºC.

5.2.2.2 Genotyping PCR

2μL of DNA from each sample was used in a 15μL PCR reaction including: 4μL 5x

Polymerase Buffer; 2μL MgCl2 25mM; 1.8μL dNTPs 12.5mM; 0.16μL Pag1-1 (5’- TTCTTTCAGAAGACAGCACGCTG-3’) and pZeo (5’-GGCCAGGGTGTTGTCCGGCACC- 3’) primers; 0.04μL Pag1-2 (5’-GCGTCCACCGGTCCCTTCTGCAG-3’) primer; and 0.4μL MangoTaq. PCR cycling conditions included an initial incubation at 96°C for 5mins, 35x cycles of [45sec at 94°C, 45sec at 55ºC and 45secs at 72°C], followed by a final extension at 72ºC for 7min. PCR amplified fragments were then separated by electrophoreses in a 3% agarose gel for 1h at 100V. Samples containing a 581bp, 473bp or both bands were

Functional characterization of the new 8q21 Asthma risk locus | 192

Chapter 5 indicative of a WT, Pag1-/- or Pag1het animal, respectively. All mice were genotyped before experimental procedures.

5.2.3 Sample collection

At the time of euthanasia by pentobarbital overdose, blood samples were collected by cardiac puncture and centrifuged for 30min at 13,000rpm to obtain serum. Serum samples were stored at -20ºC for immunoglobulin measurement. Next, lungs were flushed with 600μl of ice cold PBS to recover the bronchoalveolar lavage fluid (BALF). Samples were centrifuged at 5,000rpm for 5min with pelleted cells used for airway inflammation assessment using fluorescent-activated cell sorting (FACS) and supernatants stored at - 20ºC for cytokine measurement. Following BALF recovery, lungs were harvested and lobes used for (i) airway inflammation assessment (left lobe), (ii) fixing and sectioning for immunohistochemistry (right superior lobe); or (iii) snap freezing for mRNA and/or protein expression assessment (remaining lobes). Depending on the experiment, additional organs were also harvested including mediastinal lymph nodes (MLN), spleen and thymus, all used for ex vivo assessment of cytokine production.

5.2.4 Statistical analysis

The software GraphPad Prism version 7.0 (La Jolla, USA) was used to graph and analyse all data, using the Mann-Whitney test. Results are presented as min-to-max box plots.

5.2.5 Validating knockout of gene expression in the Pag1-/- mouse strain

5.2.5.1 Quantification of Pag1 gene expression

5.2.5.1.1 mRNA extraction mRNA from lung tissue was extracted from WT (n=3) and Pag1-/- (n=6) mice, as described in Chapter 4, section 4.2.2.2.1.

5.2.5.1.2 cDNA conversion

As described in Chapter 3, section 3.2.1.1.2 (ii).

Functional characterization of the new 8q21 Asthma risk locus | 193

Chapter 5 5.2.5.1.3 Gene expression analysis

Measured as described in Chapter 3, section 3.2.1.1.2 (iii), using the TaqMan Gene Expression assay Mm00474700_m1 (Pag1) from Life Technologies. Gene expression from each sample was measured in duplicate. Results were normalized against Gapdh

(assay Mm99999915_g1); analysed using the Comparative Ct method and shown as fold change values.

5.2.5.2 Histologic analysis of Pag1 expression (Immunofluorescence)

As described in Chapter 4, section 4.2.3.

5.2.6 Assessing the contribution of Pag1 to airway inflammation triggered in sensitised mice by acute exposure to a high dose (10μg) of allergen

5.2.6.1 Induction of allergic airway inflammation

An established model of allergic asthma was used to induce an acute inflammatory response in the lungs as illustrated in Figure 5.1. Briefly, Pag1-/- and WT mice (n=6 or 8 per group, respectively) were lightly anesthetised with isoflurane before being sensitised to house dust mite (HDM, Dermatophagoides pteronyssinus) by intranasal administration of 100μg of a HDM suspension purchased from Greer Laboratories (batch 14) in 50μl volume (PBS). Mice were then re-exposed to 10μg of the same allergen at days 14, 15, 16 and 17 (challenge phase) via the same route and euthanized 3h after the last challenge for endpoint assessment. Control mice were inoculated with PBS (n=7 or 8 per group) instead of HDM.

Figure 5.1 High dose allergen-induced airway inflammation experimental study design

5.2.6.2 Assessment of allergic airway inflammation

BALF cell pellets were obtained as described, resuspended in 200μL Gey’s lysis buffer to lyse red blood cells and washed twice in FACS buffer (1% Fetal Bovine Serum, FBS in

Functional characterization of the new 8q21 Asthma risk locus | 194

Chapter 5 PBS). Cells were then counted using a Neubauer improved chamber (Hirschman EM Techcolor) and seeded in round bottom 96-well plates, followed by an incubation with 50μL Fc block (1:50 dilution, BD Biosciences) at 4°C for 20min to block non-specific binding of Fc receptor expressing cells (i.e. myeloid and B cells). The reaction was stopped by adding 100μL FACS buffer, and the plate centrifuged at 1,600rpm for 5min at 4°C. Cells were resuspended in 50μL FACS buffer and a 50uL of a fluorescently labelled antibody cocktail was added to stain the cells. Antibodies were used in different combinations depending on the cell type of interest and are listed in table 5.1. Cells were stained for 30min at 4°C, washed twice in FACS buffer and enumerated using a BD LSR Fortessa cytometer (BD Biosciences, San Jose, CA, USA). Results were analysed using FACSDiva v8 (BD Biosciences) or FlowJo v8.8 (Treestar) softwares.

Antibody Conjugate Clone Company Ly6G FITC 1A8 CD11b PerCp Cy5.5 M1/70 CD8 PerCp 53-6.7

NKp46 V450 29A1.4 CD4 V500 RM4-5 CD90.2 V500 53-2.1 BD Biosciences APC SiglecF E50-2440 PE PE FoxP3 MF23 AF647 V500 B220 RA3-6B2 PE MHCII APC-Cy7 M5/114.15.2 eBiosciences CD19 PE eBio1D3 PE CD3 145-2C11 FITC ydTCR APC GL3 F4/80 APC BM8 CD103 PE 2E7 BioLegend CD64 BV421 X54-5/7.1

CD45 BV421 30-F11 Ly6C BV570 HK1.4 CD11c BV785 N418 Siglec H APC 551.3D3 Miltenyi Biotec Table 5.1 List of antibodies used for flow cytometry.

Functional characterization of the new 8q21 Asthma risk locus | 195

Chapter 5 5.2.6.3 Measurement of cytokines, chemokines and immunoglobulins

BALF and serum samples were obtained as described above. Lung homogenates were prepared by homogenising lung lobes in 240μL radio-immunoprecipitation assay buffer (RIPA buffer, Sigma-Aldrich) using a tissue tearor (T10 basic Ultra-Turrax, IKA). Except IL- 13, all cytokines, chemokines and immunoglobulins were measured by ELISA according to the manufacturer’s instructions. Briefly, 96-well plates were coated with an anti-mouse antibody against IL-5 (BD Biosciences); IL-17A and IFNγ (BioLegend); IL-33, CCL24 and CXCL1 (R&D Systems); or 50μg/mL HDM suspension. The plates were kept at 4°C (IL-5, IL-17A, IFNγ and HDM) or at room temperature (RT; IL-33, CCL24 and CXCL1) overnight. Wells were then incubated with 3% BSA (Bovine Serum Albumin) in PBS to block nonspecific binding. Samples were then added to the wells and incubated for 2 hours at room temperature before addition of a biotinylated detection antibody (anti-IL-5 [BD Biosciences]; anti-IL-17A and anti-IFNγ [BioLegend]; anti-IL-33, anti-CCL24, and anti- CXCL1 [R&D Systems]; and anti-IgG1 [Southern Biotech]) followed by avidin-horseradish peroxidise (HRP, BioLegend). Washing steps were performed using a 0.05% Tween- 80/PBS solution. Tetramethylbenzidine substrate (TMB, Sigma-Aldrich) was added to each well and the reaction was stopped with 1M H2SO4 (Sulfuric acid). Optical density was determined at 450nm by using a microplate reader (Sunrise Reader, Tecan, Switzerland). IL-13 was measured using a Cytometric Bead Array (CBA; BD Biosciences). Briefly, BALF samples were incubated with antibody-coated capture beads for 2h at RT. Detection reagent A was then added to the mix and samples incubated for another 2h at RT. Following incubation, the samples were washed with 100μL wash buffer and centrifuged at 1,600rpm for 5min. The supernatant was discarded and reagent B added to each tube. Samples were then incubated for 1h at RT and washed as before. Finally, samples were resuspended in wash buffer and the results acquired using a BD LSR Fortessa cytometer (BD Biosciences, San Jose, CA, USA).

5.2.6.4 Histologic analysis of mucus-secreting cells

To assess mucus production in airway epithelial cells, formalin-fixed lung lobes were used to prepare 5μm paraffin-embedded sections. Lung tissue sections were rehydrated through sequential incubations in Xylene and absolute or 70% ethanol solutions (Chem- Supply), followed by: treatment with 1% periodic acid (Sigma-Aldrich) for 10min; a 2min wash in running water; and staining with Schiffs reagent (made in house) for 25min to enumerate mucus-secreting cells. Nuclei were stained with Mayer’s hematoxylin (Dako). Functional characterization of the new 8q21 Asthma risk locus | 196

Chapter 5 Sections were then dehydrated in Xylene and 70% ethanol solutions, and mounted in Depex (DPX, Sigma-Aldrich). Slides were scanned using a digital slide scanner (Aperio Scanscope XT) and analysed with ImageScope software. Mucus secreting cells were counted in the airways (4-5 airways per mouse) and expressed as the percentage of total airway epithelial cells (AECs).

5.2.7 Assessing the contribution of Pag1 to airway inflammation triggered in naïve mice upon first exposure to allergen

5.2.7.1 Induction of allergic airway inflammation

To investigate the contribution of Pag1 specifically to the allergen sensitization phase of the experimental model described above (Figure 5.1), Pag1-/- and WT mice (n=4 per group) were administered a single dose of 100μg HDM allergen extract (Greer Laboratories, batch 14) intranasally and euthanized 72h later as illustrated in Figure 5.2. Control mice were inoculated with PBS instead of HDM (n=8 and 7 in WT and Pag1-/- groups, respectively).

Figure 5.2 Allergen sensitization experimental study design

5.2.7.2 Assessment of allergic airway inflammation

As described in section 5.2.6.2.

5.2.7.3 Measurement of cytokines

IL-5 and IL-33 were measured as described in section 5.2.6.3.

Functional characterization of the new 8q21 Asthma risk locus | 197

Chapter 5 5.2.8 Assessing the contribution of Pag1 to granulocyte recruitment into the airways

5.2.8.1 Induction of allergic airway inflammation

Pag1-/- and WT mice (n=4 per group) were intranasally exposed to 10μg of a Lipopolysaccharide suspension (LPS; Sigma-Aldrich) – also known as endotoxin and a major component of the cell membrane of Gram-negative bacteria. Control mice were inoculated with PBS (n=4 and 3 in WT and Pag1-/- groups, respectively), and all animals were euthanized 24h after inoculation, as illustrated in Figure 5.3.

Figure 5.3 LPS-induced airway inflammation experimental study design

5.2.8.2 Assessment of airway inflammation

Lung cell homogenates were obtained using a cell strainer (40μm, BD Falcon) followed by a centrifugation at 1,600rpm for 5min at 4°C. Cell pellets were resuspended in 1.5mL Gey’s lysis buffer and washed twice in FACS buffer before being counted and seeded in 96-well plates, as described above. BALF was assessed as described in section 5.2.6.2.

5.2.8.3 Measurement of dendritic cell populations

MLNs, the main site where activated dendritic cells migrate to in order to prime and activate naïve T cells, were harvested from additional HDM-exposed Pag1-/- and WT mice (n=5). MLN cell suspensions were obtained using a cell strainer (40μm, BD Falcon), pelleted, resuspended in 1mL Gey’s lysis buffer and washed twice in FACS buffer. Total cell counts, FACS staining and cell enumeration were performed as described in section 5.2.6.2.

Functional characterization of the new 8q21 Asthma risk locus | 198

Chapter 5 5.2.9 Validating findings from the model of acute experimental asthma (section 5.2.6) using an allergen batch from a different source

To validate (and help interpret) findings from the experiments described in section 5.2.6 we applied the same experimental procedure but used a different batch of HDM, obtained from CITEQ Biologics instead of Greer Laboratories.

5.2.9.1 Selection of a new HDM batch

To select a new HDM batch from CITEQ Biologics for experimental use, we compared the ability of three different batches of HDM extract (15J01, 15J02 and 15G10) to trigger airway inflammation in mice in vivo. Responses were compared against those observed with batch 12 from Greer Laboratories, a batch extensively used in the Phipps’ lab with proven powerful immunogenicity [130]. WT mice (n=4 per group) were administered a single dose of 100μg HDM suspension intranasally, euthanized 72h later and airway inflammation in BALF was assessed as described in section 5.2.1.2.

5.2.9.2 Induction of allergic airway inflammation

The new HDM batch was then used to induce airway inflammation as described in section 5.2.6.1. Briefly, Pag1-/- and WT mice (n=5 or 6 per group, respectively) were sensitised and re-exposed to a high dose of allergen (batch 10G15; CITEQ Biologics), while control mice (n=4 per group) were inoculated with PBS.

5.2.9.3 Assessment of allergic airway inflammation

As described in sections 5.2.6.2 and 5.2.8.2.

5.2.9.4 Lymph node cultures

To study the contribution of Pag1 to T cell function, MLN from mice exposed to HDM were harvested and the cells cultured ex vivo with or without the presence of HDM. MLNs are mainly populated by dendritic cells (DCs) and lymphocytes and so, when the cells are cultured in the presence of allergen, primed T cells rapidly expand and produce Th2-type cytokines. MLN cell suspensions were obtained as previously described. A total of 1x106 cells were seeded in duplicate in flat bottom 96-well plates and each sample resuspended in 200μL of ACCM media (1mM Sodium Pyruvate and 20mM HEPES (Gibco), 2mM L-

Functional characterization of the new 8q21 Asthma risk locus | 199

Chapter 5 Glutamine and 100U/mL Penicillin Streptomycin (Pen/Step) Solution (Life Technologies), 50 μM 2-Mercaptoethanol (Sigma-Aldrich), 10% FCS (Moregate Biotech) in HyClone RPMI 1640 (Gibco)) or ACCM media supplemented with 50μg/mL HDM suspension. Whenever MLN from PBS exposed mice did not contain enough cells to plate 1x106 cells per well, MLN cells from different mice were pooled to achieve this number. Cells were incubated for 4 days at 37ºC, after which the supernatant was collected for cytokine measurement.

5.2.9.5 Measurement of cytokines and immunoglobulins

IL-5, IL-17A and HDM-specific IgG1 were measured as described in section 5.2.6.3.

5.2.10 Assessing the contribution of Pag1 to airway inflammation triggered by exposure to a low dose of allergen

5.2.10.1 Induction of allergic airway inflammation

Pag1-/- and WT mice (n=5 and 6, respectively) were sensitized by intranasal administration of a suspension containing 100μg of HDM (batch 10G15, CITEQ Biologics) and subsequently challenged with 3μg of the same allergen at days 14, 15, 16 and 17 (Figure 5.4). Animals were euthanized 3h after the last challenge for endpoint assessment. Control mice (n=4 and 3 for WT and Pag1-/- groups, respectively) were inoculated with PBS.

Figure 5.4 Low dose allergen-induced airway inflammation experimental study design

5.2.10.2 Assessment of allergic airway inflammation

As described in sections 5.2.6.2 and 5.2.8.2.

5.2.10.3 Lymph node cultures

MLN suspensions were prepared and cells counted as described above. A total of 1x106 cells were seeded in duplicate in a flat bottom 96-well plates and each sample resuspended in 200μL of ACCM media or ACCM media supplemented with 20μg/mL HDM

Functional characterization of the new 8q21 Asthma risk locus | 200

Chapter 5 suspension. Cells were incubated for 4 days at 37ºC, after which the supernatant was collected for cytokine measurement.

5.2.10.4 Measurement of cytokines and chemokines

IL-5 and CCL24 were measured as described in section 5.2.6.3.

5.2.10.5 Histologic analysis of mucus-secreting cells

Mucus secreting cells were stained as described in section 5.2.6.4. Results are expressed as a histological score based on two criteria, adapted from Livraghi et al [131] – percentage of obstructed airways and estimated percentage of positive cells in airway epithelium (Table 5.2). Four or five airways are scored for each parameter and both scores are averaged per mouse. This method is faster, more reliable and reproducible than the method described in section 5.2.6.4 and thus, we adopted this approach for all subsequent analysis in this Thesis.

Parameter Percentage of obstructed Mucous secreting cell (MuSC) Score airways in lung abundance 0 no obstruction none 1 0-10% 0-10% positive AECs in airway 2 10-20% 10-30% positive AECs in airway 3 20-30% 30-50% positive AECs in airway 4 30-40% 50-80% positive AECs in airway 5 >50% >80% positive AECs in airway

Table 5.2 Histological scoring method for epithelial airway mucus secreting cell analysis.

Functional characterization of the new 8q21 Asthma risk locus | 201

Chapter 5 5.3 RESULTS

5.3.1 Validating knockout of gene expression in the Pag1-/- mouse strain

Pag1-/- animals were kindly donated by Prof Jonathan Lindquist from Otto-von-Guericke University (Magdeburg, Germany) [129]. Following strain re-derivation at the University of Queensland, WT and Pag1-/- littermate mice were obtained through Pag1het crossing. Before allocating animals for experimental use, we validated the strain by measuring Pag1 gene expression in lung tissue from WT and Pag1-/- littermate mice. The result is shown in Figure 5.5A and indicates that low levels of Pag1 expression can be detected in the knockout animals, however this expression is significantly lower than that of WT animals. In knockout systems, when the transcriptional ablation is not 100% efficient, basal gene expression levels can be found in knockout animals. This effect is called leaky expression and, in the case of high leakage, the knockout strain does not represent a reliable model for in vivo studies. To confirm that the detected Pag1 gene expression was due to leaky expression, we also assessed the expression of Pag1 protein using immunofluorescence in lung tissue sections from a Pag1-/- mouse. Results in Figure 5.5B showed only background residual staining and unspecific stained clusters due to antibody quality. For comparison, in lung sections from WT mice (used in Chapter 4, section 4.3.3), positive Pag1 staining was found in AECs above background levels, as highlighted by the white box in Figure 5.5C. No positive Pag1 staining was found within cells from Pag1-/- animals, confirming that this strain is Pag1-/-.

Functional characterization of the new 8q21 Asthma risk locus | 202

Chapter 5

Figure 5.5 Pag1-/- strain validation. RNA was extracted from WT (n=3) and Pag1-/- (n=6) lung tissue using the TRIzol method. Lung sections were prepared from a Pag1-/- mouse. (A) Gene expression was measured using TaqMan assays and Pag1-/- results were calculated as fold change from WT. *P<0.05 vs WT using the Mann-Whitney test. Bars represent mean ± SEM. Lung sections from (B) Pag1-/- (n=1) and (C) WT (n=1) mice were stained using an AF488 conjugated anti- Pag1 antibody (green) and DAPI for nucleic acids (blue).

Functional characterization of the new 8q21 Asthma risk locus | 203

Chapter 5 5.3.2 Experiment 1: assessing the contribution of Pag1 to airway inflammation triggered in sensitised mice by acute exposure to a high dose of allergen

Previously sensitised Pag1-/- and WT mice were inoculated intranasally with a suspension containing 10μg of HDM (batch 14, Greer laboratories), as illustrated in Figure 5.6A. Results indicate that BALF from Pag1-/- animals had significantly lower total number of inflammatory cells (Figure 5.6B), as well as lower neutrophil and eosinophil counts (Figure 5.6C). Of the seven cytokines or chemokines measured in BALF, significantly reduced levels in Pag1-/- animals were only observed for CCL24, a chemokine for eosinophils (Figure 5.6D and E). There were no differences in the frequency of mucus secreting airway epithelial cells (Figure 5.6F), nor in HDM-specific IgG1 levels between Pag1-/- and WT animals (Figure 5.6G).

Thus, results from this experiment suggest that Pag1 expression might be required for the recruitment of granulocytes to the airways following re-exposure to a high dose of allergen.

Functional characterization of the new 8q21 Asthma risk locus | 204

Chapter 5

Functional characterization of the new 8q21 Asthma risk locus | 205

Chapter 5

Figure 5.6 Experiment 1: assessment of airway inflammation in previously sensitised mice re-exposed to a high dose of HDM allergen. Airway inflammation triggered in WT and Pag1-/- mice according to the study design (A) was assessed by: enumerating total cells (B); and inflammatory cell populations (C) in BALF. Results are shown both as frequency (% of total cells in each sample) and absolute number calculated according to the frequencies measured. Measurement of (D) chemokines; and (E) Th1/Th2/Th17 cytokines; (F) enumeration of mucus secreting cells in airway epithelium; and (G) HDM-specific antibody measurement. n=6-8 mice per group (one WT mouse from each group in IL- 13 and one Pag1-/- mouse from the HDM groups in IL-17 and IFN-γ; were excluded from the analysis due to technical issues). ***P<0.001, **P<0.01, *P<0.05 vs. WT. §§§P<0.001, §§P<0.01, §P<0.05 vs. PBS controls.

Functional characterization of the new 8q21 Asthma risk locus | 206

Chapter 5 5.3.3 Experiment 2: assessing the contribution of Pag1 to airway inflammation triggered in naïve mice upon first exposure to allergen

In the previous experiment, we observed a significant decrease in the number of inflammatory cells in the airways of Pag1-/- mice after re-exposure to a high dose of allergen. This could arise if Pag1 expression is important in the context of innate immune responses to allergen exposure. To investigate this possibility, naïve Pag1-/- and WT mice were inoculated intranasally with a single dose of a HDM suspension (batch 14, Greer laboratories) or PBS, and euthanized 72h later for endpoint assessment (Figure 5.7A). Airway inflammation observed in this model is due to innate and not adaptive immune responses to allergen exposure.

In this experiment, the total number of inflammatory cells in BALF was comparable between Pag1-/- and WT mice (Figure 5.7B). However, the number of neutrophils and eosinophils was reduced in Pag1-/- mice, significantly so when analysed as absolute cell numbers (Figure 5.7C). On the other hand, IL-33 and IL-5 levels in BALF were similar between WT and Pag1-/- groups (Figure 5.7D), as were total cells counts (Figure 5.7E) and dendritic cell (DC) subsets in lymph nodes from HDM sensitised mice (Figure 5.7F).

Overall, results from this experiment suggest that Pag1 expression plays a role in granulocyte recruitment to the airways, even in the absence of an adaptive immune response. On the other hand, Pag1 knockout does not lead to impaired migration of DCs to the lymph nodes.

Functional characterization of the new 8q21 Asthma risk locus | 207

Chapter 5

Functional characterization of the new 8q21 Asthma risk locus | 208

Chapter 5

Figure 5.7 Experiment 2: assessment of airway inflammation in naïve mice exposed to HDM allergen for the first time. Airway inflammation triggered in WT and Pag1-/- mice according to the study design (A) was assessed by: enumerating total cells (B); and inflammatory cell populations (C) in BALF. Results are shown both as frequency (% of total cells in each sample) and absolute number calculated according to the frequencies measured. (D) cytokine measurement. n=4-8 mice per group (one WT mouse from the PBS group in IL-5 was excluded from the analysis due to a technical issue). In the lymph nodes of HDM sensitized mice, total cell number (E) and dendritic cell subsets (F) were enumerated. n=5 mice per group. *P<0.05 vs. WT. §§P<0.01, §P<0.05 vs. PBS controls.

Functional characterization of the new 8q21 Asthma risk locus | 209

Chapter 5 5.3.4 Experiment 3: assessing the contribution of Pag1 to granulocyte recruitment into the airways

In the previous experiments, we observed that both naïve and previously sensitized Pag1-/- animals had reduced numbers of neutrophils and eosinophils in BALF after allergen exposure. Furthermore, Pag1 is highly expressed in granulocytes as discussed in Chapter 4. Therefore, we hypothesized that Pag1 expression in granulocytes might be required for efficient migration of these cell types to the airways.

Granulocyte recruitment is one of the first lines of defence against bacterial infections, being triggered by the LPS content of these organisms through a TLR4-dependent mechanism [132]. Thus, to test our hypothesis, Pag1-/- and WT mice were inoculated intranasally with 10μg of LPS or PBS, and euthanized 24h later for endpoint assessment (Figure 5.8A).

In this experiment, we found no difference in neutrophil numbers between Pag1-/- and WT mice, either when considering cell counts in BALF or in lung tissue (Figure 5.8B and C). These results indicate that neutrophils with no Pag1 expression are still able to migrate to the lungs, ruling out an essential direct effect on neutrophil chemotaxis. Eosinophil recruitment was only modestly triggered by LPS exposure; despite this caveat, we observed no significant differences in eosinophil numbers between Pag1-/- and WT mice.

Thus, we conclude that the reduced numbers of granulocytes observed in Pag1-/- animals exposed to a high dose of allergen is unlikely to arise because chemotaxis is intrinsically impaired in these cell types. Instead, other indirect mechanisms are likely to be involved (e.g. impaired production of chemokines by airway epithelial cells in Pag1-/- mice).

Functional characterization of the new 8q21 Asthma risk locus | 210

Chapter 5

Figure 5.8 Experiment 3: assessment of LPS-induced airway inflammation. Airway inflammation triggered in WT and Pag1-/- mice according to the study design (A) was assessed by: enumerating total cells (B) and granulocyte cell populations (C) in BALF and lung tissue. Results are shown both as frequency (% of total cells in each sample) and absolute number calculated according to the frequencies measured. n=3-4 mice per group. *P<0.05 vs. WT. §P<0.05 vs. PBS controls.

Functional characterization of the new 8q21 Asthma risk locus | 211

Chapter 5 5.3.5 Experiment 4: Validating results from experiment 1 (sensitised mice re- challenged with a high dose of allergen)

Results from experiments 2 and 3 did not provide any insight into the mechanisms that might explain the significant reduction in granulocyte numbers observed in the airways of Pag1-/- mice re-challenged with a high dose of allergen (experiment 1). Before embarking in further experiments to attempt to identify the likely mechanisms, we first attempted to reproduce the results from experiment 1, to rule out the possibility that they were due to unaccounted experimental biases.

The HDM batch used in experiment 1 (batch 14 from Greer Laboratories) had been exhausted, and so we had to use a different HDM batch for this experiment. The timing of this experiment coincided with a change of allergen provider in the Phipps’ lab, with CITEQ Biologics replacing Greer Laboratories. Because the immunogenicity of allergen extracts from CITEQ might have been different from those sourced from Greer, our first concern was to select an allergen batch from CITEQ that elicited an inflammatory response in WT mice that was broadly comparable to that observed with batch 14 from Greer.

To this end, we obtained three batches of HDM from CITEQ to test (15J01, 15J02 and 15G10). To select the most appropriate batch for experiments, airway inflammation in WT mice was induced by intranasal inoculation of 100μg of each HDM batch from CITEQ and one batch from Greer still available (and used extensively) in the lab (batch 12). Ideally, the comparison should have been with batch 14 from Greer, but this was no longer available, as mentioned above. Mice were euthanized 72h after challenge for endpoint assessment.

Results from this experiment are shown in Figure 5.9. From the three CITEQ batches tested, 15G10 triggered an inflammatory response that was most similar to the Greer batch, with comparable frequencies and absolute numbers of granulocytes and lymphocytes measured in BALF. LPS and Der p 1 (the main immunogenic component in mites) content was also comparable between CITEQ 15G10 and Greer 12, and so we selected CITEQ batch 15G10 for experiment 4.

Functional characterization of the new 8q21 Asthma risk locus | 212

Chapter 5

Figure 5.9 Airway inflammation assessment and comparison between different batches of HDM. Airway inflammation in WT mice was assessed by enumerating total cell numbers (A) and inflammatory cell populations (B) in BALF. Results are shown by cell population both as frequency (% of total cells in each sample) and absolute number according to the frequencies measured. n=4 mice per group. *P<0.05 vs. Greer 12. #P<0.05 vs. CITEQ 15G10. (C) Comparison of LPS and protein content in each HDM batch tested. N.K. meaning “not known”.

Functional characterization of the new 8q21 Asthma risk locus | 213

Chapter 5 The study design for experiment 4 was exactly the same as described for experiment 1 (Figure 5.10A). However, unlike observed in experiment 1, airway inflammation was not dampened in Pag1-/- animals challenged with the new HDM batch from CITEQ. Specifically, in both BALF and lung tissue, there was a trend for lower overall cell numbers (Figure 5.10B) as well as neutrophil absolute numbers (Figure 5.10C) in Pag1-/- animals, but these differences were not statistically significant. No differences were also observed in broad T and B cell populations (Figure 5.10D) or dendritic cell subsets (Figure 5.10E), including total conventional DCs (cDC), cDC subsets CD11b+ and CD103+, and plasmacytoid DC (pDC). The only notable difference in immune cell numbers between groups was for CD4+ regulatory T cells (Tregs) which were increased in Pag1-/- mice (Figure 5.10F). No difference was observed in cytokine (IL-5, IL-17; Figure 5.10G) or HDM-specific IgG1 (Figure 5.10H) levels between groups.

Therefore, results from this experiment do not support our original findings that Pag1-/- mice have decreased granulocyte recruitment to the airways after re-challenge with a high dose of allergen.

Functional characterization of the new 8q21 Asthma risk locus | 214

Chapter 5

Functional characterization of the new 8q21 Asthma risk locus | 215

Chapter 5

Functional characterization of the new 8q21 Asthma risk locus | 216

Chapter 5

Functional characterization of the new 8q21 Asthma risk locus | 217

Chapter 5

Figure 5.10 Experiment 4: validation of results from experiment 1. Airway inflammation triggered in WT and Pag1-/- mice according to the study design (A) was assessed by: enumerating total cells (B); granulocytes (C); and lymphocytes (D) in BALF and lung tissue; and dendritic cell (E); and T cell (F) subsets in lung tissue. Results are shown by cell population both as frequency (% of WBC in each sample) and absolute number calculated according to the frequencies measured. Th2 and Th17 cytokine measurement in BALF and lymph node cultures (G); and HDM-specific antibody levels (H). n=4-6 mice per group (one Pag1-/- mouse from the PBS group in BALF IL-17; one WT and one Pag1-/- mouse from the HDM groups in MLN IL-17; were excluded from the analysis due to technical issues). **P<0.01, *P<0.05 vs. WT. §§P<0.01, §P<0.05 vs. PBS controls.

Functional characterization of the new 8q21 Asthma risk locus | 218

Chapter 5 5.3.6 Experiment 5: assessing the contribution of Pag1 to airway inflammation triggered in sensitised mice by acute exposure to a low dose of allergen

An important caveat of the previous experiments is that sensitised mice were later re- challenged with a high dose of allergen (10μg) over several days. This study design might induce an exaggerated inflammatory response in mice that does not resemble the processes underlying asthma in humans, who are typically exposed to lower allergen doses. For example, Dullaers et al. showed that exposing mice to a high allergen dose is not appropriate to study the contribution of B cells to the adaptive immune response triggered by HDM [133]. These findings, together with the observation that PAG1 is highly expressed in B and T cells, prompted us to investigate the contribution of PAG1 to airway inflammation triggered by exposure to a low dose of allergen.

At day 0, Pag1-/- and WT mice were sensitised to HDM through intranasal inoculation of an allergen extract (batch 10G15, CITEQ) containing 100μg of HDM. At days 14, 15, 16 and 17, mice were then re-exposed to a low dose (3μg) of the same allergen extract, and euthanized 3h after the last challenge for endpoint assessment (Figure 5.11A).

In this experiment, Pag1-/- animals had significantly increased total number of inflammatory cells in lung tissue, with a similar trend observed in BALF (Figure 5.11B). Eosinophil counts were significantly higher in BALF and lung tissue from Pag1-/- animals, while neutrophils were reduced in lung tissue from these mice (Figure 5.11C). In BALF, T and B cell counts were increased in Pag1-/- animals (Figure 5.11D). No difference was observed in T cell or DC subsets in lung tissue (Figure 5.11E and F). There was a trend for increased CCL24 (Figure 5.11G) and IL-5 (Figure 5.11H) production in Pag1-/- mice, but this was not statistically significant. On the other hand, mucus production was significantly greater in Pag1-/- animals (Figure 5.11I).

Overall, results from this experiment raise the possibility that inhibition of Pag1 expression might aggravate a Th2 immune response triggered by re-exposure to a low dose of HDM.

Functional characterization of the new 8q21 Asthma risk locus | 219

Chapter 5

Functional characterization of the new 8q21 Asthma risk locus | 220

Chapter 5

Functional characterization of the new 8q21 Asthma risk locus | 221

Chapter 5

Figure 5.11 Experiment 5: assessment of airway inflammation triggered by re-exposure to a low dose of HDM. Airway inflammation triggered in WT and Pag1-/- mice according to the study design (A) was assessed by: enumerating total cells (B); granulocytes (C); and lymphocytes (D) in BALF and lung tissue; and T cell (E) and dendritic cell (F) subsets in lung tissue. Results are shown by cell population both as frequency (% of WBC in each sample) and absolute number calculated according to the frequencies measured. Measurement of (G) chemokines; and (H) cytokine levels in BALF, lung tissue or lymph node cultures; and (I) mucus production assessment. n=3-6 mice per group (one WT mouse in the HDM group was excluded from mucus score analysis due to a technical issue). **P<0.01, *P<0.05 vs. WT. §§§P<0.001, §§P<0.01, §P<0.05 vs. PBS controls. Functional characterization of the new 8q21 Asthma risk locus | 222

Chapter 5 5.4 DISCUSSION

To our knowledge, the experiments described in this Chapter constitute the first assessment of the contribution of PAG1 to the pathophysiology of asthma, specifically to the development of allergen-induced airway inflammation.

We received Pag1-/- mice from Otto-von-Guericke University (Magdeburg, Germany), a strain that we re-derived and established using specific pathogen-free animals. Before embarking in the experimental assessment of the role PAG1 using in vivo models, littermate WT and Pag1-/- animals were generated through Pag1het crossing, and used to measure Pag1 gene expression. The results showed that residual Pag1 expression was detected in Pag1-/- mice however, at significantly lower levels compared to WT animals. Pag1-/- mice were generated by disrupting Pag1 exon 6 (aa94/nt280 – aa287/nt861) using modified bacterial artificial chromosome (BAC) technology [134]. The TaqMan assay used to measure gene expression in these mice uses primers and a probe spanning the exon 6-7 boundary, which anneals downstream of the insertion site and would therefore be able to detect the expression of the two main Pag1 variants (accession numbers NM_001195031 and NM_053182). Low basal gene expression levels, or leaky gene expression, are undesirable features in knockout systems, which are not uncommonly observed. Thus, to be confident that the gene expression detected would not give rise to functional Pag1 protein, we investigated Pag1 protein levels in Pag1-/- lung tissue. Using immunofluorescence we showed that Pag1 was not expressed in airway cells or lung parenchyma, indicating that Pag1-/- animals do not produce Pag1 protein, which is consistent with previous reports using these mice [129]. Therefore, we concluded that, as expected this strain was Pag1-/- and proceeded to use these mice for the experiments described in this Chapter.

Experiments in mice have long been used to study the pathophysiology of asthma. However, this is often challenging, given the complex multifactorial nature of the disease. A multitude of protocols have been described for animal experiments that reproduce some features of clinical asthma. Rather than a chronic allergen challenge protocol, that involves weeks or months of low allergen exposure in sensitized mice, most groups have adopted shorter acute allergen challenge protocols with administration of high allergen doses [135, 136]. These shorter models exhibit a more robust inflammatory response than the chronic models. However, they lack the airway remodelling features observed in asthmatic lungs, such as goblet cell hyperplasia, matrix deposition and increased smooth muscle cell

Functional characterization of the new 8q21 Asthma risk locus | 223

Chapter 5 number [137, 138]. There is no ideal model to study all disease aspects but rather several protocols, modified to best suit the research question. Besides the type of challenge, another critical variable in these models is the type of allergen used. Historically, ovalbumin (OVA) was the mainstay in allergic asthma murine models. Inexpensive and easy, protocols using OVA coupled with an adjuvant to sensitize and challenge mice have been widely implemented to induce and skew the immune response towards a Th2-driven phenotype [139]. However, due to poor clinical translation of the findings from these models, in the last five years the field has changed to adopt more naturally occurring allergens, such as HDM, cockroach or Alternaria alternata [140-143].

In our laboratory, an improved acute allergen model has been extensively used with proven induction of most hallmark features of asthma, including airway hyperresponsiveness, mucous cell hyperplasia and granulocytic inflammation [122]. This model, which uses a HDM allergen extract as the sensitizing (day zero) and triggering agent (days 14, 15, 16 and 17), was selected for the initial assessment of the potential role of Pag1 in asthma pathophysiology (experiment 1). Mice were euthanized 3h after the last challenge, a critical time point for cytokine detection in BALF.

Results from this experiment showed a profound reduction in BALF inflammatory cell infiltration in Pag1-/- mice, specifically neutrophils and eosinophils, which are key effector cells in asthmatic airway inflammation [144]. Evaluation of CXCL1 (or KC) and CCL24 (or eotaxin 2), the chemokines responsible for the recruitment of neutrophils and eosinophils, respectively, only showed a reduced production of the latter in Pag1-/- mice. No significant differences between groups were observed for the Th2-instructive cytokine IL-33, nor for the two Th2-type cytokines measured (IL-5 and IL-13). Overall, results from this experiment indicated that granulocyte recruitment to the airways following re-challenge with a high dose of HDM was impaired in Pag1-/- mice. If confirmed, this finding would be consistent with our observation that genetically-determined increased PAG1 expression in humans was associated with asthma risk.

Next, we investigated if the findings from experiment 1 were likely to arise due to an impaired innate immune response to HDM in Pag1-/- mice. In this second experiment, WT and Pag1-/- mice were administered a single dose of HDM and euthanized 72h later for endpoint assessment. Mirroring the results from the acute model, neutrophilia and eosinophilia were both reduced in Pag1-/- animals. However, IL-33 levels were comparable between Pag1-/- and WT and mice, as were the levels of IL-5, which is mainly produced by

Functional characterization of the new 8q21 Asthma risk locus | 224

Chapter 5 type 2 innate lymphoid cells (ILC2) in response to IL-33 [145]. Additionally, we investigated the possibility that Pag1 knockout could affect dendritic cell (DC) activation and/or migration into draining lymph nodes, and thus result in impaired T cell priming in these animals. For this, we enumerated several DC subsets in HDM-sensitized mice mediastinal lymph nodes (MLN), including: tissue resident and critical antigen presenting cells, cDC (conventional or classical DC) [146]; its subsets CD11b+ DC (typically inflammatory and involved in the induction of the Th2 subset) [147] and CD103+ DC (although controversial, this subset is potentially tolerogenic and involved in the induction of the Th1 subset) [148, 149]; and professional type-I interferon (IFN) producing pDCs (plasmacytoid DC) [150]. Results from this analysis showed no difference in the number of any of the DC subsets between WT and Pag1-/- mice, suggesting that Pag1 does not contribute to DC activation or migration. Based on these results, we conclude that Pag1 knockout is unlikely to affect T cell priming but could potentially influence granulocyte recruitment directly.

To investigate the possibility that Pag1 could be directly involved in granulocyte recruitment, we inoculated WT and Pag1-/- animals with a single dose of LPS, also known as endotoxin, which is a major component of gram-negative bacteria cell membrane and a potent neutrophil inducer [151]. Results from this third experiment showed no difference between groups in either frequency or total number of neutrophils, in both BALF and lung tissue. LPS had little effect on eosinophil recruitment, consistent with previous findings [152]. Nevertheless, similarly to the neutrophil results, no difference between groups was observed in the eosinophil population. These results demonstrate that granulocytes without Pag1 expression are able to migrate to the airways, suggesting that other indirect mechanisms (e.g. impaired production of chemokines by airway epithelial cells) are responsible for the lower granulocyte numbers observed in Pag1-/- mice in experiments 1 and 2.

To increase confidence that our key findings from experiment 1 were real, and not a result of an experimental artefact, we repeated the experiment but this time using an HDM batch from a different source. In this experiment, Pag1-/- mice again had lower neutrophil numbers in both BALF and lung tissue, but these differences did not reach statistical significance (P=0.052). On the other hand, there were no differences between groups for eosinophils, nor for most of the other inflammatory parameters measured. The exception was the observation that the frequency of Tregs was increased in Pag1-/- mice. Therefore, unlike observed in experiment 1, results from experiment 4 suggest that inhibition of Pag1

Functional characterization of the new 8q21 Asthma risk locus | 225

Chapter 5 expression does not have a major effect on the inflammatory response that develops in sensitised mice re-exposed to a high dose of allergen.

How can we explain the different results obtained in experiments 1 and 4? The only difference between the two experiments was the batch of allergen used, which were obtained from two different providers (Greer and CITEQ). Several variables are known to directly affect batch-to-batch variability between differently sourced allergens, including: growth media [153]; extraction method [154] and allergen composition [155]. Growth media components are not known for the batches used, however the extraction methods are. Whereas at Greer HDM extracts are prepared from mite bodies, yielding a fairly pure extract, CITEQ uses whole mite cultures including mite bodies, faeces, larvae, eggs, and cultivation media. For this reason, and due to additional downstream purification methods, CITEQ batches might lack important immunogenic proteins that are lost in these processes, and be enriched in inert components such as salts, starch and other biologic material (as per product data sheet). Even though the Der p 1 content in both batches is comparable, the total protein content per vial is about 10-fold higher in the Greer batch than the CITEQ batches, which might include other important immunogenic components such as Der p 4, 5, 7, 21 and 23 [156, 157]. It has been extensively reported that allergen composition in HDM extracts is critical for immune responses, including a report comparing batches from Greer, CITEQ and ALK-Abello, concluding that the Greer batch induced a more powerful immune response between the three [158]. Precisely due to the low total protein content in CITEQ vials (540μg in batch 15G10), this batch stock was diluted based on the total dry powder weight (10mg/vial). Greer batch stocks, on the other hand, are normally diluted in our laboratory based on total protein content (3-5mg/vial, 5.36mg in Greer 12), an approach not feasible for the CITEQ batch. Following this strategy, the amount used to sensitize mice (100μg) is very different between the two batches with regards to the protein content. Diluting the allergen based on protein content means that mice are given 100μg of protein when they are inoculated with the Greer batch, an amount that is 20-fold higher than what mice inoculated with the CITEQ batch are given (5.4μg of protein in 100μg of allergen suspension). Consistent with this, if the total BALF cell number in WT mice is compared between the two experiments that we performed, it is clear that the Greer batch used exerted a much more robust inflammatory response than CITEQ 15G10 (2.5-fold difference). Moreover, as previously mentioned, eosinophil recruitment was the parameter mostly impacted by batch variability (Greer vs. CITEQ) and one of the key phenotypes lost when the acute model was performed using

Functional characterization of the new 8q21 Asthma risk locus | 226

Chapter 5 the CITEQ allergen. Neutrophil recruitment differences, although not significant in the CITEQ experiment, followed the same trend in both experiments. Thus, it is technically possible that the results observed in experiments 1 and 2 are real, but not reproducible with the allergen batch obtained from CITEQ.

Besides the allergen source and its biochemical properties, some limitations of experiments 1 and 4 are noteworthy, specifically the high allergen dose used in both the sensitisation and challenge phases. The high amount of allergen availability may overload the system and lower the antigen presenting threshold significantly. In other words, most cells are pushed into an antigen presenting phenotype including those that normally would not have this role, and as a result, an overstimulation of the immune system may occur and mask milder phenotypes. Dullaers et al. showed that in a model where the amount of inhaled allergen was limiting, mice exhibited a phenotype that was not evident in a high allergen dose model [133]. Their findings mainly implicated B cells as having a critical role in the development of adaptive responses, and because PAG1 is highly expressed in this cell type, we decided to investigate its contribution to allergen-induced airway inflammation in the context of low allergen availability.

WT and Pag1-/- mice were sensitised with a single dose of 100μg of CITEQ 15G10 HDM. Lower sensitising amounts were tested using this batch, however due to the induced immune response, we decided to maintain the sensitisation dose the same as used in the experiments with the Greer allergen – not shown. At days 14, 15, 16 and 17, mice were re- exposed to a 3μg allergen dose (vs. 10μg used in experiments 1 and 4). Results from this fifth and final experiment indicated that unlike previous experiments, Pag1-/- mice exhibited higher total cell counts in lung tissue, with a similar trend observed in BALF. Furthermore, these increased cell counts in the knockout mice were supported by the higher eosinophilia in BALF and lung tissue and lymphocyte infiltration (both T and B cells) in BALF, two surprising observations considering the results from previous models. Consistent with increased eosinophil and T cell infiltration, Th2 cytokine IL-5 production was higher in Pag1-/- animals in both BALF and lymph node cultures. A similar trend for increased B cell numbers, but not T cells, was also observed in lung tissue from Pag1-/- animals. Consistently, no difference in the number of T cell subsets (CD4+ T and CD8+ T) was observed between groups. Dendritic cell subset analysis also showed similar results between groups. Additionally, unlike previous results, CCL24 level was not impaired in Pag1-/- mice and mucus production was increased in this group. The only consistent observation when compared to the previous experiments was the impaired neutrophilia in

Functional characterization of the new 8q21 Asthma risk locus | 227

Chapter 5 Pag1-/- mice lung tissue. Overall, results from this experiment suggest that eosinophilic airway inflammation in Pag1-/- mice is significantly exacerbated when the inhaled allergen dose is limiting.

The observation that lymphocyte populations were increased in Pag1-/- animals might not be entirely surprising. The main function of PAG1 is to regulate Csk recruitment, a kinase that modulates the activity of the Src family kinases (SFK), such as Src, Fyn and Lyn, which play key regulatory roles in several important cell processes [159]. This role for PAG1 was first described in human T cells, following the observation that PAG1 phosphorylation by SFK induces Csk recruitment into the lipid rafts where it binds to PAG1, inhibiting Src kinases. This mechanism was shown to implicate PAG1 in the regulation of signalling via TCR (T cell receptor) engagement. Upon TCR activation through antigen binding, PAG1 is dephosphorylated, disrupting Csk binding and activating Fyn, a constitutively PAG1-bound SFK. Fyn activation induces PAG1 re-phosphorylation and consequently Csk recruitment to inhibit Fyn and thus terminating TCR signalling [82]. In a similar way, PAG1 is also involved in signal transduction from BCR (B cell receptor) engagement in B lymphocytes [160]. Importantly, besides being implicated in downstream signalling from these receptors, PAG1 has been implicated in setting the activation threshold of both T and B lymphocytes [49, 161]. Interestingly, in the latter, Barua et al [161] showed using a computational model that BCR signalling can trigger several qualitatively distinct responses depending on the strength of the antigen signal. These observations can be extrapolated to our in vivo results from the low-dose allergen model, where the limited allergen availability together with PAG1 knockout could significantly reduce the activation threshold of lymphocytes upon antigen TCR and BCR engagement, increasing activation, proliferation and recruitment of these cells. Higher T cell numbers, which in turn produce more Th2-type cytokines, such as IL-5 (as shown), would therefore contribute to the observed airway eosinophilia (the slightly increased production of CCL24 could possibly add to this phenotype). This model would therefore explain why this phenotype is only observed within the specific context of low allergen availability.

Taken together, results from the five experiments described in this Chapter suggest that PAG1 might play a role in two distinct regulatory mechanisms. On the one hand, it might be involved in the early innate immune response, specifically in the airway epithelium response to allergen exposure. This role is suggested by the increased mucus production (experiment 5) and impaired granulocyte recruitment (experiments 1 and 2) observed in Pag1-/- mice. As we showed in Chapter 4, Pag1 is expressed in airway epithelial cells and

Functional characterization of the new 8q21 Asthma risk locus | 228

Chapter 5 although no changes were observed in the production of IL-33 in the acute models, Pag1 could affect the production of other alarmins such as IL1α, IL1β or HMGB1 (High mobility group box 1 protein). Additionally, epithelial cells also produce other proinflammatory factors such as IL-8 and IL-25, MCP-1 and MCP-4, arachidonic acid metabolites and eotaxin, which recruit inflammatory cells, increase mucus production and skew the immune response towards a Th2-type. To investigate this possibility, we propose that experiment 2 should be repeated using the CITEQ batch 15G10 to assess (i) DC subset activation and migration; (ii) epithelial alarmin, chemotaxis and cytokine production; and (iii) to investigate the possibility that Pag1 does not contribute directly to the production of these mediators but instead plays a role in airway remodelling (structural changes should also be addressed by investigating, for example, smooth muscle cell changes).

On the other hand, Pag1 may play a role in the adaptive immune response, specifically in lymphocyte activation and/or proliferation. This suggestion derives from the observation that lymphocyte numbers and IL-5 production are increased in BALF of Pag1-/- mice re- exposed to a low dose of allergen (experiment 5). Additionally, increased IL-5 levels in the lymph node cultures directly supports the implication of a T cell phenotype in Pag1-/- animals. We therefore propose that T cell polarization and activation should be assessed ex vivo in response to limited allergen availability. A consistent observation across the different experiments conducted was the observation that neutrophilia was impaired in Pag1-/- mice. Generally speaking, HDM-mediated responses are mainly eosinophilic, inducing neutrophils to a lesser extent. Therefore, to investigate the role of Pag1 in the neutrophil phenotype, we suggest that experiment 1 should be repeated using an allergen type that induces a stronger neutrophilic response (e.g. cockroach).

Finally, it is important to test the possibility that Pag1 plays a role in viral-induced asthma, a hypothesis that we investigated in Chapter 6.

In conclusion, we provide the first insight into the possible role(s) of Pag1 in the development of allergen-induced airway inflammation. However, many questions still remain unanswered, and additional experiments are warranted to address these.

Functional characterization of the new 8q21 Asthma risk locus | 229

6

The contribution of Pag1 to viral-induced airway inflammation in mice

Chapter 6 6.1 INTRODUCTION

6.1.1 Background Previous studies have shown that a lower respiratory tract infection (LRTI) or severe virus- associated bronchiolitis in early childhood substantially increases the risk of asthma development later in life [162, 163]. Impaired viral clearance, possibly due to a defective anti- viral immunity, contributes to lung damage and an aberrant immune response, collectively leading to a decline in lung function. Given the relatively high expression of PAG1 observed both in the lung and cell types that participate in anti-viral immunity (e.g. CD8+ T cells, see Chapter 4), we hypothesised that variation in PAG1 expression might contribute to asthma risk by influencing the immune response to a LRTI.

In humans, rhinovirus and respiratory syncytial virus (RSV) are the main causative agents of LRTI [164, 165]. However, even at very high doses, human RSV does not induce significant lung pathology in mice [166]. Thus, to study disease pathophysiology in mice, the field has adopted the natural rodent-specific pathogen, pneumonia virus of mice (PVM). Neonate mice infected with PVM develop disease features reminiscent of those observed in children who develop LRTI or bronchiolitis from an RSV infection [167]. Anti-viral immunity has been extensively studied in neonatal PVM mouse models, and two main cell populations have been identified as the key players in the immediate response to viral infection: airway epithelial cells (AEC) and plasmacytoid dendritic cells (pDCs).

Upon infection, viruses penetrate AECs where they replicate rapidly, inducing an anti-viral response by these cells which includes the release of anti-viral proteins such as type-III interferons (IFN; e.g. IFN-λ). This responsive state is maintained further by the quick release of vast amounts of potent anti-viral type-I IFNs (IFN-α/β), produced by activated pDCs through TLR-7-mediated viral sensing. On the one hand, this balanced immunity generates an appropriate Th1-type response which promotes viral clearance and tissue homeostasis [168, 169]. On the other hand, if pDC viral sensing ability is compromised by functional or genetic defects, low or no type-I IFN release fails to induce the AEC antiviral state. A deficient immediate anti-viral response facilitates viral replication, with increased viral burden promoting airway epithelium injury and consequently an aberrant immune response.

This aberrant immune response is mediated by the release of alarmins and Th2-instructive cytokines from the damaged epithelium such as HMGB1 and IL-33, inducing the

Functional characterization of the new 8q21 Asthma risk locus | 231

Chapter 6 recruitment and proliferation of type 2 innate lymphoid cells (ILC2) [169]. ILC2 cells then trigger the recruitment of additional immune cells such as granulocytes through cytokine release (e.g. IL-5) which in turn produce (i) pro-inflammatory cytokines such as IL-13; and (ii) growth factors, which promote epithelium reconstruction. In subsequent RSV infections this aberrant response is repeated in susceptible individuals, ultimately leading to airway remodeling and chronic Th2-type inflammation, a characteristic feature of asthma [169, 170].

In this Chapter we address the possibility that PAG1 expression influences the immune response to a viral LRTI.

6.1.2 Hypothesis

Results from the previous Chapter indicate that inhibition of PAG1 expression might predispose to a Th2-type immune response. Based on these results, we hypothesize that Pag1-/- mice have a compromised immune response to a viral LRTI in early life, and this is associated with a predisposition to Th2-type inflammation.

6.1.3 Aim

To compare features of airway inflammation in Pag1-/- and WT mice following PVM infection in early life.

Functional characterization of the new 8q21 Asthma risk locus | 232

Chapter 6 6.2 METHODS

6.2.1 Mouse strains

The strains used are described in detail in Chapter 5. WT and Pag1-/- mice for experiments were obtained from WT x WT and Pag1-/- x Pag1-/- crosses, respectively. The WT and Pag1-/- mice used in these crosses were littermates. Neonate mice were used for experiments at seven days of age. All experiments were approved and performed in accordance with UQ’s Animal Care and Ethics Committees (Brisbane, Australia).

6.2.2 Study design

Seven-day old WT and Pag1-/- mice were lightly anesthetised with isoflurane before being inoculated with 10 plaque forming units (PFU) of PVM (strain J3666) by intranasal administration of 10μl volume (10% fetal bovine serum (FBS; Sigma-Aldrich) in Dulbecco’s Modified Eagle Medium (DMEM; Invitrogen). Mice were euthanized seven (n=7 and 6 in WT and Pag1-/- groups, respectively) or ten days (n=4 and 5 in WT and Pag1-/- groups, respectively) post infection (dpi) for endpoint assessment, as illustrated in Figure 6.1. In this model, WT mice are able to clear the viral infection adequately by day 10 and do not develop Th2-type airway inflammation. As such, it is an appropriate model to test if down- regulation of Pag1 expression leads to impaired anti-viral immunity.

Figure 6.1 Viral-induced acute asthma model study design

6.2.3 Sample collection

BALF and lungs were collected as described in Chapter 5.

6.2.4 Assessment of airway inflammation

As described in Chapter 5, section 5.2.6.2.

Functional characterization of the new 8q21 Asthma risk locus | 233

Chapter 6 6.2.5 Cytokine measurement IFN-λ (IL-28A/B) and IL-33 (both R&D systems) levels were measured in BALF samples, as described in Chapter 5, section 5.2.6.3.

6.2.6 Histologic analysis of viral load and mucus production

For histologic analysis of viral load (PVM staining) and mucus production (Muc5ac staining), formalin fixed lung lobes were used to prepare 5μm paraffin-embedded sections. Lung tissue sections were deparaffinised and rehydrated through sequential incubations in Xylene and absolute or 70% ethanol solutions (both Chem-Supply). Antigen retrieval was performed by incubating the slides in citrate buffer (10mM Citric Acid, 0.05% Tween 20, pH 6.0) in a pressure cooker for 10min using the maximum heat setting of a microwave. Slides were then allowed to cool down to RT before being incubated with 0.5% TritonX- 100/PBS or 0.6% Tween-20/PBS (PBS-T; for HMGB1 staining) for 10min to permeabilize the cells. Samples were rinsed twice with 0.05% PBS-T and blocked with 10% normal goat serum (Sigma-Aldrich) for 30min, to avoid unspecific antibody binding. The blocking solution was then discarded and an (i) anti-PVM antibody (rabbit polyclonal; 1:8000 dilution in 10% FCS/PBS, kindly supplied by Dr Ursula J Buchholz, Laboratory of Infectious Diseases, National Institute of Allergy and Infectious Diseases, Bethesda, Maryland, USA), or (ii) anti-Muc5ac antibody (mouse monoclonal MA1-35706; clone 45M1; 1:1000 dilution in 10% FCS/PBS, Invitrogen), was added to stain samples overnight at RT in a humidity chamber. The following day, slides were washed with 0.6% PBS-T for 10min and then twice with 0.05% PBS-T for 3min followed by a quick rinse with 0.05% Tween- 20/TBS (TBS-T). Samples were then incubated with an Alkaline Phosphatase-conjugated anti-rabbit (for PVM detection) or anti-mouse (for Muc5a detection) secondary antibodies (AP-conjugated polyclonal goat anti-rabbit IgG A8438 or AP-conjugated polyclonal goat anti-mouse IgG A3562; both 1:200 dilution in TBS; Sigma-Aldrich) for 1 hour at RT. Slides were washed with 0.6% PBS-T for 10min and then twice with 0.05% PBS-T for 3min before being stained with Fast Red Substrate (Sigma-Aldrich). In the presence of alkaline phosphatase enzyme, Fast Red substrate acts as a chromogen, producing a red coloured product that can be visualized using bright field microscopy. Colour development was monitored (10-20min) and the reaction stopped by adding ddH2O. Sections were then counterstained with Mayer’s hematoxylin for 5min. Finally, slides were mounted in glycergel (Dako) and scanned using a digital slide scanner (Aperio ScanScope XT). The acquired images were analysed using the ImageScope software. Mucus secreting cells Functional characterization of the new 8q21 Asthma risk locus | 234

Chapter 6 were analysed as described in Chapter 5, section 5.2.10.5. Viral load was assessed by enumerating the number of AECs stained positive for PVM antigen. Four to five airways per mouse were counted and expressed as a histological score (mucus secreting cells) or as percentage of total AECs (PVM).

6.2.7 Statistical analysis

The software GraphPad Prism version 7.0 (La Jolla, USA) was used to graph and analyse all data, using the Mann-Whitney test. Results are presented as min-to-max box plots.

Functional characterization of the new 8q21 Asthma risk locus | 235

Chapter 6 6.3 RESULTS

In this experiment (Figure 6.2A) body weight increased steadily following viral inoculation in both WT (n=11 up to 7 dpi and n=4 from 7-10 dpi) and Pag1-/- mice (n=11 up to 7 dpi and n=5 from 7-10 dpi) (Figure 6.2B). These results suggest that the PVM infection did not have severe health effects on either group of mice.

We next compared the proportion of PVM-infected AECs between WT and Pag1-/- mice. At 7 dpi, the peak of viral infection [168, 170], Pag1-/- mice had significantly more infected cells than WT mice (Figure 6.2C). However, at 10 dpi, which corresponds to the resolution phase of the PVM infection, no differences were observed between groups. The virus was mainly located within airway epithelial cells with some viral particles also found scattered in the underlying mucosa and lung parenchyma (illustrative examples in Figure 6.2C).

Pag1-/- mice had an increased number of PVM-infected AECs, and so we hypothesised that the production of anti-viral cytokines might be compromised in these mice. Consistent with this possibility, there was a trend for lower BALF IFN-λ levels in Pag1-/- mice when compared to WT, both at 7 and 10 dpi (Figure 6.2D). However, these differences were not statistically significant. We also hypothesised that an increased number of infected AECs would be associated with increased release of alarmins, such as IL-33. Contrary to our prediction, IL-33 levels were comparable between groups (Figure 6.2D).

Functional characterization of the new 8q21 Asthma risk locus | 236

Chapter 6

Figure 6.2 Assessment of viral clearance and innate immune response to PVM infection. Responses measured in WT and Pag1-/- mice inoculated with PVM, according to the study design (A). (B) Measurement of animal body weight up to 10 dpi; (C) representative images of PVM-positive immunostained airways in lung sections and quantifications as percentage of positive AECs; (D) Measurement of BALF IFN-λ and IL-33 concentration. n=4-7 mice per group. **P<0.01, *P<0.05 vs. WT.

Functional characterization of the new 8q21 Asthma risk locus | 237

Chapter 6 Next, to investigate if a defect in DC numbers might underlie the increased AEC infection rate in Pag1-/- mice, we quantified DC numbers in BALF and lung tissue. However, we found no difference in the number of pDCs between WT and Pag1-/- mice (Figure 6.3).

Figure 6.3 Assessment of dendritic cell recruitment in response to PVM infection. Responses measured in WT and Pag1-/- mice inoculated with PVM, according to the study design (Figure 6.2 A). Enumeration of (A) total cells and (B) the pDC subset, in BALF and lung tissue. Results are shown both as frequency (% of WBC in each sample) and absolute number calculated according to the frequencies measured. n=4-7 mice per group.

Functional characterization of the new 8q21 Asthma risk locus | 238

Chapter 6 Finally, we explored the possibility that an increased number of PVM-infected AECs at 7 dpi would predispose Pag1-/- mice to subsequent Th2-type immune responses. We found that the number of eosinophils in both BALF and lung tissue were significantly higher at 10 dpi in Pag1-/- mice (Figure 6.4A). At the same time point, BALF neutrophil numbers were also elevated in the Pag1-/- group (Figure 6.4B). Furthermore, also at 10 dpi, we found a trend for increased mucus production in Pag1-/- mice (Figure 6.4C), however this difference was not statistically significant.

Functional characterization of the new 8q21 Asthma risk locus | 239

Chapter 6

Functional characterization of the new 8q21 Asthma risk locus | 240

Chapter 6

Figure 6.4 Assessment of granulocyte recruitment in response to PVM infection. Responses measured in WT and Pag1-/- mice inoculated with PVM, according to the study design (Figure 6.2 A). Enumeration of (A) eosinophils and (B) neutrophils, in BALF and lung tissue. Results are shown both as frequency (% of WBC in each sample) and absolute number calculated according to the frequencies measured. n=4-7 mice per group. *P<0.05 vs. WT.

Functional characterization of the new 8q21 Asthma risk locus | 241

Chapter 6 6.4 DISCUSSION

Results from this Chapter constitute the first assessment of the role of Pag1 in the pathophysiology of bronchiolitis, specifically in response to a primary infection with PVM.

The experimental design used mimics a natural RSV infection in young children, and results in similar features, such as airway sloughing and remodelling, oedema and granulocytic inflammation [168, 171]. In this model, viral burden in the lungs is elevated at 5 dpi, peaking at 7 dpi and cleared by 10 dpi. Antiviral IFN production as well as neutrophil recruitment to the lung tissue, follow a similar pattern, peaking at 7dpi. Plasmacytoid DCs (pDCs) play a critical role in the immune response to PVM, with increased numbers in BALF and lung detected as early as 24h post-inoculation and up to 7 dpi [172].

When we tested Pag1-/- mice in this model, we found that PVM-infected AECs were more common in Pag1-/- mice when compared to WT at day 7. This could arise because of an impaired anti-viral immune response in Pag1-/- animals. However, we did not observe lower pDC numbers or IFN-lambda levels in BALF or lungs of Pag1-/- mice, suggesting that Pag1 inhibition did not result in impaired DC recruitment or activation. A caveat of this analysis was that we did not measure type-I IFNs, which represent the majority of IFN produced by pDCs (e.g. IFN-α/β). IFNs play a critical role in viral infections, inducing an antiviral state in virtually all cell types [173] or severe disease if depleted [174] thus, representing an important and more relevant indicator to assess whether Pag1 modulates pDC function.

Alternatively, Pag1 knockout might compromise the function of other immune cells that play an important role in anti-viral immunity such as NK cells [172, 175, 176]. PAG1 might also directly affect the function of structural cells in the lung. PAG1 associates with caveolae – specialized lipid raft microdomains within the plasma membrane which are not fully understood [177] – through a constitutive interaction with caveolin-1 [37, 178]. Caveolae are present in both human and murine AECs [179], playing a critical role in epithelium barrier function [180]. Therefore, Pag1 might be required to establish tight junctions between AECs and so the epithelium of Pag1-/- mice might be more permeable, which would facilitate viral infiltration [181]. Future studies that address this possibility are warranted.

At 10 dpi, Pag1-/- mice had increased numbers of eosinophils in BALF and lung tissue, when compared to WT mice. We also observed a trend for increased mucus secretion at 10 dpi from AEC of Pag1-/- mice. These results indicate that Pag1-/- mice are predisposed

Functional characterization of the new 8q21 Asthma risk locus | 242

Chapter 6 to Th2-type airway inflammation after a PVM infection in early life. IL-33 was not elevated in BALF of Pag1-/- mice, either at 7 or 10 dpi, suggesting that other alarmins (e.g. HMGB1) or cytokines (e.g. IL-5) might explain the observed increase in eosinophil numbers.

Overall, our results suggest that Pag1 inhibition exacerbates airway inflammation in response to a primary infection with PVM, but the underlying mechanisms remain to be determined. Because of time constraints, we were not able to repeat the experiment described in this Chapter to confirm our findings, but this is planned for the near future.

Functional characterization of the new 8q21 Asthma risk locus | 243

7

The contribution of Zbtb10 to allergen- induced airway inflammation in mice

Chapter 7 7.1 INTRODUCTION

7.1.1 Background

ZBTB10 was identified as a target gene of 8q21 allergy risk variants (see Chapter 2) and so we were interested in performing preliminary experiments to help understand how the expression of this gene might influence asthma risk.

ZBTB10 has been very poorly studied to date: the PubMed query ‘zbtb10 OR “rin zf”’ performed on the 11th of October 2017, returned only 38 studies. Most of these (35 or 92%) were performed in the area of cancer, including 23 (61%) from the Safe lab at the Texas A&M University. In summary, these studies have found that ZBTB10: (i) is a transcription factor that represses the expression of the specificity proteins (Sp), which are up-regulated in several cancers [57, 182, 183]; and (ii) its expression influences various cancer-related mechanisms, such as angiogenesis and tumour metastasis [184], follicle stimulating hormone (FSH)-induced effects [185, 186], estrogen receptor-alpha expression and hormone responsiveness [187], tumour growth and cell apoptosis [186]. Furthermore, from the 35 cancer studies mentioned above, 22 (63%) report a role for the mir-27a- ZBTB10-Sp axis in the anticancer activity of several commercial drugs and natural compounds. Altogether, these findings suggest that ZBTB10 plays an important role in cancer-related mechanisms.

To our knowledge, ZBTB10 has not been directly linked to immune cell function or other mechanisms relevant to allergic disease. However, the following lines of evidence suggest that this is likely. First, overexpression of Sp1 (which is down-regulated by ZBTB10) in two murine cell lines induced the expression of genes involved in anti-viral immunity, such as Irf7 and Ifnb [85]. Second, our analyses of gene expression in LCLs stimulated with phenol (see Chapter 3) suggest that: (i) ZBTB10 induces the expression of immune genes that are associated with a Th2-like phenotype, such as IL-5 and CD40; and (ii) the expression of ZBTB10 is greatest after phenol exposure in LCLs from individuals homozygous for the asthma-predisposing allele. Third, as reviewed in Chapter 4, ZBTB10 is expressed in a variety of immune cell types, including alpha-beta T cells, conventional dendritic cells and naïve B cells in humans.

Collectively, these findings indicate that ZBTB10 expression might contribute to immune cell function and, in that way, contribute to asthma risk.

Functional characterization of the new 8q21 Asthma risk locus | 245

Chapter 7 7.1.2 Hypothesis

ZBTB10 expression plays a role in the (i) development of the immune system; and (ii) airway inflammation triggered by allergen exposure.

7.1.3 Aims

The experiments reported in this Chapter aimed to:

(i) Validate knockdown of gene expression in the Zbtb10het mouse strain; (ii) Compare immune cell numbers between Zbtb10het and WT mice; (iii) Assess the contribution of Zbtb10 to airway inflammation triggered by acute exposure to a high dose of allergen in mice

Functional characterization of the new 8q21 Asthma risk locus | 246

Chapter 7 7.2 METHODS

7.2.1 Mouse strains

Zbtb10het mice were generated by the Australian Phenomics Network at Monash University (Melbourne, Australia), through a contract research agreement with Dr. Ferreira. For details on the procedure used to generate the Zbtb10het mice, see the Appendix. The strain was re-derived using specific pathogen free (SPF) wild type (WT) C57Bl/6 mice purchased from UQ Biological Resources (Brisbane, Australia). Littermate WT and Zbtb10het mice were obtained through WT x Zbtb10het crossings. All mice were of C57/Bl6 background and used for experiments between eight to twelve weeks of age. All experiments were approved and performed in accordance with UQ’s Animal Care and Ethics Committees (Brisbane, Australia).

7.2.2 Genotyping

7.2.2.1 DNA extraction

DNA for genotyping was extracted from toe clippings from each animal as described in Chapter 5.

7.2.2.2 Genotyping PCR

2μL of DNA from each sample was used in a 25μL PCR reaction including 5μL 5x

Polymerase Buffer, 1.25μL MgCl2 25mM, 0.5μL dNTPs 12.5mM, 0.16μL of each primer Zbtb10_fwd (5’-GTAGAAGATTGCTCAGTGATGC-3’) and Zbtb10_rev (5’- CAGCGTGAGTGATCTTATGG-3’), and 0.2μL MangoTaq. PCR cycling conditions included an initial incubation at 50°C for 2min and then 95°C for 10mins, 45x cycles of [15sec at 95°C, 2.5min at 60ºC and 15secs at 80°C], followed by a final extension at 72ºC for 7min. PCR amplified fragments were then separated by electrophoreses in a 3% agarose gel for 1h at 100V. Samples containing a 320bp, 367bp or both bands were indicative of a WT, Zbtb10-/- or Zbtb10het animal, respectively. All mice were genotyped before experimental procedures.

Functional characterization of the new 8q21 Asthma risk locus | 247

Chapter 7 7.2.3 Sample collection

Blood, BALF, lung, lymph node, spleen and thymus samples were collected as described in Chapter 5.

7.2.4 Quantification of Zbtb10 expression and immune cell numbers in the new Zbtb10het mouse strain

7.2.4.1 Tissue collection

Following euthanasia by pentobarbital overdose, blood, lungs, spleen and thymus were harvested from WT (n=3) and Zbtb10het (n=3) mice. Lung lobes were dissected, with the left lobe of each mouse used for FACS and the rights lobes snap frozen in dry ice and stored at -80°C until needed. Cell homogenates from lung, spleen and thymus were achieved using a cell strainer pelleted, resuspended in 1mL Gey’s lysis buffer and washed twice in FACS buffer. Red blood cells in blood samples were lysed with two incubations with 2mL Gey’s lysis buffer followed by two washes in FACS buffer. Total cell counts, FACS staining and cell enumeration were performed as described in section 5.2.6.2.

7.2.4.2 Quantification of Zbtb10 gene expression

7.2.4.2.1 mRNA extraction

As described in Chapter 4, section 4.2.2.2.1.

7.2.4.2.2 cDNA conversion

As described in Chapter 3, section 3.2.1.1.2 (ii).

7.2.4.2.3 Gene expression analysis

Measured as described in Chapter 3, section 3.2.1.1.2 (iii), using the TaqMan Gene Expression assay Mm01281740_m1 (Zbtb10) from Life Technologies. Gene expression from each sample was measured in duplicate. Results were normalized against Gapdh

(assay Mm99999915_g1), analysed using the Comparative Ct method and shown as fold change values.

Functional characterization of the new 8q21 Asthma risk locus | 248

Chapter 7 7.2.4.3 Immune cell population screen

Several immune cell populations were quantified in spleen, lung, blood and thymus tissues from naïve WT (n=3) and Zbtb10het (n=3) mice, as described in Chapter 5, section 5.2.6.2.

7.2.5 Assessing the contribution of Zbtb10 to airway inflammation triggered by acute exposure to a high dose of allergen

7.2.5.1 Induction of allergic airway inflammation

WT and Zbtb10het mice were inoculated intranasally with a HDM suspension (100μg, batch 12, Greer laboratories; n=7 WT; n=5 Zbtb10het) or PBS (n=5 WT; n=3 Zbtb10het), as described in Chapter 5, section 5.2.6.1.

7.2.5.2 Assessment of allergic airway inflammation

As described in Chapter 5, section 5.2.6.2.

7.2.5.3 Lymph node cultures

Lymph nodes from PBS (n=5 WT; n=3 Zbtb10het) and HDM (n=7 WT; n=5 Zbtb10het) exposed mice were collected, processed and seeded as described in Chapter 5, section 5.2.9.4. Cells were incubated for 4 days at 37ºC in: (i) ACCM media only; or (ii) ACCM media supplemented with 50μg/mL HDM suspension (batch 12, Greer Laboratories). Cell culture volume was 200μL and after incubation the supernatant was collected for cytokine measurement.

7.2.5.4 Splenocyte cultures

Spleens were collected from PBS exposed mice only, including WT (n=3) and Zbtb10het (n=3) mice. Splenocyte homogenates were achieved using a cell strainer (40μm, BD Falcon) followed by a centrifugation at 1,600rpm for 5min at 4°C. Cell pellets were resuspended in 1 mL Gey’s lysis buffer and washed twice in FACS buffer before being counted. 1x106 cells were seeded in flat bottom 96-well plates and incubated in (i) ACCM media only; or (ii) ACCM media supplemented with different concentrations of a HDM suspension (0.001, 0.01, 0.1, 1, 10 or 100 μg/mL; CITEQ Biologics). Cell culture volume

Functional characterization of the new 8q21 Asthma risk locus | 249

Chapter 7 was 200μL and after 24h incubation at 37°C, the supernatant was collected for cytokine measurement.

7.2.6 Measurement of Cytokines

IL-4, IL-5 (both BD biosciences), IL-6 and IL-17A (both BioLegend) were measured as described in Chapter 5, section 5.2.6.3.

7.2.7 Statistical analysis

The software GraphPad Prism version 7.0 (La Jolla, USA) was used to graph and analyse all data. The Mann-Whitney test was used to compare quantitative outcomes between genotype groups. Fisher’s exact test was used to compare the observed and expected numbers of mice in each genotype class.

Functional characterization of the new 8q21 Asthma risk locus | 250

Chapter 7 7.3 RESULTS

7.3.1 Quantification of Zbtb10 expression and immune cell numbers in the new Zbtb10het mouse strain

7.3.1.1 Zbtb10 expression levels in the Zbtb10het strain

Zbtb10het animals were generated by the Australian Phenomics Network at Monash University (Melbourne, Australia). Following strain re-derivation at The University of Queensland, WT and Zbtb10het littermate mice were obtained through WT x Zbtb10het crossings. To confirm that gene expression was reduced in the Zbtb10het strain, we extracted RNA from lung tissue of WT and Zbtb10het littermate mice (n=3 mice per group). As expected, Zbtb10 expression in Zbtb10het was approximately 50% lower than in WT animals, consistent with the transcription of a single copy of Zbtb10 in Zbtb10het mice.

Figure 7.1 Zbtb10het strain validation. RNA was extracted from WT and Zbtb10het lung tissue (n=3 per group) using the TRIzol method. Gene expression was measured using TaqMan assays and Zbtb10het results were calculated as fold change from WT. Bars represent mean± SEM.

Functional characterization of the new 8q21 Asthma risk locus | 251

Chapter 7 7.3.1.2 Zbtb10 knockout is embryonic lethal

Our intention was to generate Zbtb10-/- mice for experimental use. However, no Zbtb10-/- mice were identified out of 22 animals born from seven Zbtb10het crossings (Table 7.1). Therefore, we conclude that diploid deletion of Zbtb10 is embryonic lethal, indicating that this gene plays a critical role in embryonic development.

Average litter size was 3.14 pups (SD=0.69), which was significantly lower (P=0.00016) when compared to the litter size from 28 Pag1het crossings (n=228 pups born, mean=8.14 pups per litter, SD=3.26). In total, we observed 8 (36%) Zbtb10het and 14 (64%) WT mice. The average number of WT mice observed per litter was not statistically different between Zbtb10het (mean=1.9, SD=1.1) and Pag1het (mean=1.7, SD=1.5) crossings. This indicates that WT mice were generated from Zbtb10het crossings at the expected frequency. In turn, this suggests that the reproductive system of the Zbtb10het breeders was not detrimentally affected by the 50% reduction in Zbtb10 expression.

Based on the number of WT mice, which should be ~25% of all mice produced from heterozygous crossings, we predict that approximately 56 mice (14/0.25) should have been born. If so, then we would have expected to observe 28 Zbtb10het and 14 Zbtb10-/- mice from these seven crossings. The difference between the expected and observed genotype numbers (Table 7.1) was statistically significant (Fisher’s exact test P-value = 0.001). Not only Zbtb10-/- mice were not generated, but the number of Zbtb10het mice born was also substantially lower than expected (8 vs. 28). These results suggest that haploid expression of Zbtb10 can also be embryonic lethal, although with incomplete penetrance (prediction: 20 of 28 = 71%). As we will see below, results from WT x Zbtb10het crosses support this hypothesis. There were relatively more males (64%) than expected (50%), but differences from expectation were not significant (Table 7.2).

Functional characterization of the new 8q21 Asthma risk locus | 252

Chapter 7

het het Total Genotype Zbtb10 x Zbtb10 pups WT HET KO Observed 22 14 8 0

Expected based on 56 14 28 14 extrapolation from n=14 WT

Table 7.1 Summary of genotype results from Zbtb10het crossings. The first row shows the observed number of mice in each of the three possible genotype classes, generated from seven Zbtb10het crossings. The second row shows the approximate number of mice in each genotype class (and in total) expected based on (i) Mendelian inheritance (i.e. 25% WT, 50% HET and 25% KO); (ii) no detrimental effect of Zbtb10 on embryonic survival; and (iii) the observed number of WT mice.

Gender Gender by genotype het het Total Zbtb10 x Zbtb10 WT HET pups Male Female (M) (F) M F M F Observed 22 14 8 5 9 3 5 Expected from n=14 WT - 11 11 7 7 4 4 and n=8 HET

Table 7.2 Gender breakdown of mice born from Zbtb10het crossings. The first row shows the observed number of males and females obtained from seven Zbtb10het crossings. The second row shows the number of males and females expected in total (and per genotype class) based on (i) 1:1 male to female ratio; and (ii) the observed number of WT and HET.

Functional characterization of the new 8q21 Asthma risk locus | 253

Chapter 7 7.3.1.3 Generation of littermate Zbtb10het and WT mice for experimental use

As we could not generate mice with constitutive diploid deletion of Zbtb10, and because cell-type specific Zbtb10-/- mice could not be produced within the time frame of this thesis, we opted for using Zbtb10het mice in our pilot experiments. These mice have ~50% lower expression of Zbtb10 relative to WT (Figure 5.1), so results should be interpreted in the context of partial downregulation of Zbtb10.

WT and Zbtb10het genotypes were generated by crossing WT with Zbtb10het littermate mice; these WT x Zbtb10het crossings are expected to result in 50% WT and 50% HET mice. A total of 112 pups were born from 29 crosses (mean= 3.8 pups per litter, SD= 1.5), including 39 Zbtb10het (35%) and 73 WT (65%) mice (Table 7.3). This represents a significant deviation from the expectation of 1:1 Zbtb10het to WT mice (P=0.02). Therefore, as observed in the Zbtb10het crossings, these results suggest that haploid expression of Zbtb10 is embryonic lethal in some mice (prediction: in 34 of 73 = 47%). There was a trend for a higher proportion of females (62%) than males (38%) in Zbtb10het animals, but this was not significantly different from the 50% expectation (Table 7.4).

Functional characterization of the new 8q21 Asthma risk locus | 254

Chapter 7

het Total Genotype WT x Zbtb10 pups WT HET Observed 112 73 39

Expected based on 146 73 73 extrapolation from n=73 WT

Table 7.3 Summary of genotype results from WT x Zbtb10het crossings. The first row shows the observed number of mice in each of the three possible genotype classes, generated from 29 WT x Zbtb10het crossings. The second row shows the approximate number of mice in each genotype class (and in total) expected based on (i) Mendelian inheritance (i.e. 50% WT and 50% HET); (ii) no detrimental effect of Zbtb10 on embryonic survival; and (iii) the observed number of WT mice.

Gender Gender by genotype het Total WT x Zbtb10 WT HET pups Male Female (M) (F) M F M F Observed 112 58 54 39 34 15 24

Expected from n=73 WT - 36-37 36-37 19-20 19-20 and n=39 HET 56 56

Table 7.4 Gender breakdown of mice born from WT x Zbtb10het crossings. The first row shows the observed number of males and females obtained from 29 WT x Zbtb10het crossings. The second row shows the number of males and females expected in total (and per genotype class) based on (i) 1:1 male to female ratio; and (ii) the observed number of WT and HET.

Functional characterization of the new 8q21 Asthma risk locus | 255

Chapter 7 7.3.1.4 Immune cell numbers in Zbtb10het mice

The first hypothesis we addressed in this Chapter was that Zbtb10 is required for normal immune cell function in mice. To this end, we measured 13 immune cell types in three relevant tissues (spleen, lung and blood) and seven immune cell types in an additional relevant tissue (thymus), in both WT and Zbtb10het mice (Figure 7.2). The main findings are summarised below:

(i) The relative number of conventional dendritic cells in spleen (relative to the total white blood cell count) was significantly lower in Zbtb10het mice, with a similar trend observed when considering absolute cell numbers; (ii) The absolute and relative number of macrophages/monocytes in lung tissue were significantly lower in Zbtb10het mice, with a similar trend in blood and spleen; (iii) The absolute number of NK cells in spleen was significantly lower in Zbtb10het mice, with a similar trend in lung and blood (relative numbers); (iv) There was a trend for reduced neutrophil and eosinophil numbers (absolute and relative) in lung and blood from Zbtb10het mice; (v) There was a trend for lower CD4+ and CD8+ Treg numbers (absolute) in Zbtb10het mice; (vi) There was a trend for lower B cell numbers (absolute) in spleen and lung from Zbtb10het mice; (vii) There were no obvious differences between WT and Zbtb10het mice for: a. T cell numbers (CD3+), CD4+ and CD8+ T cell subsets and gamma-delta T cells (γδ T) in all tissues studied; b. CD4+CD8+ T cells in the thymus; c. Natural killer T (NKT) cells in all tissues studied.

Therefore, overall, the results from this immune cell screen suggest that down-regulation of Zbtb10 can lead to a significant decrease in cell numbers for a wide range of immune cell types. This would be consistent with the hypothesis that ZBTB10 expression is required for normal haematopoiesis.

Functional characterization of the new 8q21 Asthma risk locus | 256

Chapter 7

Functional characterization of the new 8q21 Asthma risk locus | 257

Chapter 7

Functional characterization of the new 8q21 Asthma risk locus | 258

Chapter 7

Functional characterization of the new 8q21 Asthma risk locus | 259

Chapter 7

Figure 7.2 Immune cell population screen in relevant tissues. (A) total cells; (B) granulocytes, macrophages/monocytes, cDCs; (C) lymphocyte subsets; (D) regulatory T cells; and (E) NK cells, were enumerated in immune relevant tissues from WT and Zbtb10het mice. Tissues assessed included spleen, lung, blood and thymus and the tissue information for each cell type measured is described on the Y axis. n=3 mice per group. **P<0.005, *P<0.05 vs. WT, calculated using an Unpaired t Test.

Functional characterization of the new 8q21 Asthma risk locus | 260

Chapter 7 7.3.2 Assessing the contribution of Zbtb10 to airway inflammation triggered by acute exposure to a high dose of allergen

Results from our cellular experiments (Chapter 3) and the immune cell screen described above, suggest that decreased ZBTB10 expression might have a protective effect in allergen-induced inflammation.

To test this possibility, Zbtb10het and WT mice were inoculated intranasally with a HDM suspension at day 0 and then re-exposed to the same allergen two weeks later (Figure 7.3A). In this experiment, HDM-challenged WT and Zbtb10het mice had similar numbers of granulocytes and macrophages/monocytes in BALF (Figure 7.3C). BALF IL-5 levels were also comparable between the two groups (Figure 7.3E). On the other hand, total lymphocyte numbers were significantly elevated in HDM-challenged Zbtb10het mice (Figure 7.3D). However, a similar trend was also observed in PBS-challenged mice, suggesting that the differences were unrelated to allergen exposure. Thus, based on the relatively small number of inflammatory parameters measured, Zbtb10het mice do not appear to have reduced airway inflammation in response to HDM exposure.

Functional characterization of the new 8q21 Asthma risk locus | 261

Chapter 7

Functional characterization of the new 8q21 Asthma risk locus | 262

Chapter 7

Figure 7.3 Assessment of airway inflammation in a HDM-induced model of acute experimental asthma. Airway inflammation triggered in WT and Zbtb10het mice as per the study design (A) was assessed by: enumerating total cells (B); granulocytes (C); and (D) lymphocytes and macrophages/monocytes, in BALF. Results are shown both as frequency (% of WBC in each sample) and absolute number calculated according to the frequencies measured. (D) Measurement of IL-5 concentration in BALF. Bars represent means ± SEM, n=3-6 mice per group (one WT mouse and one ELISA data point were removed due to technical issues in BALF recovery and cytokine measurement). *P<0.05 vs. WT. §§P<0.01, §P<0.05 vs. PBS controls; calculated using a Mann-Whitney Test.

Functional characterization of the new 8q21 Asthma risk locus | 263

Chapter 7 We then characterised in more detail the contribution of Zbtb10 to the production of cytokines that are characteristic of an adaptive immune response to HDM. Lymph nodes were isolated from these mice and cultured ex vivo in the presence of HDM (50μL/mL; batch 12, Greer Laboratories) for four days. Relevant cytokines (IL-4, IL-5 and IL-17A) were then measured in the supernatant from these cultures. As expected, we found that exposure to HDM significantly induced cytokine production in lymph nodes isolated from mice previously sensitised to HDM but not from mice challenged with PBS (Figure 7.4B). However, there were no significant differences in HDM-induced cytokine levels between WT and Zbtb10het groups for any of the cytokines measured.

Functional characterization of the new 8q21 Asthma risk locus | 264

Chapter 7

Figure 7.4 Assessment of the contribution of Zbtb10 to adaptive cytokine production ex vivo. Results from lymph node cultures established from mice used in section 7.3.2. (A) Total MLN cell counts, and (B) cytokine measurements in cultures using cells from previously PBS or HDM-exposed mice. §§P<0.01, §P<0.05 vs. PBS controls; calculated using a Mann-Whitney test. n=2-3 for PBS and n=5-7 for HDM samples (one data point removed from IL-5 ELISA due to a technical issue).

Functional characterization of the new 8q21 Asthma risk locus | 265

Chapter 7 Lastly, we investigated if Zbtb10 might contribute to the activation of immune cells upon first exposure to HDM. In this case, we isolated spleen cells from WT (n=3) and Zbtb10het mice not previously exposed to HDM. The spleen constitutes a reservoir of several immune cell types and, in naïve mice, represents a greater source of cells for experiments than lymph nodes. Splenocytes were then incubated with increasing concentrations of HDM (batch 15G10, CITEQ Biologics) and the levels of IL-6 measured in the culture supernatants. We chose to measure IL-6 because it is a pleiotropic cytokine that plays a critical role in the innate to adaptive immunity interface [188]. We found that IL-6 levels measured in splenocytes exposed to low concentrations of HDM (up to 1μg/mL) were not elevated when compared to unstimulated cells, both from WT and Zbtb10het mice (Figure 7.5B). On the other hand, in cells from WT mice, the production of IL-6 increased significantly over baseline at higher concentrations of HDM (10 and 100μg/mL). The same pattern was observed in splenocytes from Zbtb10het mice, although there was a trend (P=0.2) for lower HDM-induced IL-6 production when compared to WT mice.

Functional characterization of the new 8q21 Asthma risk locus | 266

Chapter 7

Figure 7.5 Assessment of the contribution of Zbtb10 to innate cytokine production ex vivo. Results from splenocyte cultures established from naïve WT and Zbtb10het mice (A) Total splenocyte counts, and (B) IL-6 measurements in cultures with increasing concentrations of HDM (X axis). §§§P<0.001, §§P<0.01, §P<0.05 vs. no HDM; calculated using a two-way ANOVA test. n=3 mice per group.

Functional characterization of the new 8q21 Asthma risk locus | 267

Chapter 7 7.4 DISCUSSION

ZBTB10 belongs to the BTB-ZF (broad-complex, tramtrack and bric-à-brac - zinc finger) family of transcription factors. This family comprises 49 BTB-ZF proteins implicated in several critical processes including among others: immune development (ZBTB1, ZBTB7A/B, ZBTB16, ZBTB17, PATZ1, ZBTB24, BCL6/6B, ZBTB32 and ZBTB46), tumorigenesis (ZBTB7A/C, ZBTB16, BCL6, HIC1 and ZBTB33), fertility and embryogenesis (ZBTB7A, ZBTB16, ZBTB17, PATZ1, BCL6 and HIC1), and neurological development (ZBTB18, ZBTB20 and ZBTB33). Most members of this BTB-ZF family however, do not have a well characterized and described function, as is the case for ZBTB10 [1]. It is known that ZBTB10 is a transcriptional regulator that represses the specificity proteins SP1, SP3 and SP4 that in turn regulate several important cellular processes, mainly investigated in a cancer context. However, broader roles of ZBTB10 have not yet been identified.

To our knowledge Zbtb10-deficient mice had not been previously described. We contracted the Australian Phenomics Network at Monash University (Melbourne, Australia) to generate Zbtb10het mice, a strain that we then re-derived and expanded using specific pathogen-free animals. Following re-derivation, we proceeded to validate the strain by measuring the expression of Zbtb10 in WT and Zbtb10het mice. Results from this assessment demonstrated that Zbtb10 expression in Zbtb10het mice was about 0.5-fold (or 50%) lower than that observed in WT, an expected expression ratio between WT (diploid for Zbtb10) and heterozygous (haploid for Zbtb10) animals.

Next, when we attempted to generate Zbtb10-/- mice for experimental use, we found that no Zbtb10-/- mice were generated from heterozygous crossing. Furthermore, litter size was smaller than expected and the number of Zbtb10het mice obtained from the same crosses, and from the subsequent Zbtb10het x Zbtb10het crosses, was also below the expectation. Together, these results indicate that ZBTB10 plays a critical role in developmental stages, resulting in embryonic lethality in all Zbtb10-/- and some Zbtb10het (with predicted penetrance of 47%-71%) mice. We did not investigate the cause of the embryonic lethality. However, it is of interest that one of the likely targets of ZBTB10 identified in Chapter 3 – SKI, which encodes a nuclear proto-oncogene protein that represses TGF-beta signalling [189] – is highly expressed in blood vessels and the heart during embryonic development [190]. Furthermore, morpholino-based knockdown of SKI in zebrafish resulted in severe cardiac anomalies [190]. Since ZBTB10 is highly expressed in arteries and the heart

Functional characterization of the new 8q21 Asthma risk locus | 268

Chapter 7 (Chapter 4), we speculate that ZBTB10 knockdown leads to embryonic lethality because, at least in part, it leads to low expression of SKI in the developing cardiovascular system. Other possible explanations for the embryonic lethality include effects on neural development and the reproductive system, given the high expression of Zbtb10 in tissues from these organ systems.

As Zbtb10-/- mice could not be generated, we instead performed pilot experiments using WT and Zbtb10het mice. We first addressed the hypothesis that Zbtb10 knockdown could impair immune cell development. To investigate this possibility, we profiled the baseline inflammatory cell pool in naïve Zbtb10het and WT mice. A multitude of cell populations were screened including eosinophils, neutrophils, lymphocytes (CD4+ T, CD8+ T, NKT, NK, γδT cells and B cells), monocytes and dendritic cells (DCs), in several different tissues such as lung, spleen, blood and thymus (in the latter only CD4+ and CD8 T+ cells were measured). We found that for cell types for which there were significant differences between WT and Zbtb10het mice, or at least a trend for differences, Zbtb10het mice consistently had lower cell counts. Of these cell types, the largest differences between groups were observed for dendritic cells, macrophages/monocytes and NK cells. Interestingly, these cell types match the populations where Zbtb10 expression is thought to be highest, as discussed in Chapter 3 and observed in Siggs et al [1].

These results support a critical role for Zbtb10 in the development of the immune system. They also provide some clues as to which immune cell types might be affected functionally to a greater extend by a diploid deletion of Zbtb10. We will take this information into account when we attempt to generate mice with a cell-type specific deletion of Zbtb10, using the Cre-Lox and FLP-FRT systems, which is planned for the very near future.

Next, we performed pilot experiments to investigate the contribution of Zbtb10 to airway inflammation; specifically, to inflammation triggered by acute re-exposure to a high dose allergen, as discussed more extensively in Chapter 5. Results from the in vivo component of this experiment suggest that, despite having reduced numbers of relevant immune cell subsets at baseline, Zbtb10het mice were not protected from allergen-induced neutrophilic and eosinophilic recruitment to the airways. However, this observation should be interpreted with the following caveats in mind. Firstly, heterozygous mice have on average, 50% lower Zbtb10 expression when compared to WT mice. It is possible that transcription of a single copy of Zbtb10 is sufficient to adequately mount a typical allergic response to HDM. Another limitation of this study was the study design used; specifically, we used a

Functional characterization of the new 8q21 Asthma risk locus | 269

Chapter 7 high dose of allergen and only tested one type of allergen. As discussed in Chapter 5, a high dose of allergen can overload the immune system and result in an exaggerated and saturated inflammatory response, with potentially confounding effects on the experiment readouts, thus rendering mild phenotypes harder to discern. Different allergen types can also elicit different types of immune response, which might be more relevant in the context of Zbtb10 down-regulation. Therefore, results from this experiment do not definitely rule out a role for Zbtb10 in HDM-induced inflammation. To overcome these limitations, future studies should consider using a limiting dose of HDM, or a different allergen, as discussed in Chapter 5. Additionally, testing mice with cell specific deletion of Zbtb10 might be critical to characterise the contribution of this gene to airway inflammation in vivo.

In a final set of experiments, we used an ex vivo approach to test if Zbtb10 might contribute to the production of cytokines by immune cells that are required to mount an allergic response to HDM. We found that the production of Th2-type cytokines – specifically IL-4, IL-5 and IL-17A – was induced by HDM in lymph node cells from previously sensitised WT mice. This was an expected observation, which therefore validated the experimental approach used. However, when we measured the same cytokines in lymph node cultures from Zbtb10het mice, we found no significant differences compared to WT mice. These results suggest that partial down-regulation of Zbtb10 expression does not affect Th2 lymphocyte differentiation and/or function, which is consistent with our cytokine results from the in vivo experiments described above.

Lastly, we also measured IL-6 levels in splenocyte cultures from naïve mice, to investigate if Zbtb10 might contribute to mechanisms underlying allergic sensitization to HDM. We found that HDM induced IL-6 production significantly in WT mice, again validating the approach used. Interestingly, HDM-induced IL-6 production was lower in cells from Zbtb10het mice. Although differences were not statistically significant, the direction of effect is consistent with our prediction that down-regulation of Zbtb10 expression should be protective in the context of allergic inflammation. In the innate immune response, IL-6 is produced mainly by macrophages and dendritic cells [191, 192], two of the main cell types that express Zbtb10 and had reduced numbers in Zbtb10het mice. Therefore, it is possible that the reduced levels of IL-6 obtained from HDM-challenged splenocytes simply reflects the lower number of these cell types in Zbtb10het mice, and not a direct effect of Zbtb10 on IL-6 signalling. For example, DC numbers were about 25% lower in the spleen of Zbtb10het when compared to WT mice, which is very close to the reduction observed in IL-6 production (30%) in Zbtb10het mice.

Functional characterization of the new 8q21 Asthma risk locus | 270

Chapter 7 In conclusion, results from this chapter suggest that: (i) Zbtb10 is critical for embryonic development; (ii) Zbtb10 plays an important role in the development of the immune system; (iii) partial down-regulation of Zbtb10 does not result in attenuated granulocyte recruitment to the airways, or Th2 cytokine production, after exposure to HDM; (iv) Zbtb10 might influence the activation of immune cells during the allergen sensitization phase. Studies that confirm these preliminary findings are warranted.

Functional characterization of the new 8q21 Asthma risk locus | 271

8

Conclusions and future directions

Chapter 8 The overarching aim of this thesis was to elucidate the biological mechanisms underlying the association between 8q21 variants and asthma risk. Specifically, four main research questions were addressed:

1. What are the likely causal variants and the target genes of asthma risk variants in the 8q21 locus? 2. How is the expression of these genes modulated? 3. What tissues and cell types express these genes? 4. Are the target genes likely to contribute to disease pathophysiology?

In this final section, I summarise the main findings of this study and highlight some important future directions.

8.1 Main findings

1. Chapter 2. Using in silico and in vitro approaches, we were able to identify two target genes of 8q21 asthma risk variants, PAG1 and ZBTB10. We determined that the expression of both genes is modulated by enhancers that are located in the 8q21 core region of association. For PAG1, we also pinpointed one likely causal variant (rs2370615) which disrupts Foxo3a binding to the interacting enhancer. We found that the asthma risk allele was associated with increased PAG1 expression. This suggested for the first time that increased PAG1 expression might have a pro-inflammatory effect in asthma.

2. Chapter 3. The expression of both PAG1 and ZBTB10 was significantly up-regulated in LCLs by phenol exposure. These findings indicate that both genes play a role in the response to oxidative stress. We identified immune-related genes that might be up- regulated by ZBTB10, namely IL5 and CD40. These findings suggest that increased ZBTB10 expression might predispose to a Th2-type immune response.

3. Chapter 4. We identified the main tissues and immune cells that express PAG1 and ZBTB10 in humans and mice. We report, for the first time, that airway epithelial cells express PAG1, which might be up-regulated after allergen challenge. These results hint to the possibility that PAG1 expression might contribute to asthma risk by affecting the function of airway epithelial cells.

Functional characterization of the new 8q21 Asthma risk locus | 273

Chapter 8 4. Chapters 5 to 7. Results from the in vivo experiments performed for both Pag1 and Zbtb10 were not conclusive. However, they point to the possibility that inhibition of PAG1 exacerbates airway inflammation triggered by a low-dose of allergen (Chapter 5) or by a viral infection in early life (Chapter 6). On the other hand, we found that inhibition of Zbtb10 causes embryonic lethality and compromises the development of the immune system (Chapter 7). Zbtb10 expression might also play a role in the activation of immune cells during the allergen sensitization phase.

Together, these findings increase our understanding of: (i) the biological mechanisms underlying the association between 8q21 variants and asthma; (ii) the regulation, function and expression patterns of PAG1 and ZBTB10; and (iii) the contribution of both PAG1 and ZBTB10 to asthma pathophysiology. However, many questions still remain unanswered.

8.2 Future directions

Either due to time constraints or experimental caveats, both already acknowledged in each Chapter, we were unable to completely answer all the research questions posed in this project. Several outstanding questions can be used to inform future studies, and build upon the findings from this thesis:

1. We have identified a transcription factor that modulates PAG1 expression in individuals carrying the asthma protective allele for rs7009110; however we have not been able to confirm the transcription factor responsible for PAG1 up-regulation in individuals carrying the asthma risk allele. Having found a stimulus that significantly up-regulates the expression of PAG1 (phenol), repeating the ChIP assay in cells exposed to this stimulus might be informative to address this question. Similarly, phenol significantly up-regulated ZBTB10 expression; therefore, we suggest that this is important to repeat the luciferase assays after stimulating cells with phenol. This might help confirm the putative causal variant underlying the association with asthma risk. Furthermore, ChIP assays should also be employed to determine the transcription factor(s) that modulate ZBTB10 expression. To test the possibility that the asthma risk variants might influence the expression of other genes in the locus, additional 3C experiments targeting the promoter of the other genes in this locus should also be carried out.

Functional characterization of the new 8q21 Asthma risk locus | 274

Chapter 8 2. We identified a trend for a difference in phenol-induced gene expression between individuals carrying the risk and protective alleles. We suggest that these experiments should be expanded to increase the sample size, and therefore the power, to detect a significant genotype effect.

3. We identified a small number of genes that are likely to be modulated by ZBTB10. These predictions need to be confirmed and the analysis extended to all genes in the genome (e.g. using ChIP-seq).

4. The observation that Pag1 is expressed in airway epithelial cells should be confirmed in lung sections from additional mice exposed to various stimuli (e.g. PBS, house dust mite, cockroach, Alternaria or PVM). This will help determine if the up-regulation of Pag1 in these cells is reproducible and observed with different stimuli. Additionally, we suggest that the immunostaining assays be repeated including additional antibodies, to identify specific immune cell subtypes expressing Pag1 in vivo in the context of allergen- of viral-induced airway inflammation.

5. Results from our in vivo studies have not been conclusive with regards to the contribution of Pag1 to airway inflammation. Further studies to address the caveats of our studies are warranted. For example, additional studies should be performed (i) to help confirm the granulocyte phenotype observed in Pag1-/- mice; (ii) to investigate lymphocyte differentiation, expansion and cytokine production ex vivo; and (iii) to investigate the expression pattern of Pag1 in airway epithelial cells and the impact of Pag1 inhibition on the function of these cells.

6. Similarly, additional studies should also be performed to build upon our preliminary findings using Zbtb10het mice. First and foremost, it is important to generate mice with tissue specific Zbtbt10 deletion, so that a role in airway inflammation can be studied in greater detail. Additionally, future studies assessing the immune response triggered (i) by different allergens; (ii) using lower doses of these allergens in the challenge phase; (iii) in the sensitization phase to these allergens; and (iv) by a viral infection, would also be important to understand the contribution of Zbtb10 to asthma pathophysiology.

Functional characterization of the new 8q21 Asthma risk locus | 275

References

1. Siggs, O.M. and B. Beutler, The BTB-ZF transcription factors. Cell Cycle, 2012. 11(18): p. 3358-69. 2. Ecker, J.R., et al., Genomics: ENCODE explained. Nature, 2012. 489(7414): p. 52- 5. 3. Ferreira, M.A., et al., Genome-wide association analysis identifies 11 risk variants associated with the asthma with hay fever phenotype. J Allergy Clin Immunol, 2013. 4. Painter, J.N., et al., A Common Variant at the 14q32 Endometrial Cancer Risk Locus Activates AKT1 through YY1 Binding. Am J Hum Genet, 2016. 98(6): p. 1159-1169. 5. Helbig, S., et al., Functional dissection of breast cancer risk-associated TERT promoter variants. Oncotarget, 2017. 8(40): p. 67203-67217. 6. Vicente, C.T., et al., Long-Range Modulation of PAG1 Expression by 8q21 Allergy Risk Variants. Am J Hum Genet, 2015. 97(2): p. 329-36. 7. Purcell, S., et al., PLINK: a tool set for whole-genome association and population- based linkage analyses. Am J Hum Genet, 2007. 81(3): p. 559-75. 8. Genomes Project, C., et al., A global reference for human genetic variation. Nature, 2015. 526(7571): p. 68-74. 9. Lappalainen, T., et al., Transcriptome and genome sequencing uncovers functional variation in humans. Nature, 2013. 501(7468): p. 506-11. 10. Fairfax, B.P., et al., Genetics of gene expression in primary immune cells identifies cell type-specific master regulators and roles of HLA alleles. Nat Genet, 2012. 44(5): p. 502-10. 11. Dixon, A.L., et al., A genome-wide association study of global gene expression. Nat Genet, 2007. 39(10): p. 1202-7. 12. Hao, K., et al., Lung eQTLs to help reveal the molecular underpinnings of asthma. PLoS Genet, 2012. 8(11): p. e1003029. 13. Zeller, T., et al., Genetics and beyond--the transcriptome of human monocytes and disease susceptibility. PLoS One, 2010. 5(5): p. e10693. 14. Westra, H.J., et al., Systematic identification of trans eQTLs as putative drivers of known disease associations. Nat Genet, 2013. 45(10): p. 1238-43. 15. Ferreira, M.A. and S.M. Purcell, A multivariate test of association. Bioinformatics, 2009. 25(1): p. 132-3. 16. Kornberg, R.D., Chromatin structure: a repeating unit of histones and DNA. Science, 1974. 184(4139): p. 868-71. 17. Greer, E.L. and Y. Shi, Histone methylation: a dynamic mark in health, disease and inheritance. Nat Rev Genet, 2012. 13(5): p. 343-57. 18. Boyle, A.P., et al., High-resolution mapping and characterization of open chromatin across the genome. Cell, 2008. 132(2): p. 311-22. 19. Consortium, E.P., An integrated encyclopedia of DNA elements in the human genome. Nature, 2012. 489(7414): p. 57-74. 20. Ferreira, M.A., et al., Genome-wide association analysis identifies 11 risk variants associated with the asthma with hay fever phenotype. J Allergy Clin Immunol, 2014. 133(6): p. 1564-71. 21. Genomes Project, C., et al., An integrated map of genetic variation from 1,092 human genomes. Nature, 2012. 491(7422): p. 56-65. 22. Rockman, M.V. and L. Kruglyak, Genetics of global gene expression. Nat Rev Genet, 2006. 7(11): p. 862-72. 23. Ward, L.D. and M. Kellis, HaploReg: a resource for exploring chromatin states, conservation, and regulatory motif alterations within sets of genetically linked variants. Nucleic Acids Res, 2012. 40(Database issue): p. D930-4. 24. Boyle, A.P., et al., Annotation of functional variation in personal genomes using RegulomeDB. Genome Res, 2012. 22(9): p. 1790-7.

Functional characterization of the new 8q21 Asthma risk locus | 277

25. Lin, L., J.D. Hron, and S.L. Peng, Regulation of NF-kappaB, Th activation, and autoinflammation by the forkhead transcription factor Foxo3a. Immunity, 2004. 21(2): p. 203-13. 26. Malin, D., et al., Forkhead box F1 is essential for migration of mesenchymal cells and directly induces integrin-beta3 expression. Mol Cell Biol, 2007. 27(7): p. 2486- 98. 27. Bagati, A., et al., Melanoma Suppressor Functions of the Carcinoma Oncogene FOXQ1. Cell Rep, 2017. 20(12): p. 2820-2832. 28. Liu, J.Y., et al., FOXQ1 promotes cancer metastasis by PI3K/AKT signaling regulation in colorectal carcinoma. Am J Transl Res, 2017. 9(5): p. 2207-2218. 29. Wang, Y., et al., Overexpression of forkhead box J2 can decrease the migration of breast cancer cells. J Cell Biochem, 2012. 113(8): p. 2729-37. 30. Zhang, Z., et al., The prognostic role and reduced expression of FOXJ2 in human hepatocellular carcinoma. Mol Med Rep, 2016. 14(1): p. 254-62. 31. Altomonte, J., et al., Inhibition of Foxo1 function is associated with improved fasting glycemia in diabetic mice. Am J Physiol Endocrinol Metab, 2003. 285(4): p. E718- 28. 32. Nakae, J., et al., The forkhead transcription factor Foxo1 regulates adipocyte differentiation. Dev Cell, 2003. 4(1): p. 119-29. 33. Tone, M., et al., IL-10 gene expression is controlled by the transcription factors Sp1 and Sp3. J Immunol, 2000. 165(1): p. 286-91. 34. Tone, M., et al., The role of Sp1 and NF-kappa B in regulating CD40 gene expression. J Biol Chem, 2002. 277(11): p. 8890-7. 35. Kelly, R.D. and S.M. Cowley, The physiological roles of histone deacetylase (HDAC) 1 and 2: complex co-stars with multiple leading parts. Biochem Soc Trans, 2013. 41(3): p. 741-9. 36. Brdicka, T., et al., Phosphoprotein associated with glycosphingolipid-enriched microdomains (PAG), a novel ubiquitously expressed transmembrane adaptor protein, binds the protein tyrosine kinase csk and is involved in regulation of T cell activation. J Exp Med, 2000. 191(9): p. 1591-604. 37. Hrdinka, M. and V. Horejsi, PAG--a multipurpose transmembrane adaptor protein. Oncogene, 2014. 33(41): p. 4881-92. 38. Seumois, G., et al., Epigenomic analysis of primary human T cells reveals enhancers associated with TH2 memory cell differentiation and asthma susceptibility. Nat Immunol, 2014. 15(8): p. 777-88. 39. Hnisz, D., et al., Super-enhancers in the control of cell identity and disease. Cell, 2013. 155(4): p. 934-47. 40. Bonizzi, G. and M. Karin, The two NF-kappaB activation pathways and their role in innate and adaptive immunity. Trends Immunol, 2004. 25(6): p. 280-8. 41. Schubart, K., et al., B cell development and immunoglobulin gene transcription in the absence of Oct-2 and OBF-1. Nat Immunol, 2001. 2(1): p. 69-74. 42. Mok, Y., et al., MiR-210 is induced by Oct-2, regulates B cells, and inhibits autoantibody production. J Immunol, 2013. 191(6): p. 3037-48. 43. Zhao, B., et al., The NF-kappaB genomic landscape in lymphoblastoid B cells. Cell Rep, 2014. 8(5): p. 1595-606. 44. Sepulveda, M.A., A.V. Emelyanov, and B.K. Birshtein, NF-kappa B and Oct-2 synergize to activate the human 3' Igh hs4 enhancer in B cells. J Immunol, 2004. 172(2): p. 1054-64. 45. Jin, F., et al., A high-resolution map of the three-dimensional chromatin interactome in human cells. Nature, 2013. 503(7475): p. 290-4.

Functional characterization of the new 8q21 Asthma risk locus | 278

46. Itoh, K., et al., Cutting edge: negative regulation of immune synapse formation by anchoring lipid raft to cytoskeleton through Cbp-EBP50-ERM assembly. J Immunol, 2002. 168(2): p. 541-4. 47. Kalland, M.E., et al., Modulation of proximal signaling in normal and transformed B cells by transmembrane adapter Cbp/PAG. Exp Cell Res, 2012. 318(14): p. 1611-9. 48. Davidson, D., et al., Phosphorylation-dependent regulation of T-cell activation by PAG/Cbp, a lipid raft-associated transmembrane adaptor. Mol Cell Biol, 2003. 23(6): p. 2017-28. 49. Smida, M., et al., PAG/Cbp suppression reveals a contribution of CTLA-4 to setting the activation threshold in T cells. Cell Commun Signal, 2013. 11(1): p. 28. 50. Ohtake, H., et al., Cutting Edge: Transmembrane phosphoprotein Csk-binding protein/phosphoprotein associated with glycosphingolipid-enriched microdomains as a negative feedback regulator of mast cell signaling through the FcepsilonRI. J Immunol, 2002. 168(5): p. 2087-90. 51. Xu, S., et al., Cbp deficiency alters Csk localization in lipid rafts but does not affect T-cell development. Mol Cell Biol, 2005. 25(19): p. 8486-95. 52. Dobenecker, M.W., et al., The ubiquitously expressed Csk adaptor protein Cbp is dispensable for embryogenesis and T-cell development and function. Mol Cell Biol, 2005. 25(23): p. 10533-42. 53. Inomata, M., et al., Impairment in a negative regulatory system for TCR signaling in CD4+ T cells from old mice. FEBS Lett, 2007. 581(16): p. 3039-43. 54. Tedoldi, S., et al., Transmembrane adaptor molecules: a new category of lymphoid- cell markers. Blood, 2006. 107(1): p. 213-21. 55. Lloyd, C.M. and E.M. Hessel, Functions of T cells in asthma: more than just T(H)2 cells. Nat Rev Immunol, 2010. 10(12): p. 838-48. 56. Ghosh, S., et al., B lymphocytes regulate airway granulocytic inflammation and cytokine production in a murine model of fungal allergic asthma. Cell Mol Immunol, 2015. 12(2): p. 202-12. 57. Tillotson, L.G., RIN ZF, a novel zinc finger gene, encodes proteins that bind to the CACC element of the gastrin promoter. J Biol Chem, 1999. 274(12): p. 8123-8. 58. Yousuf, S., et al., Raf kinase inhibitor protein (RKIP) blocks signal transducer and activator of transcription 3 (STAT3) activation in breast and prostate cancer. PLoS One, 2014. 9(3): p. e92478. 59. Bhavsar, P., T. Ahmad, and I.M. Adcock, The role of histone deacetylases in asthma and allergic diseases. J Allergy Clin Immunol, 2008. 121(3): p. 580-4. 60. Tang, W., et al., MiR-27 as a prognostic marker for breast cancer progression and patient survival. PLoS One, 2012. 7(12): p. e51702. 61. van Dijk, T.B., et al., A composite C/EBP binding site is essential for the activity of the promoter of the IL-3/IL-5/granulocyte-macrophage colony-stimulating factor receptor beta c gene. J Immunol, 1999. 163(5): p. 2674-80. 62. Zhang, W., et al., Identification and characterization of DPZF, a novel human BTB/POZ zinc finger protein sharing homology to BCL-6. Biochem Biophys Res Commun, 2001. 282(4): p. 1067-73. 63. Liu, X., et al., Zinc finger protein ZBTB20 promotes Toll-like receptor-triggered innate immune responses by repressing IkappaBalpha gene transcription. Proc Natl Acad Sci U S A, 2013. 110(27): p. 11097-102. 64. Sathasivam, P., et al., The role of the coiled-coil motif in interactions mediated by TPD52. Biochem Biophys Res Commun, 2001. 288(1): p. 56-61. 65. Thomas, D.D., et al., Tumor protein D52 expression and Ca2+-dependent phosphorylation modulates lysosomal membrane protein trafficking to the plasma membrane. Am J Physiol Cell Physiol, 2010. 298(3): p. C725-39.

Functional characterization of the new 8q21 Asthma risk locus | 279

66. Balleine, R.L., et al., The hD52 (TPD52) gene is a candidate target gene for events resulting in increased 8q21 copy number in human breast carcinoma. Genes Chromosomes Cancer, 2000. 29(1): p. 48-57. 67. Byrne, J.A., et al., Tumor protein D52 (TPD52) is overexpressed and a gene amplification target in ovarian cancer. Int J Cancer, 2005. 117(6): p. 1049-54. 68. Rubin, M.A., et al., Overexpression, amplification, and androgen regulation of TPD52 in prostate cancer. Cancer Res, 2004. 64(11): p. 3814-22. 69. Ummanni, R., et al., Altered expression of tumor protein D52 regulates apoptosis and migration of prostate cancer cells. FEBS J, 2008. 275(22): p. 5703-13. 70. Tiacci, E., et al., Tumor protein D52 (TPD52): a novel B-cell/plasma-cell molecule with unique expression pattern and Ca(2+)-dependent association with annexin VI. Blood, 2005. 105(7): p. 2812-20. 71. Smathers, R.L. and D.R. Petersen, The human fatty acid-binding protein family: evolutionary divergences and functions. Hum Genomics, 2011. 5(3): p. 170-91. 72. Fang, L.Y., et al., Fatty-acid-binding protein 5 promotes cell proliferation and invasion in oral squamous cell carcinoma. J Oral Pathol Med, 2010. 39(4): p. 342-8. 73. Ogawa, E., et al., Epidermal FABP (FABP5) regulates keratinocyte differentiation by 13(S)-HODE-mediated activation of the NF-kappaB signaling pathway. J Invest Dermatol, 2011. 131(3): p. 604-12. 74. Mabalirajan, U., et al., Linoleic acid metabolite drives severe asthma by causing airway epithelial injury. Sci Rep, 2013. 3: p. 1349. 75. Gally, F., et al., FABP5 deficiency enhances susceptibility to H1N1 influenza A virus-induced lung inflammation. Am J Physiol Lung Cell Mol Physiol, 2013. 305(1): p. L64-72. 76. Gally, F., H.W. Chu, and R.P. Bowler, Cigarette smoke decreases airway epithelial FABP5 expression and promotes Pseudomonas aeruginosa infection. PLoS One, 2013. 8(1): p. e51784. 77. Schutze, S., et al., TNF-induced activation of NF-kappa B. Immunobiology, 1995. 193(2-4): p. 193-203. 78. Yang, C.H., et al., IFNalpha/beta promotes cell survival by activating NF-kappa B. Proc Natl Acad Sci U S A, 2000. 97(25): p. 13631-6. 79. Schorg, A., et al., Destruction of a distal hypoxia response element abolishes trans- activation of the PAG1 gene mediated by HIF-independent chromatin looping. Nucleic Acids Res, 2015. 43(12): p. 5810-23. 80. Kirshenbaum, A.S., et al., Demonstration that human mast cells arise from a progenitor cell population that is CD34(+), c-kit(+), and expresses aminopeptidase N (CD13). Blood, 1999. 94(7): p. 2333-42. 81. Laidlaw, T.M., et al., Characterization of a novel human mast cell line that responds to stem cell factor and expresses functional FcepsilonRI. J Allergy Clin Immunol, 2011. 127(3): p. 815-22 e1-5. 82. Horejsi, V., W. Zhang, and B. Schraven, Transmembrane adaptor proteins: organizers of immunoreceptor signalling. Nat Rev Immunol, 2004. 4(8): p. 603-16. 83. Palsson-McDermott, E.M. and L.A. O'Neill, Signal transduction by the lipopolysaccharide receptor, Toll-like receptor-4. Immunology, 2004. 113(2): p. 153- 62. 84. Sreevalsan, S. and S. Safe, Reactive Oxygen Species and Colorectal Cancer. Curr Colorectal Cancer Rep, 2013. 9(4): p. 350-357. 85. Dupuis-Maurin, V., et al., Overexpression of the transcription factor Sp1 activates the OAS-RNAse L-RIG-I pathway. PLoS One, 2015. 10(3): p. e0118551. 86. Hayden, M.S. and S. Ghosh, Regulation of NF-kappaB by TNF family cytokines. Semin Immunol, 2014. 26(3): p. 253-66.

Functional characterization of the new 8q21 Asthma risk locus | 280

87. Yang, C.H., et al., Interferon alpha activates NF-kappaB in JAK1-deficient cells through a TYK2-dependent pathway. J Biol Chem, 2005. 280(27): p. 25849-53. 88. Neumann, M., et al., Differential expression of Rel/NF-kappaB and octamer factors is a hallmark of the generation and maturation of dendritic cells. Blood, 2000. 95(1): p. 277-85. 89. Ammon, C., et al., Differential expression of the transcription factor NF-kappaB during human mononuclear phagocyte differentiation to macrophages and dendritic cells. Biochem Biophys Res Commun, 2000. 268(1): p. 99-105. 90. O'Sullivan, B.J., et al., RelB nuclear translocation regulates B cell MHC molecule, CD40 expression, and antigen-presenting cell function. Proc Natl Acad Sci U S A, 2000. 97(21): p. 11421-6. 91. Pfeffer, L.M., The role of nuclear factor kappaB in the interferon response. J Interferon Cytokine Res, 2011. 31(7): p. 553-9. 92. Eltzschig, H.K. and P. Carmeliet, Hypoxia and inflammation. N Engl J Med, 2011. 364(7): p. 656-65. 93. Broytman, O., et al., Effects of chronic intermittent hypoxia on allergen-induced airway inflammation in rats. Am J Respir Cell Mol Biol, 2015. 52(2): p. 162-70. 94. Baxi, S.N. and W. Phipatanakul, The role of allergen exposure and avoidance in asthma. Adolesc Med State Art Rev, 2010. 21(1): p. 57-71, viii-ix. 95. Thompson, P.J., Unique role of allergens and the epithelium in asthma. Clin Exp Allergy, 1998. 28 Suppl 5: p. 110-6; discussion 117-8. 96. Wan, H., et al., Der p 1 facilitates transepithelial allergen delivery by disruption of tight junctions. J Clin Invest, 1999. 104(1): p. 123-33. 97. Paternoster, L., et al., Multi-ancestry genome-wide association study of 21,000 cases and 95,000 controls identifies new risk loci for atopic dermatitis. Nat Genet, 2015. 47(12): p. 1449-1456. 98. Rockwood, J., M.S. Morgan, and L.G. Arlian, Proteins and endotoxin in house dust mite extracts modulate cytokine secretion and gene expression by dermal fibroblasts. Exp Appl Acarol, 2013. 61(3): p. 311-25. 99. Dickinson, B.C. and C.J. Chang, Chemistry and biology of reactive oxygen species in signaling or stress responses. Nat Chem Biol, 2011. 7(8): p. 504-11. 100. Safe, S. and R. Kasiappan, Natural Products as Mechanism-based Anticancer Agents: Sp Transcription Factors as Targets. Phytother Res, 2016. 30(11): p. 1723- 1732. 101. Bradding, P., A.F. Walls, and S.T. Holgate, The role of the mast cell in the pathophysiology of asthma. J Allergy Clin Immunol, 2006. 117(6): p. 1277-84. 102. Morgan, M.J. and Z.G. Liu, Crosstalk of reactive oxygen species and NF-kappaB signaling. Cell Res, 2011. 21(1): p. 103-15. 103. Ferber, E.C., et al., FOXO3a regulates reactive oxygen metabolism by inhibiting mitochondrial gene expression. Cell Death Differ, 2012. 19(6): p. 968-79. 104. Liu, X., et al., Ski/Sno and TGF-beta signaling. Cytokine Growth Factor Rev, 2001. 12(1): p. 1-8. 105. Chatzigeorgiou, A., et al., CD40/CD40L signaling and its implication in health and disease. Biofactors, 2009. 35(6): p. 474-83. 106. Sanderson, C.J., Interleukin-5, eosinophils, and disease. Blood, 1992. 79(12): p. 3101-9. 107. Brewer, J.A., et al., T-cell glucocorticoid receptor is required to suppress COX-2- mediated lethal immune activation. Nat Med, 2003. 9(10): p. 1318-22. 108. Diefenbach, A., et al., Requirement for type 2 NO synthase for IL-12 signaling in innate immunity. Science, 1999. 284(5416): p. 951-5. 109. Stanley, E.R., et al., Biology and action of colony--stimulating factor-1. Mol Reprod Dev, 1997. 46(1): p. 4-10.

Functional characterization of the new 8q21 Asthma risk locus | 281

110. Park, J.H., et al., Association analysis of CD40 polymorphisms with asthma and the level of serum total IgE. Am J Respir Crit Care Med, 2007. 175(8): p. 775-82. 111. Burgess, J.K., et al., CD40 and OX40 ligand are increased on stimulated asthmatic airway smooth muscle. J Allergy Clin Immunol, 2005. 115(2): p. 302-8. 112. Krimmer, D.I., et al., CD40 and OX40 ligand are differentially regulated on asthmatic airway smooth muscle. Allergy, 2009. 64(7): p. 1074-82. 113. Hong, G.U., et al., IgE production in CD40/CD40L cross-talk of B and mast cells and mediator release via TGase 2 in mouse allergic asthma. Cell Signal, 2013. 25(6): p. 1514-25. 114. Propst, S.M., et al., Proinflammatory and Th2-derived cytokines modulate CD40- mediated expression of inflammatory mediators in airway epithelia: implications for the role of epithelial CD40 in airway inflammation. J Immunol, 2000. 165(4): p. 2214-21. 115. Caux, C., et al., Activation of human dendritic cells through CD40 cross-linking. J Exp Med, 1994. 180(4): p. 1263-72. 116. O'Sullivan, B.J. and R. Thomas, CD40 ligation conditions dendritic cell antigen- presenting function through sustained activation of NF-kappaB. J Immunol, 2002. 168(11): p. 5491-8. 117. Carithers, L.J. and H.M. Moore, The Genotype-Tissue Expression (GTEx) Project. Biopreserv Biobank, 2015. 13(5): p. 307-8. 118. Noguchi, S., et al., FANTOM5 CAGE profiles of human and mouse samples. Sci Data, 2017. 4: p. 170112. 119. Huntley, M.A., et al., Complex regulation of ADAR-mediated RNA-editing across tissues. BMC Genomics, 2016. 17: p. 61. 120. Heng, T.S., M.W. Painter, and C. Immunological Genome Project, The Immunological Genome Project: networks of gene expression in immune cells. Nat Immunol, 2008. 9(10): p. 1091-4. 121. Rock, J.R., et al., Basal cells as stem cells of the mouse trachea and human airway epithelium. Proc Natl Acad Sci U S A, 2009. 106(31): p. 12771-5. 122. Phipps, S., et al., Toll/IL-1 signaling is critical for house dust mite-specific helper T cell type 2 and type 17 [corrected] responses. Am J Respir Crit Care Med, 2009. 179(10): p. 883-93. 123. Consortium, G.T., Human genomics. The Genotype-Tissue Expression (GTEx) pilot analysis: multitissue gene regulation in humans. Science, 2015. 348(6235): p. 648- 60. 124. Wu, C., et al., BioGPS: an extensible and customizable portal for querying and organizing gene annotation resources. Genome Biol, 2009. 10(11): p. R130. 125. Holtzman, M.J., et al., The role of airway epithelial cells and innate immune cells in chronic respiratory disease. Nat Rev Immunol, 2014. 14(10): p. 686-98. 126. Mantovani, A., et al., Neutrophils in the activation and regulation of innate and adaptive immunity. Nat Rev Immunol, 2011. 11(8): p. 519-31. 127. Corrigan, C.J. and A.B. Kay, T cells and eosinophils in the pathogenesis of asthma. Immunol Today, 1992. 13(12): p. 501-7. 128. Mildner, A. and S. Jung, Development and function of dendritic cell subsets. Immunity, 2014. 40(5): p. 642-56. 129. Lindquist, S., et al., Phosphoprotein associated with glycosphingolipid-enriched microdomains differentially modulates SRC kinase activity in brain maturation. PLoS One, 2011. 6(9): p. e23978. 130. Ullah, M.A., et al., Allergen-induced IL-6 trans-signaling activates gammadelta T cells to promote type 2 and type 17 airway inflammation. J Allergy Clin Immunol, 2015. 136(4): p. 1065-73.

Functional characterization of the new 8q21 Asthma risk locus | 282

131. Livraghi, A., et al., Airway and lung pathology due to mucosal surface dehydration in {beta}-epithelial Na+ channel-overexpressing mice: role of TNF-{alpha} and IL- 4R{alpha} signaling, influence of neonatal development, and limited efficacy of glucocorticoid treatment. J Immunol, 2009. 182(7): p. 4357-67. 132. Reutershan, J., et al., Sequential recruitment of neutrophils into lung and bronchoalveolar lavage fluid in LPS-induced acute lung injury. Am J Physiol Lung Cell Mol Physiol, 2005. 289(5): p. L807-15. 133. Dullaers, M., et al., House dust mite-driven asthma and allergen-specific T cells depend on B cells when the amount of inhaled allergen is limiting. J Allergy Clin Immunol, 2016. 134. Yang, Y. and B. Seed, Site-specific gene targeting in mouse embryonic stem cells with intact bacterial artificial chromosomes. Nat Biotechnol, 2003. 21(4): p. 447-51. 135. Nials, A.T. and S. Uddin, Mouse models of allergic asthma: acute and chronic allergen challenge. Dis Model Mech, 2008. 1(4-5): p. 213-20. 136. Lloyd, C.M., Building better mouse models of asthma. Curr Allergy Asthma Rep, 2007. 7(3): p. 231-6. 137. Van Hove, C.L., et al., Comparison of acute inflammatory and chronic structural asthma-like responses between C57BL/6 and BALB/c mice. Int Arch Allergy Immunol, 2009. 149(3): p. 195-207. 138. McMillan, S.J. and C.M. Lloyd, Prolonged allergen challenge in mice leads to persistent airway remodelling. Clin Exp Allergy, 2004. 34(3): p. 497-507. 139. Kumar, R.K., C. Herbert, and P.S. Foster, The "classical" ovalbumin challenge model of asthma in mice. Curr Drug Targets, 2008. 9(6): p. 485-94. 140. Arizmendi, N.G., et al., Mucosal exposure to cockroach extract induces allergic sensitization and allergic airway inflammation. Allergy Asthma Clin Immunol, 2011. 7: p. 22. 141. Cates, E.C., et al., Intranasal exposure of mice to house dust mite elicits allergic airway inflammation via a GM-CSF-mediated mechanism. J Immunol, 2004. 173(10): p. 6384-92. 142. Havaux, X., et al., A new mouse model of lung allergy induced by the spores of Alternaria alternata and Cladosporium herbarum molds. Clin Exp Immunol, 2005. 139(2): p. 179-88. 143. Debeuf, N., et al., Mouse Models of Asthma. Curr Protoc Mouse Biol, 2016. 6(2): p. 169-84. 144. Fahy, J.V., Eosinophilic and neutrophilic inflammation in asthma: insights from clinical studies. Proc Am Thorac Soc, 2009. 6(3): p. 256-9. 145. Cayrol, C. and J.P. Girard, IL-33: an alarmin cytokine with crucial roles in innate immunity, inflammation and allergy. Curr Opin Immunol, 2014. 31: p. 31-7. 146. Lambrecht, B.N. and H. Hammad, Biology of lung dendritic cells at the origin of asthma. Immunity, 2009. 31(3): p. 412-24. 147. Plantinga, M., et al., Conventional and monocyte-derived CD11b(+) dendritic cells initiate and maintain T helper 2 cell-mediated immunity to house dust mite allergen. Immunity, 2013. 38(2): p. 322-35. 148. Furuhashi, K., et al., Mouse lung CD103+ and CD11bhigh dendritic cells preferentially induce distinct CD4+ T-cell responses. Am J Respir Cell Mol Biol, 2012. 46(2): p. 165-72. 149. Nakano, H., et al., Pulmonary CD103(+) dendritic cells prime Th2 responses to inhaled allergens. Mucosal Immunol, 2012. 5(1): p. 53-65. 150. Colonna, M., G. Trinchieri, and Y.J. Liu, Plasmacytoid dendritic cells in immunity. Nat Immunol, 2004. 5(12): p. 1219-26.

Functional characterization of the new 8q21 Asthma risk locus | 283

151. Soler-Rodriguez, A.M., et al., Neutrophil activation by bacterial lipoprotein versus lipopolysaccharide: differential requirements for serum and CD14. J Immunol, 2000. 164(5): p. 2674-83. 152. Bahaie, N.S., et al., N-Glycans differentially regulate eosinophil and neutrophil recruitment during allergic airway inflammation. J Biol Chem, 2011. 286(44): p. 38231-41. 153. Fernandez-Caldas, E., Towards a more complete standardization of mite allergen extracts. Int Arch Allergy Immunol, 2013. 160(1): p. 1-3. 154. Abe, T. and A. Ishii, Comparison of Dermatophagoides pteronyssinus allergens from culture medium extract and whole body extract by using the same probe of pooled human serum. Allergy, 1987. 42(5): p. 352-8. 155. Casset, A., et al., Varying allergen composition and content affects the in vivo allergenic activity of commercial Dermatophagoides pteronyssinus extracts. Int Arch Allergy Immunol, 2012. 159(3): p. 253-62. 156. Thomas, W.R., Hierarchy and molecular properties of house dust mite allergens. Allergol Int, 2015. 64(4): p. 304-11. 157. Kauffman, H.F., et al., House dust mite major allergens Der p 1 and Der p 5 activate human airway-derived epithelial cells by protease-dependent and protease- independent mechanisms. Clin Mol Allergy, 2006. 4: p. 5. 158. Post, S., et al., The composition of house dust mite is critical for mucosal barrier dysfunction and allergic sensitisation. Thorax, 2012. 67(6): p. 488-95. 159. Parsons, S.J. and J.T. Parsons, Src family kinases, key regulators of signal transduction. Oncogene, 2004. 23(48): p. 7906-9. 160. Kurosaki, T., Regulation of B-cell signal transduction by adaptor proteins. Nat Rev Immunol, 2002. 2(5): p. 354-63. 161. Barua, D., W.S. Hlavacek, and T. Lipniacki, A computational model for early events in B cell antigen receptor signaling: analysis of the roles of Lyn and Fyn. J Immunol, 2012. 189(2): p. 646-58. 162. Kusel, M.M., et al., Early-life respiratory viral infections, atopic sensitization, and risk of subsequent development of persistent asthma. J Allergy Clin Immunol, 2007. 119(5): p. 1105-10. 163. James, K.M., et al., Risk of childhood asthma following infant bronchiolitis during the respiratory syncytial virus season. J Allergy Clin Immunol, 2013. 132(1): p. 227- 9. 164. Kwon, J.M., et al., Prevalence of respiratory viral infection in children hospitalized for acute lower respiratory tract diseases, and association of rhinovirus and influenza virus with asthma exacerbations. Korean J Pediatr, 2014. 57(1): p. 29-34. 165. Khetsuriani, N., et al., Prevalence of viral respiratory tract infections in children with asthma. J Allergy Clin Immunol, 2007. 119(2): p. 314-21. 166. You, D., et al., Building a better neonatal mouse model to understand infant respiratory syncytial virus disease. Respir Res, 2015. 16: p. 91. 167. Dyer, K.D., et al., The Pneumonia Virus of Mice (PVM) model of acute respiratory infection. Viruses, 2012. 4(12): p. 3494-510. 168. Kaiko, G.E., et al., Toll-like receptor 7 gene deficiency and early-life Pneumovirus infection interact to predispose toward the development of asthma-like pathology in mice. J Allergy Clin Immunol, 2013. 131(5): p. 1331-9 e10. 169. Lynch, J.P., et al., The plasmacytoid dendritic cell: at the cross-roads in asthma. Eur Respir J, 2014. 43(1): p. 264-75. 170. Lynch, J.P., et al., Aeroallergen-induced IL-33 predisposes to respiratory virus- induced asthma by dampening antiviral immunity. J Allergy Clin Immunol, 2016. 138(5): p. 1326-1337.

Functional characterization of the new 8q21 Asthma risk locus | 284

171. Aherne, W., et al., Pathological changes in virus infections of the lower respiratory tract in children. J Clin Pathol, 1970. 23(1): p. 7-18. 172. Davidson, S., et al., Plasmacytoid dendritic cells promote host defense against acute pneumovirus infection via the TLR7-MyD88-dependent signaling pathway. J Immunol, 2011. 186(10): p. 5938-48. 173. Durbin, R.K., S.V. Kotenko, and J.E. Durbin, Interferon induction and function at the mucosal surface. Immunol Rev, 2013. 255(1): p. 25-39. 174. Simpson, J., et al., The Absence of Interferon-beta Promotor Stimulator-1 (IPS-1) Predisposes to Bronchiolitis and Asthma-like Pathology in Response to Pneumoviral Infection in Mice. Sci Rep, 2017. 7(1): p. 2353. 175. Abboud, G., et al., Natural Killer Cells and Innate Interferon Gamma Participate in the Host Defense against Respiratory Vaccinia Virus Infection. J Virol, 2015. 90(1): p. 129-41. 176. Loh, J., et al., Natural killer cells utilize both perforin and gamma interferon to regulate murine cytomegalovirus infection in the spleen and liver. J Virol, 2005. 79(1): p. 661-7. 177. Bastiani, M. and R.G. Parton, Caveolae at a glance. J Cell Sci, 2010. 123(Pt 22): p. 3831-6. 178. Jiang, L.Q., et al., Csk-binding protein (Cbp) negatively regulates epidermal growth factor-induced cell transformation by controlling Src activation. Oncogene, 2006. 25(40): p. 5495-506. 179. Krasteva, G., et al., Caveolin-1 and -2 in airway epithelium: expression and in situ association as detected by FRET-CLSM. Respir Res, 2006. 7: p. 108. 180. Hackett, T.L., et al., Caveolin-1 controls airway epithelial barrier function. Implications for asthma. Am J Respir Cell Mol Biol, 2013. 49(4): p. 662-71. 181. Georas, S.N. and F. Rezaee, Epithelial barrier function: at the front line of asthma immunology and allergic airway inflammation. J Allergy Clin Immunol, 2014. 134(3): p. 509-20. 182. Mertens-Talcott, S.U., et al., The oncogenic microRNA-27a targets genes that regulate specificity protein transcription factors and the G2-M checkpoint in MDA- MB-231 breast cancer cells. Cancer Res, 2007. 67(22): p. 11001-11. 183. Safe, S., MicroRNA-Specificity Protein (Sp) Transcription Factor Interactions and Significance in Carcinogenesis. Curr Pharmacol Rep, 2015. 1(2): p. 73-78. 184. Tang, W., et al., miR-27a regulates endothelial differentiation of breast cancer stem like cells. Oncogene, 2014. 33(20): p. 2629-38. 185. Lai, Y., et al., The microRNA-27a: ZBTB10-specificity protein pathway is involved in follicle stimulating hormone-induced VEGF, Cox2 and survivin expression in ovarian epithelial cancer cells. Int J Oncol, 2013. 42(2): p. 776-84. 186. Jutooru, I., et al., Methyl 2-cyano-3,12-dioxooleana-1,9-dien-28-oate decreases specificity protein transcription factors and inhibits pancreatic tumor growth: role of microRNA-27a. Mol Pharmacol, 2010. 78(2): p. 226-36. 187. Li, X., et al., MicroRNA-27a Indirectly Regulates Estrogen Receptor {alpha} Expression and Hormone Responsiveness in MCF-7 Breast Cancer Cells. Endocrinology, 2010. 151(6): p. 2462-73. 188. Tanaka, T., M. Narazaki, and T. Kishimoto, IL-6 in inflammation, immunity, and disease. Cold Spring Harb Perspect Biol, 2014. 6(10): p. a016295. 189. Sun, Y., et al., Interaction of the Ski oncoprotein with Smad3 regulates TGF-beta signaling. Mol Cell, 1999. 4(4): p. 499-509. 190. Doyle, A.J., et al., Mutations in the TGF-beta repressor SKI cause Shprintzen- Goldberg syndrome with aortic aneurysm. Nat Genet, 2012. 44(11): p. 1249-54. 191. Arango Duque, G. and A. Descoteaux, Macrophage cytokines: involvement in immunity and infectious diseases. Front Immunol, 2014. 5: p. 491.

Functional characterization of the new 8q21 Asthma risk locus | 285

192. Dodge, I.L., et al., IL-6 production by pulmonary dendritic cells impedes Th1 immune responses. J Immunol, 2003. 170(9): p. 4457-64.

Functional characterization of the new 8q21 Asthma risk locus | 286

Appendix

Long-range modulation of PAG1 expression by 8q21 allergy risk variants

Cristina T Vicente,1* Stacey L Edwards,1* Kristine M Hillman,1 Susanne Kaufmann,1 Hayley Mitchell,1

Lisa Bain,1 Dylan M Glubb,1 Jason S Lee,1 Juliet D French,1* Manuel AR Ferreira1*

1QIMR Berghofer Medical Research Institute, Brisbane 4029, Australia

* Equal contributions

Corresponding authors:

Manuel AR Ferreira, PhD QIMR Berghofer Medical Research Institute Locked Bag 2000, Royal Brisbane Hospital, Herston QLD 4029 Australia Phone: +61 7 3845 3552 Fax: +61 7 3362 0101 Email: [email protected]

Juliet D French, PhD QIMR Berghofer Medical Research Institute Locked Bag 2000, Royal Brisbane Hospital, Herston QLD 4029 Australia Phone: +61 7 3845 3029 Fax: +61 7 3362 0105 Email: [email protected] ABSTRACT

The gene(s) whose expression is regulated by allergy risk variants is unknown for many loci identified through genome-wide association studies. Addressing this knowledge gap might point to new therapeutic targets for allergic disease. The aim of this study was to identify the target gene(s) and the functional variant(s) underlying the association between rs7009110 on chromosome 8q21 and allergies.

Eight genes are located within 1 Mb of rs7009110. Multivariate association analysis of publicly available exon expression levels from lymphoblastoid cell lines (LCLs) identified a significant association between rs7009110 and the expression of a single gene: PAG1 (P=0.0017), 732 kb away.

Analysis of histone modifications and DNase I hypersensitive sites in LCLs identified four putative regulatory elements (PREs) in the region. Chromosome conformation capture confirmed that two PREs interacted with the PAG1 promoter, one in allele-specific fashion. To determine if these PREs were functional, LCLs were transfected with PAG1 promoter-driven luciferase reporter constructs. PRE3 acted as a transcriptional enhancer for PAG1 exclusively when it carried the rs2370615:C allergy predisposing allele, a variant in complete linkage disequilibrium with rs7009110. As such, rs2370615, which overlaps RelA transcription factor (TF) binding in LCLs and was found to disrupt Foxo3a binding to PRE3, represents the putative functional variant in this locus. Our studies suggest that the risk-associated allele of rs2370615 predisposes to allergic disease by increasing PAG1 expression which may promote B-cell activation and have a pro-inflammatory effect. Inhibition of PAG1 expression or function may have therapeutic potential for allergic diseases. To date, genome-wide association studies (GWAS) have identified 41 genetic associations with allergic diseases (Table S1), including asthma (MIM 600807), hayfever or allergic rhinitis (MIM 607154) and atopic dermatitis or eczema (MIM 603165). The identification of allergy risk variants is expected to provide new insights into the molecular pathways involved in disease pathophysiology and, in this way, facilitate the development of new disease treatments. However, these expectations have been hard to meet and this represents a major bottleneck in the field. There are two main reasons for this. First, for many loci, there is no functional evidence linking the risk variant with changes in the expression or protein sequence of any nearby genes; the likely target gene(s) is therefore unknown. This is the case for 27 of the 41 allergy risk loci discovered to date; of note, for 10 of these 27 loci, experimental mouse models of allergic disease implicate nearby genes in disease pathophysiology. Second, often there is little or no information available to determine if and how disruption of the expression or sequence of the target gene impacts cellular function and, ultimately, disease pathophysiology. Addressing these knowledge gaps is critical to help translate GWAS findings into clinically useful information.

In a recent GWAS, we compared 6,685 individuals with both asthma and hayfever against

14,091 asthma- and hayfever-free controls and identified two new risk loci for allergic disease: 8q21 and 16p13 1. Variants in both were associated with the individual risks of asthma and hayfever, but the association was stronger with the combined asthma with hayfever phenotype. In the 16p13 locus, evidence from gene expression studies suggests that DEXI is the most likely target gene of this association 2; 3; functional studies of this gene are now warranted to understand how variation in its expression might affect disease risk. Currently, the gene whose expression is regulated by allergy risk variants in the 8q21 locus is unknown. Therefore, the aim of this study was to use both population genetics and functional approaches to identify the target gene(s) and likely causal variant(s) underlying the 8q21 association with allergy risk. The study procedures used were approved by the Human

Research Ethics Committee from the QIMR Berghofer Medical Research Institute.

Eight genes and one miRNA are located within 1 Mb of the sentinel single nucleotide polymorphism (SNP) rs7009110, which has a 36% risk allele frequency and was associated with a 1.14 per-allele odds of disease 1; the nearest gene being ZBTB10 (Figure 1A). rs7009110 was the variant with the strongest association with disease risk after imputation of unmeasured variants from the 1000

Genomes project 1. Forty-three gene expression quantitative trait loci (eQTL) have been reported in this

2 Mb region across six relevant tissue or cell types, but rs7009110 is not in linkage disequilibrium (LD) with any of these (Table S2). However, rs7009110 is in modest LD with two nearby risk variants associated with eczema (rs7000782, r2=0.51) 4 and rheumatoid arthritis (MIM 180300; rs998731, r2=0.20) 5; in Europeans, the risk alleles for these three variants (rs7009110:T, rs7000782:A and rs998731:T) are on the same haplotype. The associations at this locus with asthma, hayfever, eczema and rheumatoid arthritis suggest that the underlying disrupted molecular mechanism affects a component of the immune system that is shared between allergic and auto-immune diseases. The most plausible candidate target genes are TPD52 (MIM 604068), which is involved in B- cell maturation 6;

PAG1 (MIM 605767), involved in T- and B-cell activation 7-10; and ZBTB10, a putative repressor of the

Sp1 transcription factor (TF) 11, which regulates multiple immune-related genes 12; 13.

We first considered the possibility that the lack of a published association between rs7009110 and the expression of nearby genes could represent a false-negative finding, arising because: (i) a true association did not exceed the false discovery rate threshold in the original eQTL studies and so was not reported; and/or (ii) most published eQTL studies surveyed used expression microarrays, which have incomplete coverage of gene expression patterns. To test this possibility, we analysed RNA-seq data obtained by the Geuvadis Project 14 for lymphoblastoid cell line (LCL) samples of 373 individuals of European descent from the 1000 Genomes Project 15. LCLs are derived from peripheral blood B- cells and therefore represent a practical and effective in vitro model to study gene expression patterns relevant to immune-related conditions.

Genotype and RNA-seq data were downloaded from EBI ArrayExpress (accessions E-GEUV-1 and E-GEUV-2). As in Lappalainen 14, we selected the exon (not the gene) as the quantification unit and restricted the analysis to exons expressed in >90% of all the individuals. Read counts normalized by library depth and with technical variation removed by PEER normalization 14 were available for all exons of six of the eight genes located within a 1 Mb region around rs7009110. Read counts were quantile normalized and adjusted for ancestry informative covariates and genotype imputation status, as in Lappalainen 14. For each of these six genes, the association between rs7009110 allelic dosage and variation in the expression of all exons was tested simultaneously using a multivariate test of association that improves power over the alternative strategy of testing each exon individually 16. The remaining two genes, ZNF704 and STMN2 (MIM 600621), were only expressed in 56% and 12% of samples, respectively. Given their low relative abundance and because there was no association between dichotomised read counts (expressed vs not expressed) and rs7009110 genotype (not shown), both genes were excluded from further analysis. Likewise, the expression of miR5708 was not associated with rs7009110 (not shown) and so this miRNA was not considered further.

Multivariate association was tested between rs7009110 and the expression levels of five exons in each of HEY1 (MIM 602953), MRPS28 (MIM 611990) and FABP5 (MIM 605168); seven in

ZBTB10; nine in PAG1; and 10 in TPD52. In this analysis, rs7009110 was significantly associated with the expression of PAG1 (P=0.0017) but with no other gene (Table S3). The weights attributed to individual PAG1 exons in the multivariate analysis were consistent with an effect of rs7009110 on the expression of exons 1, 2 and 3; this observation was confirmed with individual univariate analyses of these three exons (Table S4). The rs7009110:T allele that increases the risk of allergic disease was associated with an increased expression of PAG1 (Figure S1), explaining 2.6% of the observed variation.

To help fine-map the eQTL results for PAG1, we extended the multivariate association analysis to an additional 167 common variants (MAF>=5%, call rate >95%, Hardy-Weinberg equilibrium test

P-value > 10-6) located in the 69.4 kb core region of association with allergic disease, delimited by the left-most (rs7822328, chr8:81246659) and right-most (rs4739746, chr8:81316034) variants in LD (r2≥0.6) with rs7009110. This is the region most likely to include the underlying causal variant(s); most variants were in LD with rs7009110 (70% with r2≥0.6). Multiple SNPs in high LD (r2≥0.8) with rs7009110 were associated with the expression of PAG1 (Figure 1B). In general, the strength of the association increased with the physical proximity to rs7009110, but did not exceed that seen with this

SNP. However, no SNPs in LD (r2≥0.6) with rs7009110 showed an association with the expression of the other five genes tested (P>0.05, Figure S2).

Although the association between 8q21 allergy risk variants and PAG1 expression in LCLs was relatively modest, it pointed to the possibility that a regulatory element in this region might control the expression of PAG1. Thus, we next queried genome-wide maps of epigenetic profiles to search for putative regulatory elements (PREs) in the 8q21 core region of association. Analysis of histone modifications associated with regulatory activity (e.g. H3K4me1/2) and DNase I hypersensitive sites assayed by the ENCODE Project 17 in an LCL (GM12878) identified four PREs in this region, named

PRE1 to PRE4 (Figure 1C). Consistent with these results, Seumois et al 18 detected a significant gain in H3K4me2, which marks both active and poised enhancers, in PRE1 and PRE2 when comparing primary human CD4+ memory T cells against naïve CD4+ T cells. Similarly, Hnisz et al 19 predicted

PRE3 to be an enhancer in multiple human immune cell types. Of the 118 SNPs that are in LD (r2≥0.6) with rs7009110, 35 overlap one of the four PREs identified (Table S5); rs7009110 overlaps PRE3. We therefore hypothesized that (i) one (or more) of these four PREs in this region regulates the expression of PAG1 and (ii) rs7009110 or a correlated variant disrupts the function of the PRE.

To test the first hypothesis, we used Chromosome Conformation Capture (3C) as described previously 20 to quantify the frequency of long-range chromatin interactions that take place between the core region of association and the promoter of PAG1, 710 kb apart. Briefly, 3C libraries were created by cross-linking the chromatin from LCLs of individuals homozygote for the rs7009110:T risk allele

(n=2) or the C protective allele (n=2); DNA was then digested with EcoRI which flanks 17 contiguous fragments that cover the core region of association, as well as the PAG1 promoter (Table S6); DNA was religated and decrosslinked; and quantitative PCR (qPCR) with primers for the bait (PAG1 promoter) and interactors (17 fragments) was then performed to detect the presence of ligation products, which represent gene loops. BAC clones covering the regions of interest were used to normalize for PCR efficiency.

The frequency of chromatin interactions between the PAG1 promoter and 15 of the 17 fragments that covered the core region of association was low and consistent with being due to chance

(Figure 2A). However, for two fragments – numbered F9 and F11 – the interaction frequency was significantly above the background level. The size of the interaction products amplified was confirmed by gel electrophoresis and their sequence verified by Sanger sequencing; similar results were also obtained in two additional independent replicates of each sample (Figure S3). Fragment F9 is 1.1 kb long and overlaps the centromeric end of PRE2, including the peak region of H3K4me2 gain observed by Seumois et al 18 in memory CD4+ T cells. Interactions with fragment F9 were observed in LCLs homozygote for the rs7009110:C protective allele but not (or less frequently) for the T predisposing allele. Consistent with this, allele-specific 3C from two LCLs heterozygous for rs11783496, a SNP near fragment F9 and in LD with rs7009110 (r2=0.75), indicate that the protective C-allele of rs11783496 is more strongly associated with looping of PRE2 to the PAG1 promoter (Figure 2B and Figure S4).

These results suggest that an allergy protective allele (rs11783496:C or another on the same haplotype) is associated with the establishment of long-range chromatin interactions between PRE2 and the PAG1 promoter. The interaction observed with fragment F11 – which is 10.6 kb long and overlaps both the telomeric end of PRE2 and the centromeric end of PRE3 – was detected in all four samples, irrespective of rs7009110 genotype. This suggests that allergy-associated variants do not influence the establishment of chromatin looping between the PAG1 promoter and PRE3.

Having established that two independent DNA fragments overlapping PRE2 and PRE3 physically interact with the promoter of PAG1, next we tested the hypothesis that the regulatory ability of these two PREs is modulated by allergy risk variants. To assess this, a PAG1 promoter-driven luciferase reporter construct was generated by inserting a 1,660 bp fragment containing the PAG1 promoter into pGL3-basic, as described previously 21. A fragment overlapping PRE2 (1,893 bp) or

PRE3 (1,560 bp) and containing the major (ie. allergy protective) allele for SNPs in LD (r2>0.6) with rs7009110 was inserted downstream of luciferase. The coordinates of these fragments were selected such that they coincided with the two interaction peaks observed in the 3C experiments (fragments F9 and F11) and, within these regions, with the highest density of histone modifications and DNase I hypersensitive sites from ENCODE (Figure 1D and Figure S5). To assess the effect of individual

SNPs on PRE activity, the minor (ie. allergy predisposing) allele for SNPs in LD with rs7009110 was individually incorporated into PRE2 (rs11783496, rs11786685, rs11786685 or rs13275449) and PRE3

(rs4739737, rs10957979 or rs2370615) (Figure 1D) via standard DNA cloning. All constructs were sequenced to confirm variant incorporation (AGRF, Australia). Primers used to generate all constructs are listed in Table S7. LCLs were electroporated with the reporter plasmids and luciferase activity was measured 24 hr post-transfection. To correct for any differences in transfection efficiency or cell lysate preparation, Firefly luciferase activity was normalized to Renilla luciferase. The activity of each test construct was calculated relative to PAG1 promoter construct.

Our 3C studies provided evidence for allele-specific chromatin looping between PRE2 and PAG1. Despite this, PRE2 did not enhance or silence the PAG1 promoter in reporter assays (Figure

3A) making it unclear of the functional consequence of allele-specific chromatin looping. For PRE3, which interacted with the PAG1 promoter irrespective of genotype, we found that the construct containing the major (allergy protective) allele at the three SNPs (rs4739737, rs10957979 and rs2370615) in LD with rs7009110 also did not increase PAG1 promoter activity (Figure 3B). A similar lack of regulatory activity was observed for PRE3 when the fragments cloned contained the minor

(allergy predisposing) allele for rs4739737 or rs10957979. However, the PRE3 construct containing the minor allele for rs2370615 increased PAG1 promoter activity by 1.8-fold when compared to the promoter-only construct (P=0.0051), and by 2.3-fold when compared to the construct with the promoter plus the enhancer containing the major alleles (P=0.0005). These results demonstrate that

PRE3 acts as a transcriptional enhancer in the presence of the rs2370615:C allergy predisposing allele.

Consistent with this effect, rs2370615:C was associated with increased expression of PAG1 in the eQTL analyses described above (P=0.0018); this variant is in complete LD (r2=1) with rs7009110.

Collectively, these results confirm that PAG1 is a target gene of 8q21 allergy risk variants and suggest that rs2370615 represents the underlying putative functional variant.

Based on ENCODE ChIP-seq data for LCLs 17, two transcription factors (TFs) bind to PRE3 and overlap the putative functional variant rs2370615: the RelA (or p65 [MIM 164014]) subunit of the nuclear factor kB (NF-kb) TF, which is critical for innate and adaptive immune responses 22, and the

POU domain class 2 transcription factor 2 (POU2F2 or Oct-2 [MIM 164176]), which regulates B-cell- specific genes 23. These TFs have been shown to co-occur in LCLs 24 and to synergistically regulate enhancer activity in B-cells 25. Furthermore, TNF-α (MIM 191160) -induced recruitment of RelA to enhancers involved in long-range looping interactions is associated with transcriptional induction of target genes 26. These observations suggest that binding of RelA and Oct-2 to PRE3 may be required for long-range activation of PAG1 transcription. However, despite RelA binding to PRE3 being highest precisely over rs2370615 (Figure S5), this T/C polymorphism is not predicted to disrupt the binding motifs reported for either RelA or Oct-224; 27. Based on these observations, it is possible that the rs2370615:C allele alters the binding affinity of another transcription factor, which is involved in RelA recruitment to PRE3 and so promotes long-range activation of PAG1 transcription.

Notably, the rs2370615:C allele is predicted to disrupt the binding motif (TTGTTTAC) for five

Forkhead box (Fox) TFs 28, namely Foxo3a (MIM 602681), an NF-kB antagonist that inhibits lymphocyte activation and proliferation 29. To test this prediction, we performed Chromatin

Immunoprecipitation (ChIP) using an anti-Foxo3a rabbit mAb (2497S, Cell Signaling Technology) to determine Foxo3a binding to the PRE3 region that overlaps rs2370615 (Table S8), in LCLs with the rs2370615:TT, CT or CC genotype. When compared to an IgG control antibody (SC-8334, Santa Cruz Biotechnology), we observed a significant enrichment in Foxo3a binding to PRE3 for the TT (3.4-fold,

P=0.001) and TC (1.3-fold, P=0.006) but not CC (0.94-fold, P=0.820) LCLs (Figure 4A). Differences in IgG-normalised Foxo3a binding to PRE3 were statistically significant when comparing TT to either the TC (P=0.0018) or CC (P=0.0003) LCLs. This pattern of Foxo3a binding was not observed in a control region 4kb upstream of PRE3 that does not contain the canonical Fox binding motif (Table S8).

Although there was a slight enrichment of Foxo3a binding across the three LCLs in this region (Figure

4B), suggesting that Foxo3a might bind weakly to it, there were no significant differences between genotypes after correction for multiple testing. These results confirm that Foxo3a binds with high affinity to the PRE3 region that overlaps rs2370615 and that the C allele of this polymorphism reduces

Foxo3a binding to PRE3.

Lastly, given that Foxo3a was confirmed to bind to PRE3, which regulates PAG1 expression, we hypothesized that FOXO3A expression in LCLs might be correlated with PAG1 expression. To test this hypothesis, we used gene-expression results for LCLs of 373 individuals obtained by the Geuvadis

Project 14, as described above, to measure the association between FOXO3A and PAG1 (exon 2) expression, stratified by rs2370615 genotype (Figure 5). There was a significant positive correlation between FOXO3A and PAG1 expression in TT (r=0.31, 95% CI 0.15 to 0.45, P=0.0003) and TC

(r=0.21, 95% CI 0.06 to 0.34, P=0.0055) but not CC (r=-0.12, 95% CI -0.37 to 0.15, P=0.3960) LCLs, consistent with the disruptive effect of the rs2370615:C allele on Foxo3a binding to PRE3. This interaction effect between FOXO3A levels and rs2370615 genotype on PAG1 expression was statistically significant (P=0.0157).

Collectively, our results are consistent with a model whereby: (1) PRE3 is a poised regulatory element for PAG1 in all individuals, irrespective of genotype, and promotes PAG1 transcription. (2)

PRE2 is a more dynamic regulatory element, with as yet unconfirmed induction/suppression capacity for PAG1. To become regulatory active, PRE2 might require for example (i) the presence of signals from PRE3 (eg. PRE2 might suppress PRE3); (ii) appropriate cellular activation (eg. allergen or viral stimulation); or (iii) transcription factors that have low/no expression in the LCL tested. (3) Individuals with the rs2370615:C allergy risk allele at PRE3 have overall higher PAG1 expression, perhaps because

TFs such as RelA are able to bind to PRE3 and drive high levels of PAG1 transcription (Figure S6A).

(4) Individuals with the rs2370615:T allergy protective allele at PRE3 have overall lower PAG1 expression, perhaps because (i) Foxo3a binds to PRE3 and drives lower levels of transcription (when compared to for example RelA); and/or (ii) PRE2, which physically interacts with the PAG1 promoter in these individuals, might suppress PRE3 activity (Figure S6B). Further studies that characterize in greater detail the molecular mechanisms underlying the regulation of PAG1 expression by PRE2 and

PRE3 are warranted.

Importantly, our results suggest that increased PAG1 transcription is associated with increased risk of allergic disease. To our knowledge, there are no published studies investigating the contribution of PAG1 to the pathophysiology of allergic diseases. PAG1 encodes the phosphoprotein associated with glycosphingolipid microdomains (or Csk-binding protein, Cbp), a transmembrane adaptor protein localized to lipid rafts that has highest expression in the immune system, notably in T- and B-cells 30-32.

One of the main cellular functions of PAG1 is the regulation of Csk activity 30, with direct effects on immunoreceptor signalling 33. The role of PAG1 in the development of immune responses has been studied extensively using in vitro and in vivo experimental systems, with conflicting results (reviewed in 33). Briefly, in vitro studies suggest that PAG1 overexpression inhibits T-cell, B-cell and mast cell activation 7-10; 34 and so, by extension, would be expected to have an anti-inflammatory effect. In contrast, in vivo studies have shown that T- and B-cell development and function are normal in PAG- deficient mice 35; 36.

Our finding that allergy predisposing alleles increase the transcription of PAG1 in human B-cell lines is not consistent with results from these experimental studies; instead, it suggests that PAG1 overexpression might promote B-cell activation and so have a pro-inflammatory effect. Interestingly, this possibility was proposed by Hrdinka 33 to explain the observation that PAG1 is phosphorylated upon antigen stimulation in B-cells but not T-cells. This notion is also supported by the recent observation that PAG1 inhibition in mast cells can suppress degranulation and the production of pro- inflammatory cytokines if cells are activated via an allergen-dependent mechanism 37. Given the potential role of the NF-kB pathway in PAG1 transcription, and the observation that this pathway is activated by allergen and/or pathogen engagement of Toll like receptors 38, characterization of the immune response to allergen or viral challenge in PAG1-deficient mice might help resolve the conflicting observations reported in experimental studies.

A limitation of our work is that we cannot exclude the possibility that other genes in the region are also target genes of the same variants. We focused our experimental work on PAG1 because this was the only gene for which we found a significant association between allergy risk SNPs and variation in gene expression levels. However, it is possible that despite no current available eQTL evidence, the expression of genes such as ZBTB10 or TPD52, which are expressed in relevant immune cells, might also be regulated by the PREs we identified, perhaps in different cell types.

In conclusion, we showed that an allergy risk allele on chromosome 8q21 increases PAG1 transcription by activating a poised enhancer located 732 kb away from the gene promoter. The significance of these results is two-fold. First, they highlight the fact that the target gene(s) underlying an observed genetic association can be a considerable distance away from the GWAS sentinel SNP.

Second, our results suggest that inhibition of PAG1 expression or function may have therapeutic potential to treat allergic diseases.

Figure 1. Allergy risk SNPs are associated with PAG1 expression and overlap putative regulatory elements (PREs). (A) Location of the core region of association (blue rectangle) and genes located within 1 Mb of the GWAS sentinel SNP, rs7009110. (B) Results from multivariate association analysis (N=373) between exon expression levels and variants in the core region of association. The color of each variant reflects linkage disequilibrium with rs7009110 (purple square). (C) Location of four PREs, based on histone marks and DNase I hypersensitive sites in an LCL analysed by the ENCODE project. (D) Location of the 3C interacting fragments; the two fragments cloned in the luciferase assays; and the SNPs that are in LD (r2>0.6) with rs7009110 and overlap the cloned fragments.

Figure 2. Chromatin interactions at the 8q21 risk region with PAG1. (A) 3C interaction profiles indicating frequent interactions between PRE2 and PRE3 and the PAG1 promoter region. 3C libraries were generated with EcoRI, with the anchor point set at the PAG1 promoter. A physical map of the region interrogated by 3C is shown below, with the blue bars representing the position of the PREs and the black lines representing the EcoR1 fragments interrogated. Triangles show position of variants in linkage disequilibrium (LD) with rs7009110, with color gradient representing the strength of association with PAG1 gene expression levels (cf. Figure 1B), from most associated in red to least associated in grey. Graph shows 3C profiles generated from four different LCLs assayed in duplicate. Error bars denote SD. (B) Allele-specific chromatin looping between PRE2 and the PAG1 promoter. 3C followed by direct sequencing of the PCR product was performed in two heterozygous LCLs using rs11783496 as the surrogate marker. The rs11783496:C and rs7009110:C alleles are in phase (r2 = 0.75). Chromatograms represent one of two independent 3C libraries generated and sequenced.

Figure 3. PRE3 containing the rs2370615 risk allele acts as an enhancer on the PAG1 promoter. PRE2 or PRE3 was cloned upstream of a PAG1 promoter-driven luciferase reporter with and without the candidate causal SNPs. 1 ug of each plasmid was electroporated into 1x106 cells using Amaxa NucleofectorII with the SF buffer and the program EH-100. Luciferase activity was measured after 24h using the Dual-Glo luciferase assay system on a Beckman-Coulter DTX-800 plate reader. Graphs represent three independent experiments assayed in duplicate. Error bars denote 95% CI and P-values were determined with a two-way ANOVA followed by Dunnett’s multiple comparisons test (* P<0.01, ** P<0.001).

Figure 4. Binding of the Foxo3a transcription factor to PRE3 is disrupted by the rs2370615:C allele. Chromatin Immunoprecipitation was performed as previously described 39. LCLs (1x107) with the indicated genotypes were crosslinked, sonicated and immunoprecipitated using either anti-Foxo3a antibody or control IgG antibody. Immunoprecipitates were washed and DNA was extracted after reverse-crosslinking. The enrichment of Foxo3a to PRE3 (A) and control region 4kb upstream of PRE3 (B) was examined by performing qPCR. The binding of Foxo3a is shown as fold enrichment over IgG. (horizontal line). Graphs represent three independent experiments assayed in triplicate. Error bars denote 95% CI and P-values were determined using unpaired t-test.

Figure 5. The association between FOXO3A and PAG1 expression in LCLs is modulated by rs2370615 genotype. Gene expression levels were normalized using a rank-based transformation, adjusted for potential confounders (including ancestry and 16 principal components that capture the effects of unobserved technical confounders on gene expression) and the association tested using linear regression. Horizontal solid lines show the mean expression of PAG1 for a given genotype class (cf. Figure S1).

SUPPLEMENTAL DATA DESCRIPTION

One file is included in the Supplementary Data section, with six figures and eight tables. SUPPLEMENTAL FIGURES

Figure S1. Effect of rs7009110 genotype on variation in PAG1 expression. The rs7009110:T allele increases the risk of allergic disease. Plotted are the read counts (N=373) for exon 2 of PAG1 (exon with largest weight on the multivariate analysis, cf. Table S3), previously1 normalized by library depth, with technical variation removed by PEER normalization and after quantile normalization. The association between rs7009110 (coded additively: 0, 1 or 2 copies of the T minor allele) and variation in gene expression levels was significant (P=0.0018, cf. Table S4).

Figure S2. Results from multivariate association analysis (N=373) between gene expression levels and variants in the core region of association. Color of each variant reflects linkage disequilibrium with rs7009110 (purple square). Results are shown for five of the eight genes located within 1 Mb of rs7009110; results for PAG1 are shown in Figure 1 of the main manuscript. The remaining two genes, ZNF704 and STMN2, were only expressed in 56% and 12% of samples, respectively, and so were not analysed.

Figure S3. Results from independent biological replicates for the chromatin interaction analysis between the 8q21 risk region and PAG1. For details, see legend to Figure 2A of the main manuscript.

rs11783496 rs11783496

Input

3C

C/T C/T C/T LCL5 LCL6 LCL6 Replicate 2 Replicate 1 Replicate 2

Figure S4. Independent 3C biological replicates showing allele-specific chromatin looping between PRE2 and the PAG1 promoter. 3C followed by direct sequencing of the PCR product was performed in two heterozygous LCLs using rs11783496 as the surrogate marker. Chromatograms represent one of two independent 3C libraries generated and sequenced.

Figure S5. Location of the PRE2 and PRE3 fragments used in the luciferase assay experiments. Also shown are the location of SNPs in linkage disequilibrium with rs7009110, peak 3C interaction fragments, histone modifications and DNase I hypersensitive sites from an ENCODE LCL, transcription Factor ChIP-seq from ENCODE and RelA binding based on ChIP-seq data in LCLs2. The latter track was downloaded from: ftp://ftp.ncbi.nlm.nih.gov/geo/samples/GSM1329nnn/GSM1329660/suppl/GSM1329660_RelA.bw

Figure S6. Illustration of a plausible model of PAG1 regulation by PRE2 and PRE3. (A) Individuals with the rs2370615:C allergy risk allele at PRE3 have overall higher PAG1 expression, perhaps because TFs such as RelA are able to bind to PRE3 and drive high levels of PAG1 transcription. (B) Individuals with the rs2370615:T allergy protective allele at PRE3 have overall lower PAG1 expression, perhaps because (i) Foxo3a binds to PRE3 and drives lower levels of transcription (compared to for example RelaA); and/or (ii) PRE2, which physically interacts with the PAG1 promoter in these individuals, might suppress PRE3 activity. SUPPLEMENTAL TABLES

Table S1. Published risk loci for allergic disease (asthma, eczema, allergies and hayfever)

Disease

Nearby Primary Locus gene publication(s)

Asthma Eczema

Allergies

Hayfever

Mouse model Mouse

Clinicaltrials eQTL or nsSNP eQTLor 17q12 ORMDL3 3 5q12 PDE4D 4 1q31 DENND1B 5 6p21 HLA 6 9p24 IL33 6 2q12 IL1RL1 6 15q22 SMAD3 6 22q12 IL2RB 6 8q24.11 SLC30A8 7 1q23 PYHIN1 8 5q22 TSLP 8; 9 4q31 GAB1 10 12q13 IKZF4 10 10p14 GATA3 10 11q13.5 LRRC32 11 1q21 IL6R 11 19q13 KLK3 12 15q22 RORA 13 4p14 TLR1 14 7q22 CDHR3 15 16p13 CLEC16A 16 8q21 ZBTB10 16 8q24 HAS2 17 1q21 FLG 9 5q31 IL13 18 11q13.1 OVOL1 18 19p13 ACTL9 18 3p22 CCR4 19 3q13 CD200 19 7p22 CARD11 19 10q21 EGR2 19 11p15 NLRP10 19 20q13 CYP24A1 19 5p13 PTGER4 14 2q33 PLCL1 14

14q21 FOXA1 14 20q13 NFATC2 14 3q28 BCL6 20 8q24.21 MYC 20 4q27 IL2 20 12q13 STAT6 20 P<5x10-8 P<0.05

Table S2: SNPs located within 1 Mb of rs7009110 and associated with the expression of nearby genes in published genome-wide association studies of gene expression. Index SNPs are in low linkage disequilibrium (r2<0.1) with each other; from each study queried, we selected the SNP correlated with the Index SNP (r2>0.1) that had the most significant association with gene expression.

r2 with eQTL Gene Index SNP Study Tissue Best SNP P-value rs7009110

1 FABP5 rs4739622 21 PBMCs rs4739622 2.14E-07 0 2 FAM164A rs2705496 22 Monocytes rs2705496 9.60E-04 0 3 FTHL11 rs3812459 22 Monocytes rs3812459 4.38E-06 0.01 4 rs1026830 22 Monocytes rs1026830 1.53E-05 0 5 HEY1 rs282850 23 Blood rs282850 5.81E-56 0.01 6 rs12544114 23 Blood rs12544114 7.27E-43 0 7 rs17534643 23 Blood rs17534643 4.76E-17 0 8 rs390691 23 Blood rs390691 1.38E-15 0 9 rs6996298 23 Blood rs6996298 2.00E-15 0.01 10 rs2956251 23 Blood rs2956251 8.00E-14 0.01 11 rs2029824 23 Blood rs2029824 5.44E-06 0.01 12 rs2920944 23 Blood rs2920944 5.75E-06 0 13 rs12543376 23 Blood rs12543376 6.89E-06 0 14 rs2467778 23 Blood rs2467778 1.59E-04 0.01 15 rs2278667 23 Blood rs2278667 4.51E-04 0 16 rs7845273 23 Blood rs7845273 7.10E-04 0 17 rs2979707 23 Blood rs2979707 1.09E-03 0 18 rs13281919 23 Blood rs13281919 1.19E-03 0.01 19 rs1157639 23 Blood rs1157639 1.22E-03 0 20 IL7 rs1863593 22 Monocytes rs1863593 4.87E-04 0 21 IMPA1 rs17582647 21 PBMCs rs17582647 7.38E-06 0.01 22 LOC646374 rs3812460 22 Monocytes rs3812460 7.97E-04 0 23 rs10110615 22 Monocytes rs10110615 2.63E-03 0 24 MRPS28 rs903583 22 Monocytes rs903583 1.53E-04 0.01 25 PAG1 rs13279056 24 LCLs rs4500045 5.20E-11 0 25 22 B cells rs12677218 3.37E-04 0 25 22 Monocytes rs16908663 2.85E-17 0.01 25 1 LCLs rs12677218 5.35E-08 0 25 25 Lung rs13266020 7.59E-08 0 25 23 Blood rs13279056 1.55E-97 0 25 21 PBMCs rs6473263 3.54E-38 0.01 26 rs10097731 24 LCLs rs10504730 5.00E-08 0 26 22 B cells rs10504730 2.54E-04 0 26 1 LCLs rs10504730 1.97E-08 0 26 23 Blood rs10097731 1.64E-88 0 27 rs4739791 22 Monocytes rs1896216 2.51E-03 0 27 23 Blood rs4739791 2.54E-31 0 27 21 PBMCs rs10957999 2.65E-17 0 28 rs4237093 23 Blood rs4237093 7.38E-25 0 29 rs920983 22 Monocytes rs1445558 5.20E-04 0 29 23 Blood rs920983 3.52E-11 0 30 rs11988289 23 Blood rs17494473 8.95E-04 0 30 21 PBMCs rs11988289 7.11E-07 0 31 rs9650270 22 Monocytes rs9650270 1.39E-06 0 31 23 Blood rs7009399 4.39E-06 0 32 rs7831986 21 PBMCs rs7831986 2.51E-06 0.01 33 rs2705499 23 Blood rs2705499 2.06E-05 0 34 rs7017984 23 Blood rs7017984 5.09E-04 0 35 TPD52 rs1863436 25 Lung rs10755965 9.02E-07 0.01 35 23 Blood rs1863436 6.31E-52 0 36 rs6473202 23 Blood rs6473202 3.94E-11 0.01 37 rs4440674 23 Blood rs4440674 8.81E-06 0.01 38 rs4739729 23 Blood rs4739729 1.58E-05 0.13 39 ZBTB10 rs9298338 25 Lung rs9298338 0.00E+00 0 39 1 LCLs rs201499880 1.90E-11 0.01 40 rs3863246 25 Lung rs3863246 9.76E-11 0 40 1 LCLs rs368280 5.39E-09 0.01 41 rs11781854 1 LCLs rs11781854 9.86E-09 0.02 42 ZFAND1 rs3812459 22 Monocytes rs3812459 1.71E-04 0.01 43 rs1026830 22 Monocytes rs1026830 4.19E-04 0 44 rs6473199 22 Monocytes rs6473199 9.32E-04 0.01

Table S3. Results from multivariate association analysis between rs7009110 and exon expression levels measured in LCLs from 373 individuals of European descent studied by the Geuvadis Project 1. For each gene, the association between rs7009110 (coded additively: 0, 1 and 2 copies of the T minor allele) and all exons of that gene were tested using the multivariate test described in Ferreira 26.

Weights from multivariate association analysis N P- Gene exons value Exon Exon Exon Exon Exon Exon Exon Exon Exon Exon 1 2 3 4 5 6 7 8 9 10 HEY1 5 0.7969 -0.18 -0.44 -0.51 -0.51 -0.1 - - - - - MRPS28 5 0.5247 0.29 -0.06 0.45 -0.02 0.68 - - - - - TPD52 10 0.8366 -0.25 -0.25 -0.49 -0.59 0.25 -0.15 -0.11 0.01 0.1 -0.01 ZBTB10 7 0.3411 -0.34 0.52 0.63 0.61 0.34 0.53 0.66 - - - PAG1 9 0.0017 0.51 0.61 0.46 0.33 0.14 0.24 0.01 0.25 0.16 - FABP5 5 0.3932 0.19 0.58 0.25 0.62 0.67 - - - - -

Table S4. Results from univariate association analysis between rs7009110 and the individual expression levels of exons 1, 2 and 3 from PAG1. For each exon, the association between rs7009110 (coded additively: 0, 1 and 2 copies of the T minor allele) and expression levels were tested using linear regression.

Exon Start End Beta SE P-value 3 81942231 81942324 0.165 0.07 0.0189 2 81982347 81982405 0.217 0.069 0.0018 1 82023826 82024303 0.189 0.072 0.0086

Table S5. List of variants in linkage disequilibrium (r2>0.6) with rs7009110 and located in one of the four putative regulatory elements (PREs) identified in the core region of association.

Position, r2 with Variant PRE bp rs7009110

rs76079837 81259105 0.842 1 rs10957975 81259446 0.826 1 rs13275219 81259826 0.786 1 rs4739735 81259877 0.842 1 rs11776367 81260051 0.837 1 rs3913968 81261038 0.847 1 rs3913969 81261064 0.847 1 rs62517191 81262527 0.847 1 rs62517192 81262544 0.783 1 rs62517193 81262839 0.808 1 rs7824993 81262896 0.721 1 rs5892724 81263715 0.847 1 rs7462675 81263962 0.797 1 indel:2D_8_81264051 81264051 0.765 1 rs201311214 81264052 0.759 1 rs11783496 81275652 0.751 2 rs11786685 81275835 0.753 2 rs11786704 81275860 0.856 2 rs13275449 81276113 0.792 2 rs4739736 81277369 0.784 2 rs12543811 81278885 0.792 2 rs13272328 81280496 0.781 2 rs13270496 81280666 0.868 2 rs7826521 81280778 0.858 2 rs6473225 81281007 0.868 2 rs952559 81288634 0.889 3 rs952558 81288734 0.956 3 rs952557 81288925 0.889 3 rs4739737 81289625 0.889 3 rs10957979 81289787 0.889 3 rs2370615 81290387 1 3 rs4739738 81291645 0.919 3 rs7009110 81291879 - 3 rs28656344 81304496 0.687 4 rs4739739 81304576 0.656 4

Table S6. Primers and coordinates of fragments tested in 3C experiments.

Primer 3C Interaction Fragment Fragment Primer Sequence (5' - 3') Coordinates Fragment Size (bp)

1 CTGTCAGCCTATTATTCTGCCCACCACG 81.248.205-81.248.232 81.243.654-81.248.331 4678

2 GATTGTCCTTTGCAATGGAACTGCCTAGC 81.252.072-81.252.100 81.248.338-81.252.262 3925

3 GCACACTCTGGCTTTCCTCCTAGTTAACTGG 81.254.901-81.254.931 81.252.269-81.255.043 2775

4 CCACGACTCTTCCCTCAAATGTAGATTTCTGTC 81.255.113-81.255.145 81.255.050-81.258.454 3405

5 CTCTTTCAAGATCTTGGTAAACATGATGAACATGG 81.263.132-81.263.166 81.258.553-81.263.168 4616

6 AAGAAAGCTTTTCAGGCCAGGCATGG 81.265.288-81.265.313 81.263.175-81.265.494 2320

7 GAGCTGACATTGCACCACTGCAGTCC 81.274.947-81.274.972 81.265.501-81.275.183 9683

8 CCTCTGTGGTAGGATGTCTGGTAATTATTTTGCTC 81.275.375-81.275.409 81.275.190-81.275.897 708

9 CCTCTTGTGAATTGCCTGCTCTACATGCTG 81.276.023-81.276.052 81.275.904-81.277.025 1122

10 GCATTCTCATGCTGTTGAGTGCCTATGG 81.277.126-81.277.153 81.277.032-81.279.569 2538

11 CCAGCCTGTTTCCTTTCCACTGATGC 81.279.715-81.279.740 81.279.576-81.290.220 10645

12 CCTTCTGATTTGGGCAGGACTTACTGACTCA 81.290.335- 81.290.365 81.290.227-81.299.133 8907

13 CCATTGGCTGACACCGTTGACCTCTT 81.299.290-81.299.315 81.299.140-81.304.960 5821

14 CCAGTGATGGTCTCTAGTCACCACTGTACGC 81.305.069-81.305.099 81.304.967-81.311.710 6744

15 GAGTTGCACTGAGGTCACCTCCAGTTGC 81.313.861-81.313.888 81.311.717-81.313.898 2182

16 GGATGCAAAGGGTCTCCTGATGATCATG 81.314.008-81.314.035 81.313.905-81.314.467 563

17 GCACCACCCAAGAATCAAGAGAGCCA 81.314.581-81.314.606 81.314.503-81.315.368 866 PAG1 promoter CCTGGAGCAGAGCTTCTGAAACAGTTGG 82.024.998-82.025.025 primer

Table S7. Primers used to generate luciferase assay constructs.

Primer Fragment Cloning Primer Primer Sequence (5'-3') Coordinates Size (bp) 82.025.770- PAG1promoter_fwd CCACCATTAGTCAACTCATGTCCAGG 82.025.795 1660 82.024.136- PAG1promoter_rev CAAGGGAATCACGGCTCAATTAGG 82.024.159 81.275.368- PRE2_fwd GAGGTACCACCGGTGCAAATTCCTCTGTGGTAGGATGTCTGG 81.275.395 1893 81.277.234- PRE2_rev GATCTAGACCTGCAGGGTAGGGTCCAGTGGGAGAGACACTTGC 81.277.260 81.289.408- PRE3_fwd GAGAAGCTTACCGGTCCTTTGATGGACATGATGTCACC 81.289.433 1560 81.290.942- PRE3_rev GATCTAGACCTGCAGGCACGTGGCTGCCTTCTTATGAAAAGC 81.290.967

Table S8. Primers used for ChIP experiments.

Primer Primer Sequence (5'-3') Region amplified

PRE3_test_region_fwd CTTCCCTTCTGATTTGGGCAGG 81,290,331-81,290,507 PRE3_test_region_rev TTGGCATCATCAAAAGCAACAGC control_region_fwd GAGGTCGGGAGTTTGAGAACAGC 81,283,900-81,284,074 control_region_rev GCAATCTTGGCTCAGTGCAACC

ACKNOWLEDGMENTS

We would like to thank Prof Peter Visscher and Prof Grant Montgomery for providing the lymphoblastoid cell lines used in this study. WEB RESOURCES

EMBL-EBI ArrayExpress – functional genomics data, https://www.ebi.ac.uk/arrayexpress/

Online Mendelian Inheritance in Man (OMIM), http://www.omim.org/ REFERENCES

1. Ferreira, M.A., Matheson, M.C., Tang, C.S., Granell, R., Ang, W., Hui, J., Kiefer, A.K., Duffy, D.L., Baltic, S., Danoy, P., et al. (2014). Genome-wide association analysis identifies 11 risk variants associated with the asthma with hay fever phenotype. The Journal of allergy and clinical immunology 133, 1564-1571. 2. Zeller, T., Wild, P., Szymczak, S., Rotival, M., Schillert, A., Castagne, R., Maouche, S., Germain, M., Lackner, K., Rossmann, H., et al. (2010). Genetics and beyond--the transcriptome of human monocytes and disease susceptibility. PLoS One 5, e10693. 3. Davison, L.J., Wallace, C., Cooper, J.D., Cope, N.F., Wilson, N.K., Smyth, D.J., Howson, J.M., Saleh, N., Al-Jeffery, A., Angus, K.L., et al. (2012). Long-range DNA looping and gene expression analyses identify DEXI as an autoimmune disease candidate gene. Human molecular genetics 21, 322-333. 4. Paternoster, L., Evans, D.M., Nohr, E.A., Holst, C., Gaborieau, V., Brennan, P., Gjesing, A.P., Grarup, N., Witte, D.R., Jorgensen, T., et al. (2011). Genome-wide population-based association study of extremely overweight young adults--the GOYA study. PLoS One 6, e24303. 5. Okada, Y., Wu, D., Trynka, G., Raj, T., Terao, C., Ikari, K., Kochi, Y., Ohmura, K., Suzuki, A., Yoshida, S., et al. (2014). Genetics of rheumatoid arthritis contributes to biology and drug discovery. Nature 506, 376-381. 6. Tiacci, E., Orvietani, P.L., Bigerna, B., Pucciarini, A., Corthals, G.L., Pettirossi, V., Martelli, M.P., Liso, A., Benedetti, R., Pacini, R., et al. (2005). Tumor protein D52 (TPD52): a novel B- cell/plasma-cell molecule with unique expression pattern and Ca(2+)-dependent association with annexin VI. Blood 105, 2812-2820. 7. Smida, M., Cammann, C., Gurbiel, S., Kerstin, N., Lingel, H., Lindquist, S., Simeoni, L., Brunner- Weinzierl, M.C., Suchanek, M., Schraven, B., et al. (2013). PAG/Cbp suppression reveals a contribution of CTLA-4 to setting the activation threshold in T cells. Cell communication and signaling : CCS 11, 28. 8. Itoh, K., Sakakibara, M., Yamasaki, S., Takeuchi, A., Arase, H., Miyazaki, M., Nakajima, N., Okada, M., and Saito, T. (2002). Cutting edge: negative regulation of immune synapse formation by anchoring lipid raft to cytoskeleton through Cbp-EBP50-ERM assembly. J Immunol 168, 541- 544. 9. Davidson, D., Bakinowski, M., Thomas, M.L., Horejsi, V., and Veillette, A. (2003). Phosphorylation- dependent regulation of T-cell activation by PAG/Cbp, a lipid raft-associated transmembrane adaptor. Molecular and cellular biology 23, 2017-2028. 10. Kalland, M.E., Solheim, S.A., Skanland, S.S., Tasken, K., and Berge, T. (2012). Modulation of proximal signaling in normal and transformed B cells by transmembrane adapter Cbp/PAG. Experimental cell research 318, 1611-1619. 11. Tillotson, L.G. (1999). RIN ZF, a novel zinc finger gene, encodes proteins that bind to the CACC element of the gastrin promoter. The Journal of biological chemistry 274, 8123-8128. 12. Tone, M., Powell, M.J., Tone, Y., Thompson, S.A., and Waldmann, H. (2000). IL-10 gene expression is controlled by the transcription factors Sp1 and Sp3. Journal of immunology 165, 286-291. 13. Takahashi, K., Hayashi, N., Shimokawa, T., Umehara, N., Kaminogawa, S., and Ra, C. (2008). Cooperative regulation of Fc receptor gamma-chain gene expression by multiple transcription factors, including Sp1, GABP, and Elf-1. The Journal of biological chemistry 283, 15134- 15141. 14. Lappalainen, T., Sammeth, M., Friedlander, M.R., t Hoen, P.A., Monlong, J., Rivas, M.A., Gonzalez-Porta, M., Kurbatova, N., Griebel, T., Ferreira, P.G., et al. (2013). Transcriptome and genome sequencing uncovers functional variation in humans. Nature 501, 506-511. 15. Genomes Project, C., Abecasis, G.R., Auton, A., Brooks, L.D., DePristo, M.A., Durbin, R.M., Handsaker, R.E., Kang, H.M., Marth, G.T., and McVean, G.A. (2012). An integrated map of genetic variation from 1,092 human genomes. Nature 491, 56-65. 16. Ferreira, M.A., and Purcell, S.M. (2009). A multivariate test of association. Bioinformatics 25, 132- 133. 17. Consortium, E.P. (2012). An integrated encyclopedia of DNA elements in the human genome. Nature 489, 57-74. 18. Seumois, G., Chavez, L., Gerasimova, A., Lienhard, M., Omran, N., Kalinke, L., Vedanayagam, M., Ganesan, A.P., Chawla, A., Djukanovic, R., et al. (2014). Epigenomic analysis of primary human T cells reveals enhancers associated with TH2 memory cell differentiation and asthma susceptibility. Nature immunology 15, 777-788. 19. Hnisz, D., Abraham, B.J., Lee, T.I., Lau, A., Saint-Andre, V., Sigova, A.A., Hoke, H.A., and Young, R.A. (2013). Super-enhancers in the control of cell identity and disease. Cell 155, 934-947. 20. Ghoussaini, M., Edwards, S.L., Michailidou, K., Nord, S., Cowper-Sal Lari, R., Desai, K., Kar, S., Hillman, K.M., Kaufmann, S., Glubb, D.M., et al. (2014). Evidence that breast cancer risk at the 2q35 locus is mediated through IGFBP5 regulation. Nature communications 4, 4999. 21. French, J.D., Ghoussaini, M., Edwards, S.L., Meyer, K.B., Michailidou, K., Ahmed, S., Khan, S., Maranian, M.J., O'Reilly, M., Hillman, K.M., et al. (2013). Functional variants at the 11q13 risk locus for breast cancer regulate cyclin D1 expression through long-range enhancers. American journal of human genetics 92, 489-503. 22. Bonizzi, G., and Karin, M. (2004). The two NF-kappaB activation pathways and their role in innate and adaptive immunity. Trends in immunology 25, 280-288. 23. Matthias, P. (1998). Lymphoid-specific transcription mediated by the conserved octamer site: who is doing what? Seminars in immunology 10, 155-163. 24. Zhao, B., Barrera, L.A., Ersing, I., Willox, B., Schmidt, S.C., Greenfeld, H., Zhou, H., Mollo, S.B., Shi, T.T., Takasaki, K., et al. (2014). The NF-kappaB genomic landscape in lymphoblastoid B cells. Cell reports 8, 1595-1606. 25. Sepulveda, M.A., Emelyanov, A.V., and Birshtein, B.K. (2004). NF-kappa B and Oct-2 synergize to activate the human 3' Igh hs4 enhancer in B cells. J Immunol 172, 1054-1064. 26. Jin, F., Li, Y., Dixon, J.R., Selvaraj, S., Ye, Z., Lee, A.Y., Yen, C.A., Schmitt, A.D., Espinoza, C.A., and Ren, B. (2013). A high-resolution map of the three-dimensional chromatin interactome in human cells. Nature 503, 290-294. 27. Staudt, L.M., Clerc, R.G., Singh, H., LeBowitz, J.H., Sharp, P.A., and Baltimore, D. (1988). Cloning of a lymphoid-specific cDNA encoding a protein binding the regulatory octamer DNA motif. Science 241, 577-580. 28. Ward, L.D., and Kellis, M. (2012). HaploReg: a resource for exploring chromatin states, conservation, and regulatory motif alterations within sets of genetically linked variants. Nucleic acids research 40, D930-934. 29. Lin, L., Hron, J.D., and Peng, S.L. (2004). Regulation of NF-kappaB, Th activation, and autoinflammation by the forkhead transcription factor Foxo3a. Immunity 21, 203-213. 30. Brdicka, T., Pavlistova, D., Leo, A., Bruyns, E., Korinek, V., Angelisova, P., Scherer, J., Shevchenko, A., Hilgert, I., Cerny, J., et al. (2000). Phosphoprotein associated with glycosphingolipid-enriched microdomains (PAG), a novel ubiquitously expressed transmembrane adaptor protein, binds the protein tyrosine kinase csk and is involved in regulation of T cell activation. The Journal of experimental medicine 191, 1591-1604. 31. Inomata, M., Shimada, Y., Hayashi, M., Shimizu, J., and Ohno-Iwashita, Y. (2007). Impairment in a negative regulatory system for TCR signaling in CD4+ T cells from old mice. FEBS Lett 581, 3039-3043. 32. Tedoldi, S., Paterson, J.C., Hansmann, M.L., Natkunam, Y., Rudiger, T., Angelisova, P., Du, M.Q., Roberton, H., Roncador, G., Sanchez, L., et al. (2006). Transmembrane adaptor molecules: a new category of lymphoid-cell markers. Blood 107, 213-221. 33. Hrdinka, M., and Horejsi, V. (2014). PAG--a multipurpose transmembrane adaptor protein. Oncogene 33, 4881-4892. 34. Ohtake, H., Ichikawa, N., Okada, M., and Yamashita, T. (2002). Cutting Edge: Transmembrane phosphoprotein Csk-binding protein/phosphoprotein associated with glycosphingolipid- enriched microdomains as a negative feedback regulator of mast cell signaling through the FcepsilonRI. Journal of immunology 168, 2087-2090. 35. Xu, S., Huo, J., Tan, J.E., and Lam, K.P. (2005). Cbp deficiency alters Csk localization in lipid rafts but does not affect T-cell development. Mol Cell Biol 25, 8486-8495. 36. Dobenecker, M.W., Schmedt, C., Okada, M., and Tarakhovsky, A. (2005). The ubiquitously expressed Csk adaptor protein Cbp is dispensable for embryogenesis and T-cell development and function. Mol Cell Biol 25, 10533-10542. 37. Draberova, L., Bugajev, V., Potuckova, L., Halova, I., Bambouskova, M., Polakovicova, I., Xavier, R.J., Seed, B., and Draber, P. (2014). Transmembrane adaptor protein PAG/CBP is involved in both positive and negative regulation of mast cell signaling. Molecular and cellular biology 34, 4285-4300. 38. Medzhitov, R. (2001). Toll-like receptors and innate immunity. Nature reviews Immunology 1, 135- 145. 39. Lee, J.S., Kim, Y., Kim, I.S., Kim, B., Choi, H.J., Lee, J.M., Shin, H.J., Kim, J.H., Kim, J.Y., Seo, S.B., et al. (2010). Negative regulation of hypoxic responses via induced Reptin methylation. Molecular cell 39, 71-85.

SUPPLEMENTAL REFERENCES

1. Lappalainen, T., Sammeth, M., Friedlander, M.R., t Hoen, P.A., Monlong, J., Rivas, M.A., Gonzalez- Porta, M., Kurbatova, N., Griebel, T., Ferreira, P.G., et al. (2013). Transcriptome and genome sequencing uncovers functional variation in humans. Nature 501, 506-511. 2. Zhao, B., Barrera, L.A., Ersing, I., Willox, B., Schmidt, S.C., Greenfeld, H., Zhou, H., Mollo, S.B., Shi, T.T., Takasaki, K., et al. (2014). The NF-kappaB genomic landscape in lymphoblastoid B cells. Cell reports 8, 1595-1606. 3. Moffatt, M.F., Kabesch, M., Liang, L., Dixon, A.L., Strachan, D., Heath, S., Depner, M., von Berg, A., Bufe, A., Rietschel, E., et al. (2007). Genetic variants regulating ORMDL3 expression contribute to the risk of childhood asthma. Nature 448, 470-473. 4. Himes, B.E., Hunninghake, G.M., Baurley, J.W., Rafaels, N.M., Sleiman, P., Strachan, D.P., Wilk, J.B., Willis-Owen, S.A., Klanderman, B., Lasky-Su, J., et al. (2009). Genome-wide association analysis identifies PDE4D as an asthma-susceptibility gene. American journal of human genetics 84, 581-593. 5. Sleiman, P.M., Flory, J., Imielinski, M., Bradfield, J.P., Annaiah, K., Willis-Owen, S.A., Wang, K., Rafaels, N.M., Michel, S., Bonnelykke, K., et al. (2010). Variants of DENND1B associated with asthma in children. The New England journal of medicine 362, 36-44. 6. Moffatt, M.F., Gut, I.G., Demenais, F., Strachan, D.P., Bouzigon, E., Heath, S., von Mutius, E., Farrall, M., Lathrop, M., Cookson, W.O., et al. (2010). A large-scale, consortium-based genomewide association study of asthma. The New England journal of medicine 363, 1211- 1221. 7. Noguchi, E., Sakamoto, H., Hirota, T., Ochiai, K., Imoto, Y., Sakashita, M., Kurosaka, F., Akasawa, A., Yoshihara, S., Kanno, N., et al. (2011). Genome-wide association study identifies HLA-DP as a susceptibility gene for pediatric asthma in Asian populations. PLoS genetics 7, e1002170. 8. Torgerson, D.G., Ampleford, E.J., Chiu, G.Y., Gauderman, W.J., Gignoux, C.R., Graves, P.E., Himes, B.E., Levin, A.M., Mathias, R.A., Hancock, D.B., et al. (2011). Meta-analysis of genome-wide association studies of asthma in ethnically diverse North American populations. Nature genetics 43, 887-892. 9. Sun, L.D., Xiao, F.L., Li, Y., Zhou, W.M., Tang, H.Y., Tang, X.F., Zhang, H., Schaarschmidt, H., Zuo, X.B., Foelster-Holst, R., et al. (2011). Genome-wide association study identifies two new susceptibility loci for atopic dermatitis in the Chinese Han population. Nature genetics 43, 690- 694. 10. Hirota, T., Takahashi, A., Kubo, M., Tsunoda, T., Tomita, K., Doi, S., Fujita, K., Miyatake, A., Enomoto, T., Miyagawa, T., et al. (2011). Genome-wide association study identifies three new susceptibility loci for adult asthma in the Japanese population. Nature genetics 43, 893-896. 11. Ferreira, M.A., Matheson, M.C., Duffy, D.L., Marks, G.B., Hui, J., Le Souef, P., Danoy, P., Baltic, S., Nyholt, D.R., Jenkins, M., et al. (2011). Identification of IL6R and chromosome 11q13.5 as risk loci for asthma. Lancet 378, 1006-1014. 12. Myers, R.A., Himes, B.E., Gignoux, C.R., Yang, J.J., Gauderman, W.J., Rebordosa, C., Xie, J., Torgerson, D.G., Levin, A.M., Baurley, J., et al. (2012). Further replication studies of the EVE Consortium meta-analysis identifies 2 asthma risk loci in European Americans. The Journal of allergy and clinical immunology 130, 1294-1301. 13. Ramasamy, A., Kuokkanen, M., Vedantam, S., Gajdos, Z.K., Couto Alves, A., Lyon, H.N., Ferreira, M.A., Strachan, D.P., Zhao, J.H., Abramson, M.J., et al. (2012). Genome-wide association studies of asthma in population-based cohorts confirm known and suggested loci and identify an additional association near HLA. PloS one 7, e44008. 14. Hinds, D.A., McMahon, G., Kiefer, A.K., Do, C.B., Eriksson, N., Evans, D.M., St Pourcain, B., Ring, S.M., Mountain, J.L., Francke, U., et al. (2013). A genome-wide association meta- analysis of self-reported allergy identifies shared and allergy-specific susceptibility loci. Nature genetics 45, 907-911. 15. Bonnelykke, K., Sleiman, P., Nielsen, K., Kreiner-Moller, E., Mercader, J.M., Belgrave, D., den Dekker, H.T., Husby, A., Sevelsted, A., Faura-Tellez, G., et al. (2014). A genome-wide association study identifies CDHR3 as a susceptibility locus for early childhood asthma with severe exacerbations. Nature genetics 46, 51-55. 16. Ferreira, M.A., Matheson, M.C., Tang, C.S., Granell, R., Ang, W., Hui, J., Kiefer, A.K., Duffy, D.L., Baltic, S., Danoy, P., et al. (2014). Genome-wide association analysis identifies 11 risk variants associated with the asthma with hay fever phenotype. The Journal of allergy and clinical immunology 133, 1564-1571. 17. Yatagai, Y., Sakamoto, T., Yamada, H., Masuko, H., Kaneko, Y., Iijima, H., Naito, T., Noguchi, E., Hirota, T., Tamari, M., et al. (2014). Genomewide association study identifies HAS2 as a novel susceptibility gene for adult asthma in a Japanese population. Clinical and experimental allergy : journal of the British Society for Allergy and Clinical Immunology 44, 1327-1334. 18. Paternoster, L., Evans, D.M., Nohr, E.A., Holst, C., Gaborieau, V., Brennan, P., Gjesing, A.P., Grarup, N., Witte, D.R., Jorgensen, T., et al. (2011). Genome-wide population-based association study of extremely overweight young adults--the GOYA study. PloS one 6, e24303. 19. Hirota, T., Takahashi, A., Kubo, M., Tsunoda, T., Tomita, K., Sakashita, M., Yamada, T., Fujieda, S., Tanaka, S., Doi, S., et al. (2012). Genome-wide association study identifies eight new susceptibility loci for atopic dermatitis in the Japanese population. Nature genetics 44, 1222- 1226. 20. Bonnelykke, K., Matheson, M.C., Pers, T.H., Granell, R., Strachan, D.P., Alves, A.C., Linneberg, A., Curtin, J.A., Warrington, N.M., Standl, M., et al. (2013). Meta-analysis of genome-wide association studies identifies ten loci influencing allergic sensitization. Nature genetics 45, 902- 906. 21. Zeller, T., Wild, P., Szymczak, S., Rotival, M., Schillert, A., Castagne, R., Maouche, S., Germain, M., Lackner, K., Rossmann, H., et al. (2010). Genetics and beyond--the transcriptome of human monocytes and disease susceptibility. PloS one 5, e10693. 22. Fairfax, B.P., Makino, S., Radhakrishnan, J., Plant, K., Leslie, S., Dilthey, A., Ellis, P., Langford, C., Vannberg, F.O., and Knight, J.C. (2012). Genetics of gene expression in primary immune cells identifies cell type-specific master regulators and roles of HLA alleles. Nature genetics 44, 502-510. 23. Westra, H.J., Peters, M.J., Esko, T., Yaghootkar, H., Schurmann, C., Kettunen, J., Christiansen, M.W., Fairfax, B.P., Schramm, K., Powell, J.E., et al. (2013). Systematic identification of trans eQTLs as putative drivers of known disease associations. Nature genetics 45, 1238-1243. 24. Dixon, A.L., Liang, L., Moffatt, M.F., Chen, W., Heath, S., Wong, K.C., Taylor, J., Burnett, E., Gut, I., Farrall, M., et al. (2007). A genome-wide association study of global gene expression. Nature genetics 39, 1202-1207. 25. Hao, K., Bosse, Y., Nickle, D.C., Pare, P.D., Postma, D.S., Laviolette, M., Sandford, A., Hackett, T.L., Daley, D., Hogg, J.C., et al. (2012). Lung eQTLs to help reveal the molecular underpinnings of asthma. PLoS genetics 8, e1003029. 26. Ferreira, M.A., and Purcell, S.M. (2009). A multivariate test of association. Bioinformatics 25, 132- 133.

Solutions and reagents used in Chapter 3

Western Blot gels

Loading gel – 1.0mm at 8% • 3.5mL H2O • 2mL of 30% acrylamide (Bio-Rad) • 2mL Buffer A (0.1% SDS, 1.5M Tris-HCl pH8.8) • 75μL of 10% SDS (Bio-Rad) • 75μL Ammonium Persulfate (APS, Sigma-Aldrich) • 4.5μL Tetramethylethylenediamine (TEMED, Sigma-Aldrich)

Stacking gel • 1.4mL H2O • 330μL of 30% acrylamide • 250μL Buffer B (0.1% SDS, 0.5M Tris-HCl pH6.8) • 20μL of 10% SDS • 20μL APS • 2.5μL TEMED

Running, transfer and blocking buffers

Laemmli 2X buffer/loading buffer • 4% SDS (Bio-Rad) • 10% 2-mercaptoethanol (Bio-Rad) • 20% glycerol (Chem-Supply) • 0.004% bromophenol blue (Sigma-Aldrich) • 0.125 M Tris-HCl (Sigma-Aldrich) Check the pH and adjust to 6.8

Running buffer (Tris-Glycine/SDS) • 25mM Tris base (Sigma-Aldrich) • 200mM glycine (Sigma-Aldrich) • 0.1% SDS Check the pH and adjust to 8.3

Transfer buffer (wet) • 300mM Tris base • 300mM glycine • 200mL methanol (Chem-Supply) • 300μL of 20% SDS MilliQ H20 to 1L

10x Tris-buffered saline (TBS) • 24g Tris base • 88g NaCl (Chem-Supply) • 900mL ddH20 Check the pH and adjust to 7.6. ddH20 to 1L

1x Tris-buffered saline with Tween 20 (TBST) • 100mL 10x TBS • 1mL Polysorbate 20 (Tween 20, Sigma-Aldrich) ddH20 to 1L

Blocking buffer 3–5% milk or BSA (bovine serum albumin) in TBST.

Mini Trans-Blot assembly for wet system protein transfer (sourced from Bio-Rad Mini Trans-Blot® Electrophoretic Transfer Cell Instruction Manual, url: http://www.bio- rad.com/webroot/web/pdf/lsr/literature/M1703930.pdf)

List of genes included in the gene expression arrays used in Chapter 3

Gene I.D. Gene description Assay I.D. 18S 18S ribosomal RNA Hs99999901_s1 ACE Angiotensin I converting enzyme (peptidyl-dipeptidase A) 1 Hs00174179_m1 ACTB Actin, beta Hs99999903_m1 AGTR1 Angiotensin II receptor, type 1 Hs00241341_m1 AGTR2 Angiotensin II receptor, type 2 Hs00169126_m1 BAX BCL2-associated X protein Hs00180269_m1 BCL2 B-cell CLL/lymphoma 2 Hs00153350_m1 BCL2L1 BCL2-like 1 Hs00169141_m1 C3 Complement component 3 Hs00163811_m1 CCL19 Chemokine (C-C motif) ligand 19 Hs00171149_m1 CCL2 Chemokine (C-C motif) ligand 2 Hs00234140_m1 CCL3 Chemokine (C-C motif) ligand 3 Hs00234142_m1 CCL5 Chemokine (C-C motif) ligand 5 Hs00174575_m1 CCR2 Chemokine (C-C motif) receptor 2 Hs00174150_m1 CCR4 Chemokine (C-C motif) receptor 4 Hs99999919_m1 CCR5 Chemokine (C-C motif) receptor 5 Hs00152917_m1 CCR7 Chemokine (C-C motif) receptor 7 Hs00171054_m1 CD19 CD19 antigen Hs00174333_m1 CD28 CD28 antigen (Tp44) Hs00174796_m1 CD34 CD34 antigen Hs00156373_m1 CD38 CD38 antigen (p45) Hs00233552_m1 CD3E CD3E antigen, epsilon polypeptide (TiT3 complex) Hs00167894_m1 CD4 CD4 antigen (p55) Hs00181217_m1 CD40 CD40 antigen (TNF receptor superfamily member 5) Hs00374176_m1 CD40LG CD40 ligand (TNF superfamily, member 5, hyper-IgM syndrome) Hs00163934_m1 CD68 CD68 antigen Hs00154355_m1 CD80 CD80 antigen (CD28 antigen ligand 1, B7-1 antigen) Hs00175478_m1 CD86 CD86 antigen (CD28 antigen ligand 2, B7-2 antigen) Hs00199349_m1 CD8A CD8 antigen, alpha polypeptide (p32) Hs00233520_m1 COL4A5 Collagen, type IV, alpha 5 (Alport syndrome) Hs00166712_m1 CSF1 Colony stimulating factor 1 (macrophage) Hs00174164_m1 CSF2 Colony stimulating factor 2 (granulocyte-macrophage) Hs00171266_m1 CSF3 Colony stimulating factor 3 (granulocyte) Hs00357085_g1 CTLA4 Cytotoxic T-lymphocyte-associated protein 4 Hs00175480_m1 CXCL10 Chemokine (C-X-C motif) ligand 10 Hs00171042_m1 CXCL11 Chemokine (C-X-C motif) ligand 11 Hs00171138_m1 CXCR3 Chemokine (C-X-C motif) receptor 3 Hs00171041_m1 CYP1A2 Cytochrome P450, family 1, subfamily A, polypeptide 2 Hs00167927_m1 CYP7A1 Cytochrome P450, family 7, subfamily A, polypeptide 1 Hs00167982_m1 ECE1 Endothelin converting enzyme 1 Hs00154837_m1 EDN1 Endothelin 1 Hs00174961_m1 FAS Fas (TNF receptor superfamily, member 6) Hs00163653_m1 FASLG Fas ligand (TNF superfamily, member 6) Hs00181225_m1 FN1 Fibronectin 1 Hs00365052_m1 GAPDH Glyceraldehyde-3-phosphate dehydrogenase Hs99999905_m1 GNLY Granulysin Hs00246266_m1 GUSB Glucuronidase, beta Hs99999908_m1 GZMB Granzyme B (granzyme 2, cytotoxic T-lymphocyte-associated serine esterase 1) Hs00188051_m1 HLA-DRA Major histocompatibility complex, class II, DR alpha Hs00219575_m1 HLA-DRB1 Major histocompatibility complex, class II, DR beta 1 Hs99999917_m1 HMOX1 Heme oxygenase (decycling) 1 Hs00157965_m1 ICAM1 Intercellular adhesion molecule 1 (CD54), human rhinovirus receptor Hs00164932_m1 ICOS Inducible T-cell co-stimulator Hs00359999_m1 IFNG Interferon, gamma Hs00174143_m1 IKBKB Inhibitor of kappa light polypeptide gene enhancer in B-cells, kinase beta Hs00395088_m1 IL10 Interleukin 10 Hs00174086_m1 IL12A Interleukin 12A (natural killer cell stimulatory factor 1, cytotoxic lymphocyte maturation factor 1, p35) Hs00168405_m1 IL12B Interleukin 12B (natural killer cell stimulatory factor 2, cytotoxic lymphocyte maturation factor 2, p40) Hs00233688_m1 IL13 Interleukin 13 Hs00174379_m1 IL15 Interleukin 15 Hs00174106_m1 IL17 Interleukin 17 (cytotoxic T-lymphocyte-associated serine esterase 8) Hs00174383_m1 IL18 Interleukin 18 (interferon-gamma-inducing factor) Hs00155517_m1 IL1A Interleukin 1, alpha Hs00174092_m1 IL1B Interleukin 1, beta Hs00174097_m1 IL2 Interleukin 2 Hs00174114_m1 IL2RA Interleukin 2 receptor, alpha Hs00166229_m1 IL3 Interleukin 3 (colony-stimulating factor, multiple) Hs00174117_m1 IL4 Interleukin 4 Hs00174122_m1 IL5 Interleukin 5 (colony-stimulating factor, eosinophil) Hs00174200_m1 IL6 Interleukin 6 (interferon, beta 2) Hs00174131_m1 IL7 Interleukin 7 Hs00174202_m1 IL8 Interleukin 8 Hs00174103_m1 IL9 Interleukin 9 Hs00174125_m1 LRP2 Low density lipoprotein-related protein 2 Hs00189742_m1 LTA Lymphotoxin alpha (TNF superfamily, member 1) Hs00236874_m1 MYH6 Myosin, heavy polypeptide 6, cardiac muscle, alpha (cardiomyopathy, hypertrophic 1) Hs00411908_m1 NFKB2 Nuclear factor of kappa light polypeptide gene enhancer in B-cells 2 (p49/p100) Hs00174517_m1 NOS2A Nitric oxide synthase 2A (inducible, hepatocytes) Hs00167248_m1 PGK1 Phosphoglycerate kinase 1 Hs99999906_m1 PRF1 Perforin 1 (pore forming protein) Hs00169473_m1 PTGS2 Prostaglandin-endoperoxide synthase 2 (prostaglandin G/H synthase and cyclooxygenase) Hs00153133_m1 PTPRC Protein tyrosine phosphatase, receptor type, C Hs00365634_g1 REN Renin Hs00166915_m1 RPL3L Ribosomal protein L3-like Hs00192564_m1 SELE Selectin E (endothelial adhesion molecule 1) Hs00174057_m1 SELP Selectin P (granule membrane protein 140kDa, antigen CD62) Hs00174583_m1 SKI v-ski sarcoma viral oncogene homolog (avian) Hs00161707_m1 SMAD3 SMAD, mothers against DPP homolog 3 (Drosophila) Hs00232219_m1 SMAD7 SMAD, mothers against DPP homolog 7 (Drosophila) Hs00178696_m1 STAT3 Signal transducer and activator of transcription 3 (acute-phase response factor) Hs00234174_m1 TBX21 T-box 21 Hs00203436_m1 TFRC Transferrin receptor (p90, CD71) Hs99999911_m1 TGFB1 Transforming growth factor, beta 1 (Camurati-Engelmann disease) Hs00171257_m1 TNF Tumor necrosis factor (TNF superfamily, member 2) Hs00174128_m1 TNFRSF18 Tumor necrosis factor receptor superfamily, member 18 Hs00188346_m1 VEGF Vascular endothelial growth factor Hs00173626_m1

tm1a: The targeted allele in EUCOMM ES cells is “tm1a”. ES cells are male and heterozygous for the targeted mutation. “tm1a” is a so called targeted trap allele: it has been generated by targeting, yet it functions as a gene-trap knockout. tm1c: The conditional allele is called “tm1c”. It is generated by crossing “tm1a” -mice to mice, which are ubiquitously or germline expressing Flp recombinase. “tm1c” is a functional wild -type allele. However, the so called critical exon (= targeted region) is flanked by loxP sites, which allows for excision of the critical exon with Cre recombinase. tm1d: The organ or cell type specific null allele is called “tm1d”. This null a llele is generated by crossing “tm1c” -mice to a mouse that is expressing Cre recombinase in the desired organ or cell type specific manner. Cre recombinase will excise the critical exon, which for most EUCOMM alleles will cause a frameshift and loss of downstream protein sequence due to nonsense mutations. tm1b: This allele is a lacZ tagged null allele . The critical exon has been deleted using Cre recombinase.

Zbtb10 EUCOMM project 117694 genotyping (HEPD0855_6_G04)

Primers P1 (located at the 3’ end of exon ENSMUSE00000769877) and P2 (located at the 5’ end of the 3’ homology arm) will amplify a product of 320bp from the wild type allele and 367bp from the targeted allele.

exon ENSMUSE00000769877 5' homology arm 3' homology arm

P1 P2

Zbtb10 wild type

SV40 pA loxP human beta actin promoter neomy cin En2 Exon SV40 pA En2 Intron FRT exon ENSMUSE00000769877 5' homology arm FRT lacZ loxP loxP 3' homology arm

P1 P2

Zbtb10 EUCOMM 117694 PCR result

M WT +ve H O 180 181 182 183 184 2

500bp 400bp 300bp 200bp

M 100bp marker (NEB) WT JM8.N4 wild type genomic DNA +ve DNA from Zbtb10 targeted embryonic stem cells clone HEPD0855_6_G04 H20 no DNA control 180‐184 mice 180‐184 derived from embryonic stem cell clone HEPD0855_6_G04

The assay was performed using the REDExtract‐N‐AmpTM Tissue PCR kit from SIGMA using the procedure for DNA extraction from mouse tails and the following conditions.

Reagents Cycling parameters o REDExtract‐N‐Amp PCR reaction mix 10.0l Denature 94 C 2min x1 cycle Forward oligo P1 (10M) 0.8l Reverse oligo P2 (10M) 0.8l Denature 94oC 30sec Template 4.0l Annealing 60oC 30sec x35 cycles o H2O 4.4l Extension 72 C 45sec

Total 20.0l Extension 72oC 5min x1 cycle

Primers: P1 5’ GTAGAAGATTGCTCAGTGATGC 3’ P2 5’ CAGCGTGAGTGATCTTATGG 3’

UQ Research and Innovation Director, Research Management Office Nicole Thompson Animal Ethics Approval Certificate 22-Aug-2016 Please check all details below and inform the Animal Welfare Unit within 10 working days if anything is incorrect.

Activity Details Chief Investigator: Associate Professor Simon Phipps, Biomedical Sciences Title: Role of different PRRs and its ligands in the development of different endotype of asthma AEC Approval Number: SBMS/350/13/NHMRC Previous AEC Number: Approval Duration: 16-Nov-2013 to 16-Nov-2016 Funding Body: NHMRC Group: Anatomical Biosciences Other Staff/Students: Vivian Zhang, Rhiannon Werder, Khang Duong, Stuart Mazzone, Ken Loh, Jaisy Eacharath, Jason Lynch, Sean Yang, Ashik Ullah, Cristina Vicente, Laura Durham, Jennifer Simpson Location(s): St Lucia Bldg 76 - Chemistry (SCMB)

Summary Subspecies Strain Class Gender Source Approved Remaining Mice - genetically RAGE-/- C57 Adults Mix Institutional 112 90 modified (Genia) Breeding Colony Mice - genetically TLR4-/- C57 Adults Mix Institutional 144 125 modified (Teague) Breeding Colony Mice - genetically RAGE/TLR4 Adults Mix Institutional 32 22 modified double deficient Breeding Colony C57 (Guscott) Mice - genetically Tcrd -/- (Farrell) Adults Mix Institutional 32 12 modified Breeding Colony Mice - genetically Parling (PAG1 Adults Mix Institutional 336 324 modified deficient) Breeding Colony Mice - genetically 4C13R (Gill) Adults Male Institutional 64 64 modified Breeding Colony Mice - genetically Zbtb10-/- (ford) Adults Mix Institutional 192 192 modified Breeding Colony Mice - non C57BL/6 Adults Mix Commercial 2336 2082 genetically breeding colony modified Mice - non BALB/c Adults Mix Commercial 256 256 genetically breeding colony modified

Permits

Provisos

Approval Details

Animal Welfare Unit Cumbrae-Stewart Building +61 7 336 52925 (Enquiries) [email protected] UQ Research and Innovation Research Road +61 7 334 68710 (Enquiries) uq.edu.au/research The University of Queensland Brisbane Qld 4072 Australia +61 7 336 52713 (Coordinator) Page 1 of 3 Description Amount Balance

Mice - genetically modified (4C13R (Gill), Male, Adults, Institutional Breeding Colony) 31 Dec 2015 Use in 2015 (from 2016 MAR) 0 0 29 Feb 2016 Modification #7 64 64 Mice - genetically modified (Parling (PAG1 deficient) , Mix, Adults, Institutional Breeding Colony) 3 Aug 2015 Mod #5 48 48 31 Dec 2015 Use in 2015 (from 2016 MAR) -12 36 11 May 2016 Mod #10 64 100 11 Aug 2016 Mod #11 224 324 Mice - genetically modified (RAGE-/- C57 (Genia), Mix, Adults, Institutional Breeding Colony) 13 Nov 2013 Initial approval 112 112 31 Dec 2013 Use in 2013 (from 2014 MAR) 0 112 31 Dec 2014 Use in 2014 (from 2015 MAR) -22 90 31 Dec 2015 Use in 2015 (from 2016 MAR) 0 90 Mice - genetically modified (RAGE/TLR4 double deficient C57 (Guscott), Mix, Adults, Institutional Breeding Colony) 13 Nov 2013 Initial approval 32 32 31 Dec 2013 Use in 2013 (from 2014 MAR) 0 32 31 Dec 2014 Use in 2014 (from 2015 MAR) -10 22 31 Dec 2015 Use in 2015 (from 2016 MAR) 0 22 Mice - genetically modified (Tcrd -/- (Farrell), Mix, Adults, Institutional Breeding Colony) 31 Dec 2013 Use in 2013 (from 2014 MAR) 0 0 12 Feb 2014 Modification #1 32 32 31 Dec 2014 Use in 2014 (from 2015 MAR) -20 12 31 Dec 2015 Use in 2015 (from 2016 MAR) 0 12 Mice - genetically modified (TLR4-/- C57 (Teague), Mix, Adults, Institutional Breeding Colony) 13 Nov 2013 Initial approval 144 144 31 Dec 2013 Use in 2013 (from 2014 MAR) 0 144 31 Dec 2014 Use in 2014 (from 2015 MAR) -19 125 31 Dec 2015 Use in 2015 (from 2016 MAR) 0 125 Mice - genetically modified (Zbtb10-/- (ford), Mix, Adults, Institutional Breeding Colony) 11 May 2016 Mod #10 192 192 Mice - non genetically modified (BALB/c, Mix, Adults, Commercial breeding colony) 8 Apr 2015 Mod #4 256 256 31 Dec 2015 Use in 2015 (from 2016 MAR) 0 256 Mice - non genetically modified (C57BL/6, Mix, Adults, Commercial breeding colony) 13 Nov 2013 Initial approval 336 336 31 Dec 2013 Use in 2013 (from 2014 MAR) 0 336 12 Feb 2014 Modification #1 48 384 9 Jul 2014 Mod #2 96 480 12 Nov 2014 Modification #3 48 528 31 Dec 2014 Use in 2014 (from 2015 MAR) -158 370 3 Aug 2015 Mod #5 48 418

Page 2 of 3 13 Aug 2015 Mod #6 64 482 31 Dec 2015 Use in 2015 (from 2016 MAR) -96 386 10 Feb 2016 Modification #8 320 706 13 Apr 2016 Mod #9 1376 2082

Please note the animal numbers supplied on this certificate are the total allocated for the approval duration

Please use this Approval Number: 1. When ordering animals from Animal Breeding Houses 2. For labelling of all animal cages or holding areas. In addition please include on the label, Chief Investigator's name and contact phone number. 3. When you need to communicate with this office about the project.

It is a condition of this approval that all project animal details be made available to Animal House OIC. (UAEC Ruling 14/12/2001)

The Chief Investigator takes responsibility for ensuring all legislative, regulatory and compliance objectives are satisfied for this project. This certificate supercedes all preceeding certificates for this project (i.e. those certificates dated before 22-Aug-2016)

Animal Welfare Unit Cumbrae-Stewart Building +61 7 336 52925 (Enquiries) [email protected] UQ Research and Innovation Research Road +61 7 334 68710 (Enquiries) uq.edu.au/research The University of Queensland Brisbane Qld 4072 Australia +61 7 336 52713 (Coordinator) Page 3 of 3

UQ Research and Innovation Director, Research Management Office Nicole Thompson Animal Ethics Approval Certificate 24-Nov-2016 Please check all details below and inform the Animal Welfare Unit within 10 working days if anything is incorrect.

Activity Details Chief Investigator: Associate Professor Simon Phipps, Biomedical Sciences Title: Role of different PRRs and its ligands in the development of different endotype of asthma AEC Approval Number: SBMS/393/16 Previous AEC Number: SBMS/350/13/NHMRC Approval Duration: 15-Nov-2016 to 15-Nov-2019 Funding Body: NHMRC Group: Anatomical Biosciences Other Staff/Students: Vivian Zhang, Rhiannon Werder, Jennifer Simpson, Jason Lynch, Kevin Wathen-Dunn, Cristina Vicente, Al Amin Sikder Location(s): St Lucia Bldg 75 - AIBN

Summary Subspecies Strain Class Gender Source Approved Remaining Mice - genetically PAG1-/- (Parling) Adults Mix Institutional 384 384 modified Breeding Colony Mice - genetically Zbtb10 -/- (Ford) Adults Mix Institutional 384 384 modified Breeding Colony Mice - genetically P2Y13 -/- Adults Mix Institutional 576 576 modified (Joseph) Breeding Colony Mice - genetically 4C13R (Gill) Adults Mix Institutional 192 192 modified Breeding Colony Mice - non C57/BL6 Adults Mix Institutional 1920 1920 genetically Breeding Colony modified Mice - non BALB/c Adults Mix Institutional 384 384 genetically Breeding Colony modified

Permits

Provisos Cristina Vicente is listed as a participant on this project. According to our records, we have not received a response regarding this nomination. These participants cannot work on an approved project until they have confirmed involvement via electronic signature or, via email notification being sent to the Animal Welfare Unit Administration Officer [email protected]

Online Training Module The CI is reminded to ensure that all participants have completed the Animal Ethics Online Training Module available from

Approval Details

Animal Welfare Unit Cumbrae-Stewart Building +61 7 336 52925 (Enquiries) [email protected] UQ Research and Innovation Research Road +61 7 334 68710 (Enquiries) uq.edu.au/research The University of Queensland Brisbane Qld 4072 Australia +61 7 336 52713 (Coordinator) Page 1 of 2 Description Amount Balance

Mice - genetically modified (4C13R (Gill), Mix, Adults, Institutional Breeding Colony) 15 Nov 2016 Initial approval 192 192 Mice - genetically modified (P2Y13 -/- (Joseph), Mix, Adults, Institutional Breeding Colony) 15 Nov 2016 Initial approval 576 576 Mice - genetically modified (PAG1-/- (Parling), Mix, Adults, Institutional Breeding Colony) 15 Nov 2016 Initial approval 384 384 Mice - genetically modified (Zbtb10 -/- (Ford), Mix, Adults, Institutional Breeding Colony) 15 Nov 2016 Initial approval 384 384 Mice - non genetically modified (BALB/c, Mix, Adults, Institutional Breeding Colony) 15 Nov 2016 Initial approval 384 384 Mice - non genetically modified (C57/BL6, Mix, Adults, Institutional Breeding Colony) 15 Nov 2016 Initial approval 1920 1920

Please note the animal numbers supplied on this certificate are the total allocated for the approval duration

Please use this Approval Number: 1. When ordering animals from Animal Breeding Houses 2. For labelling of all animal cages or holding areas. In addition please include on the label, Chief Investigator's name and contact phone number. 3. When you need to communicate with this office about the project.

It is a condition of this approval that all project animal details be made available to Animal House OIC. (UAEC Ruling 14/12/2001)

The Chief Investigator takes responsibility for ensuring all legislative, regulatory and compliance objectives are satisfied for this project. This certificate supercedes all preceeding certificates for this project (i.e. those certificates dated before 24-Nov-2016)

Animal Welfare Unit Cumbrae-Stewart Building +61 7 336 52925 (Enquiries) [email protected] UQ Research and Innovation Research Road +61 7 334 68710 (Enquiries) uq.edu.au/research The University of Queensland Brisbane Qld 4072 Australia +61 7 336 52713 (Coordinator) Page 2 of 2 UQ Research and Innovation Director, Research Management Office Nicole Thompson Animal Ethics Approval Certificate 28-Feb-2017 Please check all details below and inform the Animal Welfare Unit within 10 working days if anything is incorrect.

Activity Details Chief Investigator: Associate Professor Simon Phipps, Biomedical Sciences Title: Pneuovirus infections in early life and development of allergic asthma AEC Approval Number: SBMS/194/16/NHMRC Previous AEC Number: SBMS/139/16/NHMRC Approval Duration: 14-Jun-2016 to 14-Jun-2019 Funding Body: NHMRC Group: Anatomical Biosciences Other Staff/Students: Vivian Zhang, Nick Hodge, Kym French, Rhiannon Werder, Kirsty Short, Jennifer Simpson, Jason Lynch, Laura Durham, Bodie Curren, Al Amin Sikder Location(s): St Lucia Bldg 75 - AIBN

Summary Subspecies Strain Class Gender Source Approved Remaining Mice - genetically pDC-DTR (Sonny Juvenile / Weaners Mix 12288 12288 modified Bill) / Pouch animal Mice - genetically RAGE-/- (Genia) Juvenile / Weaners Mix 1472 1472 modified / Pouch animal Mice - genetically RAGE-/- (Genia) Adults Female 368 368 modified Mice - genetically pDC-DTR (Sonny Adults Female 3008 3008 modified Bill) Mice - genetically 4CR13/RAGE-/- Adults Mix 292 292 modified (Gill Genia) Mice - genetically 4CR13/RAGE-/- Juvenile / Weaners Mix 768 768 modified (Gill Genia) / Pouch animal Mice - genetically Bill Foxred Adults Mix 260 260 modified Mice - genetically Bill Foxred Juvenile / Weaners Mix 640 640 modified / Pouch animal Mice - genetically IPS1-/- (Larkham) Adults Mix 128 128 modified Mice - genetically Ripk1k45A Adults Mix 256 256 modified (Cusiter) Mice - genetically Ripk1k45A/IRF7 Adults Mix 400 400 modified -/- (Gray) Mice - genetically 4C13R/pDC DTR Adults Female 768 768 modified (Gill Bill) Mice - genetically 4C13R/pDC DTR Juvenile / Weaners Mix 3072 3072 modified (Gill Bill) / Pouch animal

Animal Welfare Unit Cumbrae-Stewart Building +61 7 336 52925 (Enquiries) [email protected] UQ Research and Innovation Research Road +61 7 334 68710 (Enquiries) uq.edu.au/research The University of Queensland Brisbane Qld 4072 Australia +61 7 336 52713 (Coordinator) Page 1 of 6 Mice - genetically 4C13R/OT-II Adults Mix 100 100 modified (Smith) Mice - genetically RAGE/TLR4 Adults Female 32 32 modified double deficient (Guscott) Mice - genetically WT C57/IRF7-/- Adults Female 80 80 modified heterozygous Mice - genetically TLR4-/- (Teague) Adults Female 32 32 modified Mice - genetically IPS1-/- (Larkham) Juvenile / Weaners Mix 128 128 modified / Pouch animal Mice - genetically TLR4-/- (Teague) Juvenile / Weaners Mix 128 128 modified / Pouch animal Mice - genetically Ripk1k45A Juvenile / Weaners Mix 640 640 modified (Cusiter) / Pouch animal Mice - genetically 4C13R/IRF7-/- Juvenile / Weaners Mix 640 640 modified (Read) / Pouch animal Mice - genetically RAGE/TLR4 Juvenile / Weaners Mix 128 128 modified double deficient / Pouch animal (Guscott) Mice - genetically 4C13R/IRF7-/- Adults Mix 160 160 modified (Read) Mice - genetically WT C57/IRF7-/- Juvenile / Weaners Mix 320 320 modified heterozygous / Pouch animal Mice - genetically Ripk1k45A/IRF7 Juvenile / Weaners Mix 1216 1216 modified -/- (Gray) / Pouch animal Mice - genetically IRF7-/- [Cueto] Juvenile / Weaners Mix 1664 1664 modified / Pouch animal Mice - genetically IRF7-/- [Cueto] Adults Mix 512 512 modified Mice - genetically IRF3 deficient Juvenile / Weaners Mix Institutional 128 128 modified (Lawes) / Pouch animal Breeding Colony Mice - genetically IRF3 deficient Adults Mix Institutional 32 32 modified (Lawes) Breeding Colony Mice - genetically IRF3/7 deficient Juvenile / Weaners Mix Institutional 128 128 modified (Young) / Pouch animal Breeding Colony Mice - genetically IRF3/7 deficient Adults Mix Institutional 32 32 modified (Young) Breeding Colony Mice - genetically P2Y14 deficient Adults Mix 576 576 modified (Nowell) Mice - genetically Akita (Ins2Akita) Adults Mix 96 96 modified Mice - genetically Parling (PAG1 Neonates Mix 128 128 modified deficient) Mice - genetically Parling (PAG1 Adults Female 32 32 modified deficient) Mice - non 4C13R (Gill) Juvenile / Weaners Mix 848 848 genetically / Pouch animal modified Mice - non BALB/c Juvenile / Weaners Mix 2512 2512 genetically / Pouch animal modified Mice - non BALB/c Adults Female 612 612 genetically modified

Page 2 of 6 Mice - non 4C13R (Gill) Adults Female 212 212 genetically modified Mice - non C57BL/6 Adults Mix Commercial 1026 1026 genetically breeding colony modified Mice - non C57BL/6 Juvenile / Weaners Mix Commercial 2256 2256 genetically / Pouch animal breeding colony modified Mice - non C57BL/6 Neonates Mix Commercial 128 128 genetically breeding colony modified

Permits

Provisos The AEC approves a maximum 20% weight loss on the proviso that oximeters are used and animals continue to be weighed. The Committee required that the CI provide a report to the AEC following the first trial.

Approval Details

Description Amount Balance

Mice - genetically modified (4C13R/IRF7-/- (Read), Mix, Adults, ) 8 Jun 2016 Initial approval 80 80 29 Aug 2016 Mod #4 80 160 Mice - genetically modified (4C13R/IRF7-/- (Read), Mix, Juvenile / Weaners / Pouch animal, ) 8 Jun 2016 Initial approval 320 320 29 Aug 2016 Mod #4 320 640 Mice - genetically modified (4C13R/OT-II (Smith), Mix, Adults, ) 8 Jun 2016 Initial approval 100 100 Mice - genetically modified (4C13R/pDC DTR (Gill Bill), Female, Adults, ) 8 Jun 2016 Initial approval 768 768 Mice - genetically modified (4C13R/pDC DTR (Gill Bill), Mix, Juvenile / Weaners / Pouch animal, ) 8 Jun 2016 Initial approval 3072 3072 Mice - genetically modified (4CR13/RAGE-/- (Gill Genia), Mix, Adults, ) 8 Jun 2016 Initial approval 292 292 Mice - genetically modified (4CR13/RAGE-/- (Gill Genia), Mix, Juvenile / Weaners / Pouch animal, ) 8 Jun 2016 Initial approval 768 768 Mice - genetically modified (Akita (Ins2Akita), Mix, Adults, ) 20 Dec 2016 Mod #14 96 96 Mice - genetically modified (Bill Foxred, Mix, Adults, ) 8 Jun 2016 Initial approval 260 260 Mice - genetically modified (Bill Foxred, Mix, Juvenile / Weaners / Pouch animal, ) 8 Jun 2016 Initial approval 640 640 Mice - genetically modified (IPS1-/- (Larkham), Mix, Adults, ) 8 Jun 2016 Initial approval 128 128 Mice - genetically modified (IPS1-/- (Larkham), Mix, Juvenile / Weaners / Pouch animal, ) 8 Jun 2016 Initial approval 128 128

Page 3 of 6 Mice - genetically modified (IRF3 deficient (Lawes), Mix, Adults, Institutional Breeding Colony) 14 Sep 2016 Mod #6 32 32 Mice - genetically modified (IRF3 deficient (Lawes), Mix, Juvenile / Weaners / Pouch animal, Institutional Breeding Colony) 14 Sep 2016 Mod #6 128 128 Mice - genetically modified (IRF3/7 deficient (Young), Mix, Adults, Institutional Breeding Colony) 14 Sep 2016 Mod #6 32 32 Mice - genetically modified (IRF3/7 deficient (Young), Mix, Juvenile / Weaners / Pouch animal, Institutional Breeding Colony) 14 Sep 2016 Mod #6 128 128 Mice - genetically modified (IRF7-/- [Cueto], Mix, Adults, ) 8 Jun 2016 Initial approval 512 512 Mice - genetically modified (IRF7-/- [Cueto], Mix, Juvenile / Weaners / Pouch animal, ) 8 Jun 2016 Initial approval 1664 1664 Mice - genetically modified (P2Y14 deficient (Nowell), Mix, Adults, ) 12 Oct 2016 Mod #7 576 576 Mice - genetically modified (Parling (PAG1 deficient) , Female, Adults, ) 15 Feb 2017 Mod #16 32 32 Mice - genetically modified (Parling (PAG1 deficient) , Mix, Neonates, ) 15 Feb 2017 Mod #16 128 128 Mice - genetically modified (pDC-DTR (Sonny Bill), Female, Adults, ) 8 Jun 2016 Initial approval 3008 3008 Mice - genetically modified (pDC-DTR (Sonny Bill), Mix, Juvenile / Weaners / Pouch animal, ) 8 Jun 2016 Initial approval 12032 12032 29 Aug 2016 Mod #5 128 12160 14 Sep 2016 Mod #1 128 12288 Mice - genetically modified (RAGE-/- (Genia), Female, Adults, ) 8 Jun 2016 Initial approval 328 328 9 Nov 2016 Mod #10 24 352 20 Dec 2016 Mod #13 16 368 Mice - genetically modified (RAGE-/- (Genia), Mix, Juvenile / Weaners / Pouch animal, ) 8 Jun 2016 Initial approval 992 992 14 Sep 2016 Mod #6 320 1312 9 Nov 2016 Mod #10 96 1408 20 Dec 2016 Mod #13 64 1472 Mice - genetically modified (RAGE/TLR4 double deficient (Guscott), Female, Adults, ) 8 Jun 2016 Initial approval 32 32 Mice - genetically modified (RAGE/TLR4 double deficient (Guscott), Mix, Juvenile / Weaners / Pouch animal, ) 8 Jun 2016 Initial approval 128 128 Mice - genetically modified (Ripk1k45A (Cusiter), Mix, Adults, ) 8 Jun 2016 Initial approval 176 176 29 Aug 2016 Mod #4 80 256 Mice - genetically modified (Ripk1k45A (Cusiter), Mix, Juvenile / Weaners / Pouch animal, )

Page 4 of 6 8 Jun 2016 Initial approval 320 320 29 Aug 2016 Mod #4 320 640 Mice - genetically modified (Ripk1k45A/IRF7-/- (Gray), Mix, Adults, ) 8 Jun 2016 Initial approval 320 320 29 Aug 2016 Mod #4 80 400 Mice - genetically modified (Ripk1k45A/IRF7-/- (Gray), Mix, Juvenile / Weaners / Pouch animal, ) 8 Jun 2016 Initial approval 896 896 29 Aug 2016 Mod #4 320 1216 Mice - genetically modified (TLR4-/- (Teague), Female, Adults, ) 8 Jun 2016 Initial approval 32 32 Mice - genetically modified (TLR4-/- (Teague), Mix, Juvenile / Weaners / Pouch animal, ) 8 Jun 2016 Initial approval 128 128 Mice - genetically modified (WT C57/IRF7-/- heterozygous, Female, Adults, ) 8 Jun 2016 Initial approval 80 80 Mice - genetically modified (WT C57/IRF7-/- heterozygous, Mix, Juvenile / Weaners / Pouch animal, ) 8 Jun 2016 Initial approval 320 320 Mice - non genetically modified (4C13R (Gill), Female, Adults, ) 8 Jun 2016 Initial approval 212 212 Mice - non genetically modified (4C13R (Gill), Mix, Juvenile / Weaners / Pouch animal, ) 8 Jun 2016 Initial approval 848 848 Mice - non genetically modified (BALB/c, Female, Adults, ) 8 Jun 2016 Initial approval 580 580 29 Aug 2016 Mod #3 32 612 Mice - non genetically modified (BALB/c, Mix, Juvenile / Weaners / Pouch animal, ) 8 Jun 2016 Initial approval 2320 2320 29 Aug 2016 Mod #3 192 2512 Mice - non genetically modified (C57BL/6, Mix, Adults, Commercial breeding colony) 8 Jun 2016 Initial approval 556 556 29 Aug 2016 Mod #2 270 826 29 Aug 2016 Mod #5 32 858 14 Sep 2016 Mod #1 32 890 9 Nov 2016 Mod #10 24 914 1 Dec 2016 Mod #9 80 994 15 Feb 2017 Mod #16 32 1026 Mice - non genetically modified (C57BL/6, Mix, Juvenile / Weaners / Pouch animal, Commercial breeding colony) 8 Jun 2016 Initial approval 1840 1840 9 Nov 2016 Mod #10 96 1936 1 Dec 2016 Mod #9 320 2256 Mice - non genetically modified (C57BL/6, Mix, Neonates, Commercial breeding colony) 15 Feb 2017 Mod #16 128 128

Page 5 of 6 Please note the animal numbers supplied on this certificate are the total allocated for the approval duration

Please use this Approval Number: 1. When ordering animals from Animal Breeding Houses 2. For labelling of all animal cages or holding areas. In addition please include on the label, Chief Investigator's name and contact phone number. 3. When you need to communicate with this office about the project.

It is a condition of this approval that all project animal details be made available to Animal House OIC. (UAEC Ruling 14/12/2001)

The Chief Investigator takes responsibility for ensuring all legislative, regulatory and compliance objectives are satisfied for this project. This certificate supercedes all preceeding certificates for this project (i.e. those certificates dated before 28-Feb-2017)

Animal Welfare Unit Cumbrae-Stewart Building +61 7 336 52925 (Enquiries) [email protected] UQ Research and Innovation Research Road +61 7 334 68710 (Enquiries) uq.edu.au/research The University of Queensland Brisbane Qld 4072 Australia +61 7 336 52713 (Coordinator) Page 6 of 6 194/16 Modification #20

05/14

The University of Queensland Animal Ethics Modification Form Animal Welfare Unit, UQ Research & Innovation

Committee consideration: AEC Group: Date Received 02/03/2017 / / ABS Date CI Without change: / / Advised Approved subject to: Resubmit: Modification No: # 20 Date/Signature Chairperson: Comment:

/ /

This form, together with required number of copies, should be forwarded to the Animal Welfare Unit, UQ Research & Innovation, Cumbrae Stewart Building (72), The University of Queensland, St Lucia, QLD. Please Note:  Approval of a modification does not affect the expiry date of the project’s current ethical clearance. The ethical clearance is valid only for the period shown on the Animal Ethics Approval Certificate.  This form should only be used for a modification to an existing approved project if it remains within the same aims of the original project. Any major changes that fall outside the original scope of the project should be submitted as a new application. If in doubt, please contact the Animal Welfare Unit.  If you are altering an approved project’s experimental design, it is recommended to submit a flowchart illustrating where this modification takes place within the original scope of the project. This will assist Committee members when considering this modification, especially for projects with large numbers of experiments and modifications.  Do not elaborate or repeat items in this modification that are already approved by the AEC to be utilised under this protocol, simply indicate “as per original application”. Project details:

Title: Pneumovirus infections in early life and development of allergic asthma

Current AEC No: SBMS/194/16 We have developed a model of viral bronchiolitis and subsequent asthma Brief overview of original project including airway remodelling in WT mice following neonatal inoculation to be modified (max. 250words): with pneumovirus of mice (PVM) and cockroach allergen (CRE). Brief details of changes to be made: (This should be dot point and not include full details e.g. We now wish to use ZBTB10 deficient mice to investigate if absence of  change of location ZBTB10 is associated with exaggerated viral bronchiolitis and subsequent  change of personnel  additional experiment asthma.  minor changes to experimental design ) Chief Investigator: Dr Simon Phipps

Institute/Department/School: School of Biomedical Sciences

Contact Phone No: 3365 2785

Email Address: [email protected]

Modification Form Page 1 of 4 194/16 Modification #20

05/14

Type of change Details of change Justification for change

Number of animals required are additional to those approved: 16 neonates/group x 3 [vehicle, PVM, PVM+CRE] x 4 [time points = x3 in primary infection and x1 in re-infection] x 2 groups Change to numbers of animals A total of 384 neonatal ZBTB10 deficient (FORD) and 96 [ZBTB10+/- and ZBTB10+/+] = 384 neonatal mice. 384/4 = 96 required: pregnant ZBTB10 deficient (FORD) mice will be needed. pregnant ZBTB10+/- mice. The littermate control will be used as Must indicate control.  Whether additional to original animals or in place of original animals. Statistical justification. We calculated the number of neonates  Replacing which strains where required for this study using a standard power analysis of airway appropriate smooth muscle remodeling (a key asthma endpoint) from historical  In which experiment. data with the following equation: standardised effect size  Include transport, housing and (signal/noise ratio) = (Mean1-Mean2)/standard deviation. Mean location of experiment (if area of smooth muscle in asthma mouse (pDC-DTR) = 3; healthy changed). control (WT) = 0.5. Standard deviation = 2.4. Hence, standardized  Do not include experimental effect = 1, number of animals = 16. details in this section Therefore 16 mice per group is necessary for statistical comparisons (with 80% power) between treatment groups in this study.

Experiments are additional to those approved. Results from genetic studies carried out by one of Dr Phipps’ collaborator (Dr Ferreira at QIMR Berghofer) showed that ZBTB10 As per the figure, neonatal mice of each strain will be is a transcriptional repressor and increased ZBTB10 expression in administered intranasally vehicle or PVM (5 PFU; i.n. 10 µl) at 7 day of age and vehicle or PVM (100 PFU; intranasal route. 50 µl) humans is associated with lower risk of asthma. We now would like to test the hypothesis that whether the absence of ZBTB10 Addition of experiment: on day 49 of age (42 days post primary infection, DPI) under light Must indicate anesthesia with isoflurane using an induction chamber (AIBN expression exaggerates bronchiolitis and subsequent asthma  Experiment number animal house), dose rate (mg/kg body weight): 1-5% in O2 (1- phenotype in mice. This study will reveal an anti-inflammatory role  Statistical justification 3L/min) as per the original modification. Using this same protocol for ZBTB10 in asthma and so justify the development of ZBTB10  Indicate where and how this we will administer cockroach extract (CRE; 1μg; i.n. 10μL for agonists for asthma treatment. additional experiment fits into neonatal mice and 50 μL for adult mice) on the days indicated in the original approval the figure below.

Mice will be sacrificed (using sodium pentobarbitone overdose according to SOP AHT-39: Euthanasia of Rats) at a total of 4 time points as indicated in the figure.

Modification Form Page 2 of 4 194/16 Modification #20

05/14 Figure: Schema indicating how these experiments fit in with the overall

project:

Change to existing experimental design: Must indicate N/A N/A  Which experiment you are changing  Statistical justification Other:  Indicate relationship to original N/A N/A approval

Impact on animals:

Please describe the likely impact of the proposed changes on individual animals and explain what steps will be taken to minimise this change.

We do not anticipate any significant morbidity associated with PVM infection of this strain. Consistent with our initial application, mice will be monitored on a daily basis for at least one week following each infection. If any morbidities are evident mice will be monitored more frequently and euthanized if any signs of pain or discomfort are evident.

If GM animals, indicate the genotype/phenotype and any welfare implications?

N/A

Modification Form Page 3 of 4 194/16 Modification #20

05/14 Type of change Details of change Justification for change Change to personnel: N/A Must include:  Qualifications  Experience N/A  Role in project, +/- training required  Contact details.  Signed declaration below

DECLARATION OF CONSENT BY ADDITIONAL PARTICIPANTS Name Signature

Signature of Chief Date: 02 / 03 / 2017 Investigator:

Name of Chief Dr Simon Phipps

Investigator: (please print or type)

Modification Form Page 4 of 4