Genetic Susceptibility to Childhood Bronchiolitis

D 1461 OULU 2018 D 1461

UNIVERSITY OF OULU P.O. Box 8000 FI-90014 UNIVERSITY OF OULU FINLAND ACTA UNIVERSITATIS OULUENSIS ACTA UNIVERSITATIS OULUENSIS ACTA

DMEDICA Anu Pasanen Anu Pasanen University Lecturer Tuomo Glumoff GENETIC SUSCEPTIBILITY University Lecturer Santeri Palviainen TO CHILDHOOD

Postdoctoral research fellow Sanna Taskila BRONCHIOLITIS

Professor Olli Vuolteenaho

University Lecturer Veli-Matti Ulvinen

Planning Director Pertti Tikkanen

Professor Jari Juga

University Lecturer Anu Soikkeli

Professor Olli Vuolteenaho UNIVERSITY OF OULU GRADUATE SCHOOL; UNIVERSITY OF OULU, FACULTY OF MEDICINE; Publications Editor Kirsti Nurkkala PEDEGO RESEARCH UNIT; MEDICAL RESEARCH CENTER OULU; ISBN 978-952-62-1898-4 (Paperback) OULU UNIVERSITY HOSPITAL ISBN 978-952-62-1899-1 (PDF) ISSN 0355-3221 (Print) ISSN 1796-2234 (Online)

ACTA UNIVERSITATIS OULUENSIS D Medica 1461

ANU PASANEN

GENETIC SUSCEPTIBILITY TO CHILDHOOD BRONCHIOLITIS

Academic dissertation to be presented with the assent of the Doctoral Training Committee of Health and Biosciences of the University of Oulu for public defence in Auditorium F101 of the Faculty of Biochemistry and Molecular Medicine (Aapistie 7), on 25 May 2018, at 12 noon

Supervised by Professor Mika Rämet Professor Mikko Hallman Doctor Minna Karjalainen

Reviewed by Docent Liisa Myllykangas Professor Ville Peltola

Opponent Professor Johanna Schleutker

ISBN 978-952-62-1898-4 (Paperback) ISBN 978-952-62-1899-1 (PDF)

ISSN 0355-3221 (Printed) ISSN 1796-2234 (Online)

Cover Design Raimo Ahonen

JUVENES PRINT TAMPERE 2018 Pasanen, Anu, Genetic susceptibility to childhood bronchiolitis. University of Oulu Graduate School; University of Oulu, Faculty of Medicine; PEDEGO Research Unit; Medical Research Center Oulu; Oulu University Hospital Acta Univ. Oul. D 1461, 2018 University of Oulu, P.O. Box 8000, FI-90014 University of Oulu, Finland

Abstract Bronchiolitis is an infection of the small airways of the lung and is a common reason for infant hospitalizations. The most common causative pathogen is the respiratory syncytial virus (RSV). Genetic factors are thought to influence the risk of bronchiolitis, and better knowledge of bronchiolitis genetics will likely help to elucidate the disease process. Severe bronchiolitis in childhood may predispose to asthma. Therefore, an effective treatment of bronchiolitis may affect the present-day as well as lifelong respiratory health. In this project, we aimed to identify genetic loci of bronchiolitis susceptibility by a genome- wide association study (GWAS) and suitable follow-up studies, and to study a previously asthma- associated CDHR3 variant for association across five bronchiolitis populations by meta-analysis. We performed the GWAS on a Finnish-Swedish case-control population and identified several loci below the suggestive genome-wide significance level. Of these, three variants showed nominal associations in a replication population from the Netherlands. One of the loci affected KCND3 expression, and two others were intergenic variants with putative regulatory potential. In a follow-up study conducted on a GWAS sub population, we identified the NKG2D locus as a candidate of susceptibility to bronchiolitis. The genomic region encompassing NKG2D variants was reportedly associated with NKG2D mRNA and protein abundance. We validated the association between NKG2D genotypes and protein expression with flow cytometry. The association between NKG2D and bronchiolitis was supported by a Finnish replication study. The meta-analysis was performed on populations from Denmark, Finland, Sweden, Germany, and the Netherlands. A potential virus-specific role for the CDHR3 variant was detected in a population that comprised mostly RSV-negative cases. In conclusion, we identified new candidates of bronchiolitis susceptibility in GWAS and subsequent studies. We found the CDHR3 variant was a potential susceptibility factor in severe non-RSV bronchiolitis and asthma. Our preliminary results provide interesting starting points for further studies. In the future, better understanding of the disease mechanisms and the relationship of bronchiolitis and asthma could provide means to design new therapeutic options.

Keywords: asthma, bronchiolitis, gene expression, genetic susceptibility, genome-wide association study, infant, lower respiratory infection, meta-analysis, RSV

Pasanen, Anu, Geneettinen alttius lapsuusajan bronkioliitille. Oulun yliopiston tutkijakoulu; Oulun yliopisto, Lääketieteellinen tiedekunta; PEDEGO- tutkimusyksikkö; Medical Research Center Oulu; Oulun yliopistollinen sairaala Acta Univ. Oul. D 1461, 2018 Oulun yliopisto, PL 8000, 90014 Oulun yliopisto

Tiivistelmä Bronkioliitti on viruksen aiheuttama alahengitystieinfektio, joka usein johtaa pienten lasten sai- raalahoitoon. Yleisin bronkioliitin aiheuttaja lapsilla on respiratory syncytial -virus (RSV). Perintötekijöiden arvellaan altistavan bronkioliitille, joten uusi tieto altistavista geeneistä voi auttaa ymmärtämään taudin taustalla olevia biologisia mekanismeja. Lapsuusiän bronkioliitin ajatellaan voivan altistaa astmalle, joten bronkioliitin tehokas hoito voi vaikuttaa merkittävästi hengitysterveyteen myös pitkällä aikavälillä. Työssä pyrittiin selvittämään lapsuusajan bronkioliitille altistavia geneettisiä tekijöitä gen- ominlaajuisella assosiaatiokartoituksella, joka toteutettiin suomalais-ruotsalaisessa tapaus-ver- rokkiväestössä. Löydökset pyrittiin varmentamaan soveltuvilla jatkotutkimuksilla. Lisäksi tar- kastelimme astmalle altistavaa CDHR3-geenin polymorfismia viidessä eurooppalaisessa bron- kioliittikohortissa käyttäen meta-analyysia. Assosiaatiokartoituksessa havaittiin useita mahdollisia bronkioliittialttiuteen vaikuttavia gee- nikohtia. Näistä kolme sai tukea hollantilaisessa väestössä tehdyssä assosiaatioanalyysissä, jos- sa testattiin assosiaatiokartoituksen lupaavimmat löydökset. Yksi altistavista polymorfismeista vaikutti KCND3-geenin ilmentymiseen, ja kaksi muuta olivat geenien välisiä, mahdollisesti gee- ninsäätelyyn osallistuvia variantteja. Assosiaatiokartoituksen osa-analyysissä NKG2D tunnistet- tiin mahdolliseksi bronkioliitille altistavaksi geeniksi. NKG2D-immuunireseptorin alentunut ilmentyminen voi tulostemme perusteella altistaa vakavalle bronkioliitille. Meta-analyysissä, jonka tutkimuskohortit olivat peräisin Tanskasta, Suomesta, Ruotsista, Saksasta ja Hollannista, todettiin mahdollinen yhteys CDHR3-geenin polymorfismin ja muun viruksen kuin RSV:n aihe- uttaman bronkioliitin välillä. Toteutimme tässä työssä ensimmäisen genominlaajuisen bronkioliittialttiutta koskevan assosiaatiokartoituksen. Assosiaatiokartoituksessa, sitä seuranneissa jatkotutkimuksissa ja meta-ana- lyysissä tunnistimme useita lupaavia alttiusgeenejä, mutta tuloksemme vaativat varmentamista suuremmissa tutkimusväestöissä.

Asiasanat: alahengitystieinfektio, astma, bronkioliitti, geenin ilmentyminen, genominlaajuinen assosiaatiokartoitus, lapsi, meta-analyysi, perinnöllinen alttius, RSV

Acknowledgements

This work was carried out at the department of children and adolescents and the Medical Research Center Oulu, University of Oulu and Oulu University Hospital, during the years 2013-2018. I wish to acknowledge all the people who have contributed to this work one way or another and supported me during these years. I would like to sincerely thank the trio of my supervisors including Professor Mika Rämet, Professor Mikko Hallman, and Doctor Minna Karjalainen. My principal supervisor Mika Rämet is especially acknowledged for his insightful comments and practical advice in various situations as well as for his positive and encouraging but down-to-earth attitude. I am grateful to Mikko Hallman for giving me the opportunity to work in this group and I value his optimism, curiosity and open-mindedness towards research questions and life. I would like to thank my third supervisor and dear friend Minna Karjalainen for introducing me into this project as well as for her invaluable guidance regarding scientific work and thinking as well as encouragement and advice throughout this project. I am grateful to Professor Matti Korppi for his insightful remarks regarding bronchiolitis and asthma. I wish to sincerely thank all other co-authors for their valuable contributions to the original publications. Göran Wennergren, Emma Goksör, Louis Bont, Kuldeep Kumawat, Marja Ruotsalainen, Eija Piippo- Savolainen, Ilkka Junttila, Laura Kummola, Hennie Hodemaekers, Riny Janssen, Kirsi Nuolivirta, Tuomas Jartti, Anders Husby, Johannes Waage, Astrid Sevelsted, Jakob Stokholm, Bo Chawes, Nadja Vissing, Andrea Heinzmann, Hans Bisgaard, and Klaus Bønnelykke are warmly acknowledged. It was a pleasure to work with such professional and friendly people. I would like to thank the pre-examiners of this thesis, Professor Ville Peltonen and Adjunct Professor Liisa Myllykangas, for their constructive comments to improve the contents of this thesis. I am grateful to Professor Johanna Schleutker for agreeing to be my opponent and I look forward to our discussion. Kira Heller and Deborah Kaska are acknowledged for revising the English language of the thesis. Anna Vuolteenaho is acknowledged for revising the language of the Finnish abstract. I warmly thank all former and present members of our research group. I am grateful for all the good times and educational moments we have shared during these years. I wish to thank Maarit Haarala, Johanna Huusko, Antti Haapalainen, Heli Tiensuu, Tomi Määttä, Juhani Metsola, Annamari Salminen, Marja Ojaniemi, Ravindra Daddali, and others, for assistance, peer support and intriguing

7 conversations in laboratory, coffee breaks, and many other situations. Special thanks to Marjatta Paloheimo for help in various occasions. I also wish to thank the people of CRC for friendly atmosphere and interesting discussions. I wish to thank the members of my follow-up group, Docent Minna Männikkö and Docent Terhi Tapiainen, for time and encouraging comments. I always enjoyed our meetings and left with full of fresh thoughts and perspectives. I am deeply grateful to all my long term as well as newer friends. You have been very important to me during different periods of life. I am grateful for all joyful times, conversations, travels, and laughs and cries we have shared. I want to express warmest thanks to my family. I wish to thank my parents Anne and Pekka for support, and give special thanks to Tepi and Tuula for tolerating and loving them. My hearfelt thanks go to my sibligs Eini and Jussi-Pekka and my grandparents mummu and tuta. Finally, I wish to give my giant thanks to my beloved Jari-Pekka and our kissue of Nella and Luna. This work was financially supported by the MRC Oulu Doctoral Programme, the Alma and K.A. Snellman Foundation, the Sohlberg Foundation, and the Research Foundation of the Pulmonary Diseases.

Oulu, March 2018 Anu Pasanen

Abbreviations

1000 GP 1000 Genomes Project bp Base pair CDHR3 Cadherin related family member 3 CNV Copy number variant CX3CR1 C-X-3C motif chemokine receptor 1 DNA Deoxyribonucleic acid eQTL Expression quantitative trait locus GTEx Genotype-tissue expression project GWAS Genome-wide association study HWE Hardy-Weinberg equilibrium ICAM-1 Intercellular adhesion molecule 1 IL Interleukin INF Interferon kb Kilo base pair KCND3 Potassium voltage-gated channel subfamily D member 3 LD Linkage disequilibrium LRI Lower respiratory infection MAF Minor allele frequency MFI Mean fluorescence intensity mRNA Messenger RNA ncRNA Noncoding RNA NK Natural killer NKG2D NKG2-D type II integral membrane protein PCA Principal component analysis PCR Polymerase chain reaction PIV Parainfluenza virus pQTL Protein quantitative trait locus PAMP Pathogen associated molecular pattern PRR Pattern recognition receptor PRS Polygenic risk score QC Quality control QTL Quantitative trait locus RSV Respiratory syncytial virus RV Rhinovirus SNP Single nucleotide polymorphism

9 Th1 T helper 1 Th2 T helper 2 TLR Toll-like receptor VDR Vitamin D receptor WES Whole-exome sequencing WGS Whole-genome sequencing

List of original publications

This thesis is based on the following publications, which are referred to throughout the text by Roman numerals:

I Pasanen, A., Karjalainen, MK., Bont, L., Piippo-Savolainen, E., Ruotsalainen, M., Goksör, E., Kumawat, K., Hodemaekers, H., Nuolivirta, K., Jartti, T., Wennergren, G., Hallman, M., Rämet, M., & Korppi, M. (2017). Genome-wide association study of polymorphisms predisposing to bronchiolitis. Sci Rep 7:41653. II Pasanen, A., Karjalainen, M.K., Kummola, L., Waage, J., Bønnelykke, K., Ruotsalainen, M., Piippo-Savolainen, E., Goksör, E., Nuolivirta, K., Chawes, B., Vissing, N., Bisgaard, H., Jartti, T., Wennergren, G., Junttila, I., Hallman, M., Korppi, M., & Rämet, M. (2017). NKG2D gene variation and susceptibility to viral bronchiolitis in childhood. Manuscript. III Husby, A., Pasanen, A., Waage, J., Sevelsted, A., Hodemaekers, H., Janssen, R., Karjalainen, M.K., Stokholm, J., Chawes, B.L., Korppi, M., Wennergren, G., Heinzmann, A., Bont, L., Bisgaard, H., & Bønnelykke, K. (2017). CDHR3 gene variation and childhood bronchiolitis. J Allergy Clin Immunol 140:1469-1471.e7. .

Contents

Abstract Tiivistelmä Acknowledgements 7 Abbreviations 9 List of original publications 11 Contents 13 1 Introduction 15 2 Review of literature 17 2.1 Studying genetics of complex traits ...... 17 2.1.1 Genetic variation in humans ...... 17 2.1.2 Role of genetic variation in disease ...... 19 2.1.3 Genome-wide association studies ...... 21 2.1.4 Interpreting GWAS signals ...... 25 2.2 Bronchiolitis ...... 27 2.2.1 Definition...... 27 2.2.2 Treatment and prevention ...... 27 2.2.3 Epidemiology ...... 28 2.2.4 Immune response to respiratory viruses ...... 30 2.2.5 Bronchiolitis and asthma ...... 33 2.3 Genetic factors of bronchiolitis susceptibility ...... 33 2.3.1 Genetic variants associated with bronchiolitis in previous studies ...... 34 2.3.2 Genetic variants associated with bronchiolitis and asthma ...... 37 3 Aims of the study 39 4 Materials and methods 41 4.1 Ethical considerations ...... 41 4.2 Study populations in genetic studies ...... 41 4.2.1 Study I ...... 41 4.2.2 Study II ...... 44 4.2.3 Study III ...... 45 4.3 DNA sample preparation and genotyping ...... 45 4.3.1 DNA samples (I-III) ...... 45 4.3.2 Genome-wide genotyping (I, II) ...... 45 4.3.3 Targeted SNP genotyping (I, II) ...... 46 4.3.4 Genotyping of CDHR3 SNPs (III) ...... 46 4.4 Processing of the genome-wide data ...... 46

13 4.4.1 Quality control (I, II) ...... 46 4.4.2 Genotype imputation (I) ...... 46 4.5 GWAS ...... 47 4.5.1 Principal component analysis (I) ...... 47 4.5.2 Association tests (I, II) ...... 47 4.5.3 Selection of variants for further study (I, II) ...... 48 4.6 Analysis of allele specific expression (II) ...... 48 4.7 Association analyses in replication cohorts (I, II) ...... 49 4.8 Analysis of CDHR3 SNP (III) ...... 49 5 Results 51 5.1 Discovery analyses for the combined genome-wide data (I) ...... 51 5.1.1 Best associating loci ...... 51 5.1.2 Association of variants previously linked to bronchiolitis or asthma in the current GWAS ...... 53 5.2 Analysis of shared loci between results from the GWAS subpopulation and reported QTLs (II)...... 54 5.3 Follow-up studies based on the genome-wide analyses ...... 55 5.3.1 Associations of the best variants from the combined GWAS in replication populations (I) ...... 55 5.3.2 Validation of the effect of the NKG2D genotype on NKG2D expression (II) ...... 56 5.3.3 Replication analysis of NKG2D SNPs (II) ...... 57 5.4 Analysis of CDHR3 SNP rs6967330 in European bronchiolitis cohorts (III) ...... 58 6 Discussion 59 6.1 Potential bronchiolitis susceptibility loci from GWAS (I) ...... 59 6.1.1 Variants that were associated with bronchiolitis or asthma in previous studies ...... 60 6.1.2 Replication analysis of the best loci in GWAS ...... 61 6.2 NKG2D as a candidate of bronchiolitis susceptibility (II) ...... 62 6.2.1 Follow-up studies on NKG2D ...... 63 6.3 Potential virus-specific role of CDHR3 variant in bronchiolitis (III) ...... 64 6.4 Prospects for future studies ...... 64 7 Conclusions 67 References 69 Original publications 85

1 Introduction

Bronchiolitis is a common lower respiratory infection (LRI) that frequently leads to infant hospitalizations worldwide (Meissner, 2016). Viral infection is a typical cause of bronchiolitis, and respiratory syncytial virus (RSV) infections account for most bronchiolitis cases. Prematurity, immunodeficiency, heart disease, and exposure to cigarette smoke are factors that predispose children to severe bronchiolitis (Florin, Plint, & Zorc, 2017). In addition to these risk factors, the variations observed in disease phenotypes are likely due at least in part to genetic factors. Previous genetic studies of bronchiolitis have generally focused on the association of immune-involved candidate genes with bronchiolitis (Miyairi & DeVincenzo, 2008). These studies have identified some associated loci, but knowledge of the genetic variation contributing to bronchiolitis susceptibility is incomplete. Individuals affected by severe viral bronchiolitis in infancy appear to have an elevated risk of becoming asthmatic (Jartti & Gern, 2017). Asthma is a chronic, heterogeneous lung disorder with high heritability, and there are several polymorphisms associated with asthma subtypes (Ullemar et al., 2016). It has been suggested that bronchiolitis either triggers asthma pathogenesis or has shared risk factors with asthma, or both (Edwards, Bartlett, Hussell, Openshaw, & Johnston, 2012). The observed connection between LRIs in infancy and subsequent asthma has prompted a search for shared genetic factors to unravel disease mechanisms and to identify predictive strategies and preventive therapies for both disorders (Larkin & Hartert, 2015). In this work, we aimed to identify novel genetic polymorphisms associated with severe bronchiolitis and common genetic determinants of bronchiolitis and asthma. To achieve this, we performed a hypothesis-free genome-wide association study (GWAS) of bronchiolitis, further characterized a promising GWAS candidate locus of bronchiolitis susceptibility, and conducted a meta-analysis to assess the role of an asthma-associated polymorphism in bronchiolitis.

2 Review of literature

2.1 Studying genetics of complex traits

2.1.1 Genetic variation in humans

Deoxyribonucleic acid (DNA) contains the information that organisms need to develop, live, and reproduce. In human genomes, DNA is assembled into 22 autosomal chromosome pairs and a pair of sex-determining chromosomes. The information written in the DNA sequence is encoded as alterations in the base-pair (bp) composition of the DNA. The first draft sequence covered >90% of the human genome and became available in the early 2000s; it refined the position and structure of genes (Lander et al., 2001; Venter et al., 2001). The sequence of the human genome guides the formation of approximately 20,500 protein-coding genes and the transcription of multiple nonprotein-coding sequences (1000 Genomes Project Consortium et al., 2015). Since the first sequences of human genomes became available, advances in genotyping and high-throughput sequencing, as well as usage of these methodologies in genetic studies conducted within and between human populations, have improved our understanding of the genetic diversity and complexity of the human genome (1000 Genomes Project Consortium et al., 2015; International HapMap Consortium, 2003; Lek et al., 2016). It is now estimated that only up to 3% of the human genome encodes proteins, although approximately 80% of the genome is actively transcribed (Dykes & Emanueli, 2016; Harrow et al., 2012). Several types of noncoding RNA (ncRNA) species comprise most of the transcribed genome, including long noncoding RNA (lncRNA), micro RNA (miRNA), small interfering RNA (siRNA), and piwi-interacting RNA (piRNA). Variations in nonprotein-coding sequence are implicated in the regulation of various biological processes, such as alterations in tissue and context-specific gene expression (Hrdlickova, de Almeida, Borek, & Withoff, 2014). Noncoding variants can affect gene expression; for example, by altering the function of enhancers or affecting miRNA binding sites (L. W. Barrett, Fletcher, & Wilton, 2012). Single-nucleotide polymorphisms (SNPs) are the most basic variations in the genomic sequence. SNPs are changes at a single bp position of the DNA sequence that give rise to multiple alleles at that position. Other types of genetic variants include structural variants such as insertions, deletions, inversions, translocations, substitutions, copy number variants (CNVs), and microsatellites (Feuk, Carson, &

17 Scherer, 2006; Sudmant et al., 2015). All of these are genomic rearrangements in which stretches of DNA are added, removed, inverted, moved to another genomic location, or tandemly repeated. Recent whole-genome sequencing (WGS) and whole-exome sequencing (WES) studies conducted on large populations with diverse ancestries have identified 88 million polymorphic sites in human genomes (1000 Genomes Project Consortium et al., 2015). More than 7 million variants, or approximately one variant for every eight bp, were identified across the exomes of >60,000 people (Lek et al., 2016). Of these, about 50% were singleton sites (i.e. sites at which a polymorphism occurred only once). A typical individual genome harbors on average 4 to 5 million polymorphic sites that differ from the reference sequence. The majority (>99%) of the variations are SNPs and short insertions/deletions. In addition, the average human genome contains about 1,000 large deletions, 160 CNVs, and approximately 1,500 inversions, which, taken together, encompass more sites than the SNPs in a single genome (1000 Genomes Project Consortium et al., 2015). In pooled human genome data, the majority of variants are rare. In the exome sequences of >60,000 individuals, 99% of the identified variants were rare, and occurred with frequencies of <1%. (Lek et al., 2016). Across whole-genome sequences of 2,054 individuals, about 75% of the identified variants were rare, with frequencies of <0.5%, and about 14% of variants occurred at low frequencies (0.5– 5%). However, within individual genomes, most of the variants were common and only a fraction (1–4%) of the identified variants occurred at frequencies of <0.5%. Variants shared among populations are mostly common variants (1000 Genomes Project Consortium et al., 2015). The occurrence of alleles and their frequencies for SNPs and other variants can vary among different ethnic groups or according to geographical location. Genetic diversity is highest in African populations and lower in other continents, which is consistent with out-of-Africa hypothesis of human origins (Tishkoff et al., 2009). Allele frequencies have been shaped by demographic events and evolutionary processes as a response to historical environmental pressures (Akey et al., 2004). Deleterious variants are expected to occur at lower allele frequencies than neutral variants due to purifying selection, which removes them from populations. Balancing selection and directional selection maintain advantageous alleles or drive them into fixation, whereas slightly deleterious alleles are lost or become more common by random genetic drift (Blekhman et al., 2008; Quintana-Murci, 2016). Small populations are prone to genetic drift, and it is common in populations that have undergone population bottlenecks (i.e., the population has undergone a

drastic reduction in size). In small populations, genetic drift may result in loss of rare alleles and decrease the gene pool. At the same time, rare deleterious alleles may become more common in a population by chance (Quintana-Murci, 2016). For example, the Finnish population history, which comprises founder effects, population isolates, and population bottlenecks, has enabled genetic drift to shape the gene pool (Peltonen, Palotie, & Lange, 2000). This is reflected in specific genetic structure including long stretches of linkage disequilibrium (LD) and regional genetic differences (Jakkula et al., 2008; Kerminen et al., 2017). Finnish demographic history has also led to the enrichment of rare deleterious variants, which constitute the Finnish disease heritage (Peltonen et al., 2000).

2.1.2 Role of genetic variation in disease

Many human diseases have a genetic background, in which an individual’s genetic make-up gives rise to or increases the likelihood of developing a certain disease. In monogenic disorders, a mutation in one gene is sufficient to cause the disease, whereas multifactorial diseases result from a combination of numerous variants in multiple genes and environmental factors (Smith & Lusis, 2002). Some present- day complex diseases may reflect a mismatch between adaptation to past selective pressures and the current environment (Blekhman et al., 2008). Variants across the allele frequency spectrum can contribute to disease risk (Gibson, 2012). In general, the effect size distribution of variants tends to change in relation to allele frequency, with rare alleles conferring larger effects than more common alleles. This inverse correlation has been observed for many traits, and it is in accordance with population genetic models of differential selective forces that act on rare and common variants (Bomba, Walter, & Soranzo, 2017). Monogenic rare diseases with Mendelian inheritance patterns arise as a result of a mutation in a single gene. The deleterious variants in such diseases are usually nonsynonymous changes located in exons. Nonsynonymous variants alter the protein-coding nucleotide sequence of DNA, which leads to damage in the protein structure or function (Chong et al., 2015). Currently, 3,466 single-gene defects are known to give rise to 5,075 monogenic disorders or phenotypes (Online Mendelian Inheritance in Man [OMIM], January 2018) (McKusick, 2007). For example, a mutation in survival of motor neuron 1 (SMN1) causes spinal muscular atrophy (SMA), in which degeneration of motor neurons in the spinal cord ultimately results in muscle atrophy. SMA is an autosomal recessive condition, which means that the onset of disease requires a defect in both copies of SMN1 (Wirth, 2000). Another

19 example is the Finnish disease heritage, which includes 36 disorders that are caused by a single-gene defect (Polvi et al., 2013). These are diseases that are specifically enriched in Finland due to demographic history, in which founder effects and population isolation occurred (Peltonen, Jalanko, & Varilo, 1999). Complex diseases are caused by multiple factors, including environmental risk factors, lifestyle factors, genetics, and epigenetics (Bookman et al., 2011). Environmental factors may modify DNA methylation patterns and chromatin states, and sometimes these epigenetic changes are transmitted to the offspring (Feil & Fraga, 2012). Complex phenotypes result from interactions among these various genetic and environmental factors, each of which typically have a small influence on the outcome (Figure 1). Most of the common human diseases, such as type 2 diabetes, cardiovascular disease, many neurological disorders, and cancer, have a complex polygenic background (Bookman et al., 2011). Complex disorders do not have clear inheritance patterns but may cluster in families. Typically, heritability of a multifactorial common disease is 30–60% (Price, Spencer, & Donnelly, 2015). Many genes that cause Mendelian diseases have been successfully localized with linkage analysis (Jorde, 2000). Linkage mapping identifies genetic markers that cosegregate with a given trait in different family configurations (Baron, 2001).

Fig. 1. Complex phenotypes are influenced by an interplay of environmental, genetic, and epigenetic factors that individually may have modest effects (modified from Oksenberg & Baranzini, 2010).

Application of linkage analysis to complex traits has had some success, such as identifying risk variants for Mendelian subforms of common diseases (e.g., BRCA1, DNA repair associated (BRCA1) and BRCA2, DNA repair associated (BRCA2) in breast cancer). Linkage analysis has also identified some genetic variants with large effects, such as apolipoprotein E (APOE) in Alzheimer disease (Baron, 2001; Jorde, 2000). In general, linkage studies of complex disorders have been less successful due to the heterogeneous nature of these disorders and the lack of power to detect relatively common variants with modest effects (Silverman & Palmer, 2000). Association studies that investigated the variations of candidate gene polymorphisms in case-control populations of usually unrelated samples were an early alternative to linkage mapping to study the genetic basis of complex disease (Hirschhorn & Daly, 2005; Silverman & Palmer, 2000). Candidate gene association studies focused on certain genes based on plausible biology in view of the phenotype under consideration (Daly & Day, 2001). Compared to linkage studies, association studies had more power to detect risk alleles with lower effects, but they were prone to confounding by population structure and nonreplicable results (Price et al., 2015). GWASs, which screen a genome-wide array of SNPs for associations with a studied trait in a hypothesis-free manner, became available in the mid-2000s (Hirschhorn & Daly, 2005).

2.1.3 Genome-wide association studies

Improvements in the human genome map and genotyping arrays and technologies facilitated the development of GWASs. GWASs test hundreds of thousands to millions of polymorphisms for association with a particular phenotype (Evangelou & Ioannidis, 2013).

Capturing common variation

The detection of causal variants in a GWAS is largely based on LD (i.e., correlation between nearby variants resulting from population history and evolutionary forces) (International HapMap Consortium, 2005). Genotyped variants capture the underlying unobserved causal variants if they are in LD with the observed variants (Visscher et al., 2017). Genotyping of approximately 500,000 SNPs captures about 80% of common variation (minor allele frequency [MAF] >5%) in European populations (Price et al., 2015). The statistical power to identify disease-associated

21 regions is governed by the sample size of the study, the effect size and frequency distributions of causal variants, and LD between genetic markers and causal variants (Hong & Park, 2012). Accurate case definition is also important to avoid excess genetic and environmental heterogeneity, which may reduce the power to detect an effect (Zondervan & Cardon, 2007). However, in most phenotypes, adequate sample size is statistically more important than exact phenotype definition, and large sample size may balance the noise from less-precise phenotype definitions (Price et al., 2015). Genotype data from publicly available population controls can be used to increase the statistical power of a GWAS (Zhuang et al., 2010). Genotype imputation (i.e., the statistical prediction of unobserved genotypes) can be used to increase the resolution and power of a GWAS (B. Howie, Fuchsberger, Stephens, Marchini, & Abecasis, 2012). Imputation methods leverage a reference panel of densely characterized individuals. Haplotype segments can be extrapolated from the reference panel into a more sparsely genotyped study population that has been characterized at a subset of the reference panel samples (B. Howie, Marchini, & Stephens, 2011). Imputation increases the amount of SNPs to be analyzed and enables identification of trait-associated low-frequency variants that would not otherwise be detected. Another advantage of imputation is that it facilitates the combination of genome-wide datasets that have been genotyped in different platforms and also aids meta-analysis of such data (B. N. Howie, Donnelly, & Marchini, 2009).

Study design and quality control

Careful study design will help to identify true associations and replicate the findings of a discovery study in independent replication cohorts. The influence of population stratification increases in proportion to the number of samples (Patterson, Price, & Reich, 2006). In unadjusted analysis, the observed allele frequency differences between cases and controls may reflect systematic ancestry differences rather than true associations (Turner et al., 2011). In large GWASs, stratification may be controlled for at the study design level and at the analysis stage. Careful selection of cases and controls can minimize unwanted population structure (Zondervan & Cardon, 2007). Controls should come from the same population from which the cases were collected, and population stratification can be reduced by matching the cases and controls for ethnicity (Turner et al., 2011). Matching according to gender may be helpful if there are gender differences in

disease incidence. Over-matching should be avoided since it will eventually lead to loss of power by having to account for all matched factors in statistical analysis (Zondervan & Cardon, 2007). One way to adjust for underlying population structure in an admixed GWAS population is to incorporate principal components into the association analysis. Principal components summarize the multidimensional data into the most important axes of variation (Peres-Neto, Jackson, & Somers, 2005). Quality control (QC) of the data is a vital step to avoid spurious association signals. For example, an uneven frequency distribution of missing genotype calls across samples and relatives in data can lead to false-positive associations (Laurie et al., 2010). These issues can be avoided by filtering out SNPs with poor call or genotyping rate, deviation from Hardy–Weinberg equilibrium (HWE), and MAF below a certain threshold (often 1%) (Miyagawa et al., 2008). Related individuals, clear population outliers, and samples with large portions of missing genotypes can be traced from genome-wide data and excluded (Laurie et al., 2010). Replication studies aim to demonstrate the validity of an association and are necessary for both candidate gene studies and GWASs (NCI-NHGRI Working Group on Replication in Association Studies et al., 2007). Variants associated with a trait in a GWAS and replication study are considered robust phenotype-associated loci (Marian, 2012). Overall, GWASs that have reached genome-wide significance in the discovery phase have been highly replicable across different studies. To date, >10,000 variants have been strongly associated (p < 10−8) with several complex traits such as cardiovascular diseases, immune system diseases, hematological measurements, and lipid or lipoprotein measurements (MacArthur et al., 2016). GWAS analyses have shed light on the genetic architecture of complex diseases in humans, given insight into biological processes of diseases, and lead to discoveries of novel therapeutic strategies (Visscher et al., 2017).

Meta-analysis

Meta-analyses to produce a summarized result over multiple studies can identify further associated risk loci (Evangelou & Ioannidis, 2013). For example, a recent trans-ethnic meta-analysis identified novel loci and replicated several previous associations between genetic variants and systemic sclerosis (Terao et al., 2017). Meta-analysis of GWASs may be performed under a fixed effects or random effects model. Fixed effects meta-analysis assumes that risk variants across data sets have similar effect sizes, and it is generally used in the discovery phase and prioritization

23 of disease-associated variants. Random effects meta-analysis may be used to estimate generalizability and the average effect of an associated variant in different populations (Evangelou & Ioannidis, 2013).

Architecture of complex disease and missing heritability

GWASs have demonstrated that complex diseases are highly polygenic and that most complex disease–associated variants have small individual effects (Visscher et al., 2017). Together, the associated variants explain varying fractions of the estimated heritability of a particular disease (Bomba et al., 2017). The rest of the phenotypic variation between individuals estimated to be due to genetic differences (i.e., “missing heritability”) may be explained by several factors including unobserved rare and common variants underlying a given disease, as well as epigenetics, gene–gene interactions (epistasis), and gene–environment interactions (Manolio et al., 2009). By design, a GWAS does not detect rare variants due to lack of LD structure among common variants in the genotyping chip and rare (MAF < 1%) causal variants with modest effect on the trait (Q. Wang, Lu, & Zhao, 2015). Genotype imputation usually leads to identification of some of these rare variants. Other means to access rare and low-frequency variants include custom genotyping arrays and WGS or WES in a population dataset or family cohorts (Bomba et al., 2017; Q. Wang et al., 2015). Little is known about the impact of rare variants on complex disease. Interestingly, autism spectrum disorders may have more causal variants from the rare spectrum compared to other complex diseases (Bomba et al., 2017). The heritability explained by common variants will likely increase in future years, when larger GWASs and meta-analyses will capture more variants with even smaller effects (Manolio et al., 2009). In 2008, for example, 40 genome-wide significant variants were known to be associated with height, and together explained approximately 5% of heritability. By 2015, the number of associated variants was about 700, and the explained heritability was >20% (Visscher et al., 2017). Linear mixed models to estimate the heritability explained by common variants have shown that common SNPs typically explain about half of the narrow sense (additive) heritability estimated from twin studies. These models also include variants that individually do not reach genome-wide significance (Price et al., 2015). However, defining the total number of variants with nonzero effects is challenging. A recent omnigenic model suggested that many complex diseases are driven largely by the activity in disease-relevant tissues of multiple small-effect regulatory

variants located in peripheral genes without direct relevance to disease. The contribution of the regulatory variants to disease would then be transmitted through cell-specific regulatory networks to a more limited number of core genes with direct effects (E. A. Boyle, Li, & Pritchard, 2017). Complex diseases appear to be highly pleiotropic; that is, distinct phenotypes share a proportion of causal variants (Gratten & Visscher, 2016). The same genetic variants can be significantly associated with different diseases in individual populations. For example, hundreds of loci have been found to be associated with autoimmune disease in the last 5 years, and cross-disease studies of different auto immune disorders have led to the identification of further associated loci (Visscher et al., 2017). Polygenic risk scores (PRS), constructed on the basis of large GWASs or meta-analyses that combine results across multiple GWASs, have been useful in examining a pleiotropic basis for related traits. The first successful application of a PRS analysis was using a PRS based on the GWAS of schizophrenia to study the risk of several other disorders; the PRS was associated with the risk of bipolar disorder but not with a number of nonpsychiatric diseases (International Schizophrenia Consortium et al., 2009). In addition to demonstrating pleiotropy, PRSs may be used to increase the accuracy of predictions of disease risk, and to classify groups as having a minor or major risk for disease (Visscher et al., 2017). For example, a PRS constructed on the basis of cardiovascular disease (CVD)– associated variants was observed to be a useful in prediction CVD prevalence within a group of individuals with familial hypercholesterolemia (Paquette et al., 2017).

2.1.4 Interpreting GWAS signals

Common disease-associated SNPs in exons may exert their effects by, for example, causing a premature stop codon, changing a codon sequence, shifting a reading frame, or affecting a splice site (Lek et al., 2016; Pagani & Baralle, 2004). As a consequence, the stability or function of a protein or mRNA may be altered. The majority (approximately 80%) of the associated variants identified by GWASs are located in noncoding regions of the genome (M. J. Li, Yan, Sham, & Wang, 2015). Unraveling the path from such association signals to plausible biology and function can be challenging. In addition, resolving the causal variants or DNA elements through which GWAS SNPs influence the trait is not straightforward, due to the complex LD structure between tagging SNPs and causal variants (Zhu et al., 2016). A growing body of genomic and other data, such as tissue-specific gene expression

25 data, transcription factor binding sites, active enhancers, knowledge of conservation across genomes, and circulating levels of metabolites in genetically diverse individuals, may shed light on the role of disease-associated variants identified by GWASs (Elkon & Agami, 2017). Comparisons of GWAS data with regulatory data have shown that variants identified through GWASs are enriched for regulatory elements (Lowe & Reddy, 2015; Tak & Farnham, 2015). Approximately two thirds of disease-associated variants identified through GWASs map to functional regulatory elements listed in Encyclopedia of DNA Elements (ENCODE) (Civelek & Lusis, 2013). One way to interpret and prioritize disease-associated variants from GWAS is to integrate GWAS signals with expression quantitative trait loci (eQTLs; Figure 2) (Albert & Kruglyak, 2015; Hormozdiari et al., 2016). Currently available eQTL resources or databases include the portal for the genotype–tissue expression (GTEx) project, which has gene expression data across about 50 tissues; the blood eQTL browser; and the NCBI eQTL browser, which contains cis-eQTLs from liver, brain, and lymphoblastoid cells (Tak & Farnham, 2015). Combining disease variants from GWAS with eQTL data may help identify genes that are influenced by disease- associated genetic variation. Further, combining disease SNPs with tissue-specific eQTLs can pinpoint the particular tissue or cell type in which disease variants play a role (Pai, Pritchard, & Gilad, 2015). There are several advantages to investigating GWASs and eQTL studies in combination rather than directly comparing gene expression alterations with disease. One advantage is the independence of GWAS SNPs from environmental confounders, such as medication and nutrition (Hauberg et al., 2017).

Fig. 2. eQTLs are genetic loci that influence expression levels of mRNA transcripts. Cis eQTLs typically regulate expression of nearby genes and cause allele-specific expression. Trans eQTLs affect overall mRNA levels of distal genes. The same locus can affect mRNA expression in cis and in trans (adapted from Nica & Dermitzakis, 2013).

Several methods to combine GWASs and eQTL studies have been developed, some of which can be performed on summary statistics. One such approach is Mendelian randomization, which uses genetic variants as instrumental variables to assess if an exposure (e.g., protein level or gene expression) is causally related to an outcome (e.g., disease) (Zhu et al., 2016). Protein QTLs (pQTLs) are genetic loci that influence protein levels of a particular gene (Albert & Kruglyak, 2015). Human pQTLs are not as widely studied as eQTLs. Approximately 50% of the pQTLs in the plasma proteome have an underlying eQTL (Suhre et al., 2017). Testing GWAS SNPs for pQTL colocalization can be used as a further means to identify and prioritize potential causal variants and to explore how genetic variants might regulate disease- associated gene expression (Chick et al., 2016).

2.2 Bronchiolitis

2.2.1 Definition

Bronchiolitis is a common lower respiratory tract infection (LRI) in infants (Drysdale, Green, & Sande, 2016). It is typically caused by common respiratory viruses and leads to inflammation of bronchioles and adjacent tissue (Jartti & Gern, 2017). Bronchiolitis is generally defined as the first wheezing episode or expiratory breathing difficulty in children <2 years of age, but the upper age limit varies from 6 to 24 months among studies (Korppi, Koponen, & Nuolivirta, 2012). Other clinical characteristics typical of bronchiolitis include lower respiratory symptoms such as dry cough, tachypnea, auscultative crackling sounds, and chest retractions (Meissner, 2016).

2.2.2 Treatment and prevention

The current recommendation for treatment and management of bronchiolitis is supportive care, which includes assisted hydration and oxygen supply when reduction in fluid intake or oxygen saturation occurs (Ralston et al., 2014). There have been many efforts to treat bronchiolitis with interventions including bronchodilators, corticosteroids, and hypertonic saline (Petrarca, Jacinto, & Nenna, 2017). However, there is limited evidence for the comprehensive effectiveness of medical interventions in bronchiolitis treatment and thus, routine use of these

27 therapies is not recommended (Schweitzer & Justice, 2018). Some infants with RV induced bronchiolitis appear to benefit from corticosteroid treatment (Jartti & Gern, 2017). Other subgroups of children with viral LRI may also benefit from some of these treatments; therefore, more studies are needed (Drysdale et al., 2016). Palivizumab is an effective but relatively expensive monoclonal antibody targeted against the RSV fusion (F) glycoprotein. Immunoprophylaxis with palivizumab can be used to prevent severe RSV bronchiolitis in high risk groups including infants with heart disease or chronic lung disease (Ralston et al., 2014; The IMpact-RSV Study Group, 1998). There is no known prevention for RV induced bronchiolitis (Jacobs, Lamson, St. George & Walsha, 2013). Although palivizumab is currently the only licensed therapy against RSV infection, new promising vaccines and antiviral therapies are constantly being developed (Drysdale et al., 2016; Graham, 2017).

2.2.3 Epidemiology

The prevalence of bronchiolitis is approximately 20% during the first 2 years of life (Jartti & Gern, 2017). Bronchiolitis almost always results from infection with common respiratory viruses. The most common causative pathogen is respiratory syncytial virus (RSV), which is implicated in 50–80% of children hospitalized for bronchiolitis (Zorc & Hall, 2010). Other causative agents, from most common to less prevalent, include rhinovirus (RV), parainfluenza virus (PIV), metapneumovirus, adenovirus, coronavirus, and influenza A and B viruses. Approximately 5–25% of viral LRIs are caused by RV or PIV (Meissner, 2016). The proportions of causal agents fluctuate annually and seasonally, and infections are usually more common during cold or rainy periods. However, RV-induced bronchiolitis can occur throughout the year (Paul et al., 2017). The general clinical presentation of patients with bronchiolitis due to distinct viral etiologies is relatively similar (Singh, Moore, Gern, Lemanske, & Hartert, 2007). However, some differences in the clinical characteristics of bronchiolitis patients with different causal viruses, mainly RSV and RV, have been recognized. Children with RV-induced wheezing tend to be older than those with RSV disease and are also more likely to have eosinophilia and atopic dermatitis during infection (Korppi, Kotaniemi-Syrjanen, Waris, Vainionpaa, & Reijonen, 2004). Some studies have linked RSV infection with more-severe illness, and a higher viral load has been associated with a longer duration of hospitalization (Hasegawa et al., 2015). Coinfection with RSV and RV may result in more-severe disease compared to

bronchiolitis induced by a single virus type (Singh et al., 2007). Finally, studies of therapeutic strategies for bronchiolitis have noted differences in responses among patients with distinct viral etiologies. For example, some wheezing infants with RV LRI respond to treatment with systemic corticosteroids by showing a reduced risk for a recurrent wheezing phenotype after LRI (Lehtinen et al., 2007). Passive immunization with the monoclonal antibody palivizumab reduces the severity of RSV-induced LRIs. Further, immunoprophylaxis may reduce the risk of developing recurrent wheezing after RSV infection has passed (Blanken et al., 2013). In most cases, children infected with common respiratory viruses have mild symptoms, such as runny nose or cough, that resemble symptoms of a common cold and resolve uneventfully (Tregoning & Schwarze, 2010). However, in some cases viral infection causes severe acute bronchiolitis that necessitates hospital admission. RSV bronchiolitis is a frequent cause of hospitalization of babies and young children worldwide (Scheltema et al., 2017). While >99% of mortalities due to severe RSV bronchiolitis occur in developing countries, severe bronchiolitis is also a major health care burden in developed countries (Shi et al., 2017). For example, virus-induced bronchiolitis is the most common cause of hospital admission among children <12 months of age living in the United States (Florin et al., 2017). In general, about 2–3% of infants are admitted to hospital with severe bronchiolitis during their first year of life (Smyth & Openshaw, 2006). Several risk factors are known to make children susceptible to a severe course of bronchiolitis. The most important risk factors include chronic lung disease of prematurity, immunodeficiency, and congenital heart disease (Florin et al., 2017). Chronic lung diseases known to confer a large risk for severe bronchiolitis are bronchopulmonary dysplasia and cystic fibrosis (Ramet, Korppi, & Hallman, 2011). Very young age (<3 months) and extremely preterm birth (gestation at <29 weeks) also predispose infants to severe bronchiolitis that may present with apnea (Florin et al., 2017; Meissner, 2016). Several studies have suggested that male sex may predispose children to more-severe viral bronchiolitis. Environmental factors linked to a risk of severe bronchiolitis include exposure to cigarette smoke, maternal tobacco use at the time of pregnancy, populous living circumstances, undernourishment, and older sisters or brothers in the family (Ramet et al., 2011). The proportion of hospitalizations due to severe bronchiolitis is greatest among previously healthy children (Meissner, 2016).

29 2.2.4 Immune response to respiratory viruses

The characteristics of the causal virus as well as the host immune response are thought to be involved in bronchiolitis pathogenesis and variations in clinical disease phenotypes (Figure 3) (Singh et al., 2007). Viral infection and replication can cause injury by direct effects to the epithelium, such as necrosis. Immune responses are also believed to influence disease severity, and inflammatory cytokines are thought to be involved in the tissue damage that occurs during viral LRI (Tregoning & Schwarze, 2010). Respiratory viruses invade the airway epithelium via interactions with particular receptors and begin replicating in the epithelial cells (Jartti & Gern, 2017).

Fig. 3. Pathogenesis of respiratory syncytial virus (RSV) or rhinovirus (RV) induced bronchiolitis. Bronchiolitis is caused by direct viral damage and immune mediated effects. RSV is thought to inflict more direct damage to the epithelium than RV. Interaction of viral factors, genetics, environmental factors, and congenital or immunological factors contribute to the outcome and severity of bronchiolitis (based on Troy & Bosco, 2016; Tregoning & Schwarze, 2010; Jartti & Gern, 2017).

Pattern recognition receptors (PRRs) are crucial mediators of the innate immune system that recognize viral infection by detecting conserved pathogen-associated molecular patterns (PAMPs). PRRs, such as Toll-like receptors (TLRs), nucleotide- binding oligomerization domain (NOD)-like receptors (NLRs), and C-type lectin receptors (CLRs), are expressed on epithelial surfaces and innate immune cells (Troy & Bosco, 2016). Early innate immune responses induced by RSV or RV occur shortly after viral infection of the epithelium (Jartti & Gern, 2017). PRR recognition of viral infection activates transcription factor signaling cascades that result in expression of inflammatory interferons and cytokines (Troy and Bosco, 2016). Both RV and RSV infections induce the release of chemokines, such as C- C motif chemokine 5 (CCL5) and interlukin-8 (IL-8/CXCL8), and cytokines, including IL1, IL6, and tumor necrosis factor (TNF) (Jartti & Gern, 2017). As a consequence, innate cells such as natural killer cells (NK cells), dendritic cells (DCs), monocytes, and macrophages are recruited to sites of infection. Innate immune responses stimulate pathways that activate adaptive immune responses, which will eventually lead to viral clearance and resolution of the infection (Sabroe et al., 2002).

Respiratory syncytial virus and immune response

RSV is a single-stranded negative-sense RNA virus of the Paramyxoviridae family, and it has two major antigenic subgroups, A and B. The RSV genome gives rise to 11 mRNAs that encode transmembrane surface proteins (G, F, and SH), structural virus proteins, and nonstructural protein products (Domachowske & Rosenberg, 1999). RSV F protein plays a role in membrane fusion, which enables viral entry into cells and mediates host cell–cell fusion (Collins & Graham, 2007). Fusion of infected cells with adjacent cells results in the formation of multi nucleated syncytia. G-protein is the other major surface glycoprotein of RSV, and it mediates viral attachment to cilial airway cells (McLellan, Ray, & Peeples, 2013). RSV G protein uses CX3C chemokine receptor 1 (CX3CR1) to infect epithelial cells, whereas F protein can interact with intercellular adhesion molecule 1 (ICAM-1), toll-like receptor 4 (TLR4), and nucleolin (Johnson et al., 2015; Lambert, Sagfors, Openshaw, & Culley, 2014). RSV F and G proteins are thought to be the major targets of the antibody response. The monoclonal antibody palivizumab was designed against the F protein of RSV and functions by inhibiting viral attachment (Jartti & Gern, 2017). The less-variable F protein is also a target of ongoing RSV vaccine development (Graham, 2017).

31 RSV infection inflicts damage upon the airway epithelium, including apoptosis and necrosis of infected cells, mucus discharge, and infiltration of neutrophils and lymphocytes, which ultimately results in airway obstruction (Troy & Bosco, 2016). RSV directly damages the lung epithelium, and high viral load (i.e., high amounts of RSV genomes) is associated with more-severe illness (Hasegawa et al., 2015; Skjerven et al., 2016). In addition to the effect of viral replication on epithelial injury, the inflammatory response to RSV infection has also been linked to epithelial damage (Openshaw, 2005). DCs are crucial in mobilizing the adaptive immune response to RSV infection. These immune cells migrate from the infected airways into lungs and communicate with CD4+ and CD8+ T cells that are specific to RSV infection (Schmidt & Varga, 2017). Activated CD4+ and CD8+ T lymphocytes can be classified into two sub groups on the basis of the cytokines that they produce. T helper 1 (Th1) responses are characterized by interferon γ (IFNγ), IL1, TNFα, and IL12 production, whereas the T helper 2 (Th2) cytokine signature includes interleukin 4 (IL4), IL5, IL10, and IL13 (R. A. Pinto, Arredondo, Bono, Gaggero, & Diaz, 2006; Russell, Unger, Walton, & Schwarze, 2017). An imbalance in Th1/Th2-mediated cytokine profiles may be associated with RSV disease severity. Proinflammatory Th1 cytokines including IFNγ are involved in cell-mediated immunity and are vital to RSV clearance and resolving infection, whereas the Th2 response plays a role in the development of humoral immunity, including production of IgE and eosinophil responses (Russell et al., 2017). A shift toward Th2-mediated cytokine production may be linked to a more-severe course of RSV infection, asthma, and allergy (Hoebee et al., 2003; Russell et al., 2017).

Rhinovirus and immune response

RV is a single-stranded, positive-sense RNA virus that belongs to the Picornaviridae family. There are >160 known RV strains, which are categorized into three groups based on DNA sequence: RV-A, RV-B, and RV-C (Troy & Bosco, 2016). RVs interact with the airway epithelium through specific receptors. RV-B and most RV-As bind the ICAM-1 receptor, although some RV-As attach to low density lipoprotein receptor (LDLR). RV-C binds to cadherin-related family member 3 receptor (CDHR3) (Bochkov & Gern, 2016). RV-As and RV-Cs are associated with more-severe illness in infants with bronchiolitis than RV-Bs (Vandini, Calamelli, Faldella, & Lanari, 2017). Outside the RSV season, RV

infection is the major cause of hospitalizations due to bronchiolitis (Jartti & Gern, 2017). RV infection is detected by PRRs including TLR2, TLR3, and TLR7, and RIG- I. The detection of RV-related PAMPs activates signaling cascades that lead to the release of multiple IFNs and inflammatory cytokines (Jartti & Gern, 2017). RV impairs the function of the epithelial barrier during infection, which potentially increases the inflammatory response. This may lead to airway immune injury and obstruction and bronchial hyperresponsiveness (Vandini et al., 2017). RV elicits a serotype-specific humoral immune response that has hampered vaccine development (Jartti & Gern, 2017).

2.2.5 Bronchiolitis and asthma

Children who have undergone severe viral bronchiolitis are at risk for recurrent wheeze and asthma. Up to 48% of infants who are admitted to hospital with RSV- induced bronchiolitis become asthmatic in childhood (Larkin & Hartert, 2015). Viral LRI in infancy may influence the Th1/Th2 balance toward a Th2-weighted immune response, which elevates the risk of asthma and allergy (Ramet et al., 2011). Asthma is a chronic airway disease characterized by symptoms such as airway obstruction, wheezing, and shortness of breath (Edwards et al., 2012). It is a heterogeneous disorder that is affected by a combination of multiple environmental and host factors. These include host genetics, virus-induced wheezing, use of antibiotics, and increased hygiene (Jartti & Gern, 2017). RV infection is the major triggering agent for asthma exacerbations, especially in children with atopy and allergic sensitization (Troy & Bosco, 2016). RV-induced LRI in infancy appears to be associated with airway hyperresponsiveness and an obstructive pattern of lung function (Ramet et al., 2011). Severe RSV bronchiolitis in infancy is also associated with reduced lung function, recurrent wheeze, and asthma. It has been suggested that viral infections activate the causal pathways that lead to asthma after viral LRI. There may be a common causal background (e.g., similar host genetics) that makes individuals susceptible to both viral bronchiolitis and asthma (Larkin & Hartert, 2015).

2.3 Genetic factors of bronchiolitis susceptibility

Most children hospitalized for severe bronchiolitis do not have known or potential risk factors that would predispose them to the condition. Bronchial

33 hyperresponsiveness is one factor that might explain the severe course of virus- induced LRI among otherwise healthy children. Furthermore, genetic factors are likely responsible for some of the variation in disease severity within otherwise healthy population (Chawes, Poorisrisak, Johnston, & Bisgaard, 2012; Meissner, 2016). A Danish twin study observed a higher similarity of hospitalization patterns between identical twins with bronchiolitis, and the heritable proportion of bronchiolitis was estimated to be 22% (Thomsen et al., 2008). Another study conducted with the Dutch twin register yielded a higher heritability estimate of 30– 40% (Kimman, 2001). Bronchiolitis and asthma appear to have some shared heritability, supporting the idea of common genetic determinants that underlie both diseases (Thomsen et al., 2008).

2.3.1 Genetic variants associated with bronchiolitis in previous studies

Studies to identify bronchiolitis-associated polymorphisms have usually assessed associations between RSV bronchiolitis and variants in candidate genes of plausible function (Larkin & Hartert, 2015). The innate immune response and expression characteristics of inflammatory cytokines are believed to affect the course and severity of RSV infection. Thus, variants in immune-related genes have been of particular interest in candidate gene association studies of bronchiolitis (Miyairi & DeVincenzo, 2008). Table 1 presents an example compilation of previous genetic association studies of bronchiolitis, including some of the most comprehensive genetic studies, as well as a couple of smaller studies. Polymorphisms associated with RSV bronchiolitis in previous studies include variants in interferon alpha 5 (IFNA5), surfactant protein D (SFTPD), many ILs, and genes encoding receptors such as TLR4, C-C chemokine receptor type 5 (CCR5), and CX3CR1 (Amanatidou, Apostolakis, & Spandidos, 2009; Lahti et al., 2002; Miyairi & DeVincenzo, 2008). One of the largest genetic studies of bronchiolitis evaluated 384 polymorphisms in 220 genes classified into five groups: airway mucosal response, innate immunity, adaptive immunity, chemotaxis and known allergic asthma genes (Janssen et al., 2007). There was enrichment of associated polymorphisms in the innate immunity gene group. Genetic studies of RV-induced LRI have found associated polymorphisms in genes including IL10, and GSDMB, and IKZF3 (Caliskan et al., 2013; Helminen et al., 2008).

Table 1. Examples of previous genetic association studies of bronchiolitis.

Primary phenotype Study population Analyzed polymorphisms Associated genes Study RSV bronchiolitis 470 cases (97% <12 months) and 1,008 384 SNPs within 220 candidate genes VDR*, IFNA5, JUN, Janssen et al., population controls from The Netherlands NOS2*, FCER1A 2007 RSV bronchiolitis 296 cases (<24 months) and 113 heatlhy Four SNPs in TLR4, JUN, and VDR VDR* Kresfelder et al., controls from South Africa 2011 RSV bronchiolitis 782 cases (<12 months) and 1,045 113 SNPs in 5q31 cytokine cluster across IL13–IL4 haplotype, IL4*, Forton et al., 2009 population controls from Europe IL3, GMCSF, P4HA2, RIL, SLC22A4, SLC22A5* SLC22A5, IRF-1, IL5, RAD50, IL13, and IL4 RSV bronchiolitis 580 cases (<12 months) and 580 controls Five CCR5 promoter polymorphisms Association for 2 SNPs in Hull et al., 2003 from England CCR5* RSV bronchiolitis 207 cases and 774 population controls Three IL4/IL4Rα polymorphisms Association for 2 SNPs in Hoebee et al., from The Netherlands IL4*/IL4Rα* 2003 Viral bronchiolitis 247 cases and 190 healthy controls from Three ORMDL3 polymorphisms Association detected for 1 Liu et al., 2015 (RSV, RV or other) China ORMDL3* SNP Viral bronchiolitis Two cohorts with 700 children from USA Five SNPs in 17q21 region accross IKZF3*, ZPBP2, and Çalışkan et al., (RV and RSV) and Denmark IKZF3, ZPBP2, and GSDMB GSDMB* with RV 2013 bronchiolitis Viral bronchiolitis, 181 cases (>24 months) and 536 healthy SNPs in TLR4, TLR2, TLR9, VDR, NOS2, CCL5* with RSV; NOS2* Alvarez et al., (RSV, RV) controls from Brazil IFNA5, JUN, and CCL5 with RV bronchiolitis 2018 *Genes also suggested to play a role in asthma or post-bronchiolitis wheezing phenotypes, as in Larkin & Hartert, 2015; and Jartti & Gern, 2017. 35

Identified associations have not been consistently replicated in follow-up studies. The reasons for this may include small sample sizes, differences in bronchiolitis definitions, interaction with environment, and diverse genetic architectures among populations (Miyairi & DeVincenzo, 2008). A few of the reported associations have been replicated across studies and provided insight into underlying disease mechanism. For example, several studies have found that the haplotype encompassing IL13–IL4 sequences associated with a risk of severe RSV bronchiolitis (Forton et al., 2009; Puthothu, Krueger, Forster, & Heinzmann, 2006). The risk locus is linked to increased IL13 production, and the same haplotype is also implicated in atopic sensitization (Forton et al., 2009). An association between bronchiolitis susceptibility and a variant in VDR has also been seen in several studies. Functional studies indicate that the potential mechanism of the variant in bronchiolitis susceptibility is related to the immunosuppressive function of vitamin D (Janssen et al., 2007; Kresfelder, Janssen, Bont, Pretorius, & Venter, 2011; Stoppelenburg, von Hegedus, Huis in't Veld, Bont, & Boes, 2014). Both RSV and RV may reduce VDR expression in cells of bronchial epithelium (Telcian et al., 2017). RSV infection is recognized by TLR4, and TLR4 polymorphisms have been connected to RSV bronchiolitis severity in several studies (Mandelberg et al., 2005; Puthothu, Forster, Heinzmann, & Krueger, 2006; Tal et al., 2004). However, not all studies have found association (Douville et al., 2010; Lofgren, Marttila, Renko, Ramet, & Hallman, 2010). Moreover, a recent meta-analysis of TLR4 polymorphisms in European and African samples did not detect associations (Zhou et al., 2016). Polymorphism in SPD, IL8, and IL10 have been associated with bronchiolitis. SPD is involved in immunomodulatory functions and may be important in the clearance of RSV infection. SPD and IL8 polymorphisms have been found to be associated with severe bronchiolitis in diverse ethnic groups (Ampuero, Luchsinger, Tapia, Palomino, & Larranaga, 2011; Hull, Thomson, & Kwiatkowski, 2000; Lahti et al., 2002; L. A. Pinto et al., 2017; Puthothu et al., 2007; M. Tian, Zhao, Wen, Shi, & Chen, 2007). However, the effect of the detected association in SPD was opposite among infants from Finland and South America (Ampuero et al., 2011; Lahti et al., 2002). Finally, several studies have found that IL10 variants are associated with severe RSV and RV bronchiolitis (Helminen et al., 2008; Hoebee et al., 2004; Wilson et al., 2005).

2.3.2 Genetic variants associated with bronchiolitis and asthma

The relationship between viral bronchiolitis and asthma or recurrent wheezing phenotypes implies that there may be shared genetic determinants underlying both diseases. Heritability estimates for asthma vary from 35% to 80%, and a family history of asthma is the greatest risk factor associated with the disorder (Mersha, 2015). Unlike bronchiolitis, asthma has been studied in several large-scale and hypothesis-free genome-wide studies, and several loci, including variants near ORMDL sphingolipid biosynthesis regulator 3 (ORMDL3) and gasdermin B (GSDMB), have been associated across studies and ethnic groups (Yan et al., 2017). Currently, approximately 200 loci have been identified that show significant (p < 5 × 10−8) associations with phenotypes under the “asthma” definition (which includes asthma related traits) in the NHGRI-EBI GWAS Catalog (December 2017). Using a more-specific asthma phenotype, 39 genetic loci were associated with a risk of asthma at a genome-wide significance level (p < 5 × 10−8) (Vicente, Revez, & Ferreira, 2017). Several genetic polymorphisms have been associated with both bronchiolitis and asthma. Genes suggested to play a role in both RSV bronchiolitis and asthma include VDR, TLR4, and ADAM metallopeptidase domain 33 (ADAM33) (Larkin & Hartert, 2015). Regarding RV bronchiolitis, 17q21 polymorphisms are associated with both RV wheezing phenotype and subsequent asthma (Caliskan et al., 2013). CDHR3 encodes a cadherin-family transmembrane protein that is expressed by the epithelial cells in lung tissue. CDHR3 variant rs6967330 was associated with asthma in a phenotype-specific GWAS in which asthma in childhood was characterized by recurrent exacerbations (Bonnelykke et al., 2014). This finding was supported by another recent study in which CDHR3 variant rs6959584 was associated with a similar asthma phenotype (r2 = 0.72 with 6967330) (Pickrell et al., 2016). Interestingly, the CDHR3 receptor mediates RV-C entry to the cell (Bochkov et al., 2015). Thus, it may be a plausible candidate gene for bronchiolitis susceptibility and a potential shared genetic determinant of both asthma and viral LRI.

3 Aims of the study

The primary objective of this work was to shed light into the inadequately characterized genetic background of viral bronchiolitis in childhood. Another aim was to characterize potential shared genetic factors of bronchiolitis and asthma. The specific aims were:

1. To perform a hypothesis-free genome-wide association study (GWAS) to identify genetic loci associated with severe childhood bronchiolitis. 2. To determine if results of a GWAS subpopulation would overlap with loci known to regulate gene and protein expression. Further, to study the identified loci for association in independent bronchiolitis case-control populations, and verify the reported effect on expression in primary human immune cells. 3. To investigate if a previously asthma-associated variant in CDHR3 is associated with bronchiolitis in a meta-analysis across five European case- control populations.

4 Materials and methods

An overview of the substudies in this thesis is presented in Table 2.

4.1 Ethical considerations

The studies were approved by ethical boards of the participating institutes, namely the ethics committees of Tampere University Hospital and Kuopio University Hospital, Finland; the regional ethics committee of Gothenburg, Sweden; the ethical committees of the Wilhelmina Children’s Hospital and the Sophia Children’s Hospital, The Netherlands; the Ethical Commission of the University of Freiburg, Germany; and The Copenhagen Ethics Committee and The Danish Data Protection Agency, Denmark. Informed consent was obtained and research was subject to the guidelines of the World Medical Association (WMA) Declaration of Helsinki.

4.2 Study populations in genetic studies

All studies were conducted on partially overlapping sets of samples. Study populations, with sample sizes and their respective studies, are shown in Figure 4.

4.2.1 Study I

The main GWAS population consisted of a discovery phase population and a replication population. The discovery study was conducted on a case-control population of 217 cases and 776 controls with samples from Kuopio and Tampere, Finland, and Gothenburg, Sweden. A part of the controls were Finnish population controls from the regions of Helsinki and Kuopio, obtained from the Nordic control database (NordicDB) (Leu et al., 2010). The replication population was collected in the Netherlands and Turku, Finland (n = 1,050).

42 42

Table 2. Summary of the substudies in this thesis.

Study Purpose Approaches Study populations I Genome-wide association study (GWAS) to GWAS discovery study followed by replication GWAS, bronchiolitis cohorts from Finland and identify bronchiolitis-susceptibility loci in a studies in bronchiolitis case-control populations to Sweden; replication populations from The hypothesis-free manner verify GWAS findings Netherlands and Finland

II GWAS subset study to identify functional GWAS loci within gene and protein expression- Matched GWAS subpopulation from Finland and variants associated with bronchiolitis linked sites studied in bronchiolitis case-control Sweden; replication populations from Finland and populations, flow cytometry experiments in NK Denmark; NK cell donors from Finland cells III Analysis of an asthma-associated CDHR3 Meta-analysis across five European bronchiolitis Bronchiolitis case-control populations from variant for association with bronchiolitis case-control populations Denmark, The Netherlands, Germany, Finland, and Sweden

Fig. 4. Study populations in discovery (gray boxes) and replication (black boxes) populations. Black arrows indicate replication studies, whereas gray arrows mark shared data between studies. Some study populations did not have their own controls (controls marked with an asterisk). Finnish GWAS replication population was tested against Finnish population controls that were also a part of GWAS controls, and Turku and Tampere cases were compared to controls of the GWAS subpopulation.

GWAS discovery population

Cases from Kuopio were obtained from two prospective cohorts during 1981–1982 and 1992–1993 (Korppi, Halonen, Kleemola, & Launiala, 1986; Reijonen, Korppi, Pitkakangas, Tenhola, & Remes, 1995). Cases from Gothenburg were recruited from longitudinal cohorts during 1984-1985 (Goksor et al., 2015; Wennergren et al., 1992). All cases were full term children of native Finnish or Swedish origin hospitalized with bronchiolitis diagnosis at less than 24 months of age. Healthy controls were recruited and matched for collection site, age (+/- 2 months), and gender. The case population from Tampere consisted of healthy full term infants less than 12 months of age at the time of bronchiolitis diagnosis. The infants were enrolled between 2001 and 2004 in Tampere University Hospital (Koponen, Helminen, Paassilta, Luukkaala, & Korppi, 2012). Blood samples were collected for genetic analyses. Genotype data for anonymous population controls that originated from Helsinki and Kuopio regions were obtained from the Nordic control database (NordicDB) (Leu et al., 2010). Detailed exclusion criteria, viral

detection, and age distribution are presented in the original articles. Overall, 72 % of the cases were RSV positive, and over 80 % were younger than 12 months.

Replication population

The Dutch population comprised 416 bronchiolitis cases and 432 controls. The cases were native Dutch children hospitalized for RSV-bronchiolitis during 1992- 2006. (Janssen et al., 2007). Most of the cases (>95%) were less than 12 months of age. Unselected controls born in the Netherlands were obtained from a Regenbook health examination survey (Viet AL , Kroesbergen HT , Verhoeff AP , Seidell JC , Otten F , Veldhuizen H van, 2004). Blood samples, buccal swaps, and buffy coats were collected to obtain DNA for genetic studies. The replication case-population from the Turku district included 202 children originating from two cohorts diagnosed with bronchiolitis at 3-23 months of age (Jartti et al., 2015; Lehtinen et al., 2007). Sample collection took place between 2000 and 2010. Whole blood samples were drawn for genetic studies. Half of the cases from Turku were under 12 months of age, and RSV was detected in 31 %.

4.2.2 Study II

The study was based on a subpopulation of the main GWAS. We selected the subpopulation from Kuopio and Gothenburg as a primary population in this study because the population’s 93 cases and 94 controls were matched for age, gender, and collection site. Further, samples in this subset were genotyped on the same platform and virus distribution, age characteristics, and gender distributions were similar among populations from Kuopio and Gothenburg. As a replicate, cases that were less than 12 months of age from Tampere (n = 124) and Turku (n = 101) were tested against the same controls from Kuopio and Gothenburg. As a further replication population we had study cohorts from Denmark (n = 976). The Danish replication population comprised two prospective birth cohorts, COPSAC2000 and COPSAC2010 (COPSAC, The Copenhagen Prospective Study on Asthma in Childhood). Month old babies born to asthmatic mothers were enrolled in the COPSAC2000 study (Bisgaard, 2004). The main exclusion criteria included previous LRI, birth before 36 weeks of gestation, and congenital illness. COPSAC2010 was an unselected longitudinal birth cohort study (Bisgaard et al., 2013). All bronchiolitis cases with acute LRI symptoms were hospitalized at less than 12 months of age, and individuals from the birth cohorts without bronchiolitis

before the age of 2 years were defined as controls (Vissing, Chawes, & Bisgaard, 2013). Approximately 85% of the cases had RSV induced bronchiolitis.

4.2.3 Study III

Study III was a meta-analysis of five European case-control populations. The populations, which comprised a total of 743 cases and 1,647 controls, originated from Kuopio, Finland (n = 126); Gothenburg, Sweden (n = 57); Utrecht, The Netherlands (n = 1,352); Denmark (COPSAC, n = 324); and Freiburg, Germany (n = 531). The subjects from the populations from Finland, Sweden, Utrecht, and Denmark were also used in studies I or II, or both. Danish samples were from the COPSAC2000 birth cohort. Samples originating from Germany were enrolled between 1998-2005 (Puthothu et al., 2007). The cases were children who were hospitalized with RSV bronchiolitis before the age of 2 years. RSV infection was detected with PCR or antigen testing. Infants with congenital heart illness, immune deficiency, or chromosomal defects were excluded.

4.3 DNA sample preparation and genotyping

4.3.1 DNA samples (I-III)

DNA was extracted from whole blood samples, buccal swabs, or buffy coats following standard methods, which are described in detail in the original publications.

4.3.2 Genome-wide genotyping (I, II)

Genome-wide genotyping of the DNA samples was conducted in the Technology Centre of the Institute for Molecular Medicine in Finland (FIMM). Genotyping was conducted with genome-wide genotyping chips from Illumina (Illumina, San Diego, CA, USA): HumanOmniExpress BeadChip (populations from Kuopio and Gothenburg), and Infinium HumanCore- Exome BeadChip (study population from Tampere). Genotypes of the Finnish population controls had been resolved with HumanCNV-370-v1.0 (Illumina).

4.3.3 Targeted SNP genotyping (I, II)

GWAS replication genotyping for the selected SNPs was performed with the Sequenom iPLEX Gold assay (San Diego, CA, USA) in FIMM (I). Polymorphisms of the NKG2D locus were resolved with Sanger sequencing (rs10772271) and RFLP-PCR genotyping with DdeI (rs1049174) (Tampere, Turku), or acquired from Illumina HumanOmniExpressExome BeadChip genotyped at the AROS Applied Biotechnology AS center, in Aarhus, Denmark (COPSAC) (II).

4.3.4 Genotyping of CDHR3 SNPs (III)

Genotypes for the variant rs6967330 were determined with various methods. In the populations from Kuopio, Finland, and Gothenburg, Sweden, the SNP was imputed with IMPUTE2 using the 1000 GP phase 3 reference panel that is based on sequence data for 2,504 individuals. The SNP had an imputation info score of 0.97. In the Danish COPSAC2000 cohort, genotypes for the SNP were acquired from genome-wide genotyping chips (Illumina Infinium II HumanHap550 v1, v3, or Quad BeadChip platform). The chip genotyping was done at the Children’s Hospital of Philadelphia’s Center for Applied Genomics. In the populations from Freiburg and Utrecht, genotyping was performed commercially by LGC Genomics (LGC Genomics, Hoddesdon, UK).

4.4 Processing of the genome-wide data

4.4.1 Quality control (I, II)

Plink 1.9 was used to perform QC of the GWAS data (Chang et al., 2015). The data were subject to basic QC measures, which included removing samples with > 3% missing genotypes and excluding variants with MAF < 1%, genotyping rate < 90% or deviation from HWE (p < 0.001). In addition, population outliers and relatives (pihat > 0.15) were excluded on the basis of IBD (identical by descend) and IBS (identical by state) distances.

4.4.2 Genotype imputation (I)

We performed pre-phasing and genotype imputation to statistically infer unobserved and missing genotypes in our data (B. N. Howie et al., 2009). Densely

genotyped haplotypes based on sequence data for 2,504 individuals from 1000 GP phase 3 were used as a cosmopolitan reference panel (1000 Genomes Project Consortium et al., 2015; Delaneau, Marchini, 1000 Genomes Project Consortium, & 1000 Genomes Project Consortium, 2014). Prior to imputation, we performed strand alignment and conversion of genome coordinates to the Genome Reference Consortium Human Build 37 (GRCh37) with plink1.07 or 1.9, shapeit2, UCSC LiftOver script, and tools included in the GenGen package. Pre-phasing was performed with the shapeit2 algorithm and the estimated haplotypes were then passed on to impute2 to predict missing genotypes. After imputation, we carried out basic QC by excluding rare variants with MAF <0.01, SNPs under a genotyping rate threshold of 0.9, and SNPs deviating from HWE (p < 0.001). Moreover, we filtered out variants with an imputation info score <0.7. We used the Breslow–Day test to assess how uniform the odds ratios are between the GWAS cohorts. Variants with p <0.01 were excluded. The overall genotyping rate of the 5,304,232 variants after imputation and QC was 0.97.

4.5 GWAS

4.5.1 Principal component analysis (I)

To account for population structure, we conducted PCA for nonimputed combined GWAS data that were pruned for LD. Three foremost principal components were included in the discovery association analysis as covariates. The number of principal components was inferred on the basis of genome-wide inflation factor lambda (λ), eigenvalues, and significance of covariate values in association analysis. PCA was conducted with Plink 1.9.

4.5.2 Association tests (I, II)

Logistic regression was used to test associations of 5,304,232 variants with bronchiolitis in the combined GWAS population of 217 bronchiolitis cases and 778 controls. A subset of RSV-positive bronchiolitis cases (n = 121) vs. the controls was also tested. Further subanalyses included testing previously bronchiolitis- associated polymorphisms and previously asthma-associated SNPs for association in the current bronchiolitis population. Associations were tested with Plink 1.9, and

47 Manhattan plots were drawn with a “qqman” package in R (R Development Core Team, 2008). Genotypes of the matched subpopulation from Kuopio and Gothenburg were tested for associations with the basic Chi square allelic test. The association analysis was restrained to variants with MAF > 10% (n = 511,132) due to a small population size. Haplotype level association analyses were conducted with Haploview (J. C. Barrett, Fry, Maller, & Daly, 2005).

4.5.3 Selection of variants for further study (I, II)

SNPs for replication genotyping were chosen from loci with p < 5 × 10−4 in the combined discovery GWAS. Here, the loci were defined as aggregates of variants in moderate to high LD (r2 = 0.5−1) that were formed with default options of LD based clumping in Plink1.9. We aimed to select one SNP from each locus, and excluded isolated imputed SNPs. Several annotation databases were used to help select the SNPs. Predicted or empirically defined regulatory potential of the variants, including eQTLs, promoter or enhancer activity, or protein binding sites, were checked from databases including HaploReg v. 4.1, RegulomeDB, GTEx, and dbSNP (A. P. Boyle et al., 2012; GTEx Consortium, 2013; Sherry et al., 2001; Ward & Kellis, 2012). Of the variants selected for replication, 77 were successfully genotyped in a Dutch replication population. Variants with nominal association (p < 0.05) with the effect in the right direction were selected for a further replication test in a Finnish population. Variants in the GWAS that was conducted on the genotyped subpopulation from Kuopio and Gothenburg were prioritized based on overlap with regions linked to regulation of gene expression or protein levels, i.e. QTLs. First, we checked which GWAS variants with p < 10−4 coincided with eQTLs in the relevant tissues from GTEx (GTEx Consortium, 2013). Next, we explored the remaining loci for further eQTLs in other GTEx tissues. Finally, we checked the identified loci for pQTL colocalization in the blood pGWAS server (Suhre et al., 2017). Selected GWAS variants that overlaid both eQTL and pQTL were selected for follow-up study.

4.6 Analysis of allele specific expression (II)

Variants from the GWAS of the matched subpopulation, that overlapped QTLs associated with mRNA and protein abundance, were studied with flow cytometry

to validate allele specific expression. Flow cytometry was conducted on NK cells extracted from 33 healthy Finnish volunteers. Flow cytometric analysis was conducted with BD FACS Canto II (Becton, Dickinson and Company, Franklin Lakes, NJ). Fluorescent-labeled antibodies that were used in immunostaining and the gating strategy for NK cells are described in original article II. Briefly, we investigated CD19-, CD14-, and CD3-negative lymphocytes for CD56 and CD16 staining, and finally for NKG2D. Protein expression level was determined by mean fluorescence intensity (MFI). MFI of isotype control was utilized in normalization of NKG2D-specific MFI values with FlowJo software (FlowJo LLC, Ashland, OR, USA). MFI values were then assessed for Gaussian distribution with SPSS (IBM Corp. Released 2015. IBM SPSS Statistics for Windows, Version 23. Armonk, NY: IBM Corp.). Finally, association of studied genotypes with protein expression was analyzed by linear regression with Plink 1.9.

4.7 Association analyses in replication cohorts (I, II)

GWAS replication populations from the Netherlands and Finland were tested for association with logistic regression assuming an additive model, and gender was included in the analysis as a covariate. Replication cohorts for study II were analyzed for case-control allele frequency differences with a basic Chi square allelic test under an additive model. Regional association plots were drawn with a single nucleotide polymorphisms annotator (SNiPA) (Arnold, Raffler, Pfeufer, Suhre, & Kastenmuller, 2015).

4.8 Analysis of CDHR3 SNP (III)

The variant rs6967330 was associated with asthma typified by exacerbations in childhood in a previous GWAS (Bonnelykke et al., 2014). LRIs are known to be able to trigger asthma exacerbations, and rs6967330 was hypothesized to associate with asthma by increasing susceptibility to LRIs. An asthma-associated allele was expected to increase the risk of bronchiolitis, and the association with bronchiolitis was tested with logistic regression under an additive model in five bronchiolitis cohorts with a total of 743 cases and 1,647 controls. Random effects meta-analysis was performed with the “meta” package in R version 3.1.1 (Schwarzer, 2007).

5 Results

5.1 Discovery analyses for the combined genome-wide data (I)

Association of genetic variants with bronchiolitis was assessed in the imputed GWAS population of 217 cases and 778 controls. The GWAS population was further studied in batches that included cases with RSV-induced bronchiolitis or SNPs that were suggested in previous genetic studies of viral LRI or asthma.

5.1.1 Best associating loci

Various polymorphisms were associated with bronchiolitis or RSV bronchiolitis with a suggestive level of significance (p < 10−5). The stringent threshold of genome-wide significance of GWAS (p < 10−8) was not attained. Figure 5 and figure 6 present the overview of the results as Manhattan plots for the GWAS and RSV GWAS, respectively. A detailed description of the best associating variants is presented in the original article I (tables 1-2 and table S1).

Fig. 5. Manhattan plot of the GWAS of bronchiolitis conducted on a study population of 217 cases and 778 controls for 5,304,232 polymorphisms. Various association signals in the suggestive level of significance (p < 10−5), indicated by the blue line in the figure, were seen. Five of the best associating loci with their nearest corresponding genes are marked by the arrows. The two strongest loci with suggestive associations were detected on chromosomes 7 and 10. These regions were also among the five best associating loci in the RSV GWAS.

Fig. 6. Manhattan plot of the GWAS population that included only RSV-induced bronchiolitis cases (n = 121) tested against the GWAS controls (n = 778). Several of the tested, approximately 5,190,000 variants, were associated with RSV bronchiolitis at a suggestive significance level (p < 10−5; blue line in the figure). Five best associating loci with their nearest corresponding genes are indicated in the figure by the arrows. The strongest suggestive association was detected for a locus on chromosome 10.

5.1.2 Association of variants previously linked to bronchiolitis or asthma in the current GWAS

We screened the SNPs (+/- 20 kb) that were linked to bronchiolitis or asthma phenotypes in previous studies (Table S2 and S3 in original article I). We detected mild associations (p < 0.01) in the covered regions, whereas the exact variants that were reported in the previous studies were not associated with bronchiolitis in the current GWAS. SNP rs56039226 in the CX3CR1 intron showed the strongest signal (p = 5.7 × 10−5) in our GWAS. The variant was located at an 11 kb distance compared to a polymorphism rs3732378 that was associated with RSV bronchiolitis in a previous study (Amanatidou, Sourvinos, Apostolakis, Tsilimigaki, & Spandidos, 2006). In addition to CX3CR1 SNP, variants near or within genes, including nitric oxide synthase 2 (NOS2), IL17A1, and surfactant protein C (SFTPC), showed mild associations (p < 6 × 10−4) in the GWAS or in the RSV GWAS.

Regarding previously asthma-associated variants, the best suggestive signals (p < 10−4−10−5) were detected in the regions of the genes versican (VCAN), major histocompatibility complex, class II, DQ alpha 1 (HLA-DQA1), and ninein (NIN). These genes play roles in processes including cytokine formation and antigen engagement (NCBI Resource Coordinators, 2016).

5.2 Analysis of shared loci between results from the GWAS subpopulation and reported QTLs (II)

In a genotyped GWAS subpopulation (n = 187), study subjects were matched for age, site, and sex. Case-control analysis of the GWAS subpopulation inspected associations between 511,132 polymorphisms and bronchiolitis. 36 SNPs had p values below the preselected threshold (p < 10−4) (Table S1 in the original article II). We inspected for overlap of these SNPs with eQTLs in blood and lung tissues, and further screened the identified loci for additional eQTLs (GTEx Consortium et al., 2017). Following this procedure, we ended up with three GWAS loci that harbored significant eQTLs associated with genes wntless Wnt ligand secretion mediator (WLS), ankyrin repeat domain 22 (ANKRD22), lipase family member N (LIPN), and four genes in the NK complex encoding for NKG2C, -D, -E and -F. (Table 1 in original article II). The Locus near the NKG2D gene (official symbol KLRK1) further colocalized with a pQTL associated with NKG2D levels in blood (Table 3) (Suhre et al., 2017). Bronchiolitis susceptibility allele-A of rs1077227 near NKG2D was associated with decreased levels of NKG2D mRNA and protein.

Table 3. Loci from GWAS subpopulation analysis that resided within QTLs. One of the screened GWAS loci with p < 10-4 overlapped with eQTLs associated with NKGD2 mRNA levels. Moreover, the locus colocalized with a pQTL affecting NKG2D protein abundance in blood plasma.

GWAS variant/ locus eQTL information pQTL information Beta p value Beta p value rs10772271-A – 0.341 8.1×10–8 – 0.26 7.8×10–9 /NKG2D – 0.362 2.2×10–8 Beta for eQTL in 1subcutaneous adipose and 2skeletal muscle.

5.3 Follow-up studies based on the genome-wide analyses

5.3.1 Associations of the best variants from the combined GWAS in replication populations (I)

In the combined GWAS, 77 polymorphisms were assessed in the primary replication cohort that originated from the Netherlands (Table S1 in original article I). Of the SNPs with p < 0.05 in the replication analysis, three had directionally similar Odds ratios (ORs) compared to the discovery study. The outcome of the RSV GWAS was very similar. These three SNPs were then assessed in another replication set consisting of cases from Turku, tested against allele frequencies of Finnish population controls. In this analysis, one variant showed nominal association with bronchiolitis. The results for the three variants in the replication populations as well as in the discovery GWAS are shown in Table 4. The SNP rs269094 was located in an intergenic region on chromosome 1. The variant resided in a regulatory region and was in high LD with its surrounding SNPs, as indicated by the regional association plot in figure 7. The SNP was an eQTL for potassium voltage-gated channel subfamily D member 3 (KCND3) (p = 1.4 × 10−12) and long intergenic nonprotein-coding RNA 1750 (LINC01750/RP11-88H9) (p = 1.0×10−8) in spleen (GTEx Consortium et al., 2017). The SNP rs9591920 showed nominal association in a second replication study. The variant was quite intergenic with the nearest gene located >200kb away. The third SNP, rs1537091 was also distant from any characterized gene but had some predicted regulatory capacity. Both variants correlated with nearby polymorphisms that had some putative regulatory effects.

Table 4. GWAS variants with nominal associations in the Dutch replication group. Three GWAS SNPs had p < 0.05 and consistent Odds ratios in the primary Dutch replication population. Of these, rs9591920 showed nominal association also in the other replication group.

SNP ID GWAS Replication, The Netherlands Replication, Finland p OR p OR p OR rs269094 5.3×10−5 1.67 0.029 1.28 0.989 0.94 rs9591920 2.0×10−6 2.16 0.023 1.41 0.045 1.47 rs1537091 1.3×10−5 1.69 0.041 1.24 0.229 1.17

Fig. 7. Regional association plot of the genome-wide association results around rs269094. Rs269094 was chosen for replication analysis, and showed nominal association in the primary replication population from the Netherlands. The locus had several correlating variants with regulatory potential, and was an eQTL influencing expression levels of KCND3.

5.3.2 Validation of the effect of the NKG2D genotype on NKG2D expression (II)

The A-allele of rs10772271, which was more common in bronchiolitis cases than controls, was related to smaller quantities of NKG2D gene expression and protein product in the QTL data. The effect on mRNA expression was detected in many tissues (e.g., adipose and muscle), whereas the influence on NKG2D protein expression was measured in samples from blood plasma (GTEx Consortium et al., 2017; Suhre et al., 2017). Thus, decreased NKG2D expression would predispose children to bronchiolitis. In order to verify the association between NKG2D genotypes and expression, we scrutinized NKG2D receptor expression in natural killer (NK) cells with flow cytometry. In addition to rs10772271 genotypes, we conducted flow cytometric experiments for another NKG2D variant, rs1049174, which also was a significant eQTL and pQTL for NKG2D. Rs1049174 was in LD

Table 5. NKG2D genotypes and NKG2D expression in NK cells. The studied genotypes near the NKG2D locus were associated with NKG2D receptor expression in a primary cytotoxic NK1 cell population and in two cell populations consisting of mainly immunoregulatory NK cells (NK2 and NK3).

Variant ID NK1 cell population NK2 cell population NK3 cell population p Beta p Beta p Beta rs10772271-A 6.0×10−4 – 0.57 1.3×10−6 – 0.73 0.0053 – 0.47 rs1049174-C 4.3×10−4 – 0.58 1.8×10−6 – 0.73 0.0034 – 0.50 with rs10772271 (r2 = 0.71 in our data), and the major regional haplotype spanning both variants was more common in bronchiolitis cases compared to the healthy controls (p = 8.0 × 10−4). Three major NK cell populations were analyzed. These were a primary cytotoxic CD56dimCD16bright NK cell population (from now on called “NK1” cell population), and two immunoregulatory NK cell populations, CD56brightCD16− (“NK2”) and CD56brightCD16dim (“NK3”). Genotypes of the studied loci were associated with NKG2D expression in all these populations in linear regression analysis, and the directions of the effects were consistent with those reported in the QTL data (Table 5).

5.3.3 Replication analysis of NKG2D SNPs (II)

In the GWAS subpopulation, the detected allele-frequency difference for NKG2D SNPs rs10772271 and rs1049174 between cases and controls was high (19% and 13%, respectively). We tested these variants for association in three replication groups. The results are shown in Table 6. We detected nominal association for rs10772271 in the replication population from Tampere. There was no allele frequency difference between cases and controls for rs1049174 in the Tampere population. No associations were detected in the analyses of two other replication groups.

Table 6. Association results for NKG2D variants in the GWAS subpopulation and in the replication studies.

Study rs10772271-A rs1049174-C Frequency p OR Frequency p OR cases, controls cases, controls GWAS 0.73, 0.54 9.9×10-5 2.34 0.77, 0.65 0.0011 1.80 subpopulation Finnish replication 0.64, 0,54 0.049 1.50 0.66, 0.65 0.78 1.06 (Tampere) Finnish replication 0.61, 0.54 0.15 1.43 0.65, 0.65 0.92 1.02 (Turku) Replication cohort 0.56, 0.62 0.40 0.82 0.74, 0.72 0.71 1.10 from Denmark

5.4 Analysis of CDHR3 SNP rs6967330 in European bronchiolitis cohorts (III)

The previously asthma-associated SNP rs6967330 in the CDHR3 gene was studied for association in five bronchiolitis cohorts. In the meta-analysis, the polymorphism did not show association with bronchiolitis triggered by RSV only (OR = 1.2, p = 0.26). However, a meta-analysis including bronchiolitis populations from Finland and Sweden detected an association (OR = 2.5, p = 0.0058). These cohorts comprised mostly RSV-negative cases. In the overall meta-analysis with RSV positive and RSV negative bronchiolitis cases there was a mild trend (OR = 1.4, p = 0.08).

6 Discussion

Knowledge of the role of host genetics in bronchiolitis is mostly based on candidate gene studies. Thus, new information of the genetic factors involved in bronchiolitis pathogenesis is important to better understand the disease process. In this work, the aim was to identify novel susceptibility polymorphisms associated with a risk of childhood bronchiolitis by hypothesis-generating GWAS and follow-up studies. Another objective was to detect potential shared genetic loci underlying viral bronchiolitis and asthma.

6.1 Potential bronchiolitis susceptibility loci from GWAS (I)

We detected many associations with a suggestive significance level in the GWAS. The strongest signals resided in genomic regions that were not linked to bronchiolitis previously. Such results were anticipated, because previous studies of bronchiolitis genetics were conducted on preselected polymorphisms, often missense variants, based on what is already known about their respective genes. In our GWAS, and in the association analysis that contained only the individuals with RSV-induced bronchiolitis as cases, the best associating loci were near genome- wide significant. The region with one of the strongest signals in the GWAS and RSV GWAS was in an intergenic region on chromosome 10. The signal encompassed several correlating regulatory variants. Regulatory functions for this locus in annotation databases included histone marks associated with enhancer and promoter activity. The locus resided 7 and 9 kb away from its nearest ncRNA and protein coding genes, respectively. The nearest protein coding gene was called chromosome 10 open reading frame 71 (C10orf71). Interestingly, the gene was very recently aliased as CEFIP in a study that characterized its protein product, cardiac- enriched FHL2-interacting protein, for the first time (Dierck et al., 2017). In that study, CEFIP was implicated in calcineurin signaling in cardiomyocytes, and it was suggested to play a role in cardiac disease. One of the best associating variants in the GWAS locus on chromosome 10, rs76728164, was an eQTL for C10orf71 in heart tissue. However, the gene is also expressed in other tissues including skin and pancreas (GTEx Consortium et al., 2017). Some other variants in the locus were eQTLs for chromosome 10 open reading frame 128 (C10orf128). C10orf128 has been validated as a protein coding gene, which is expressed in multiple tissues, but it does not have any known functions yet (GTEx Consortium et al., 2017; NCBI

Resource Coordinators, 2016). Although the region did not show association in the replication analysis, the direction of the effect was similar. Among the regions with the strongest suggestive associations in the GWAS was also an intergenic locus on chromosome 7. The locus harbored regulatory potential including enhancers or histone marks of putative enhancer activity in several cell types. The locus was surrounded by many uncharacterized ncRNA genes, and there was no known connection with bronchiolitis. Associations detected in the GWAS were not significant at the stringent genome-wide level, which may be due to a limited population size. In addition, our study populations comprised cases with varying viral etiologies and ages. Our candidate findings need to be validated in further studies. As sample sizes of bronchiolitis cohorts continue to grow, the data of our cases may be useful in genetic studies that analyze virus-specific or age stratified bronchiolitis groups.

6.1.1 Variants that were associated with bronchiolitis or asthma in previous studies

We screened bronchiolitis-associated variants reported in candidate gene studies, and their surroundings (+/- 20 kb). A role for the CX3CR1 receptor was supported by this analysis. CX3CR1 is a chemokine receptor involved in leukocyte trafficking and cell adhesion (Panek et al., 2015). The RSV G protein interacts with the CX3CR1 receptor during LRI. The interaction between RSV G proteins’ CX3C chemokine motif and CX3CR1 is potentially important in LRI and immune responses it elicits on the airway epithelium (Chirkova et al., 2015). The interaction of the RSV G protein and CX3CR1 has been suggested to mediate CD4 and CD8 T cell recruitment to the lung (Schmidt & Varga, 2017). There was also some support for a role of variants near or within genes including NOS2 and IL17A1 in the GWAS or in the RSV GWAS. These genes are involved in mediating several immune processes, such as antimicrobial response or autoimmune disease (Jin & Dong, 2013; NCBI Resource Coordinators, 2016). The NOS2 variant was among the strongest associations in a genetic study of RSV bronchiolitis (Janssen et al., 2007). A recent study showed evidence suggesting that NOS2 is also involved in RV bronchiolitis (Alvarez et al., 2018). Pathogens invading the respiratory tract can induce nitric oxide synthase expression. Nitric oxide (NO) dilates airways and has roles in several lines of general defense against microbes, such as nitrosylation of cysteine proteases, which many viruses need to replicate (Ricciardolo, Sterk, Gaston, & Folkerts, 2004). IL17A encodes a

proinflammatory cytokine that can also increase NO release (Miljkovic & Trajkovic, 2004). Variants in this gene have also been suggested to play a role in asthma and the development of asthma after childhood bronchiolitis (Chen et al., 2010; Holster et al., 2018). IL-17 has an important function in pathogen evasion but, on the other hand, has also been shown to contribute to immunopathology in certain inflammatory diseases (Tan & Rosenthal, 2013). Variants in VCAN, HLA-DQA1, and NIN of the previously asthma-associated genes showed the strongest signals in our bronchiolitis GWAS. The genes play roles in processes including cytokine formation and antigen engagement (NCBI Resource Coordinators, 2016). VCAN is a proteoglycan component of the extracellular matrix that functions in mediation of processes that include cell adhesion, proliferation and inflammation (Andersson-Sjöland et al., 2014). The amount of versican produced in the airways appears to correlate with asthma phenotype, with more VCAN production occurring in severe asthma and hyperresponsiveness. It was suggested that viral infection with respiratory viruses could influence versican abundance. HLA-DQA1 belongs to a family of human leukocyte antigen (HLA) complex genes known to have extensive functions in the immune system, and it has been associated with asthma among some other HLAs (Lasky-Su et al., 2012). A recent GWAS and fine-mapping study of the HLA region observed that diseases triggered by viral agents tended to be associated with class I molecules, whereas diseases including bacterial infection were mainly associated with class II molecules (C. Tian et al., 2017). NIN is important for the function of microtubules in epithelial cells (NCBI Resource Coordinators, 2016). It is only associated with a few phenotypes in the GWAS Catalog, occupational asthma being one, although not genome-wide significant, associated trait (Yucesoy et al., 2015).

6.1.2 Replication analysis of the best loci in GWAS

We genotyped the 77 most promising GWAS loci in the replication population from the Netherlands. The associations for three loci were nominally supported in the Dutch population, when considering the variants with consistent Odds ratios. The first, in numerical order from chromosome 1-22, was a variant rs269094 located in an intergenic region on chromosome 1. The locus was associated with KCND3 expression and LINC01750 expression in spleen (GTEx Consortium et al., 2017). Interestingly, a variant in proximity of KCND3 was associated with occupational asthma in a previous GWAS (Yucesoy et al., 2015). KCND3 plays a role in regulation of heart rate, insulin secretion, and neurotransmitter secretion (NCBI

Resource Coordinators, 2016). The gene has also been associated with INF-alpha response within a group of smallpox vaccine recipients (Kennedy et al., 2012). LINC01750 is a long ncRNA with unknown function (NCBI Resource Coordinators, 2016). The second locus, tagged by rs9591920, on chromosome 13 showed nominal association also in a secondary replication study. The region was quite intergenic, with no genes in a range of 200 kb. There were some putative regulatory elements in the locus, and the SNP was in perfect LD with a variant that was previously associated with metabolite levels (Illig et al., 2010). The third variant, rs1537091, resided on chromosome 21, and it was encircled by long intergenic ncRNAs at various distances. The surrounding transcripts did not have any characterized functions. The locus had some putative regulatory effects, but as for rs9591920 on chromosome 13, we could not find anything that would directly link the locus to our phenotypes of interest without further study.

6.2 NKG2D as a candidate of bronchiolitis susceptibility (II)

Based on a GWAS subgroup analysis that was overlaid on available eQTL and pQTL data (GTEx Consortium et al., 2017; Suhre et al., 2017), we suggest that lower levels of NKG2D mRNA and protein expression predispose children to virus triggered bronchiolitis. NKG2D is an activating killer cell lectin-like immune receptor that is expressed on surfaces of NK cells, γδ T cells, CD8+ T cells, iNKT cells, and some CD4+ T cells (Lanier, 2015). Production of conjugate ligands for NKG2D is induced when cells undergo stress due to hyperproliferation, transformation or invading viruses. Interaction of NKG2D receptors with their ligands advances lytic or cytokine mediated elimination of the NKG2D ligand- expressing cells. Some studies have suggested that RSV infection is detected via the NKG2D receptor (Ebihara et al., 2007; Farhadi et al., 2014; F. Li, Zhu, Sun, Wei, & Tian, 2012). The accurate function of NKG2D during infection is not clearly defined, and it may be affected by the surrounding cellular environment (Zhang, Basher, & Wu, 2015). In our GWAS data, the major allele of the rs10772271 susceptibility variant, which was associated with lower NKG2D quantities in the QTL databases, was more common in bronchiolitis cases than controls. The variant belonged to a same haplotype with rs1049174 that was also a protein and gene level-associated QTL for NKG2D. The mode of function for this variant was recently clarified in a study that found rs1049174 alleles have dissimilar adherence towards micro-RNA

mir-1245, which downregulates NKG2D (Espinoza et al., 2016). The haplotype that contained the major alleles of both NKG2D variants was more common in bronchiolitis cases than controls. The haplotype including our candidate susceptibility alleles has been previously characterized as a low cytotoxicity (LNK) haplotype, associated with lower cytotoxicity of NK cells and greater cancer susceptibility (Hayashi et al., 2006). The smaller subgroup of the combined GWAS was selected as a starting population for this study for several reasons. The combined GWAS included general population controls, whereas in this subpopulation, controls were carefully selected to match the cases in relation to age, site, and sex. Further, the study samples were genotyped in the same genome-wide bead chip, which makes analyzing them unambiguous.

6.2.1 Follow-up studies on NKG2D

We conducted a flow cytometry examination in NK cells to authenticate the link between NKG2D genotypes and NKG2D expression in a suitable tissue. NKG2D genotypes were associated with NKG2D receptor abundance in all three detected NK cell populations. Thus, we confirmed the reported pQTL effect, and it was directionally consistent compared to the information in the database, with major alleles associated to lower NKG2D content. We further conducted association analysis for the NKG2D variants in replication populations, of which two originated from Finland and one from Denmark. In the Danish population, the effect was not consistent, but the number of cases was so limited that we cannot confirm a negative result. NKG2D variant rs10772271 was nominally associated in one Finnish population but not in the other, although the effect was in the same direction. Nevertheless, there was a large allele frequency difference between cases and controls in the original GWAS subanalysis (19%) as well as in the positive replication analysis (10%). Allele frequency differences in the negative Finnish replication cohort were 7%. We found no evidence of association between the other NKG2D variant, rs1049174, and bronchiolitis in the replication analyses. The NKG2D receptor can bind to various ligands, which may trigger homologous or contrasting immune pathways (Hayashi et al., 2006). As bronchiolitis may result from infection by various respiratory viruses, it would be interesting to study the NKG2D receptor within virus-specific bronchiolitis studies. Further, it might be useful to conduct NKG2D expression analysis in cells extracted

from bronchiolitis patients. Those results could then be compared to our flow cytometry analysis, in which NKG2D expression was measured in cells from healthy donors and analyzed by genotype.

6.3 Potential virus-specific role of CDHR3 variant in bronchiolitis (III)

The CDHR3 variant rs6967330 was associated with asthma in previous studies (Bonnelykke et al., 2014; Pickrell et al., 2016). The studied asthma-phenotype was characterized by exacerbations in childhood, and the association was not detected in samples from the UK biobank with broader asthma definitions (Vicente et al., 2017). Meta-analysis did not detect an association between the CHDR3 variant and overall bronchiolitis, although the effect was in the right direction. There was evidence of high heterogeneity among the study populations, which may dilute the potential effect. There was no evidence for involvement of the CDHR3 SNP in RSV bronchiolitis. Instead, an association was detected in a subpopulation comprising mostly RSV-negative bronchiolitis cases. Thus, the studied CDHR3 variant may predispose for bronchiolitis that is caused by a pathogen other than RSV. Interestingly, CDHR3 is a known RV-C receptor. Our study variant, rs6967330-A, has been shown to enhance RV-C binding and progeny production (Bochkov et al., 2015). RV infection is also known to be the major triggering agent of asthma exacerbations, especially if children have had atopy and allergic sensitization (Troy & Bosco, 2016).

6.4 Prospects for future studies

We detected several new candidate loci of bronchiolitis-susceptibility in the GWAS and subsequent replication analyses. The preliminary findings of our hypothesis- free study of bronchiolitis should be assessed in larger sample sizes to validate the potential roles of the GWAS candidate loci in bronchiolitis. It would be beneficial to perform these future studies in virus-specific bronchiolitis populations, ideally further classified by age. This may be possible, as bronchiolitis cohorts continue to be collected by certain research groups, and many causal viruses can be easily detected by PCR. Many of the strongest suggestively associated loci in the GWAS resided in intergenic regions that were surrounded by uncharacterized transcripts. It will be interesting to see whether these associations will be replicated in genetic studies of

bronchiolitis. Further, it will be intriguing to observe what functions will be assigned to those uncharacterized genes in future studies. NKG2D signaling has been shown to be involved in battling viral infection, and a couple of studies have suggested that RSV interacts with the NKG2D receptor during viral LRI. The accurate function of NKG2D during infection is not well characterized and appears to be context-specific. The potential role of the NKG2D receptor in RSV bronchiolitis is also far from being resolved, and it could be studied in various settings to better characterize the pathways triggered by the potential virus-receptor -interaction. The meta-analysis of a polymorphism in CDHR3 detected an association between non-RSV bronchiolitis and the CDHR3 variant. The locus has been previously associated with a particular asthma phenotype in two large studies. Further studies on the shared genetic basis among virus-specific bronchiolitis populations and specific asthma phenotypes may provide insight into the relationship between bronchiolitis and asthma.

7 Conclusions

Knowledge of the genetic and molecular mechanisms playing a role in bronchiolitis is limited. In this work, we performed the first hypothesis-free GWAS to identify novel risk loci for childhood bronchiolitis. We further conducted a large meta- analysis in which a previously asthma-associated variant in the CDHR3 gene was tested for association across bronchiolitis studies. The GWAS suggested several new variants as candidates of bronchiolitis susceptibility. The best associating loci were similar in the population containing all virus etiologies and in the population that was stratified for RSV-induced bronchiolitis. Many of the best suggestive association signals were in uncharacterized genomic regions; thus, we were unable to assign them with any obvious role in bronchiolitis susceptibility without further study. The GWAS provided some support for associations between genetic variants and bronchiolitis detected in previous association studies that used a candidate-gene approach. In an analysis conducted on a GWAS subpopulation, we identified polymorphisms located within or near the NKG2D locus as potential variants that predispose to bronchiolitis. We further concluded that reduced expression of the NKG2D immune receptor may be associated with a risk of bronchiolitis. In the meta- analysis, we found a potential association between the CDHR3 variant and non- RSV bronchiolitis. CDHR3 is a receptor for RV. Thus, the studied variant may play a role in RV bronchiolitis. The results of this project provide new information on the genetic background of bronchiolitis. These observations may help us understand the genetics and biological mechanisms underlying bronchiolitis susceptibility, and shed light into the relationship of viral bronchiolitis and asthma. Better characterization of genetic factors that predispose to bronchiolitis, together with accumulating –omics data, clinical data, and patient material, may enable the identification of diverse biological profiles associated with specific bronchiolitis subtypes. This may help to identify biomarkers to target the already existing treatments, which are not broadly effective for all bronchiolitis patients, to appropriate subgroups. Further, such information will likely help to develop and refine treatment strategies that are currently under evaluation as well as yet to be identified therapeutic interventions.

References

1000 Genomes Project Consortium, Auton, A., Brooks, L. D., Durbin, R. M., Garrison, E. P., Kang, H. M., . . . Abecasis, G. R. (2015). A global reference for human genetic variation. Nature, 526(7571), 68-74. 10.1038/nature15393 Akey, J. M., Eberle, M. A., Rieder, M. J., Carlson, C. S., Shriver, M. D., Nickerson, D. A., & Kruglyak, L. (2004). Population history and natural selection shape patterns of genetic variation in 132 genes. PLoS Biology, 2(10), e286. 10.1371/journal.pbio.0020286 Albert, F. W., & Kruglyak, L. (2015). The role of regulatory variation in complex traits and disease. Nature Reviews.Genetics, 16(4), 197-212. 10.1038/nrg3891 Alvarez, A. E., Marson, F. A. L., Bertuzzo, C. S., Bastos, J. C. S., Baracat, E. C. E., Brandao, M. B., . . . Ribeiro, J. D. (2018). Association between single nucleotide polymorphisms in TLR4, TLR2, TLR9, VDR, NOS2 and CCL5 genes with acute viral bronchiolitis. Gene, 645, 7-17. S0378-1119(17)31075-2 Amanatidou, V., Apostolakis, S., & Spandidos, D. A. (2009). Genetic diversity of the host and severe respiratory syncytial virus-induced lower respiratory tract infection. The Pediatric Infectious Disease Journal, 28(2), 135-140. 10.1097/INF.0b013e31818c8d17 Amanatidou, V., Sourvinos, G., Apostolakis, S., Tsilimigaki, A., & Spandidos, D. A. (2006). T280M variation of the CX3C receptor gene is associated with increased risk for severe respiratory syncytial virus bronchiolitis. The Pediatric Infectious Disease Journal, 25(5), 410-414. 10.1097/01.inf.0000214998.16248.b7 Ampuero, S., Luchsinger, V., Tapia, L., Palomino, M. A., & Larranaga, C. E. (2011). SP- A1, SP-A2 and SP-D gene polymorphisms in severe acute respiratory syncytial infection in chilean infants. Infection, Genetics and Evolution : Journal of Molecular Epidemiology and Evolutionary Genetics in Infectious Diseases, 11(6), 1368-1377. 10.1016/j.meegid.2011.04.033 Andersson-Sjöland, A., Hallgren, O., Rolandsson, S., Weitoft, M., Tykesson, E., Larsson- Callerfelt, A., . . . Westergren-Thorsson, G. (2014). Versican in inflammation and tissue remodeling: The impact on lung disorders. Glycobiology, 25(3), 243-251. 10.1093/glycob/cwu120 Arnold, M., Raffler, J., Pfeufer, A., Suhre, K., & Kastenmuller, G. (2015). SNiPA: An interactive, genetic variant-centered annotation browser. Bioinformatics (Oxford, England), 31(8), 1334-1336. 10.1093/bioinformatics/btu779 Baron, M. (2001). The search for complex disease genes: Fault by linkage or fault by association? Molecular Psychiatry, 6(2), 143-149. 10.1038/sj.mp.4000845 Barrett, J. C., Fry, B., Maller, J., & Daly, M. J. (2005). Haploview: Analysis and visualization of LD and haplotype maps. Bioinformatics (Oxford, England), 21(2), 263- 265. 10.1093/bioinformatics/bth457 Barrett, L. W., Fletcher, S., & Wilton, S. D. (2012). Regulation of eukaryotic gene expression by the untranslated gene regions and other non-coding elements. Cellular and Molecular Life Sciences : CMLS, 69(21), 3613-3634. 10.1007/s00018-012-0990-9

Bisgaard, H. (2004). The copenhagen prospective study on asthma in childhood (COPSAC): Design, rationale, and baseline data from a longitudinal birth cohort study. Annals of Allergy, Asthma & Immunology : Official Publication of the American College of Allergy, Asthma, & Immunology, 93(4), 381-389. S1081-1206(10)61398-1 Bisgaard, H., Vissing, N. H., Carson, C. G., Bischoff, A. L., Folsgaard, N. V., Kreiner- Moller, E., . . . Bonnelykke, K. (2013). Deep phenotyping of the unselected COPSAC2010 birth cohort study. Clinical and Experimental Allergy : Journal of the British Society for Allergy and Clinical Immunology, 43(12), 1384-1394. 10.1111/cea.12213 Blanken, M. O., Rovers, M. M., Molenaar, J. M., Winkler-Seinstra, P. L., Meijer, A., Kimpen, J. L., . . . Dutch RSV Neonatal Network. (2013). Respiratory syncytial virus and recurrent wheeze in healthy preterm infants. The New England Journal of Medicine, 368(19), 1791-1799. 10.1056/NEJMoa1211917 Blekhman, R., Man, O., Herrmann, L., Boyko, A. R., Indap, A., Kosiol, C., . . . Przeworski, M. (2008). Natural selection on genes that underlie human disease susceptibility. Current Biology : CB, 18(12), 883-889. 10.1016/j.cub.2008.04.074 Bochkov, Y. A., & Gern, J. E. (2016). Rhinoviruses and their receptors: Implications for allergic disease. Current Allergy and Asthma Reports, 16(4), 7. 10.1007/s11882-016- 0608-7 Bochkov, Y. A., Watters, K., Ashraf, S., Griggs, T. F., Devries, M. K., Jackson, D. J., . . . Gern, J. E. (2015). Cadherin-related family member 3, a childhood asthma susceptibility gene product, mediates rhinovirus C binding and replication. Proceedings of the National Academy of Sciences of the United States of America, 112(17), 5485-5490. 10.1073/pnas.1421178112 Bomba, L., Walter, K., & Soranzo, N. (2017). The impact of rare and low-frequency genetic variants in common disease. Genome Biology, 18(1), 77. 10.1186/s13059-017-1212-4 Bonnelykke, K., Sleiman, P., Nielsen, K., Kreiner-Moller, E., Mercader, J. M., Belgrave, D., . . . Bisgaard, H. (2014). A genome-wide association study identifies CDHR3 as a susceptibility locus for early childhood asthma with severe exacerbations. Nature Genetics, 46(1), 51-55. 10.1038/ng.2830 Bookman, E. B., McAllister, K., Gillanders, E., Wanke, K., Balshaw, D., Rutter, J., . . . for the NIH G × E Interplay Workshop participants. (2011). Gene-environment interplay in common complex diseases: Forging an integrative Model—Recommendations from an NIH workshop. Genetic Epidemiology, 35(4), 217-225. 10.1002/gepi.20571 Boyle, A. P., Hong, E. L., Hariharan, M., Cheng, Y., Schaub, M. A., Kasowski, M., . . . Snyder, M. (2012). Annotation of functional variation in personal genomes using RegulomeDB. Genome Research, 22(9), 1790-1797. 10.1101/gr.137323.112 Boyle, E. A., Li, Y. I., & Pritchard, J. K. (2017). An expanded view of complex traits: From polygenic to omnigenic. Cell, 169(7), 1177-1186. S0092-8674(17)30629-3 Caliskan, M., Bochkov, Y. A., Kreiner-Moller, E., Bonnelykke, K., Stein, M. M., Du, G., . . . Ober, C. (2013). Rhinovirus wheezing illness and genetic risk of childhood-onset asthma. The New England Journal of Medicine, 368(15), 1398-1407. 10.1056/NEJMoa1211592

Casper, J., Zweig, A. S., Villarreal, C., Tyner, C., Speir, M. L., Rosenbloom, K. R., . . . Kent, W. J. (2017). The UCSC genome browser database: 2018 update. Nucleic Acids Research, 10.1093/nar/gkx1020 Chang, C. C., Chow, C. C., Tellier, L. C., Vattikuti, S., Purcell, S. M., & Lee, J. J. (2015). Second-generation PLINK: Rising to the challenge of larger and richer datasets. GigaScience, 4, 8. eCollection 2015. 10.1186/s13742-015-0047-8 Chawes, B. L., Poorisrisak, P., Johnston, S. L., & Bisgaard, H. (2012). Neonatal bronchial hyperresponsiveness precedes acute severe viral bronchiolitis in infants. The Journal of Allergy and Clinical Immunology, 130(2), 61.e3. 10.1016/j.jaci.2012.04.045 Chen, J., Deng, Y., Zhao, J., Luo, Z., Peng, W., Yang, J., . . . Liu, E. (2010). The polymorphism of IL-17 G-152A was associated with childhood asthma and bacterial colonization of the hypopharynx in bronchiolitis. Journal of Clinical Immunology, 30(4), 539-545. 10.1007/s10875-010-9391-8 Chick, J. M., Munger, S. C., Simecek, P., Huttlin, E. L., Choi, K., Gatti, D. M., . . . Gygi, S. P. (2016). Defining the consequences of genetic variation on a proteome-wide scale. Nature, 534(7608), 500-505. 10.1038/nature18270 Chirkova, T., Lin, S., Oomens, A. G., Gaston, K. A., Boyoglu-Barnum, S., Meng, J., . . . Anderson, L. J. (2015). CX3CR1 is an important surface molecule for respiratory syncytial virus infection in human airway epithelial cells. The Journal of General Virology, 96(9), 2543-2556. 10.1099/vir.0.000218 Chong, J., Buckingham, K., Jhangiani, S., Boehm, C., Sobreira, N., Smith, J., . . . Bamshad, M. (2015). The genetic basis of mendelian phenotypes: Discoveries, challenges, and opportunities. American Journal of Human Genetics, 97(2), 199-215. 10.1016/j.ajhg.2015.06.009 Civelek, M., & Lusis, A. J. (2013). Systems genetics approaches to understand complex traits. Nature Reviews.Genetics, 15(1), 34-48. 10.1038/nrg3575 Collins, P. L., & Graham, B. S. (2007). Viral and host factors in human respiratory syncytial virus pathogenesis. Journal of Virology, 82(5), 2040-2055. 10.1128/JVI.01625-07 Daly, A. K., & Day, C. P. (2001). Candidate gene case-control association studies: Advantages and potential pitfalls. British Journal of Clinical Pharmacology, 52(5), 489-499. 10.1046/j.0306-5251.2001.01510.x Delaneau, O., Marchini, J., 1000 Genomes Project Consortium, & 1000 Genomes Project Consortium. (2014). Integrating sequence and array data to create an improved 1000 genomes project haplotype reference panel. Nature Communications, 5, 3934. 10.1038/ncomms4934 Delaneau, O., Marchini, J., & Zagury, J. (2012). A linear complexity phasing method for thousands of genomes. Nature Methods, 9(2), 179-181. 10.1038/NMETH.1785 Dierck, F., Kuhn, C., Rohr, C., Hille, S., Braune, J., Sossalla, S., . . . Frey, N. (2017). The novel cardiac z-disc protein CEFIP regulates cardiomyocyte hypertrophy by modulating calcineurin signaling. The Journal of Biological Chemistry, 292(37), 15180-15191. 10.1074/jbc.M117.786764

Domachowske, J. B., & Rosenberg, H. F. (1999). Respiratory syncytial virus infection: Immune response, immunopathogenesis, and treatment. Clinical Microbiology Reviews, 12(2), 298-309. Retrieved from http://www.ncbi.nlm.nih.gov/pmc/articles/PMC88919/ Douville, R. N., Lissitsyn, Y., Hirschfeld, A. F., Becker, A. B., Kozyrskyj, A. L., Liem, J., . . . HayGlass, K. T. (2010). TLR4 Asp299Gly and Thr399Ile polymorphisms: No impact on human immune responsiveness to LPS or respiratory syncytial virus. PloS One, 5(8), e12087. 10.1371/journal.pone.0012087 Drysdale, S. B., Green, C. A., & Sande, C. J. (2016). Best practice in the prevention and management of paediatric respiratory syncytial virus infection. Therapeutic Advances in Infectious Disease, 3(2), 63-71. 10.1177/2049936116630243 Dykes, I. M., & Emanueli, C. (2016). Transcriptional and post-transcriptional gene regulation by long non-coding RNA. Genomics, Proteomics & Bioinformatics, 15(3), 177-186. 10.1016/j.gpb.2016.12.005 Ebihara, T., Masuda, H., Akazawa, T., Shingai, M., Kikuta, H., Ariga, T., . . . Seya, T. (2007). Induction of NKG2D ligands on human dendritic cells by TLR ligand stimulation and RNA virus infection. International Immunology, 19(10), 1145-1155. dxm073 Edwards, M. R., Bartlett, N. W., Hussell, T., Openshaw, P., & Johnston, S. L. (2012). The microbiology of asthma. Nature Reviews.Microbiology, 10(7), 459-471. 10.1038/nrmicro2801 Elkon, R., & Agami, R. (2017). Characterization of noncoding regulatory DNA in the human genome. Nature Biotechnology, 35(8), 732-746. 10.1038/nbt.3863 Espinoza, J. L., Nguyen, V. H., Ichimura, H., Pham, T. T. T., Nguyen, C. H., Pham, T. V., . . . Nakao, S. (2016). A functional polymorphism in the NKG2D gene modulates NK-cell cytotoxicity and is associated with susceptibility to human papilloma virus-related cancers. Scientific Reports, 6, 10.1038/srep39231. 10.1038/srep39231 Evangelou, E., & Ioannidis, J. P. (2013). Meta-analysis methods for genome-wide association studies and beyond. Nature Reviews.Genetics, 14(6), 379-389. 10.1038/nrg3472 Farhadi, N., Lambert, L., Triulzi, C., Openshaw, P. J. M., Guerra, N., & Culley, F. J. (2014). Natural killer cell NKG2D and granzyme B are critical for allergic pulmonary inflammation. The Journal of Allergy and Clinical Immunology, 133(3), 835.e3. S0091- 6749(13)01555-8 Feil, R., & Fraga, M. F. (2012). Epigenetics and the environment: Emerging patterns and implications. Nature Reviews.Genetics, 13(2), 97-109. 10.1038/nrg3142 Feuk, L., Carson, A. R., & Scherer, S. W. (2006). Structural variation in the human genome. Nature Reviews.Genetics, 7(2), 85-97. nrg1767 Florin, T. A., Plint, A. C., & Zorc, J. J. (2017). Viral bronchiolitis. Lancet (London, England), 389(10065), 211-224. S0140-6736(16)30951-5 Forton, J. T., Rowlands, K., Rockett, K., Hanchard, N., Herbert, M., Kwiatkowski, D. P., & Hull, J. (2009). Genetic association study for RSV bronchiolitis in infancy at the 5q31 cytokine cluster. Thorax, 64(4), 345-352. 10.1136/thx.2008.102111 Gibson, G. (2012). Rare and common variants: Twenty arguments. Nature Reviews.Genetics, 13(2), 135-145. 10.1038/nrg3118

Goksor, E., Amark, M., Alm, B., Ekerljung, L., Lundback, B., & Wennergren, G. (2015). High risk of adult asthma following severe wheezing in early life. Pediatric Pulmonology, 50(8), 789-797. 10.1002/ppul.23071 Graham, B. S. (2017). Vaccine development for respiratory syncytial virus. Current Opinion in Virology, 23, 107-112. 10.1016/j.coviro.2017.03.012 Gratten, J., & Visscher, P. M. (2016). Genetic pleiotropy in complex traits and diseases: Implications for genomic medicine. Genome Medicine, 8, 78. 10.1186/s13073-016- 0332-x GTEx Consortium. (2013). The genotype-tissue expression (GTEx) project. Nature Genetics, 45(6), 580-585. 10.1038/ng.2653 GTEx Consortium, Laboratory, Data Analysis &Coordinating Center (LDACC)-Analysis Working Group, Statistical Methods groups-Analysis Working Group, Enhancing GTEx (eGTEx) groups, NIH Common Fund, NIH/NCI, . . . Montgomery, S. B. (2017). Genetic effects on gene expression across human tissues. Nature, 550(7675), 204-213. 10.1038/nature24277 Harrow, J., Frankish, A., Gonzalez, J. M., Tapanari, E., Diekhans, M., Kokocinski, F., . . . Hubbard, T. J. (2012). GENCODE: The reference human genome annotation for the ENCODE project. Genome Research, 22(9), 1760-1774. 10.1101/gr.135350.111 Hasegawa, K., Jartti, T., Mansbach, J. M., Laham, F. R., Jewell, A. M., Espinola, J. A., . . . Camargo, C. A.,Jr. (2015). Respiratory syncytial virus genomic load and disease severity among children hospitalized with bronchiolitis: Multicenter cohort studies in the united states and finland. The Journal of Infectious Diseases, 211(10), 1550-1559. 10.1093/infdis/jiu658 Hauberg, M. E., Zhang, W., Giambartolomei, C., Franzen, O., Morris, D. L., Vyse, T. J., . . . Roussos, P. (2017). Large-scale identification of common trait and disease variants affecting gene expression. American Journal of Human Genetics, 100(6), 885-894. S0002-9297(17)30161-1 Hayashi, T., Imai, K., Morishita, Y., Hayashi, I., Kusunoki, Y., & Nakachi, K. (2006). Identification of the NKG2D haplotypes associated with natural cytotoxic activity of peripheral blood lymphocytes and cancer immunosurveillance. Cancer Research, 66(1), 563-570. 10.1158/0008-5472 Helminen, M., Nuolivirta, K., Virta, M., Halkosalo, A., Korppi, M., Vesikari, T., & Hurme, M. (2008). IL-10 gene polymorphism at -1082 A/G is associated with severe rhinovirus bronchiolitis in infants. Pediatric Pulmonology, 43(4), 391-395. 10.1002/ppul.20793 Hirschhorn, J. N., & Daly, M. J. (2005). Genome-wide association studies for common diseases and complex traits. Nature Reviews.Genetics, 6(2), 95-108. nrg1521 Hoebee, B., Bont, L., Rietveld, E., van Oosten, M., Hodemaekers, H. M., Nagelkerke, N. J., . . . Kimman, T. G. (2004). Influence of promoter variants of interleukin-10, interleukin-9, and tumor necrosis factor-alpha genes on respiratory syncytial virus bronchiolitis. The Journal of Infectious Diseases, 189(2), 239-247. JID31038

Hoebee, B., Rietveld, E., Bont, L., Oosten, M., Hodemaekers, H. M., Nagelkerke, N. J., . . . Kimman, T. G. (2003). Association of severe respiratory syncytial virus bronchiolitis with interleukin-4 and interleukin-4 receptor alpha polymorphisms. The Journal of Infectious Diseases, 187(1), 2-11. JID020714 Holster, A., Terasjarvi, J., Lauhkonen, E., Tormanen, S., Helminen, M., Koponen, P., . . . Nuolivirta, K. (2018). IL-17A gene polymorphism rs2275913 is associated with the development of asthma after bronchiolitis in infancy. Allergology International : Official Journal of the Japanese Society of Allergology, 67(1), 109-113. S1323- 8930(17)30064-3 Hong, E. P., & Park, J. W. (2012). Sample size and statistical power calculation in genetic association studies. Genomics & Informatics, 10(2), 117-122. 10.5808/GI.2012.10.2.117 Hormozdiari, F., van de Bunt, M., Segre, A. V., Li, X., Joo, J. W., Bilow, M., . . . Eskin, E. (2016). Colocalization of GWAS and eQTL signals detects target genes. American Journal of Human Genetics, 99(6), 1245-1260. S0002-9297(16)30439-6 Howie, B. N., Donnelly, P., & Marchini, J. (2009). A flexible and accurate genotype imputation method for the next generation of genome-wide association studies. PLoS Genetics, 5(6), e1000529. 10.1371/journal.pgen.1000529 Howie, B., Fuchsberger, C., Stephens, M., Marchini, J., & Abecasis, G. R. (2012). Fast and accurate genotype imputation in genome-wide association studies through pre-phasing. Nature Genetics, 44(8), +. 10.1038/ng.2354 Howie, B., Marchini, J., & Stephens, M. (2011). Genotype imputation with thousands of genomes. G3-Genes Genomes Genetics, 1(6), 457-469. 10.1534/g3.111.001198 Hrdlickova, B., de Almeida, R. C., Borek, Z., & Withoff, S. (2014). Genetic variation in the non-coding genome: Involvement of micro-RNAs and long non-coding RNAs in disease. Biochimica Et Biophysica Acta, 1842(10), 1910-1922. S0925-4439(14)00071- 4 Hull, J., Thomson, A., & Kwiatkowski, D. (2000). Association of respiratory syncytial virus bronchiolitis with the interleukin 8 gene region in UK families. Thorax, 55(12), 1023- 1027. 10.1136/thorax.55.12.1023 Illig, T., Gieger, C., Zhai, G., Romisch-Margl, W., Wang-Sattler, R., Prehn, C., . . . Suhre, K. (2010). A genome-wide perspective of genetic variation in human metabolism. Nature Genetics, 42(2), 137-141. 10.1038/ng.507 International HapMap Consortium. (2003). The international HapMap project. Nature, 426(6968), 789-796. 10.1038/nature02168 International HapMap Consortium. (2005). A haplotype map of the human genome. Nature, 437(7063), 1299-1320. 10.1038/nature04226 International Schizophrenia Consortium, Purcell, S. M., Wray, N. R., Stone, J. L., Visscher, P. M., O'Donovan, M. C., . . . Sklar, P. (2009). Common polygenic variation contributes to risk of schizophrenia and bipolar disorder. Nature, 460(7256), 748-752. 10.1038/nature08185

Jakkula, E., Rehnstrom, K., Varilo, T., Pietilainen, O. P., Paunio, T., Pedersen, N. L., . . . Peltonen, L. (2008). The genome-wide patterns of variation expose significant substructure in a founder population. American Journal of Human Genetics, 83(6), 787- 794. 10.1016/j.ajhg.2008.11.005 Janssen, R., Bont, L., Siezen, C. L., Hodemaekers, H. M., Ermers, M. J., Doornbos, G., . . . Hoebee, B. (2007). Genetic susceptibility to respiratory syncytial virus bronchiolitis is predominantly associated with innate immune genes. The Journal of Infectious Diseases, 196(6), 826-834. JID38563 Jartti, T., & Gern, J. E. (2017). Role of viral infections in the development and exacerbation of asthma in children. The Journal of Allergy and Clinical Immunology, 140(4), 895- 906. S0091-6749(17)31345-3 Jartti, T., Nieminen, R., Vuorinen, T., Lehtinen, P., Vahlberg, T., Gern, J., . . . Ruuskanen, O. (2015). Short- and long-term efficacy of prednisolone for first acute rhinovirus- induced wheezing episode. The Journal of Allergy and Clinical Immunology, 135(3), 8.e9. 10.1016/j.jaci.2014.07.001 Jin, W., & Dong, C. (2013). IL-17 cytokines in immunity and inflammation. Emerging Microbes & Infections, 2(9), e60. 10.1038/emi.2013.58 Johnson, S. M., McNally, B. A., Ioannidis, I., Flano, E., Teng, M. N., Oomens, A. G., . . . Peeples, M. E. (2015). Respiratory syncytial virus uses CX3CR1 as a receptor on primary human airway epithelial cultures. PLoS Pathogens, 11(12), e1005318. 10.1371/journal.ppat.1005318 Jorde, L. B. (2000). Linkage disequilibrium and the search for complex disease genes. Genome Research, 10(10), 1435-1444. Kennedy, R. B., Ovsyannikova, I. G., Pankratz, V. S., Haralambieva, I. H., Vierkant, R. A., & Poland, G. A. (2012). Genome-wide analysis of polymorphisms associated with cytokine responses in smallpox vaccine recipients. Human Genetics, 131(9), 1403-1421. 10.1007/s00439-012-1174-2 Kerminen, S., Havulinna, A. S., Hellenthal, G., Martin, A. R., Sarin, A., Perola, M., . . . Pirinen, M. (2017). Fine-scale genetic structure in finland. G3: Genes|Genomes|Genetics, 7(10), 3459-3468. 10.1534/g3.117.300217 Kimman, T. G. (2001). Genetics of infectious disease susceptibility (1st ed., p. 14) Springer Netherlands. Koponen, P., Helminen, M., Paassilta, M., Luukkaala, T., & Korppi, M. (2012). Preschool asthma after bronchiolitis in infancy. The European Respiratory Journal, 39(1), 76-80. 10.1183/09031936.00040211 Korppi, M., Halonen, P., Kleemola, M., & Launiala, K. (1986). Viral findings in children under the age of two years with expiratory difficulties. Acta Paediatrica Scandinavica, 75(3), 457-464. 10.1111/j.1651-2227.1986.tb10230.x Korppi, M., Koponen, P., & Nuolivirta, K. (2012). Upper age limit for bronchiolitis: 12 months or 6 months? The European Respiratory Journal, 39(3), 9. 10.1183/09031936.00170511

Korppi, M., Kotaniemi-Syrjanen, A., Waris, M., Vainionpaa, R., & Reijonen, T. M. (2004). Rhinovirus-associated wheezing in infancy: Comparison with respiratory syncytial virus bronchiolitis. The Pediatric Infectious Disease Journal, 23(11), 995-999. 10.1097/01.inf.0000143642.72480.53 Kresfelder, T. L., Janssen, R., Bont, L., Pretorius, M., & Venter, M. (2011). Confirmation of an association between single nucleotide polymorphisms in the VDR gene with respiratory syncytial virus related disease in south african children. Journal of Medical Virology, 83(10), 1834-1840. 10.1002/jmv.22179 Lahti, M., Lofgren, J., Marttila, R., Renko, M., Klaavuniemi, T., Haataja, R., . . . Hallman, M. (2002). Surfactant protein D gene polymorphism associated with severe respiratory syncytial virus infection. Pediatric Research, 51(6), 696-699. 10.1203/00006450- 200206000-00006 Lambert, L., Sagfors, A. M., Openshaw, P. J., & Culley, F. J. (2014). Immunity to RSV in early-life. Frontiers in Immunology, 5, 466. 10.3389/fimmu.2014.00466 Lander, E. S., Linton, L. M., Birren, B., Nusbaum, C., Zody, M. C., Baldwin, J., . . . International Human Genome Sequencing Consortium. (2001). Initial sequencing and analysis of the human genome. Nature, 409(6822), 860-921. 10.1038/35057062 Lanier, L. L. (2015). NKG2D receptor and its ligands in host defense. Cancer Immunology Research, 3(6), 575-582. 10.1158/2326-6066.CIR-15-0098 Larkin, E. K., & Hartert, T. V. (2015). Genes associated with RSV lower respiratory tract infection and asthma: The application of genetic epidemiological methods to understand causality. Future Virology, 10(7), 883-897. 10.2217/fvl.15.55 Lasky-Su, J., Himes, B. E., Raby, B. A., Klanderman, B. J., Sylvia, J. S., Lange, C., . . . SHARP investigators. (2012). HLA-DQ strikes again: Genome-wide association study further confirms HLA-DQ in the diagnosis of asthma among adults. Clinical and Experimental Allergy : Journal of the British Society for Allergy and Clinical Immunology, 42(12), 1724-1733. 10.1111/cea.12000 Laurie, C. C., Doheny, K. F., Mirel, D. B., Pugh, E. W., Bierut, L. J., Bhangale, T., . . . for the, G. I. (2010). Quality control and quality assurance in genotypic data for genome- wide association studies. Genetic Epidemiology, 34(6), 591-602. 10.1002/gepi.20516 Lehtinen, P., Ruohola, A., Vanto, T., Vuorinen, T., Ruuskanen, O., & Jartti, T. (2007). Prednisolone reduces recurrent wheezing after a first wheezing episode associated with rhinovirus infection or eczema. The Journal of Allergy and Clinical Immunology, 119(3), 570-575. S0091-6749(06)02338-4 Lek, M., Karczewski, K. J., Minikel, E. V., Samocha, K. E., Banks, E., Fennell, T., . . . Exome Aggregation Consortium. (2016). Analysis of protein-coding genetic variation in 60,706 humans. Nature, 536(7616), 285-291. 10.1038/nature19057 Leu, M., Humphreys, K., Surakka, I., Rehnberg, E., Muilu, J., Rosenstrom, P., . . . Ripatti, S. (2010). NordicDB: A nordic pool and portal for genome-wide control data. European Journal of Human Genetics : EJHG, 18(12), 1322-1326. 10.1038/ejhg.2010.112 Li, F., Zhu, H., Sun, R., Wei, H., & Tian, Z. (2012). Natural killer cells are involved in acute lung immune injury caused by respiratory syncytial virus infection. Journal of Virology, 86(4), 2251-2258. 10.1128/JVI.06209-11

Li, M. J., Yan, B., Sham, P. C., & Wang, J. (2015). Exploring the function of genetic variants in the non-coding genomic regions: Approaches for identifying human regulatory variants affecting gene expression. Briefings in Bioinformatics, 16(3), 393-412. 10.1093/bib/bbu018 Liu, X. T., Ren, L., Zhou, L. L., Xiao, Q. Y., Deng, Y., & Liu, E. M. (2015). ORMDL3 variants associated with bronchiolitis susceptibility in a chinese population. Genetics and Molecular Research : GMR, 14(4), 19155-19162. 10.4238/2015.December.29.25 Lofgren, J., Marttila, R., Renko, M., Ramet, M., & Hallman, M. (2010). Toll-like receptor 4 Asp299Gly polymorphism in respiratory syncytial virus epidemics. Pediatric Pulmonology, 45(7), 687-692. 10.1002/ppul.21248 Lowe, W. L., & Reddy, T. E. (2015). Genomic approaches for understanding the genetics of complex disease. Genome Research, 25(10), 1432-1441. 10.1101/gr.190603.115 MacArthur, J., Bowler, E., Cerezo, M., Gil, L., Hall, P., Hastings, E., . . . Parkinson, H. (2016). The new NHGRI-EBI catalog of published genome-wide association studies (GWAS catalog). Nucleic Acids Research, 45, D901. 10.1093/nar/gkw1133 Mandelberg, A., Tal, G., Naugolny, L., Cesar, K., Oron, A., Houri, S., . . . Somekh, E. (2005). Lipopolysaccharide hyporesponsiveness as a risk factor for intensive care unit hospitalization in infants with respiratory syncitial virus bronchiolitis. Clinical and Experimental Immunology, 144(1), 48-52. 10.1111/j.1365-2249.2006.03030.x Manolio, T. A., Collins, F. S., Cox, N. J., Goldstein, D. B., Hindorff, L. A., Hunter, D. J., . . . Visscher, P. M. (2009). Finding the missing heritability of complex diseases. Nature, 461(7265), 747-753. 10.1038/nature08494 Marian, A. J. (2012). Molecular genetic studies of complex phenotypes. Translational Research : The Journal of Laboratory and Clinical Medicine, 159(2), 64-79. 10.1016/j.trsl.2011.08.001 McKusick, V. A. (2007). Mendelian inheritance in man and its online version, OMIM. American Journal of Human Genetics, 80(4), 588-604. Retrieved from http://www.ncbi.nlm.nih.gov/pmc/articles/PMC1852721/ McLellan, J. S., Ray, W. C., & Peeples, M. E. (2013). Structure and function of RSV surface glycoproteins. Current Topics in Microbiology and Immunology, 372, 83-104. 10.1007/978-3-642-38919-1_4 Meissner, H. C. (2016). Viral bronchiolitis in children. N Engl J Med, 374(1), 62-72. 10.1056/NEJMra1413456 Mersha, T. B. (2015). Mapping asthma-associated variants in admixed populations. Frontiers in Genetics, 6, 292. 10.3389/fgene.2015.00292 Miljkovic, D., & Trajkovic, V. (2004). Inducible nitric oxide synthase activation by interleukin-17. Cytokine & Growth Factor Reviews, 15(1), 21-32. S1359610103000996 Miyagawa, T., Nishida, N., Ohashi, J., Kimura, R., Fujimoto, A., Kawashima, M., . . . Tokunaga, K. (2008). Appropriate data cleaning methods for genome-wide association study. Journal of Human Genetics, 53(10), 886-893. 10.1007/s10038-008-0322-y Miyairi, I., & DeVincenzo, J. P. (2008). Human genetic factors and respiratory syncytial virus disease severity. Clinical Microbiology Reviews, 21(4), 686-703. 10.1128/CMR.00017-08

NCBI Resource Coordinators. (2016). Database resources of the national center for biotechnology information. Nucleic Acids Research, 44(D1), 7. 10.1093/nar/gkv1290 NCI-NHGRI Working Group on Replication in Association Studies, Chanock, S. J., Manolio, T., Boehnke, M., Boerwinkle, E., Hunter, D. J., . . . Collins, F. S. (2007). Replicating genotype-phenotype associations. Nature, 447(7145), 655-660. 10.1038/447655a Nica, A. C., & Dermitzakis, E. T. (2013). Expression quantitative trait loci: Present and future. Philosophical Transactions of the Royal Society of London.Series B, Biological Sciences, 368(1620), 20120362. 10.1098/rstb.2012.0362 Oksenberg, J. R., & Baranzini, S. E. (2010). Multiple sclerosis genetics--is the glass half full, or half empty? Nature Reviews.Neurology, 6(8), 429-437. 10.1038/nrneurol.2010.91 Openshaw, P. J. (2005). Antiviral immune responses and lung inflammation after respiratory syncytial virus infection. Proceedings of the American Thoracic Society, 2(2), 121-125. 10.1513/pats.200504-032AW Pagani, F., & Baralle, F. E. (2004). Genomic variants in exons and introns: Identifying the splicing spoilers. Nature Reviews.Genetics, 5(5), 389-396. 10.1038/nrg1327 Panek, C. A., Ramos, M. V., Mejias, M. P., Abrey-Recalde, M. J., Fernandez-Brando, R. J., Gori, M. S., . . . Palermo, M. S. (2015). Differential expression of the fractalkine chemokine receptor (CX3CR1) in human monocytes during differentiation. Cellular & Molecular Immunology, 12(6), 669-680. 10.1038/cmi.2014.116 Paquette, M., Chong, M., Theriault, S., Dufour, R., Pare, G., & Baass, A. (2017). Polygenic risk score predicts prevalence of cardiovascular disease in patients with familial hypercholesterolemia. Journal of Clinical Lipidology, 11(3), 732.e5. S1933- 2874(17)30089-2 Patterson, N., Price, A. L., & Reich, D. (2006). Population structure and eigenanalysis. PLoS Genetics, 2(12), e190. 06-PLGE-RA-0101R3 Paul, S. P., Mukherjee, A., McAllister, T., Harvey, M. J., Clayton, B. A., & Turner, P. C. (2017). Respiratory-syncytial-virus- and rhinovirus-related bronchiolitis in children aged <2 years in an English district general hospital. The Journal of Hospital Infection, 96(4), 360-365. S0195-6701(17)30239-6 Peltonen, L., Jalanko, A., & Varilo, T. (1999). Molecular genetics of the finnish disease heritage. Human Molecular Genetics, 8(10), 1913-1923. ddc212 Peltonen, L., Palotie, A., & Lange, K. (2000). Use of population isolates for mapping complex traits. Nature Reviews.Genetics, 1(3), 182-190. 10.1038/35042049 Peres-Neto, P., Jackson, D. A., & Somers, K. M. (2005). How many principal components? stopping rules for determining the number of non-trivial axes revisited. Computational Statistics & Data Analysis, 49(4), 974-997. //doi.org/10.1016/j.csda.2004.06.015 Petrarca, L., Jacinto, T., & Nenna, R. (2017). The treatment of acute bronchiolitis: Past, present and future. Breathe, 13(1), e26. 10.1183/20734735.000717 Pickrell, J. K., Berisa, T., Liu, J. Z., Segurel, L., Tung, J. Y., & Hinds, D. A. (2016). Detection and interpretation of shared genetic influences on 42 human traits. Nature Genetics, 48(7), 709-717. 10.1038/ng.3570

Pinto, L. A., De, A. L., Mocellin, M., Acosta, P., Caballero, M. T., Libster, R., . . . De Souza, A. P. (2017). IL-8/IL-17 gene variations and the susceptibility to severe viral bronchiolitis. Epidemiology and Infection, 145(4), 642-646. 10.1017/S0950268816002648 Pinto, R. A., Arredondo, S. M., Bono, M. R., Gaggero, A. A., & Diaz, P. V. (2006). T helper 1/T helper 2 cytokine imbalance in respiratory syncytial virus infection is associated with increased endogenous plasma cortisol. Pediatrics, 117(5), 878. 10.1542/peds.2005-2119 Polvi, A., Linturi, H., Varilo, T., Anttonen, A., Byrne, M., Fokkema, I., . . . Muilu, J. (2013). The finnish disease heritage database (FinDis) update-A database for the genes mutated in the finnish disease heritage brought to the next-generation sequencing era 10.1002/humu.22389 Price, A. L., Spencer, C. C., & Donnelly, P. (2015). Progress and promise in understanding the genetic basis of common diseases. Proceedings.Biological Sciences, 282(1821), 20151684. 10.1098/rspb.2015.1684 Purcell, S., Neale, B., Todd-Brown, K., Thomas, L., Ferreira, M., Bender, D., . . . Sham, P. (2007). PLINK: A tool set for whole-genome association and population-based linkage analyses. American Journal of Human Genetics, 81(3), 559-575. 44631 Retrieved from http://pngu.mgh.harvard.edu/purcell/plink/ Puthothu, B., Forster, J., Heinzmann, A., & Krueger, M. (2006). TLR-4 and CD14 polymorphisms in respiratory syncytial virus associated disease. Disease Markers, 22(5-6), 303-308. 10.1155/2006/865890 Puthothu, B., Krueger, M., Forster, J., Heinze, J., Weckmann, M., & Heinzmann, A. (2007). Interleukin (IL)-18 polymorphism 133C/G is associated with severe respiratory syncytial virus infection. The Pediatric Infectious Disease Journal, 26(12), 1094-1098. 10.1097/INF.0b013e3181453579 Puthothu, B., Krueger, M., Forster, J., & Heinzmann, A. (2006). Association between severe respiratory syncytial virus infection and IL13/IL4 haplotypes. The Journal of Infectious Diseases, 193(3), 438-441. JID35423 Quintana-Murci, L. (2016). Understanding rare and common diseases in the context of human evolution. Genome Biology, 17, 225. 10.1186/s13059-016-1093-y R Development Core Team. (2008). R: A language and environment for statistical computing. R foundation for statistical computing, vienna, austria. ISBN 3-900051-07- 0, URL http://Www.R-project.org. Foundation for Statistical Computing, Vienna, Austria. Ralston, S. L., Lieberthal, A. S., Meissner, H. C., Alverson, B. K., Baley, J. E., Gadomski, A. M., . . . American Academy of Pediatrics. (2014). Clinical practice guideline: The diagnosis, management, and prevention of bronchiolitis. Pediatrics, 134(5), 1474. 10.1542/peds.2014-2742 Ramet, M., Korppi, M., & Hallman, M. (2011). Pattern recognition receptors and genetic risk for rsv infection: Value for clinical decision-making? Pediatric Pulmonology, 46(2), 101-110. 10.1002/ppul.21348

Reijonen, T., Korppi, M., Pitkakangas, S., Tenhola, S., & Remes, K. (1995). The clinical efficacy of nebulized racemic epinephrine and albuterol in acute bronchiolitis. Archives of Pediatrics & Adolescent Medicine, 149(6), 686-692. 10.1001/archpedi.1995.02170190096017 Ricciardolo, F. L., Sterk, P. J., Gaston, B., & Folkerts, G. (2004). Nitric oxide in health and disease of the respiratory system. Physiological Reviews, 84(3), 731-765. 10.1152/physrev.00034.2003 Russell, C. D., Unger, S. A., Walton, M., & Schwarze, J. (2017). The human immune response to respiratory syncytial virus infection. Clinical Microbiology Reviews, 30(2), 481-502. 10.1128/CMR.00090-16 Sabroe, I., Lloyd, C. M., Whyte, M. K., Dower, S. K., Williams, T. J., & Pease, J. E. (2002). Chemokines, innate and adaptive immunity, and respiratory disease. The European Respiratory Journal, 19(2), 350-355. 10.1183/09031936.02.00253602 Scheltema, N. M., Gentile, A., Lucion, F., Nokes, D. J., Munywoki, P. K., Madhi, S. A., . . . Bont, L. J. (2017). Global respiratory syncytial virus-associated mortality in young children (RSV GOLD): A retrospective case series. The Lancet.Global Health, 5(10), e991. S2214-109X(17)30344-3 Schmidt, M. E., & Varga, S. M. (2017). Modulation of the host immune response by respiratory syncytial virus proteins. Journal of Microbiology (Seoul, Korea), 55(3), 161-171. 10.1007/s12275-017-7045-8 Schwarzer, G. (2007). Meta: An R package for meta-analysis. https://Cran.r- project.org/doc/rnews/Rnews_2007-3.pdf. [meta: An R package for meta-analysis] R News, 7(3), 40-45, 7(3), 41, 42, 43, 44, 45. Schweitzer, J. W., & Justice, N. A. (2018). Respiratory syncytial virus infection (RSV). StatPearls. Treasure Island (FL): StatPearls Publishing LLC.NBK459215 Sherry, S. T., Ward, M., Kholodov, M., Baker, J., Phan, L., Smigielski, E. M., & Sirotkin, K. (2001). dbSNP: The NCBI database of genetic variation. Nucleic Acids Research, 29(1), 308-311. PMC29783 Shi, T., McAllister, D. A., O'Brien, K. L., Simoes, E. A. F., Madhi, S. A., Gessner, B. D., . . . RSV Global Epidemiology Network. (2017). Global, regional, and national disease burden estimates of acute lower respiratory infections due to respiratory syncytial virus in young children in 2015: A systematic review and modelling study. Lancet (London, England), 390(10098), 946-958. S0140-6736(17)30938-8 Silverman, E. K., & Palmer, L. J. (2000). Case-control association studies for the genetics of complex respiratory diseases. Am J Respir Cell Mol Biol, 22(6), 645-648. 10.1165/ajrcmb.22.6.f191 Singh, A. M., Moore, P. E., Gern, J. E., Lemanske, R. F.,Jr, & Hartert, T. V. (2007). Bronchiolitis to asthma: A review and call for studies of gene-virus interactions in asthma causation. American Journal of Respiratory and Critical Care Medicine, 175(2), 108-119. 200603-435PP

Skjerven, H. O., Megremis, S., Papadopoulos, N. G., Mowinckel, P., Carlsen, K. H., Lodrup Carlsen, K. C., & ORAACLE Study Group. (2016). Virus type and genomic load in acute bronchiolitis: Severity and treatment response with inhaled adrenaline. The Journal of Infectious Diseases, 213(6), 915-921. 10.1093/infdis/jiv513 Smith, D. J., & Lusis, A. J. (2002). The allelic structure of common disease. Human Molecular Genetics, 11(20), 2455-2461. 10.1093/hmg/11.20.2455 Smyth, R. L., & Openshaw, P. J. (2006). Bronchiolitis. Lancet (London, England), 368(9532), 312-322. S0140-6736(06)69077-6 Stoppelenburg, A. J., von Hegedus, J. H., Huis in't Veld, R., Bont, L., & Boes, M. (2014). Defective control of vitamin D receptor-mediated epithelial STAT1 signalling predisposes to severe respiratory syncytial virus bronchiolitis. The Journal of Pathology, 232(1), 57-64. 10.1002/path.4267 Sudmant, P. H., Rausch, T., Gardner, E. J., Handsaker, R. E., Abyzov, A., Huddleston, J., . . . Korbel, J. O. (2015). An integrated map of structural variation in 2,504 human genomes. Nature, 526(7571), 75-81. 10.1038/nature15394 Suhre, K., Arnold, M., Bhagwat, A. M., Cotton, R. J., Engelke, R., Raffler, J., . . . Graumann, J. (2017). Connecting genetic risk to disease end points through the human blood plasma proteome. Nature Communications, 8, 14357. 10.1038/ncomms14357 Tak, Y. G., & Farnham, P. J. (2015). Making sense of GWAS: Using epigenomics and genome engineering to understand the functional relevance of SNPs in non-coding regions of the human genome. Epigenetics & Chromatin, 8(1), 57. 10.1186/s13072- 015-0050-4 Tal, G., Mandelberg, A., Dalal, I., Cesar, K., Somekh, E., Tal, A., . . . Amariglio, N. (2004). Association between common toll-like receptor 4 mutations and severe respiratory syncytial virus disease. The Journal of Infectious Diseases, 189(11), 2057-2063. 10.1086/420830 Tan, H. L., & Rosenthal, M. (2013). IL-17 in lung disease: Friend or foe? Thorax, 68(8), 788-790. 10.1136/thoraxjnl-2013-203307 Telcian, A. G., Zdrenghea, M. T., Edwards, M. R., Laza-Stanca, V., Mallia, P., Johnston, S. L., & Stanciu, L. A. (2017). Vitamin D increases the antiviral activity of bronchial epithelial cells in vitro. Antiviral Research, 137, 93-101. S0166-3542(16)30369-2 Terao, C., Kawaguchi, T., Dieude, P., Varga, J., Kuwana, M., Hudson, M., . . . Allanore, Y. (2017). Transethnic meta-analysis identifies GSDMA and PRDM1 as susceptibility genes to systemic sclerosis. Annals of the Rheumatic Diseases, 76(6), 1150-1158. 10.1136/annrheumdis-2016-210645 The IMpact-RSV Study Group. (1998). Palivizumab, a humanized respiratory syncytial virus monoclonal antibody, reduces hospitalization from respiratory syncytial virus infection in high-risk infants. the IMpact-RSV study group. Pediatrics, 102(3 Pt 1), 531-537. Thomsen, S. F., Stensballe, L. G., Skytthe, A., Kyvik, K. O., Backer, V., & Bisgaard, H. (2008). Increased concordance of severe respiratory syncytial virus infection in identical twins. Pediatrics, 121(3), 493-496. 10.1542/peds.2007-1889

Tian, C., Hromatka, B. S., Kiefer, A. K., Eriksson, N., Noble, S. M., Tung, J. Y., & Hinds, D. A. (2017). Genome-wide association and HLA region fine-mapping studies identify susceptibility loci for multiple common infections. Nature Communications, 8(1), 5. 10.1038/s41467-017-00257-5 Tian, M., Zhao, D. Y., Wen, G. Y., Shi, S. Y., & Chen, R. H. (2007). Association between interleukin-8 gene-251 locus polymorphism and respiratory syncytial virus bronchiolitis and post-bronchiolitis wheezing in infants. Zhonghua Er Ke Za Zhi = Chinese Journal of Pediatrics, 45(11), 856-859. Tishkoff, S. A., Reed, F. A., Friedlaender, F. R., Ehret, C., Ranciaro, A., Froment, A., . . . Williams, S. M. (2009). The genetic structure and history of africans and african americans. Science (New York, N.Y.), 324(5930), 1035-1044. 10.1126/science.1172257 Tregoning, J. S., & Schwarze, J. (2010). Respiratory viral infections in infants: Causes, clinical symptoms, virology, and immunology. Clinical Microbiology Reviews, 23(1), 74-98. 10.1128/CMR.00032-09 Troy, N. M., & Bosco, A. (2016). Respiratory viral infections and host responses; insights from genomics. Respiratory Research, 17, 156. 10.1186/s12931-016-0474-9 Turner, S., Armstrong, L. L., Bradford, Y., Carlson, C. S., Crawford, D. C., Crenshaw, A. T., . . . Ritchie, M. D. (2011). Quality control procedures for genome wide association studies. Current Protocols in Human Genetics / Editorial Board, Jonathan L.Haines ...[Et Al.], CHAPTER, Unit1.19. 10.1002/0471142905.hg0119s68 Ullemar, V., Magnusson, P. K., Lundholm, C., Zettergren, A., Melen, E., Lichtenstein, P., & Almqvist, C. (2016). Heritability and confirmation of genetic association studies for childhood asthma in twins. Allergy, 71(2), 230-238. 10.1111/all.12783 Vandini, S., Calamelli, E., Faldella, G., & Lanari, M. (2017). Immune and inflammatory response in bronchiolitis due to respiratory syncytial virus and rhinovirus infections in infants. Paediatric Respiratory Reviews, 24, 60-64. S1526-0542(17)30001-5 Venter, J. C., Adams, M. D., Myers, E. W., Li, P. W., Mural, R. J., Sutton, G. G., . . . Zhu, X. (2001). The sequence of the human genome. Science (New York, N.Y.), 291(5507), 1304-1351. 10.1126/science.1058040 Vicente, C. T., Revez, J. A., & Ferreira, M. A. R. (2017). Lessons from ten years of genome- wide association studies of asthma. Clinical & Translational Immunology, 6(12), n/a. 10.1038/cti.2017.54 Viet AL , Kroesbergen HT , Verhoeff AP , Seidell JC , Otten F , Veldhuizen H van. (2004). Risk factors and health in the netherlands, a survey by municipal public; regional results. RIVM Report 260854005. Visscher, P. M., Wray, N. R., Zhang, Q., Sklar, P., McCarthy, M. I., Brown, M. A., & Yang, J. (2017). 10 years of GWAS discovery: Biology, function, and translation. American Journal of Human Genetics, 101(1), 5-22. S0002-9297(17)30240-9 Vissing, N. H., Chawes, B. L., & Bisgaard, H. (2013). Increased risk of pneumonia and bronchiolitis after bacterial colonization of the airways as neonates. American Journal of Respiratory and Critical Care Medicine, 188(10), 1246-1252. 10.1164/rccm.201302- 0215OC

Wang, K., Li, M., & Bucan, M. (2007). Pathway-based approaches for analysis of genomewide association studies. American Journal of Human Genetics, 81(6), 1278- 1283. Retrieved from http://www.ncbi.nlm.nih.gov/pmc/articles/PMC2276352/ Wang, Q., Lu, Q., & Zhao, H. (2015). A review of study designs and statistical methods for genomic epidemiology studies using next generation sequencing. Frontiers in Genetics, 6, 149. 10.3389/fgene.2015.00149 Ward, L. D., & 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(Database issue), 930. 10.1093/nar/gkr917 Wennergren, G., Hansson, S., Engstrom, I., Jodal, U., Amark, M., Brolin, I., & Juto, P. (1992). Characteristics and prognosis of hospital-treated obstructive bronchitis in children aged less than two years. Acta Paediatrica (Oslo, Norway : 1992), 81(1), 40- 45. 10.1111/j.1651-2227.1992.tb12076.x Wilson, J., Rowlands, K., Rockett, K., Moore, C., Lockhart, E., Sharland, M., . . . Hull, J. (2005). Genetic variation at the IL10 gene locus is associated with severity of respiratory syncytial virus bronchiolitis. The Journal of Infectious Diseases, 191(10), 1705-1709. JID33932 Wirth, B. (2000). An update of the mutation spectrum of the survival motor neuron gene (SMN1) in autosomal recessive spinal muscular atrophy (SMA). Human Mutation, 15(3), 228-237. 10.1002/(SICI)1098-1004(200003)15:3<228::AID- HUMU3>3.0.CO;2-9 Yan, Q., Brehm, J., Pino-Yanes, M., Forno, E., Lin, J., Oh, S. S., . . . Celedon, J. C. (2017). A meta-analysis of genome-wide association studies of asthma in puerto ricans. The European Respiratory Journal, 49(5), 2016. Print 2017 May 10.1183/13993003.01505- 2016. Yucesoy, B., Kaufman, K. M., Lummus, Z. L., Weirauch, M. T., Zhang, G., Cartier, A., . . . Bernstein, D. I. (2015). Genome-wide association study identifies novel loci associated with diisocyanate-induced occupational asthma. Toxicological Sciences : An Official Journal of the Society of Toxicology, 146(1), 192-201. 10.1093/toxsci/kfv084 Zhang, J., Basher, F., & Wu, J. D. (2015). NKG2D ligands in tumor immunity: Two sides of a coin. Frontiers in Immunology, 6, 97. 10.3389/fimmu.2015.00097 Zhou, J., Zhang, X., Liu, S., Wang, Z., Chen, Q., Wu, Y., . . . Huang, Z. (2016). Genetic association of TLR4 Asp299Gly, TLR4 Thr399Ile, and CD14 C-159T polymorphisms with the risk of severe RSV infection: A meta-analysis. Influenza and Other Respiratory Viruses, 10(3), 224-233. 10.1111/irv.12378 Zhu, Z., Zhang, F., Hu, H., Bakshi, A., Robinson, M. R., Powell, J. E., . . . Yang, J. (2016). Integration of summary data from GWAS and eQTL studies predicts complex trait gene targets. Nature Genetics, 48(5), 481-487. 10.1038/ng.3538 Zhuang, J. J., Zondervan, K., Nyberg, F., Harbron, C., Jawaid, A., Cardon, L. R., . . . Morris, A. P. (2010). Optimizing the power of genome-wide association studies by using publicly available reference samples to expand the control group. Genetic Epidemiology, 34(4), 319-326. 10.1002/gepi.20482

Zondervan, K. T., & Cardon, L. R. (2007). Designing candidate gene and genome-wide case- control association studies. Nature Protocols, 2(10), 2492-2501. 10.1038/nprot.2007.366 Zorc, J. J., & Hall, C. B. (2010). Bronchiolitis: Recent evidence on diagnosis and management. Pediatrics, 125(2), 342-349. 10.1542/peds.2009-2092

Original publications

I Pasanen, A., Karjalainen, M.K., Bont, L., Piippo-Savolainen, E., Ruotsalainen, M., Goksör, E., Kumawat, K., Hodemaekers, H., Nuolivirta, K., Jartti, T., Wennergren, G., Hallman, M., Rämet, M., & Korppi, M. (2017). Genome-wide association study of polymorphisms predisposing to bronchiolitis. Sci Rep 7:41653. II Pasanen, A., Karjalainen, M.K., Kummola, L., Waage, J., Bønnelykke, K., Ruotsalainen, M., Piippo-Savolainen, E., Goksör, E., Nuolivirta, K., Chawes, B., Vissing, N., Bisgaard, H., Jartti, T., Wennergren, G., Junttila, I., Hallman, M., Korppi, M., & Rämet, M. (2017). NKG2D gene variation and susceptibility to viral bronchiolitis in childhood. Manuscript. III Husby, A., Pasanen, A., Waage, J., Sevelsted, A., Hodemaekers, H., Janssen, R., Karjalainen, M.K., Stokholm, J., Chawes, B.L., Korppi, M., Wennergren, G., Heinzmann, A., Bont, L., Bisgaard, H., & Bønnelykke, K. (2017). CDHR3 gene variation and childhood bronchiolitis. J Allergy Clin Immunol 140:1469-1471.e7. Reprinted with permission from Springer Nature (I) and Elsevier (III).

Original publications are not included in the electronic version of the dissertation.

86 ACTA UNIVERSITATIS OULUENSIS SERIES D MEDICA

1445. Alaraudanjoki, Viivi (2018) Erosive tooth wear and associated factors in Northern Finland Birth Cohort 1966 1446. Sinikumpu, Suvi-Päivikki (2018) Skin diseases and their association with systemic diseases in the Northern Finland Birth Cohort 1966 1447. Nortunen, Simo (2018) Stability assessment of isolated lateral malleolar supination-external rotation-type ankle fractures 1448. Härönen, Heli (2018) Collagen XIII as a neuromuscular synapse organizer : roles of collagen XIII and its transmembrane form, and effects of shedding and overexpression in the neuromuscular system in mouse models 1449. Kajula, Outi (2018) Periytyvän rintasyöpäalttiusmutaation (BRCA1/2) kantajamiesten hypoteettinen perinnöllisyysneuvontamalli 1450. Jurvelin, Heidi (2018) Transcranial bright light : the effect on human psychophysiology 1451. Järvimäki, Voitto (2018) Lumbar spine surgery, results and factors predicting outcome in working-aged patients 1452. Capra, Janne (2018) Differentiation and malignant transformation of epithelial cells : 3D cell culture models 1453. Panjan, Peter (2018) Innovative microbioreactors and microfluidic integrated biosensors for biopharmaceutical process control 1454. Saarela, Ulla (2018) Novel culture and organoid technologies to study mammalian kidney development 1455. Virtanen, Mari (2018) The development of ubiquitous 360° learning environment and its effects on students´ satisfaction and histotechnological knowledge 1456. Vuollo, Ville (2018) 3D imaging and nonparametric function estimation methods for analysis of infant cranial shape and detection of twin zygosity 1457. Tervaskanto-Mäentausta, Tiina (2018) Interprofessional education during undergraduate medical and health care studies 1458. Peroja, Pekka (2018) Oxidative stress in diffuse large B-cell lymphoma and follicular lymphoma, and TP53 mutations and translocations of MYC, Bcl-2 and Bcl-6 in diffuse large B-cell lymphoma 1459. Bose, Muthiah (2018) Molecular and functional characterization of ABRAXAS and PALB2 genes in hereditary breast cancer predisposition 1460. Klemola, Tero (2018) Flexible hallux valgus : Results of a new surgical technique Book orders: Granum: Virtual book store http://granum.uta.fi/granum/ D 1461 OULU 2018 D 1461