The Interaction Between Environmental Risk Factors, Especially Vitamin D Exposure, Epstein‐Barr Virus Infection and DNA

THE INTERACTION BETWEEN ENVIRONMENTAL RISK

FACTORS AND DNA METHYLATION IN MULTIPLE

SCLEROSIS

Lawrence T C Ong

A thesis submitted in fulfilment of the requirement for the degree of Doctor of Philosophy

FACULTY OF MEDICINE AND HEALTH

THE UNIVERSITY OF SYDNEY

2020

TABLE OF CONTENTS

Preface……………………………………………………………………..…………………..…………………..……………………. v Acknowledgements……………………………………………………………..…………………..…………………..………… vi Publications and presentations……………………………………………………………..…………………..…………. viii List of abbreviations……………………………………………..…………………..…………………………………………….. x Abstract……………………………………………..…………………..……………………………………………………..……… xiii

CHAPTER ONE Introduction………………………………..…………………..………………..……….……..…………………………………………. 1 1.1 Introduction……………………………………………..…………………..………………..……….……..………………… 2 1.2 Multiple sclerosis……………………………………………..…………………..………………..……….………………… 3 1.2.1 Genetic risk factors……………………………………………..…………………..………………..……….…… 4 1.2.2 Environmental risk factors……………..…………………..……………..……………………………………. 5 1.2.2.1 Latitude, ultraviolet radiation and vitamin D……………..……………………………. 6 1.2.2.2 Epstein‐Barr virus……………..…………………..……………..……………………………….... 9 1.2.2.3 A critical period for MS risk……………..………………….…..…………………………….. 12 1.3 Vitamin D…………………………………………..…………………..………………..……….…………………………….. 12 1.3.1 The biology of vitamin D……………………………………..…………………..………………..………….. 12 1.3.2 Vitamin D receptors……………………………………………..…………………..………………..……….… 14 1.3.3 Vitamin D and the immune system……………..…………………..………..………………………….. 15 1.4 Epigenetics……………………………………………..…………………..………………..……….……………………….. 16 1.4.1 Non‐coding RNA……………………………………………..…………………..………………..……….……… 17 1.4.2 Histone marks……………………………………………..…………………..………………..……….………… 18 1.4.3 DNA methylation……………………………………………..…………………..………………..……….……. 19 1.4.3.1 Mechanisms underlying DNA methylation and demethylation………………. 20 1.4.3.2 Consequences of DNA methylation……………………………………………………….. 23 1.4.3.3 DNA methylation in cellular development and differentiation………………. 23 1.4.3.4 DNA methylation plasticity and developmental epigenetics………………….. 25 1.4.3.5 DNA methylation assays………………………………………………………………………… 25 1.5 The environment, DNA methylation and disease…………………………………………………………….. 27 1.5.1 Chemical exposures……………………………………………..…………………..………………..…………. 28 1.5.2 Nutrient intake……………………………………………..…………………..………………..……….………. 28

TABLE OF CONTENTS

1.5.3 Vitamin D……………………………………………..…………………..………………..……….……………….. 29 1.5.4 Exposure to microbes and infections………..…………………..………..……………………………. 32 1.5.5 Epstein‐Barr virus………..…………………..………..…………………..………..………………………….. 33 1.6 Summary and aims………..…………………..………..…………………..………..…………………………………… 33 1.7 References………..…………………..………..…………………..………..……………………………………………….. 38

CHAPTER TWO Regulation of the methylome in differentiation from adult stem cells may underpin vitamin D risk in multiple sclerosis ……..…………………..………..…………………..………..……………………………………………..……. 49 Co‐author contributions…….……………..………..…………………..………..……………………………………………….. 78

CHAPTER THREE Age‐dependent VDR peak DNA methylation as a mechanism for latitude‐dependent MS risk……. 79 Co‐author contributions…….…………..………..…………………..………..………………………………………………… 106

CHAPTER FOUR Transcribed B lymphocyte genes and multiple sclerosis risk genes are underrepresented in Epstein‐ Barr Virus hypomethylated regions…………..………..…………………..………..……………………………………… 107 Co‐author contributions..…………..………..…………………..………..……………………………………………………. 133

CHAPTER FIVE Summary and conclusions…………..………..…………………..………..…………………………………………………… 134 5.1 Summary of findings…………..………..…………………..………..…………………………………………………. 135 5.1.1 Recapitulation of haematopoietic progenitor DNA methylation in progeny cells…. 135 5.1.1.1 Significance of these findings…………..………..…………………..………..………….. 136 5.1.1.2 Study limitations……..………..…………………..………..………..…………………..……. 136 5.1.2 Effects of calcitriol on DNA methylation in myeloid cells……..………..……………………. 137 5.1.2.1 Significance of these findings……..………..…………………..………..………..……… 138 5.1.2.2 Study limitations……..………..…………………..………..………..…………………..……. 138 5.1.3 DNA methylation of LCLs at MS risk genes……..………..…………………..…………………….. 139 5.1.3.1 Significance of these findings……..………..…………………..…………………………. 140 5.1.3.2 Study limitations………..………..…………………..……..………..…………………..……. 140

TABLE OF CONTENTS

5.2 Future directions……..………..…………………..………..………..…………………..……………………………… 141 5.3 Conclusions…..………..…………………..………..………..…………………..…………..…………………………… 143 5.4 References…..………..…………………..………..………..…………………..…………………………………………. 144

APPENDICES Appendix One – Supplementary material for Chapter Two………..…………………..………..………..…….. 146 Appendix Two – Supplementary material for Chapter Three…..…………………..………..………..……….. 170 Appendix Three – Supplementary material for Chapter Four…..…………………..………..………..………. 178 Appendix Four – Vitamin D and DNA methylation in development, aging and disease (manuscript in preparation)…..………………..……….…..…………………..………..………..………………………..………..………..…… 189 Appendix Five – LINE‐1 DNA methylation in response to aging and vitamin D (manuscript in preparation)…..………………..……….…..…………………..………..………..………………………..………………………. 225

iii

PREFACE

Declaration

This thesis is submitted to the University of Sydney in fulfilment of requirements for the degree of

Doctor of Philosophy. The work presented in this thesis is original except as acknowledged in the text. I, Lawrence Ong, hereby declare that I have not submitted this material, either in full or in part, for a degree at this or any other institution.

Signature:…………………………………………….……….. Date:……………………………………………..……………

ACKNOWLEDGEMENTS

This endeavour would not have been possible without the support of so many, both directly and indirectly. To my team of dedicated supervisors, this body of work is a tribute to your patience, trust and support. Prof David Booth, for inspiring me and allowing me to wander off with your ideas in ways you never imagined possible, I owe you a great debt of gratitude. For helping me fit my research endeavours into the greater puzzle of a clinician researcher, Prof Sanjay Swaminathan – thank you. In more immediate and practical ways, Dr Grant Parnell, you have been a constant source of support throughout my time at WIMR.

Over the last five years, the Centre for Immunology and Allergy Research and the greater community at the Westmead Institute for Medical Research have become a family to me, providing a guiding hand whenever it was needed. Sincere gratitude to Stephen Schibeci, Samantha Law, Ali Afrasiabi,

Prue Gatt, Dr Ya Wang, Dr Maryam Shojaei, Jeremy Keane, Dr Jo Gamble, Monica Basuki, Dr Fiona

Mackay, Dr Ming Wei Lin, A/Prof Ben Tang, Prof Graeme Stewart, A/Prof Scott Byrne and Prof David

Brown. In particular, I would like to thank my constant desk mates Dr Velma Herwanto and Dr Nicole

Fewings for truly enriching my PhD journey.

This work would not have been possible without the financial support provided by a co‐funded

National Health and Medical Research Council, MS Research Australia and Trish MS Foundation

Scholarship. In the final years of my project, I was also supported by a NSW Health Pathology

Postgraduate Fellowship.

Last, but not least, I would like to extend my heartfelt gratitude to my family and friends, without whom this work could not have reached its full potential.

ACKNOWLEDGEMENTS

To all those mentioned above, and to those countless individuals deserving of but not mentioned, the following work is dedicated to you.

vii

PUBLICATIONS AND PRESENTATIONS

Manuscripts

1. Ong, L. T. C., Parnell, G. P., Afrasiabi, A., Stewart, G. J., Swaminathan, S., Booth, D. R. (2019)

Transcribed B lymphocyte genes and multiple sclerosis risk genes are underrepresented in

Epstein‐Barr Virus hypomethylated regions, Genes and Immunity (2019); 16, pp. 1‐9

2. Ong, L. T. C., Parnell, G. P., Veale, K., Stewart, G. J., Liddle, C., Booth, D. R. Regulation of the

methylome in differentiation from adult stem cells may underpin vitamin D risk in MS.

Submitted

3. Ong, L. T. C., Schibeci, S. D., Fewings, N. L., Booth, D. R., Parnell, G. P. Age‐dependent VDR

peak DNA methylation as a mechanism for latitude‐dependent MS risk. Submitted

4. Ong, L. T. C., Booth, D. R., Parnell, G. P. Vitamin D and DNA methylation in development,

aging and disease. Manuscript in preparation

5. Ong, L. T. C., Schibeci, S. D., Fewings, N. L., Booth, D. R., Parnell, G. P. LINE‐1 DNA

methylation in response to aging and vitamin D. Manuscript in preparation

viii

PUBLICATIONS AND PRESENTATIONS

Posters and presentations

1. Ong, L. T. C., Parnell, G. P., Stewart, G. J., Booth, D. R. MS risk loci are underrepresented in

EBV induced DNA hypomethylation. European Committee for Treatment and Research in

Multiple Sclerosis Congress, Stockholm, Sweden, September 2019. (Oral presentation)

2. Ong, L. T. C., Parnell, G. P., Stewart, G. J., Booth, D. R. Can the effects of vitamin D on MS risk

be explained by its effect on DNA methylation? European Committee for Treatment and

Research in Multiple Sclerosis Congress, Stockholm, Sweden, September 2019. (Poster)

3. Ong, L. T. C., Parnell, G. P., Stewart, G. J., Booth, D. R. The adult haematopoietic stem cell

CpG island methylome is almost entirely recapitulated in progeny cells: Implications for

multiple sclerosis. International Society of Neuroimmunology Congress, Brisbane, Australia,

October 2018. (Poster)

4. Ong, L.T.C., Parnell, G.P., Stewart, G., Booth, D.R. DNA methylation at the interface between

environment and multiple sclerosis risk. European Association for Allergy and Clinical

Immunology Congress, Munich, Germany, May 2018. (Poster prize – Autoimmunity section)

5. Ong, L., Parnell, G., Booth, D. Environmental risk factors in Multiple Sclerosis: Epigenetics at

work? Progress in MS Research Scientific Conference, Sydney, Australia, November 2017

(Poster)

LIST OF ABBREVIATIONS

5hmC 5‐hydroxymethyl cytosine

5mC 5‐methyl cytosine

ATP adenosine triphosphate

AUD Australian dollars

CD cluster of differentiation

CGI CpG island

CIS clinically isolated syndrome

CNS central nervous system

DC dendritic cell

DMR differentially methylated region

DNMT DNA methyltransferase(s)

DNMT1 DNA methyltransferase 1

DNMT3a DNA methyltransferase 3a

DNMT3L DNA methyltransferase 3L

DOHaD Developmental origin of health and disease

EBNA Epstein‐Barr nuclear antigen

EBV Epstein‐Barr virus

EDSS Extended Disability Status Scale eQTL expression quantitative trait locus

LIST OF ABBREVIATIONS

GO gene ontology

GWAS genome wide association study

HAT histone acetyltransferase(s)

HDAC histone deacetylase(s)

HLA human leukocyte antigen

IL interleukin

IU international units

LCL lymphoblastoid cell line

LMP latent membrane protein lncRNA long non‐coding RNA

MBD methyl‐CpG‐binding domain

MHC major histocompatibility complex miRNA micro RNA mQTL methylation quantitative trait locus

MRI magnetic resonance imaging

MS multiple sclerosis

NK natural killer

OR odds ratio

RCT randomised controlled trial

LIST OF ABBREVIATIONS

RRMS relapsing remitting multiple sclerosis

RXR retinoid X receptor

PBMC peripheral blood mononuclear cell piRNA PIWI‐interacting RNA

PML progressive multifocal leukoencephalopathy

PPMS primary progressive multiple sclerosis siRNA small interfering RNA

SNP single nucleotide polymorphism

SPMS secondarily progressive multiple sclerosis

TET ten‐eleven translocase

Th2 T‐helper 2

Tregs regulatory T cells

UV ultraviolet

UVR ultraviolet radiation

VDR vitamin D receptor

VDRE vitamin D respons(iv)e element

xii

ABSTRACT

Multiple sclerosis is an autoimmune and neurodegenerative disease that displays complex risk associations with genetic and environmental factors. Although the identified genetic risk factors point towards a strong immune signature in this disease, the mechanism by which environmental factors increase risk are unclear. The influence of environmental risk factors in multiple sclerosis is complicated by the presence of a critical period, during which certain risk factors exert their effects.

Epigenetic marks such as DNA methylation are important in regulating cell development, but appreciation of their role in mediating environmental effects on the organism are increasingly being recognised. This thesis explores the interaction between two well known multiple sclerosis risk factors, vitamin D exposure and Epstein‐Barr virus infection and DNA methylation, to determine whether DNA methylation mediates disease risk and the potential mechanisms by which this occurs.

In order to establish the presence of a mechanism for transmitting immune cell mediated risk, the

DNA methylation signature of immune cells from 11 healthy controls was determined. Specifically,

CD34+ haematopoietic progenitors, CD14+ monocytes and CD56+ NK cells were interrogated by reduced representation bisulfite sequencing to examine whether DNA methylation was recapitulated from progenitor to progeny cells. Using a 25% difference as a cut‐off for differential methylation, CpG islands demonstrated almost complete recapitulation of DNA methylation with

<1.5% of CpG islands being differentially methylated between any of the three subsets. It was also established that whilst small, variation in DNA methylation between the eleven individuals studied was detectable at CpG islands, providing a substrate for individual differences in disease risk. These results suggest that DNA methylation at CpG islands might provide a mechanism for transmission of environmentally influenced MS risk.

xiii

ABSTRACT

To determine whether vitamin D was involved in altering DNA methylation state differently between adults and children, DNA methylation of CD14+ myeloid cells cultured from CD34+ haematopoietic progenitors in the presence or absence of calcitriol was determined. Using whole genome bisulfite sequencing, there was a dearth of differentially methylated CpG sites found between cells cultured with and without calcitriol. Despite this, differentially methylated sites differed between cells of adult and paediatric origin. Examination of vitamin D receptor binding sites found marked differences in DNA methylation between CD14+ cells of adult and paediatric origin, with a propensity toward hypomethylation at these sites in cells originating from children. RNA sequencing found a quarter of differentially methylated vitamin D receptor binding sites to be associated with differentially expressed genes. These genes were enriched for biological pathways associated with adaptive immune system regulation and myeloid cell differentiation. The results suggest that children may be more susceptible to the effects of vitamin D, and that this susceptibility may form the basis for multiple sclerosis risk in later life.

Epstein‐Barr virus infection, another well known environmental risk factor for the development of multiple sclerosis, is known to cause widespread DNA hypomethylation in lymphoblastoid cell lines

(an in vitro model of Epstein‐Barr virus infection in B cells). To determine whether Epstein‐Barr virus associated multiple sclerosis risk was mediated by changes in DNA methylation at previously established multiple sclerosis risk loci, an analysis of existing whole genome bisulfite sequencing data from lymphoblastoid cell lines and activated B cells was conducted. This study showed that differential methylation was profoundly underrepresented at multiple sclerosis risk loci and B cell loci compared to an unbiased selection of genomic loci. Further analysis found these multiple sclerosis risk and B cell loci to be hypomethylated in activated B cells, suggesting that constitutive hypomethylation of these loci was the likely explanation for this observation. This study suggested

xiv

ABSTRACT

that Epstein‐Barr virus associated multiple sclerosis risk was unlikely to be mediated by DNA methylation at multiple sclerosis risk loci.

In summary, the studies in this thesis examine potential mechanisms by which DNA methylation may mediate environmental risk in multiple sclerosis. They show i) a potential mechanism for vitamin D mediated multiple sclerosis risk in childhood, ii) a possible mechanism for transmission of DNA methylation states in immune cells (that might predispose to autoimmunity) and iii) that Epstein‐

Barr virus induced DNA hypomethylation is unlikely act through multiple sclerosis risk loci to increase disease risk. Although further work is required to clarify these mechanisms, a thorough understanding might provide better targeted approaches to treatment and prevention of multiple sclerosis.

CHAPTER ONE

INTRODUCTION

CHAPTER ONE INTRODUCTION

1.1 INTRODUCTION

Multiple sclerosis (MS) is a chronic, autoimmune demyelinating and neurodegenerative condition that affects the central nervous system. It is currently estimated to affect 2.5 million people worldwide (1). Disease onset occurs typically between 18 and 45 years of age and clinical episodes may manifest with weakness, paraesthesias and visual disturbance. Recurrent exacerbations may lead to cognitive deficits and irreversible chronic physical disability, such that most patients will require a wheelchair or become bed bound 18 and 28 years after diagnosis respectively (2). A majority of patients with MS will die from the disease or its complications (3), and the median time to death is 30 years from disease onset, representing a reduction in life expectancy of 5‐10 years (4).

Because the onset of disability occurs at a relatively early age, MS is a source of significant social, psychological and economic burden. In economic terms, the annual cost of MS per person in

Australia in 2017 (direct and indirect) was AUD 68,382, ranging from AUD 30,561 for those without disability, to AUD 114,813 for those with severe disability (5).

Several risk factors are known to influence the likelihood of an individual developing MS, and the most prominent genetic and environmental risk factors have been shown to predict the majority of cases of MS onset (6). Exposures to environmental risk factors such as ultraviolet radiation (UVR), vitamin D levels and Epstein‐Barr Virus (EBV) infection are potentially modifiable, making these important targets for disease prevention. However, the specific mechanisms linking these risk factors to disease is unclear, and therefore the optimal risk modification measures are also unknown. It is now becoming increasingly clear that environmental risk factors in many diseases act through epigenetic mechanisms such as DNA methylation, histone marks and non‐coding RNA, to exert their influence on disease risk. This thesis will examine in more detail, the interaction between environmental risk factors, especially vitamin D exposure, Epstein‐Barr virus infection and DNA

CHAPTER ONE INTRODUCTION

methylation in MS risk. Consideration is also given, to the interaction between environmental and genetic risk factors.

1.2 MULTIPLE SCLEROSIS

The natural history of multiple sclerosis is related to disease subtype. The first episode of neurologic symptoms related to central nervous system (CNS) demyelination or inflammation is termed clinically isolated syndrome (CIS). This is considered to be the precursor to MS, which is diagnosed on subsequent episodes of CNS demyelination/inflammation. The majority of patients (~85%) with

MS (7) will suffer from relapsing‐remitting disease (RRMS), where clinical deteriorations are punctuated by periods of relatively quiescent or stable disease. The remaining patients will suffer from a primary progressive form (PPMS), which is defined as a gradually progressive and unremitting loss of neurological function observed for greater than one year. Patients with RRMS may develop secondary progressive MS (SPMS) which occurs in 50‐60% (8, 9) of RRMS patients usually between

10 to 30 years following disease onset. Like PPMS, SPMS manifests as a gradual, but progressive worsening of neurological function especially with regards to gait.

The treatments targeted at modifying disease in MS have increased exponentially in recent years, along with efficacy and improvements in side effect profiles. These treatments have focused on modulation of the immune system through various mechanisms including alteration of immune cell trafficking (10‐12) and cytotoxicity (13, 14). Despite these improvements, understanding and ameliorating risk factors is of key importance due to the shortcomings of disease modifying treatments. Firstly, there are no curative treatments currently in routine clinical use. Secondly, disease modifying therapies require ongoing use to maintain their effects, along with their attendant costs. These are not trivial, particularly with biologic agents such as alemtuzumab and ocrelizumab

(that are capable of depleting mature lymphocytes and B lymphocytes respectively) amongst others.

CHAPTER ONE INTRODUCTION

Thirdly, disease modifying therapies are only partially effective and lastly, disease modifying treatments have potentially significant side effects. For example, development of progressive multifocal leukoencephalopathy, a known complication of natalizumab therapy, has been estimated to occur in 1 in 44 patients possessing all three risk factors for its acquisition (15). More than 20% of

MS patients treated with alemtuzumab developed a novel autoimmune disease including thyroid, haematologic and cutaneous conditions between 2 weeks and 60 months after initial treatment (16).

Although not currently in routine clinical use, autologous haematopoietic stem cell transplant has shown significant promise in a subset of MS patients, especially those of younger age, with relapsing disease and lower Extended Disability Status Scale (EDSS) score (17).

1.2.1 Genetic risk factors

The contributors to MS risk are manifold and can be divided into genetic and environmental risk factors. Knowledge of the genetic risks underpinning MS have been elucidated to a large extent by genome wide association studies (GWAS). So far, greater than 230 common genetic variants have been identified by this method, which explain almost 20% of genetic heritability in MS (18). Thirty‐ two of these variants map to the classical human leukocyte antigen (HLA) region and account for

7.5% of heritability (19‐21). Most recently, a non‐GWAS based approach found four rare genetic variants not previously identified by GWAS to explain a further 5% of genetic heritability (18). Their associated genes (NLRP8, PRF1, HDAC7, PRKRA), like many of the common genetic variants have roles in immune function, providing support for the notion of MS as an immune mediated disease.

Although MS has primarily been thought of as a T cell mediated disease, the MS risk genes also suggest the importance of myeloid and NK cells in MS pathogenesis. With regard to myeloid cells, the foremost risk factor for MS, HLA‐DRB1*15:01 is a class II MHC molecule that is present on

CHAPTER ONE INTRODUCTION

antigen presenting cells, namely monocytes/macrophages, B cells and dendritic cells. This molecule is directly involved in the presentation of antigen to T cells, and provides specificity regarding those antigens that are presented. Mononuclear phagocytic cells and NK cells also demonstrate the highest expression of a large proportion of MS risk genes (22). ZMIZ1 is an MS risk gene that is most highly expressed in monocytes, underexpressed in MS (23, 24) and whose expression is increased on exposure with calcitriol in mononuclear phagocytes (24). IRF8 is another MS risk gene and transcription factor involved in promoting differentiation of monocytes into inflammatory dendritic cells (25). With regard to NK cells, class I MHC molecules are also implicated in modulating MS risk.

Many of these are most highly expressed by CD8+ T cells and NK cells, often highest in NK cells , perhaps indicating their importance in regulating an NK cell response affecting MS pathogenesis

(22). Key transcription factors of cytotoxic NK and CD8+ T cells are also known MS risk genes

(EOMES, TBET), usually more highly expressed in NK cells than CD8+ T cells. Their expression is lower in whole blood of MS patients (23, 26), and normalises with natalizumab treatment (26).

1.2.2 Environmental risk factors

Various environmental and lifestyle factors have been found to modify disease risk in MS. Well established factors include ultraviolet radiation, vitamin D exposure, smoking (27‐29) and EBV infection. More recently, epidemiologic studies have begun to recognise other less well known risk factors including adolescent obesity (30‐34), organic solvent exposure (35) and night shift work (36‐

38). In some cases, these risk factors have been found to interact with genetic factors, such as the

HLA allele, resulting in synergistic effects far outweighing the additive effects of each individual risk factor (39). Smoking, regardless of HLA status has been found to increase the risk of MS with estimated odds ratios of ~1.5 (27, 28). Carriers of the HLA‐DRB1*15 haplotype not bearing the HLA‐

A*02 haplotype displayed an OR of 5.7 for MS regardless of smoking status, whereas smokers with the corresponding haplotype had an OR of 12.7 for MS (40). Interaction effects have also been found

CHAPTER ONE INTRODUCTION

with passive smoking (41), adolescent obesity (31) and EBV infection (42, 43). The environmental risk factors that will be considered further in this thesis are ultraviolet (UV) radiation/vitamin D and EBV infection.

1.2.2.1 Latitude, ultraviolet radiation and vitamin D

Several epidemiologic studies have found a strong correlation between the latitudes at which an individual resides and their risk of developing MS. Specifically, individuals living further away from the equator have been shown to be at increased risk for MS (Figure 1). Exceptions to the latitude gradient have been found in Italy and northern Scandinavia, although the latitude gradient is restored once the prevalence of high‐risk HLA‐DRB1 alleles were corrected for in the Italian studies

(44). The reverse gradient in northern Scandinavia has been attributed to the high levels of dietary vitamin D intake (44). UV radiation and vitamin D levels are known to correlate with latitude and are therefore thought to be candidate mechanisms by which latitude of residence may confer MS risk.

Because these factors are closely linked, it has been difficult to ascribe MS risk definitively to either factor, and both are considered to be important risk factors.

CHAPTER ONE INTRODUCTION

Figure 1. The latitude gradient relating latitude of residence to MS prevalence. From (44)

Several studies have examined the relationship between vitamin D and the development of MS.

Individuals supplemented with >400 IU per day of vitamin D were found to have a lower risk of developing MS than those who were not supplemented (45). In addition, low vitamin D levels also appear to be associated with increased risk of developing MS (46) and increased risk of MS relapse

(47, 48).

That both UVR and vitamin D are known to have potent immunomodulatory actions lends further support to the idea that one or both of these factors plausibly contribute to MS risk. UVR is utilised in the treatment of psoriasis and atopic dermatitis, where it demonstrates clear anti‐inflammatory and immunosuppressive effects. In yet other situations, it appears to trigger immune hypersensitivity such as in systemic and cutaneous lupus and polymorphic light eruption (49). The effects of UVR on the immune system appear to be mediated by four key mechanisms. Stimulation of the innate immune system occurs with keratinocyte production of inflammatory mediators (e.g.

CHAPTER ONE INTRODUCTION

ATP, biolipids, chemokines), surface markers (e.g. Toll‐like receptors and RANK‐Ligand), cyclooxygenase‐2 and antimicrobial peptides. Reduced antigen specific responses have been shown with the decreased generation of antigen specific effector and memory T cells following application of antigen to UV irradiated skin (49). The induction of calcitriol production occurs when previtamin

D3 in skin is converted under the action of UV radiation, which in turn has immunomodulatory effects on immune cells. Lastly UVR stimulates the production of other immunomodulatory molecules such as nitric oxide, that induces the generation of suppressive T regulatory cells (50) and cis‐UCA, which stimulates higher FoxP3+ Treg proportions, increased IL‐10 production and decreased interferon gamma production (51) . Further details on the immunomodulatory effects of vitamin D can be found in section 1.3.3.

There is indirect molecular evidence linking MS risk to vitamin D. Enzymes involved in activation and metabolism of vitamin D (see Figure 3) have been identified as MS risk genes, for example 25‐ hydroxylase (CYP2R1; that catalyses conversion of vitamin D into calcidiol), 1α‐hydroxylase

(CYP27B1; that catalyses conversion of calcidiol to calcitriol) and 24‐hydroxylase (CYP24A1; that catalyses breakdown of calcitriol) have all been identified as MS risk genes. CYP27B1 (1α‐ hydroxylase) expression is also highest in DC1 (inflammatory) and DC2 (tolerogenic) dendritic cell subsets (52). This finding perhaps implicates not only vitamin D, but also myeloid cells in MS risk.

Further indirect evidence for vitamin D in MS risk has come from molecular studies examining the relationship between vitamin D, its receptor (the vitamin D receptor; VDR) and genes implicated in

MS risk. Firstly, VDR binding sites appear to coincide with multiple transcription factors that are products of MS risk genes, suggesting that vitamin D plays a role in mediating immune tolerance

(53). Another MS risk gene, ZMIZ1 demonstrates increased expression in response to a vitamin D analogue, calcipotriol, and altered expression in response to disease modifying treatments including

CHAPTER ONE INTRODUCTION

fingolimod and dimethylfumarate (24). Finally, MS risk genes ZMIZ1 and IRF8 regulate mononuclear phagocytic cell differentiation in response to vitamin D with expression of ZMIZ1 increasing and IRF8 decreasing in response to vitamin D exposure (25). Taken together, these data suggest that in MS,

UVR has immunomodulatory effects that may be mediated by vitamin D.

Despite evidence linking vitamin D to MS risk and disease activity, clinical trials of vitamin D for the treatment of MS have so far failed to indicate any beneficial effects. A systematic review of 12 randomised controlled trials (RCT) and quasi‐RCTs encompassing 933 subjects, did not find evidence of benefit for vitamin D in reducing MS disease activity (54). Outcome measures assessed in this study included EDSS, disease relapse or new gadolinium enhancing lesions on T1 weighted MRI.

Although the trials of vitamin D in MS are continuing, the lack of therapeutic effect points towards a need to understand the vitamin D related mechanisms underlying disease activity and risk in MS.

Potential reasons for treatment failure include limited activity of enzymes involved activating supplemental vitamin D (i.e. CYP2R1 and CYP27B1 – both MS risk genes, see Figure 3) and the association of the risk allele of CYP24A1 with inactivation of calcitriol in MS patients. Another little understood possibility is that the effect of vitamin D is mediated via DNA methylation. This will be considered further later in this thesis.

1.2.2.2 Epstein‐Barr Virus

Also known as human herpes virus 4, EBV Is recognised for its ability to infect, activate and persist in

B cells in a latent state. Once infected, the virus persists for the lifetime of the infected individual.

EBV is generally transmitted via saliva where it infects naïve B cells in the tonsil. This results in the B cell expression of viral latency proteins, including the Epstein‐Barr nuclear antigens (EBNA) 1, 2, 3A,

3B, 3C, LP and latent membrane proteins (LMP) 1, 2A and 2B. This pattern of gene expression in B

CHAPTER ONE INTRODUCTION

cells is known as latency III. Activation of these resting B cells drives them to become B cell blasts.

After entering the germinal centre, the expression of most of these latency proteins is downregulated (latency II), allowing differentiation of the B cell into a memory B cell (55). The latently infected memory B cell avoids immune detection as it does not express viral proteins

(latency 0) except during cell division, at which point EBNA1 is expressed (latency I) (56). When these latent cells return to the tonsil they can terminally differentiate into plasma cells, triggering viral replication (lytic phase) (55, 57) and promoting transmission to other individuals (58) or infection of other naïve B cells within the host (see Figure 2).

Figure 2. The Epstein‐Barr virus life cycle. Adapted from (57). EBNA – Epstein Barr nuclear antigen.

LMP – latent membrane protein.

There are several lines of evidence that link EBV infection with increased risk of developing MS.

Individuals who have suffered from infectious mononucleosis (a syndrome associated with acute onset EBV infection with fevers, pharyngitis, lymphadenopathy and malaise) have an increased risk of developing multiple sclerosis (RR 2.17‐2.3) (59, 60). This risk appears to be even higher (OR 7) in individuals who are HLA‐DR2 positive (where HLA‐DR2 is the serotype that includes HLA‐DRB1*15

CHAPTER ONE INTRODUCTION

and HLA‐DRB1*16 allele groups) (61). The risk for developing MS has also been found to increase with the magnitude of antibody responses directed against EBV (62, 63). Although EBV infection is an almost ubiquitous disease, it appears to be a prerequisite for the development of MS. One study could not find an MS patient without evidence of EBV infection (64). Another study found a pooled odds ratio of 0.06 for being EBV seronegative amongst MS patients versus control subjects (65). The fact that EBV appears to underscore MS risk is highlighted in children, where seropositivity rates in children with MS are much higher than those without the disease (86‐99% in MS vs 64‐72% in those without) (66‐68).

Antibody responses to EBV also appear to correlate with MS disease activity. IgM and IgA responses against EBV have been found in patients with clinically active but not quiescent MS (69). In addition, levels of anti EBNA1 specific IgG were also found to predict progression from clinically isolated syndrome (a single MS like clinical event) to MS (70). Secondly, anti EBNA1 IgG antibodies have been found to correlate with the number of demyelinating brain lesions as determined by MRI (70, 71).

There is significant molecular evidence linking EBV infection to multiple sclerosis risk. EBV infection in B cells results in dysregulated expression of 139 of 255 MS risk genes (72). The same study found the genotype at 44 of 201 MS risk loci to be associated with proximal gene expression in lymphoblastoid cell lines (LCLs; an in vitro model of B cell EBV infection), with 35 of these showing stronger genotype effects than in whole blood. In addition, EBV transcription factor binding sites were found to be overrepresented amongst MS risk genes and EBNA2 expression was associated with expression levels of five MS risk genes in LCLs in a genotype dependent fashion. Although not implicating EBV conclusively in MS risk, these results suggest that EBV infection plays at least a facilitative role in MS risk.

CHAPTER ONE INTRODUCTION

1.2.2.3 A critical period for MS risk

Some environmental risk factors in MS appear to be mediated by a critical period, such that susceptibility to risk factors occurs during childhood/adolescence but not during adulthood, or to a much lesser extent. Although this phenomenon does not necessarily apply to all environmental risk factors, the effects of EBV infection and to some extent UV radiation and vitamin D appear to be mediated by a critical period. The earliest evidence for this, comes from migration studies showing that the risk of developing MS increased for those migrating from low prevalence to high prevalence regions, but only if they migrated prior to adolescence (73‐77). These data provide support for UV radiation or vitamin D influences on disease risk prior to adolescence. Evidence for a critical period effect of EBV infection can be seen with infectious mononucleosis, which has its peak incidence between 15 and 24 years of age (78).

Other environmental risk factors also appear to be mediated by a critical period, including the aforementioned obesity (30, 34), which only appears to be a risk factor in childhood and adolescence, but not adulthood. Participating in shift work before the age of 20 (OR 1.5) is more highly associated with development of MS than when commencing shift work at 20 years of age or older (OR 1.2) (36).

1.3 VITAMIN D

1.3.1 The biology of vitamin D

Very little vitamin D is contained in foods, such that humans derive ~90% of their Vitamin D requirements from exposure to sunlight (79). UVB radiation (a constituent of sunlight) in particular, causes 7‐dehydrocholesterol to be converted into previtamin D3, which isomerises into vitamin D3.

Vitamin D3 is released from the cellular plasma membrane and binds to vitamin D binding protein

CHAPTER ONE INTRODUCTION

(DBP) in the circulation. It is subsequently hydroxylated in the liver by 25‐hydroxylase (CYP2R1) to

25‐hydroxyvitamin D3 (25(OH)D3). This prohormone is converted predominantly by the kidney through the action of 1α‐hydroxylase (CYP27B1) into 1,25(OH)2D3 or calcitriol (Figure 3). An extraordinary number of these enzymes are encoded by MS risk genes, pointing to the importance of this process in MS pathogenesis. In humans, the concentration of 25(OH)D3 is almost 1000 times higher than 1,25(OH)2D3 and its half‐life is also markedly longer (80). In addition, 1,25(OH)2D3 concentrations are maintained by homeostatic mechanisms within a physiological range for as long as possible, such that concentrations of 25(OH)D3 and 1,25(OH)2D3 do not correlate in normal individuals (81).

CHAPTER ONE INTRODUCTION

Figure 3. The metabolic pathway involved in the calcitriol synthesis and degradation. Adapted from

(82). Genes in blue denote MS genetic risk loci.

1.3.2 Vitamin D receptors

Calcitriol exerts its genomic effects through binding the vitamin D receptor (VDR), which is comprised of the N‐terminal DNA binding domain and C‐terminal ligand‐binding domain. Binding of calcitriol to the ligand‐binding domain results in heterodimerisation of the VDR with the retinoid X receptor (RXR). This heterodimer binds regions of DNA known as vitamin D response elements

(VDREs), which lie in the promoter regions of vitamin D responsive genes and lead to subsequent

CHAPTER ONE INTRODUCTION

upregulation or suppression of DNA transcription. Interestingly, VDREs have been found near the site of genes associated with MS risk (e.g. IRF8) (83) which may in part explain the latitude gradient in this disease.

1.3.3 Vitamin D and the immune system

Although calcitriol is known to be important in bone formation and calcium homeostasis, it also has important roles in immune function. There are several lines of evidence which support this. Firstly, several immune cell subsets express 1α‐hydroxylase which permits conversion of inactive 25(OH)D3 to calcitriol (84‐87). Secondly, most immune cell subsets express VDRs, and their expression appears to change with cellular maturation and differentiation, suggesting variable responsiveness to calcitriol across these stages. Thirdly, vitamin D has been shown to modulate a variety of immune cell functions, especially in dendritic cells (DCs) and lymphocytes. For example, calcitriol has been found to inhibit DC maturation through inhibition of CD1a, HLA‐DR and co‐stimulatory molecule expression (CD40, CD80 & CD86) (88, 89). In addition, DC cytokine secretion is also modulated by exposure to calcitriol, resulting in decreased IL‐12 and IL‐23 secretion (pro‐inflammatory) and increased IL‐10 secretion (anti‐inflammatory) (89). This pattern of activity has been shown to limit inflammatory Th1 and Th17 responses, whilst favouring the development of T regulatory cells

(Tregs). With regard to lymphocytes, VDR expression is known to increase significantly with T cell activation, although studies of its effects on T cell proliferation have yielded conflicting results (90‐

93). Calcitriol does however, appear to induce IL‐10 secreting Tregs in the absence of antigen presenting cells (91, 94). In activated B cells, calcitriol appears to inhibit proliferation, differentiation and effector functions such as immunoglobulin secretion (84).

CHAPTER ONE INTRODUCTION

1.4 EPIGENETICS

Epigenetics is the study of the regulation of the human genome through mechanisms that are independent of the DNA sequence (95). Epigenetic studies have already been heralded for their importance in understanding the interaction between genetics, environmental exposures and MS risk (39). As the fundamental DNA sequence does not vary between cell types within an individual

(with very few exceptions e.g. B cell and T cell receptors), gene expression directed by epigenetic regulation is pivotal to the acquisition of tissue and lineage specific characteristics. The mechanisms responsible for this regulation include non‐coding RNAs, histone marks and DNA methylation. These elements are involved in a complex interplay with each other and conventional genetic machinery, and their settings are known to change depending on the developmental stage of the cell.

A primary concern of the epigenetic machinery is the control of DNA accessibility in order to regulate gene expression. Understanding DNA accessibility requires an understanding of the higher order organisation of DNA. Histones are positively charged proteins that bind negatively charged DNA.

Nucleosomes are an octameric structure consisting of two of each of the four core histone subunits, namely H2A, H2B, H3 and H4. DNA is wound around the nucleosome, forming the chief component of chromatin. Chromatin can adopt a highly compacted conformation (heterochromatin) where transcription is repressed, or an open conformation (euchromatin) that is favourable for transcription (Figure 4).

CHAPTER ONE INTRODUCTION

Figure 4. Targets of epigenetic modifications, and their relative structures. Adapted from Yan et al.

(96)

1.4.1 Non‐coding RNA

Non‐coding RNAs include micro RNA (miRNA), small interfering RNA (siRNA), PIWI‐interacting RNA

(piRNA) and long non‐coding RNA (lncRNA). Unlike messenger RNA (mRNA), they do not encode for proteins. miRNA, siRNA and piRNA predominantly exert their effects in conjunction with active proteins, that then bind and/or cleave complementary transcripts, thereby repressing or silencing corresponding genes (97). They are also known to interact with or recruit other components of the epigenetic machinery, including those involved in histone modifications or DNA methylation (98‐

100). The major difference between miRNA and siRNA is that the latter has only one mRNA target, where miRNA are capable of silencing multiple targets (101). Unlike miRNA and siRNA, piRNA are processed from single‐stranded RNA precursors, are active predominantly in stem cells and the germ line, and have a major role in silencing transposable elements (97), which would otherwise result in genomic instability.

CHAPTER ONE INTRODUCTION

Long non‐coding RNAs are defined as transcripts with a primary sequence greater than 200 nucleotides in length, that lack protein‐coding potential (102). They function in the form of RNA‐ protein complexes to influence gene expression. Firstly, they exhibit a guide function by recruiting epigenetic complexes to specific genomic sites (e.g. Xist, which recruits PRC2 to lay down the repressive H3K27 methylation mark in random X chromosome inactivation) (103, 104). They can also act as scaffolds to coordinate the function of distinct transcription regulatory complexes (104) which are capable of acting in cis or trans.

1.4.2 Histone marks

Post translational modifications of amino acid residues on histone tails are known to be associated with open, closed or bivalent chromatin states, which are differentially accessible to transcription factors and enhancers. The covalent additions are denoted by the histone and amino acid residue where the modification occurs and the type of modification e.g. acetylation is denoted by “ac” and methylation is denoted by “me”, followed by the number of covalent modifications. H3K27me3 refers to trimethylation of the lysine reside at position 27 of the H3 histone, which is known to be repressive. These marks are made and erased by a variety of histone modifiers, such as the abovementioned PRC2. The overall effect of the modification is dependent not only on the type of modification, but also its location. Histone modifications are generally considered to be readily reversible, in comparison to DNA methylation which is generally considered to be stable (105).

The effects of histone marks on gene expression are mediated by the recruitment of chromatin remodellers. These ATP dependent factors utilise energy from ATP hydrolysis to disrupt DNA ‐ histone interactions, thereby mobilising nucleosomes, resulting in chromatin compaction or opening. These factors are also capable of nucleosome removal, a process known as nucleosome eviction/disassembly. There are three classes of chromatin remodellers that recognise different

CHAPTER ONE INTRODUCTION

histone tail variants, including the SWI‐SNF complex (that recognises acetylated histones), ISWI

(whose target is currently unknown) and CHD (which recognises methylated histone tails) (106).

1.4.3 DNA methylation

DNA methylation involves the covalent addition of a methyl group to cytosine residues, typically in the context of a guanine nucleotide i.e. a CpG dinucleotide. Cytosine residues may also exist in CHG and CHH contexts (where H refers to A, T or C nucleotides), although these are not generally methylated in mammals (107). DNA methylation is known to be important in multiple physiologic functions, such as random X inactivation in females, and genomic imprinting. It is also thought to play a role in stabilising the human genome by preventing the unwanted transposition of transposon derived repetitive elements which would otherwise result in significant genomic disruption and mutagenesis (108, 109). These sequences are relatively CpG rich, and uniformly methylated to prevent this from occurring (110).

The human genome has a lower density of CpG dinucleotides than expected (111). Genomic regions with high CpG density are designated as CpG islands (CGI), and these are found in the promoter regions of approximately 60% of human genes (112). CGIs are significant in that most are sites of transcription initiation, regardless of their remoteness from currently annotated promoters (113). As a result, most CGI are unmethylated.

DNA methylation may be heritable and fixed (for example at imprinted loci); heritable but change in the process of cell differentiation (114) and with aging (115); or non‐heritable, changing in response to environmental factors. In addition, genetic differences between individuals can influence methylation states (116, 117).

CHAPTER ONE INTRODUCTION

1.4.3.1 Mechanisms underlying DNA methylation and demethylation

Different enzymes are involved in the maintenance of existing DNA methylation states and de novo

DNA methylation. During mitosis, DNA methyltransferase 1 (DNMT1) maintains DNA methylation by replicating the methylation pattern of hemimethylated DNA (one strand of newly synthesised DNA is devoid of methylation marks; see Figures 5B & 6A) with its obligate partner, ubiquitin‐like plant homeodomain and RING finger domain 1 (UHRF1) (118). DNMT3A, DNMT3B and DNMT3L participate in de novo DNA methylation (119). In both maintenance and de novo DNA methylation, S‐ adenosyl‐methionine functions as a methyl donor (120) and is therefore a co‐factor in the reaction.

Although the enzymes involved in DNA methylation have been known for some time, the processes and mediators of DNA demethylation have only been more recently elucidated. The family of enzymes designated as ten‐eleven translocases (TET) are involved in the active demethylation of methylated DNA. Ascorbic acid is also known to increase the rate at which TET mediated demethylation occurs. TET enzymes catalyse the conversion of 5‐methylcytosine (5mC) to 5‐ hydroxymethylcytosine (5hmC) (121, 122) with subsequent oxidisation to 5‐formylcytosine (5fC) and

5‐carboxycytosine (5caC) (123‐125). Thymine DNA glycosylase (TDG) is then involved in the excision of 5fC or 5caC, which coupled with base excision repair mechanisms, leads to replacement with an unmethylated cytosine residue (126) (Figure 5A). Passive DNA demethylation occurs with replication‐dependent loss of 5mC or oxidised 5mC, in the absence of the DNA methylation maintenance mediated by DNMT1/UHRF1 (Figure 5B).

CHAPTER ONE INTRODUCTION

Figure 5. A) Stepwise TET mediated active demethylation of DNA. The final step involved in DNA methylation involves base excision repair mechanisms not depicted here. B) A diagram depicting active and passive DNA demethylation. Active demethylation is driven primarily by the family of ten‐ eleven‐translocase enzymes (TET), whereas passive demethylation occurs in the absence of maintenance DNA methylation mechanisms following DNA replication. In each diagram, the top and bottom strands are forward and reverse strands respectively. Figure 5A adapted from (127)

The regulatory mechanisms involved in recruiting the DNA methylation/demethylation machinery are complex and less well understood. Interactions with histone marks are known to influence DNA methylation and vice versa. For example, current models propose that during early development, methylation of H3K4 occurs before de novo DNA methylation under the influence of RNA polymerase II binding. This in turn recruits H3K4 specific methyltransferases (128). Because RNA

CHAPTER ONE INTRODUCTION

polymerase II is mostly bound to CGI in early development, these regions tend to carry the H3K4 methylation mark. DNMT3L is involved in recruiting de novo methyltransferase enzymes DNMT3A and 3B, and is inhibited by H3K4 methylation (129), resulting in decreased DNA methylation at CGI.

Figure 6. A) Hemimethylated DNA occurs when only a single DNA strand is methylated, typically following DNA synthesis. The methylated strand serves as a guide for DNMT1 to faithfully replicate methylation on the newly synthesised strand. B) DNA methylation is typically associated with chromatin compaction (heterochromatin), lack of access to transcriptional machinery and therefore transcriptional repression, whereas its absence is associated with an open chromatin state

(euchromatin) and transcription activity. This is particularly true of gene promoters and transcription start sites. C) Methyl‐CpG‐binding domain proteins such as MeCP1 bind methylated DNA and recruit transcriptional corepressors that participate in chromatin compaction and silence transcription. D)

Methylated CpGs can also inhibit the binding of transcription factors and therefore transcription.

CHAPTER ONE INTRODUCTION

MBD – methyl‐CpG‐binding domain protein. In C) and D), filled circles indicate methylated CpGs, whereas open circles are unmethylated. Figure 6B adapted from (130).

1.4.3.2 Consequences of DNA methylation

Changes in DNA methylation are important due to their association with gene expression through alterations in the conformational state of DNA. This can result in an increase or decrease in gene transcription by affecting transcription factor activity. Generally, increased DNA methylation is associated with compaction of chromatin filaments, reduced transcription factor access and repression of gene expression at that region (Figure 6B). This is particularly true in the case of gene promoters and transcription start sites (116). However, the opposite is thought to occur in gene bodies and increased methylation at these sites is generally thought to be associated with increased transcription (111).

Transcriptional repression resulting from DNA methylation occurs via multiple mechanisms, foremost of which are the methyl‐CpG‐binding domain (MBD) proteins e.g. MeCP1 & 2. This family of 11 proteins all harbour a MBD domain, that has the ability to bind single symmetrically methylated CpG nucleotides (131, 132). Most of the members of this family also have a transcriptional repression domain which can recruit other factors that condense chromatin such as histone modifiers (133) (Figure 6C). A secondary mechanism which is probably less important in DNA methylation related transcriptional repression, is the inhibition of transcription factor binding (107)

(Figure 6D). This mechanism is likely to operate at CpG poor promoters.

1.4.3.3 DNA methylation in cellular development and differentiation

In the earliest stages of mammalian development, genomic DNA undergoes two phases of large‐ scale demethylation followed by remethylation. The first wave of demethylation occurs during

CHAPTER ONE INTRODUCTION

migration of primordial germ cells, with remethylation in prospermatogonia and developing oocytes

(134). After fertilisation and prior to implantation, there is rapid demethylation of the zygote such that it reaches a nadir in DNA methylation just prior to the blastocyst stage (135). Both maternal and paternal DNA methylation is completely erased, with the exception of imprinting control regions, which avoid demethylation in both primordial germ cells and in the preimplantation zygote (Figure

7).

Figure 7. DNA methylation in early mammalian development from (135). GV – germinal vesicle, MII – second meiotic division, PGCs=primordial germ cells

DNA methylation also plays a key role in cellular differentiation. In haematopoiesis for example, the process of lineage commitment and differentiation is accompanied by stepwise alterations in DNA methylation (136). Many of these alterations correspond to functional pathways important in lineage maintenance or subset specific function (137‐139). Methylation maps of various haematopoietic lineages demonstrate similarities between cells of the myeloid lineage, which are quite distinct from lymphoid lineage cells (140). Cells of the haematopoietic lineage demonstrate greater similarity in DNA methylation to each other than to cells of non‐haematopoietic origin such

CHAPTER ONE INTRODUCTION

as skin cells or connective tissue (140). Distinctive DNA methylation signatures also differentiate other human tissues (141).

1.4.3.4 DNA methylation plasticity and developmental epigenetics

Although DNA methylation is thought to be a relatively stable epigenetic mark, it has also been shown to be change with time. The methylation state of some loci have been shown to change after short term exposures to environmental stimuli (142, 143). Others change over much longer periods of time, such as in aging, where small, but discernible global decreases in DNA methylation (115) and repetitive element methylation have been found (144), in conjunction with bidirectional, locus specific changes (145, 146). The rate of change in DNA methylation, or plasticity, also varies with age. Gestation and the first few years of life, are developmental periods thought to be sensitive to environmental influences at least in part due to the susceptibility of epigenetic marks during tissue differentiation. Once differentiated, epigenetic marks although still susceptible to change, are somewhat more stable (147). One such example is peripheral blood DNA methylation in children, which has been found to change more rapidly over time in comparison to adults (148). There are also differences in the susceptibility period of different tissues based on when they mature (147).

These findings suggest that susceptibility to changes in DNA methylation (such as those due to environment) may be greater in early life than in adulthood. Thus, DNA methylation has the potential to act as a substrate for the influence of environmental exposures during critical periods, and more broadly, during early life. Environmental effects on early life DNA methylation and disease will be further considered in the next section.

1.4.3.5 DNA methylation assays

DNA methylation assays can broadly be divided into genome wide and regional approaches.

Regional approaches include pyrosequencing and mass array and are useful when a region of

CHAPTER ONE INTRODUCTION

interest is known. These techniques are typically suited to profiling large numbers of samples.

Genome wide approaches can be divided into restriction enzyme digestion, affinity enrichment and bisulfite conversion methods (149). Enzyme digestion methods utilise methylation specific enzymes that cleave only unmethylated target sequences e.g. BstUI, HpaII. The unmethylated fraction can then be size selected and sequenced (MRE‐seq) or hybridised to an array and quantitated

(comprehensive high‐throughput arrays for relative methylation; CHARM). Affinity enrichment‐ based methods use MBD proteins or antibodies to bind methylated DNA sequences that can then be sequenced (MBDCap‐seq, methylated DNA immunoprecipitation sequencing; MeDIP‐seq) or profiled using an array (MBD‐chip). Bisulfite conversion‐based methods rely on the treatment of genomic

DNA with sodium bisulfite. This leads to deamination of unmethylated cytosines to uracils, whilst methylated cytosines remain unaffected (150). Bisulfite modified DNA fragments can then be hybridised to arrays that distinguish between methylated and unmethylated fragments (e.g. Infinium

MethylationEPIC BeadChip), selected on the basis of fragment size and CpG density prior to next generation sequencing (e.g. reduced representation bisulfite sequencing; RRBS) or subjected to next generation sequencing without enrichment (whole genome bisulfite sequencing; WGBS).

There are several factors to be considered when choosing between DNA methylation assays. The breadth of genomic coverage or specific genomic regions to be profiled may dictate which method is used. For example, RRBS enriches for CpG islands and promoter regions due to the relatively high density of CpG residues at these regions. Probes in array‐based methods may target specific regions of interest. A further consideration is the amount of input DNA required for a specific assay. For example, bisulfite conversion significantly degrades DNA, and may therefore require greater starting amounts to achieve desired sequencing coverage. This can be problematic if target DNA is only available in small quantities e.g. rare cell subsets or cultured cells.

CHAPTER ONE INTRODUCTION

Cost is a significant factor in DNA methylation profiling, and experiments examining DNA methylation need to consider the relative importance of breadth of coverage versus number of biological replicates. For example, WGBS experiments might involve fewer biological replicates, but provide data on most genome wide CpGs (~28 million in humans). This can be contrasted with bisulfite modification arrays e.g. Infinium MethylationEPIC BeadChip (Illumina; >850,000 CpGs) and pyrosequencing (300‐500bp), where a higher number of biological replicates can be sequenced at much lower genomic coverage. Finally, bioinformatic resources should also be considered amongst the various DNA methylation profiling methods. Genome wide, next‐generation sequencing methods will typically require sophisticated bioinformatic pipelines and computing infrastructure to ascertain sequencing quality and perform sequence alignment, methylation calling and analysis.

1.5 THE ENVIROMENT, DNA METHYLATION AND DISEASE

Genetic variation, including those due to single nucleotide polymorphisms (SNPs), is thought to account for a large proportion of the variation in DNA methylation in a phenomenon known as methylation quantitative trait loci (mQTL) (151). However, environmental exposures are also known to contribute to changes in DNA methylation with subsequent alterations in gene expression, phenotype (152), and the development of disease. The degree of variation due to environmental exposures also varies based on the exposure level, chronicity and type.

The effects of the environment on DNA methylation, also form the basis of developmental epigenetics. Developmental epigenetics examines the effects of nutrition, infections, drugs, chemicals and other stressors on the epigenome of young individuals, and their association with disease. Work in this field was originally based on Barker’s hypothesis on the influence of early life environment on chronic disease risk, also known as developmental origin of health and disease

(DOHaD) (153). Later work suggested that these environmental influences might act at least in part,

CHAPTER ONE INTRODUCTION

through epigenetic mechanisms such as DNA methylation (147, 154). Studies so far have identified three critical periods, where environmental exposures have been shown to affect DNA methylation, individual development and therefore propensity to disease, namely embryonic development, early life and adolescence (155). The effects of various environmental exposures on DNA methylation will now be considered in turn.

1.5.1 Chemical exposures

The effect of environmental chemicals and toxins on DNA methylation have been extensively studied. Tobacco smoke (156, 157), asbestos (158‐160), benzene (161‐163) and air pollution (164,

165) are known to cause changes in DNA methylation. For example, specific CpG dinucleotides have been found in multiple studies to be differentially methylated in the whole blood of current versus never smokers (166). In adults and newborns exposed to tobacco smoke, a number of whole blood differentially methylated regions were also found to be differentially methylated in lung cancer

(156), suggesting that changes in DNA methylation may play a role in the pathogenesis of smoking related lung cancer.

1.5.2 Nutrient intake

The effects of altered nutrient intake on DNA methylation is another area of intense research interest. Different nutrients may act by inhibiting enzymes involved in the alteration of epigenetic marks e.g. DNMT, histone deacetylases (HDAC), histone acetyltransferases (HAT) or by altering substrate availability for these enzymes (167). Studies in mice have shown that protein restriction led to marked decreases in PPARα and glucocorticoid receptor methylation which was accompanied by increases in expression of these genes (168). Furthermore, dietary restriction in mice has been found to cause transcriptional changes in enzymes involved in DNA methylation (Dnmt3b, Tet2,

Tet3) and to protect from age related changes in DNA methylation. Perhaps unsurprisingly, DNA methylation changes were also found in genes involved in lipid metabolism (169). In humans,

CHAPTER ONE INTRODUCTION

reversible decreases in DNA methylation have been ascribed to methyl‐deficient diets (167). This finding may also have functional consequences, with DNA hypomethylation due to low dietary folate associated with increased risk of pancreatic (170) and colorectal cancers (171).

Decreased nutrient intake in the form of prenatal famine exposure has also been found to have lasting effects on DNA methylation. Individuals exposed to famine in utero during the Dutch Hunger

Winter, had decreased DNA methylation at the IGF2 gene compared with same sex unexposed siblings, six decades following the famine (172). The protein encoded by this gene has growth factor activity, although the phenotypic effects on affected versus control siblings was not examined in this study.

1.5.3 Vitamin D

Interest in the effects of Vitamin D on DNA methylation has grown in recent years due to its broad ranging effects and links with chronic disease. As expected with a molecule with pleiotropic effects, the effects on DNA methylation gathered to date have been varied, compounded by the heterogeneity of cell types being studied, different forms of vitamin D measured or supplemented and methylation assays utilised. Nevertheless, studies into this area have yielded interesting insights into organism development, genome stability and chronic disease risk. A comprehensive review of vitamin D and DNA methylation is provided in Appendix 4.

Animal models studying the effect of prenatal vitamin D deficiency have found effects on imprinted genes that are transmitted through multiple generations (173). Another study found generally lower

DNA methylation in mouse sperm at several sites associated with vitamin D deficiency including at genes related to development and metabolism such as cadherin, Wnt and PDGF (174). Effects have

CHAPTER ONE INTRODUCTION

also been noted on genes associated with physiological regulation such as inflammation (through alterations in IκBα methylation) (175), body fat (176) and blood pressure (177).

In humans, observational studies on the effects of vitamin D on DNA methylation in cord blood have generally shown a lack of effect (178‐180), however randomised control trials have been more promising. One study showed DNA methylation differed between vitamin D and control subjects at

<0.01% of assayed sites. Differential methylation occurred in both directions. At four to six weeks postpartum, mothers had methylation gain/loss at 200/102 CpGs, whilst children had gain/loss at

217/213 CpGs. Interestingly, genes associated with methylation loss in children were involved in regulation of apoptosis and antigen presentation (181). Umbilical cord tissue RXRA DNA methylation was found to be reduced in those mothers supplemented with vitamin D, with a mean difference of

~2%. The authors of this study proposed RXRA methylation as a surrogate for the action of vitamin D on other tissues following birth (182).

In immune disease, two studies have examined the effect of vitamin D on CD4+ T cells in an animal model of multiple sclerosis known as experimental autoimmune encephalomyelitis (EAE). Zeitelhofer et al. (183) found lower global DNA methylation and predominantly lower regional DNA methylation with cholecalciferol supplementation in mice with EAE. Moore et al. (184) using a different method to determine global CD4+ T cell DNA methylation, found two‐fold increased levels in CD4+ T cells cultured with calcitriol for 3 days compared to those that were cultured in the absence of calcitriol.

The differences in global DNA methylation between the two studies may perhaps be related to the different methods of ascertaining DNA methylation.

CHAPTER ONE INTRODUCTION

In humans, several studies have examined the effect of vitamin D on immune cell DNA methylation.

An observational study comparing 11 vitamin D deficient boys with age matched controls found only

2 CpG sites (in a panel of >27000) to be differentially methylated in leukocytes (185). These sites corresponded to microtubule associated protein MAPRE2 (37% lower in vitamin D deficient), and

DIO3 (24.2% higher methylation in vitamin D deficient), that has important functions in the maturation and function of the thyroid axis. A randomised controlled trial of 70 individuals found cholecalciferol supplementation over 16 weeks to be associated with dose‐dependent increases in global leukocyte DNA methylation (186). These changes correlated with serum calcidiol levels. Junge et al. (187) examined cord blood of 3 children with high and low serum calcidiol levels. They found

508 DMRs, of which 311 were hypomethylated in those with high calcidiol levels. The authors found the methylation of the TSLP enhancer, to be correlated in a separate validation cohort, but only in those with high serum calcidiol levels, perhaps demonstrating an association between elevated serum vitamin D and predisposition to allergic disease. Chavez‐Valencia et al. (188) did not find any statistically significant differentially methylated CpGs in adult peripheral blood mononuclear cells cultured with calcitriol for 120 hours. They speculated that this may have been related to decreased methylomic plasticity (as the cells were of adult origin) and masking of methylation changes in a heterogeneous cell population.

The mechanisms underlying the effects of vitamin D on the methylome are unclear, but may occur due to multiple mechanisms acting in unison. These include the ability of vitamin D to alter the expression of genes that directly or indirectly affect DNA methylation. For example Bhmt1 expression is increased in CD4+ T cells of mice supplemented with cholecalciferol (184). This enzyme catalyses the generation of methionine from homocysteine. Methionine is converted to S‐ adenosylmethionine, which is a methyl‐donor utilised by DNA methyltransferases in DNA methylation. Decreased DNA methyltransferase expression was found by Zeitelhofer and colleagues

CHAPTER ONE INTRODUCTION

(183) in CD4+ T cells of cholecalciferol supplemented mice. Changes in expression of enzymes that alter DNA methylation may be related to the presence of VDR binding sites proximal to DNMT1,

DNMT3, TET2 and TET3 (189). Specific targeting mechanisms would then be required to localise active and passive alterations of DNA methylation at genic regions. Vitamin D may influence DNA methylation indirectly through other epigenetic mechanisms. For example, targeting of DNMT3A and DNMT3B appears to be dependent on methylation of H3K36 (190, 191) which may be triggered by VDR dependent alterations in chromatin (e.g. through interactions with nuclear proteins and altering transcription of chromatin modifying/remodelling genes; reviewed in (192)).

1.5.4 Exposure to microbes and infection

Epidemiological studies have shown decreased prevalence of asthma and allergic disease in individuals living in rural areas (193). This may perhaps be associated with microbial exposures and their influence on the epigenome. One study examining cord blood DNA methylation of the FOXP3 gene found decreased DNA methylation in mothers exposed to farming environments during pregnancy (194). FOXP3 is the master regulator of T regulatory cells, which regulate the function of

Th2 cells that are involved in allergen sensitisation and atopic disease. Consistent with this result, children exposed to unpasteurised milk before five years of age were found to have higher T regulatory cell numbers and decreased rates of atopic disease (195).

Infections have been shown to be associated with DNA methylation changes in several immune cell subsets. Human cytomegalovirus infection has been found to result in diversification of NK cell subsets and loss of signalling proteins corresponding to changes in DNA methylation at promoter regions of these molecules (196). Bacterial infection remodels the DNA methylation landscape of human dendritic cells, particularly at distal enhancer elements. This is associated with changes in

CHAPTER ONE INTRODUCTION

expression levels of nearby genes (197). These alterations, especially if they persist into later life, may form the basis for aberrant immune responses, including those targeting self‐antigens.

In mice exposed prenatally to virus‐like immune activation, dysregulation of DNA methylation at multiple distinct genomic regions in brain tissue has been found, including those relevant for

GABAergic differentiation, signalling and neural development in comparison to control mice This also corresponded to impairments in behaviour (198).

1.5.5 Epstein‐Barr virus

With regard to EBV specific alterations in DNA methylation, genome‐wide analyses of LCLs have mostly found DNA methylation to be decreased relative to comparator subsets such as whole blood

(199, 200) and peripheral blood lymphocytes or leukocytes (201), perhaps due to decreased DNMT1 expression (202). This includes hypomethylation at promoter regions corresponding to B cell biological pathways in comparison to resting B cells (203). In addition, EBV infection tends to increase DNA methylation at high CpG content promoters and decrease DNA methylation in low CpG content promoters (202). Finally, EBV infection has been found to result in hypomethylation of over two‐thirds of the genome (204).

1.6 SUMMARY AND AIMS

AIMS

The overall goal of the studies comprising this thesis is to determine if DNA methylation functions as a mechanism through which environmental risk factors, namely vitamin D and Epstein‐Barr virus infection, confer risk in multiple sclerosis. This might enable more precise use of vitamin D and anti‐

CHAPTER ONE INTRODUCTION

EBV therapeutics in the treatment of MS. These goals are addressed in the following hypotheses and aims:

Chapter 2

Hypothesis 1: i) A DNA methylation setting exists in haematopoietic precursors that is transmitted to progeny cells. ii) The DNA methylation setting differs between individuals, providing the basis for varying propensity to disease.

Aim 1: To determine if a DNA methylation setting exists in haematopoietic precursors that is transmitted to progeny, and the degree to which DNA methylation varies between individuals.

Rationale: In order to demonstrate transmission of a vitamin D related DNA methylation signature from progenitor to progeny cells, it was important to show that baseline methylation state is transmitted from progenitor to progeny cells. Existence of transmitted methylation in regions known to be vitamin D regulated i.e. VDR binding sites, would provide direct evidence of conservation of

DNA methylation state for VDR targets. Variation in DNA methylation between individuals would provide a plausible mechanism for individual differences that may contribute to autoimmune and

MS disease risk.

Approach: Modified reduced representation bisulfite sequencing was used to determine the DNA methylation of CpG islands and other regions in purified CD34+, CD14+ and CD56+ cells. The DNA methylation at these regions was compared. Extant CD14+ VDR binding peak regions were also compared.

CHAPTER ONE INTRODUCTION

This aim is addressed in a submitted manuscript

Ong, L. T. C., Parnell, G. P., Veale, K., Stewart, G. J., Liddle, C., Booth, D. R. Regulation of the methylome in differentiation from adult stem cells may underpin vitamin D risk in multiple sclerosis.

Chapter 3

Hypothesis 2: i) Calcitriol alters DNA methylation in differentiating haematopoietic progenitors in an age‐dependent manner. ii) DNA methylation at VDR peaks varies between differentiating haematopoietic precursors of adult and paediatric origin.

Aim 2: To determine whether vitamin D alters DNA methylation in differentiating haematopoietic precursors and the influence of chronological age on the breadth and magnitude of these changes.

Rationale: There is limited data concerning the effects of vitamin D on the human DNA methylome in immune cells. In vitro studies to date have shown no effect of vitamin D supplementation on DNA methylation in adult, peripheral blood mononuclear cells cultured for three days in the presence of calcitriol.

Approach: CD34+ haematopoietic precursors were isolated from adult peripheral blood PBMCs and cord blood. These were cultured in serum free media in the presence and absence of calcitriol, before isolation of CD14+ progeny. Whole genome bisulfite sequencing was performed and differentially methylated regions (DMRs) were calculated between vitamin D and non‐vitamin D supplemented samples as well as between adult and paediatric samples.

CHAPTER ONE INTRODUCTION

This aim is addressed in a submitted manuscript:

Ong, L. T. C., Schibeci, S., Fewings, N., Booth, D. R., Parnell, G. Age‐dependent VDR peak DNA methylation as a mechanism for latitude‐dependent MS risk.

Chapter 4

Hypothesis 3: MS risk loci are enriched in regions of hypomethylation in EBV infected B cells.

Aim 3: To determine whether EBV induced dysregulation of the B cell methylome might contribute to EBV related MS risk.

Rationale: Previous studies have shown widespread DNA hypomethylation following EBV infection, reminiscent of patterns seen in certain human malignancies. The mechanism by which this dysregulation affects MS risk, if at all, is unclear. Previous work has shown gene expression in EBV infected B cell lines (lymphoblastoid cell lines) to be associated with MS risk loci.

Approach: Extant whole genome bisulfite sequencing data was interrogated to determine local DNA methylation states proximal to MS risk loci and other unbiased gene regions to compare the degree of DNA hypomethylation occurring at these loci.

CHAPTER ONE INTRODUCTION

This aim is addressed in a published manuscript:

Ong, L. T. C., Parnell, G. P., Afrasiabi, A., Stewart, G. J., Swaminathan, S., Booth, D. R. (2019)

Transcribed B lymphocyte genes and multiple sclerosis risk genes are underrepresented in Epstein‐

Barr Virus hypomethylated regions, Genes and Immunity. 2019. 16:1‐9.

CHAPTER ONE INTRODUCTION

1.7 REFERENCES

1. Dendrou CA, Fugger L, Friese MA. Immunopathology of multiple sclerosis. Nat Rev Immunol. 2015;15(9):545‐58. 2. Scalfari A, Neuhaus A, Degenhardt A, Rice GP, Muraro PA, Daumer M, et al. The natural history of multiple sclerosis: a geographically based study 10: relapses and long‐term disability. Brain. 2010;133(Pt 7):1914‐29. 3. Compston A, Coles A. Multiple sclerosis. Lancet (London, England). 2008;372(9648):1502‐17. 4. Brønnum‐Hansen H, Koch‐Henriksen N, Stenager E. Trends in survival and cause of death in Danish patients with multiple sclerosis. Brain. 2004;127(4):844‐50. 5. Ahmad H, Palmer AJ, Campbell JA, Van der Mei I, Taylor B. Health economic impacto of multiple sclerosis in Australia in 2017: An analysis of MS Research Australia's platform ‐ the Australian MS Longitudinal Study (AMSLS). Multiple Sclerosis Research Australia; 2018. 6. Van der Mei I, Lucas RM, Taylor B, Valery P, Dwyer T, Kilpatrick TJ, et al. Population attributable fractions and joint effects of key risk factors for multiple sclerosis. Multiple Sclerosis Journal. 2016;22(4):461‐9. 7. Howard J, Trevick S, Younger DS. Epidemiology of multiple sclerosis. Neurologic clinics. 2016;34(4):919‐39. 8. Tremlett H, Zhao Y, Devonshire V. Natural history of secondary‐progressive multiple sclerosis. Multiple Sclerosis Journal. 2008;14(3):314‐24. 9. Lublin FD, Reingold SC. Defining the clinical course of multiple sclerosis: results of an international survey. Neurology. 1996;46(4):907‐11. 10. Rice GP, Hartung H‐P, Calabresi PA. Anti‐α4 integrin therapy for multiple sclerosis: mechanisms and rationale. Neurology. 2005;64(8):1336‐42. 11. Noda H, Takeuchi H, Mizuno T, Suzumura A. Fingolimod phosphate promotes the neuroprotective effects of microglia. Journal of neuroimmunology. 2013;256(1‐2):13‐8. 12. Wegner C, Stadelmann C, Pförtner R, Raymond E, Feigelson S, Alon R, et al. Laquinimod interferes with migratory capacity of T cells and reduces IL‐17 levels, inflammatory demyelination and acute axonal damage in mice with experimental autoimmune encephalomyelitis. Journal of neuroimmunology. 2010;227(1‐2):133‐43. 13. Buzzard K, Broadley S, Butzkueven H. What do effective treatments for multiple sclerosis tell us about the molecular mechanisms involved in pathogenesis? International journal of molecular sciences. 2012;13(10):12665‐709. 14. Cox AL, Thompson SA, Jones JL, Robertson VH, Hale G, Waldmann H, et al. Lymphocyte homeostasis following therapeutic lymphocyte depletion in multiple sclerosis. European journal of immunology. 2005;35(11):3332‐42. 15. Berger JR, Fox RJ. Reassessing the risk of natalizumab‐associated PML. Journal of neurovirology. 2016;22(4):533‐5. 16. Cossburn M, Pace A, Jones J, Ali R, Ingram G, Baker K, et al. Autoimmune disease after alemtuzumab treatment for multiple sclerosis in a multicenter cohort. Neurology. 2011;77(6):573‐9. 17. Muraro PA, Pasquini M, Atkins HL, Bowen JD, Farge D, Fassas A, et al. Long‐term outcomes after autologous hematopoietic stem cell transplantation for multiple sclerosis. JAMA neurology. 2017;74(4):459‐69. 18. Mitrovič M, Patsopoulos NA, Beecham AH, Dankowski T, Goris A, Dubois B, et al. Low‐ frequency and rare‐coding variation contributes to multiple sclerosis risk. Cell. 2018;175(6):1679‐87. e7. 19. Moutsianas L, Jostins L, Beecham AH, Dilthey AT, Xifara DK, Ban M, et al. Class II HLA interactions modulate genetic risk for multiple sclerosis. Nature genetics. 2015;47(10):1107.

CHAPTER ONE INTRODUCTION

20. Patsopoulos N, Baranzini SE, Santaniello A, Shoostari P, Cotsapas C, Wong G, et al. The Multiple Sclerosis Genomic Map: Role of peripheral immune cells and resident microglia in susceptibility. BioRxiv. 2017:143933. 21. Patsopoulos NA, Barcellos LF, Hintzen RQ, Schaefer C, Van Duijn CM, Noble JA, et al. Fine‐ mapping the genetic association of the major histocompatibility complex in multiple sclerosis: HLA and non‐HLA effects. PLoS genetics. 2013;9(11):e1003926. 22. Parnell GP, Booth DR. The multiple sclerosis (MS) genetic risk factors indicate both acquired and innate immune cell subsets contribute to MS pathogenesis and identify novel therapeutic opportunities. Frontiers in immunology. 2017;8:425. 23. Parnell GP, Gatt PN, Krupa M, Nickles D, McKay FC, Schibeci SD, et al. The autoimmune disease‐associated transcription factors EOMES and TBX21 are dysregulated in multiple sclerosis and define a molecular subtype of disease. Clinical immunology. 2014;151(1):16‐24. 24. Fewings N, Gatt P, McKay F, Parnell G, Schibeci S, Edwards J, et al. The autoimmune risk gene ZMIZ1 is a vitamin D responsive marker of a molecular phenotype of multiple sclerosis. Journal of autoimmunity. 2017;78:57‐69. 25. Parnell GP, Schibeci SD, Fewings NL, Afrasiabi A, Law SP, Samaranayake S, et al. The latitude‐ dependent autoimmune disease risk genes ZMIZ1 and IRF8 regulate mononuclear phagocytic cell differentiation in response to vitamin D. Human molecular genetics. 2019;28(2):269‐78. 26. McKay FC, Gatt PN, Fewings N, Parnell GP, Schibeci SD, Basuki MA, et al. The low EOMES/TBX21 molecular phenotype in multiple sclerosis reflects CD56+ cell dysregulation and is affected by immunomodulatory therapies. Clinical Immunology. 2016;163:96‐107. 27. Hawkes C. Smoking is a risk factor for multiple sclerosis: a metanalysis. Multiple Sclerosis Journal. 2007;13(5):610‐5. 28. Handel AE, Williamson AJ, Disanto G, Dobson R, Giovannoni G, Ramagopalan SV. Smoking and multiple sclerosis: an updated meta‐analysis. PloS one. 2011;6(1):e16149. 29. Hedström AK, Bäärnhielm M, Olsson T, Alfredsson L. Tobacco smoking, but not Swedish snuff use, increases the risk of multiple sclerosis. Neurology. 2009;73(9):696‐701. 30. Gianfrancesco MA, Acuna B, Shen L, Briggs FB, Quach H, Bellesis KH, et al. Obesity during childhood and adolescence increases susceptibility to multiple sclerosis after accounting for established genetic and environmental risk factors. Obesity research & clinical practice. 2014;8(5):e435‐e47. 31. Hedström AK, Bomfim IL, Barcellos L, Gianfrancesco M, Schaefer C, Kockum I, et al. Interaction between adolescent obesity and HLA risk genes in the etiology of multiple sclerosis. Neurology. 2014;82(10):865‐72. 32. Langer‐Gould A, Brara SM, Beaber BE, Koebnick C. Childhood obesity and risk of pediatric multiple sclerosis and clinically isolated syndrome. Neurology. 2013;80(6):548‐52. 33. Munger KL, Bentzen J, Laursen B, Stenager E, Koch‐Henriksen N, Sørensen TI, et al. Childhood body mass index and multiple sclerosis risk: a long‐term cohort study. Multiple Sclerosis Journal. 2013;19(10):1323‐9. 34. Munger KL, Chitnis T, Ascherio A. Body size and risk of MS in two cohorts of US women. Neurology. 2009;73(19):1543‐50. 35. Barragán‐Martínez C, Speck‐Hernández CA, Montoya‐Ortiz G, Mantilla RD, Anaya J‐M, Rojas‐ Villarraga A. Organic solvents as risk factor for autoimmune diseases: a systematic review and meta‐ analysis. PloS one. 2012;7(12):e51506. 36. Hedström A, Åkerstedt T, Olsson T, Alfredsson L. Shift work influences multiple sclerosis risk. Multiple Sclerosis Journal. 2015;21(9):1195‐9. 37. Hedström AK, Åkerstedt T, Hillert J, Olsson T, Alfredsson L. Shift work at young age is associated with increased risk for multiple sclerosis. Annals of neurology. 2011;70(5):733‐41. 38. Gustavsen S, Søndergaard H, Oturai D, Laursen B, Laursen JH, Magyari M, et al. Shift work at young age is associated with increased risk of multiple sclerosis in a Danish population. Multiple sclerosis and related disorders. 2016;9:104‐9.

CHAPTER ONE INTRODUCTION

39. Olsson T, Barcellos LF, Alfredsson L. Interactions between genetic, lifestyle and environmental risk factors for multiple sclerosis. Nature Reviews Neurology. 2017;13(1):25. 40. Hedström AK, Katsoulis M, Hössjer O, Bomfim IL, Oturai A, Sondergaard HB, et al. The interaction between smoking and HLA genes in multiple sclerosis: replication and refinement. European journal of epidemiology. 2017;32(10):909‐19. 41. Hedström AK, Bomfim IL, Barcellos LF, Briggs F, Schaefer C, Kockum I, et al. Interaction between passive smoking and two HLA genes with regard to multiple sclerosis risk. International journal of epidemiology. 2014;43(6):1791‐8. 42. Sundqvist E, Sundström P, Linden M, Hedström A, Aloisi F, Hillert J, et al. Epstein‐Barr virus and multiple sclerosis: interaction with HLA. Genes and immunity. 2012;13(1):14. 43. Simon K, Van der Mei I, Munger K, Ponsonby A, Dickinson J, Dwyer T, et al. Combined effects of smoking, anti‐EBNA antibodies, and HLA‐DRB1* 1501 on multiple sclerosis risk. Neurology. 2010;74(17):1365‐71. 44. Simpson S, Blizzard L, Otahal P, Van der Mei I, Taylor B. Latitude is significantly associated with the prevalence of multiple sclerosis: a meta‐analysis. Journal of Neurology, Neurosurgery & Psychiatry. 2011;82(10):1132‐41. 45. Munger KL, Zhang SM, O’Reilly E, Hernán MA, Olek MJ, Willett WC, et al. Vitamin D intake and incidence of multiple sclerosis. Neurology. 2004;62(1):60‐5. 46. Munger KL, Levin LI, Hollis BW, Howard NS, Ascherio A. Serum 25‐hydroxyvitamin d levels and risk of multiple sclerosis. JAMA. 2006;296(23):2832‐8. 47. Mowry EM, Krupp LB, Milazzo M, Chabas D, Strober JB, Belman AL, et al. Vitamin D status is associated with relapse rate in pediatric‐onset multiple sclerosis. Annals of Neurology. 2010;67(5):618‐24. 48. Simpson Jr S, Taylor B, Blizzard L, Ponsonby AL, Pittas F, Tremlett H, et al. Higher 25‐ hydroxyvitamin D is associated with lower relapse risk in multiple sclerosis. Annals of Neurology. 2010;68(2):193‐203. 49. Hart PH, Norval M, Byrne SN, Rhodes LE. Exposure to ultraviolet radiation in the modulation of human diseases. Annual Review of Pathology: Mechanisms of Disease. 2019;14:55‐81. 50. Yu C, Fitzpatrick A, Cong D, Yao C, Yoo J, Turnbull A, et al. Nitric oxide induces human CLA+ CD25+ Foxp3+ regulatory T cells with skin‐homing potential. Journal of Allergy and Clinical Immunology. 2017;140(5):1441‐4. e6. 51. Correale J, Farez MF. Modulation of multiple sclerosis by sunlight exposure: role of cis‐ urocanic acid. Journal of neuroimmunology. 2013;261(1‐2):134‐40. 52. Shahijanian F, Parnell GP, McKay FC, Gatt PN, Shojoei M, O'Connor KS, et al. The CYP27B1 variant associated with an increased risk of autoimmune disease is underexpressed in tolerizing dendritic cells. Human molecular genetics. 2014;23(6):1425‐34. 53. Booth D, Ding N, Parnell G, Shahijanian F, Coulter S, Schibeci S, et al. Cistromic and genetic evidence that the vitamin D receptor mediates susceptibility to latitude‐dependent autoimmune diseases. Genes and immunity. 2016;17(4):213. 54. Jagannath VA, Filippini G, Di Pietrantonj C, Asokan GV, Robak EW, Whamond L, et al. Vitamin D for the management of multiple sclerosis. Cochrane Database of Systematic Reviews. 2018(9). 55. Thorley‐Lawson DA, Gross A. Persistence of Epstein‐Barr virus and the origins of associated lymphomas. New England Journal of Medicine. 2004;350:1328‐37. 56. Hochberg D, Middeldorp JM, Catalina M, Sullivan JL, Luzuriaga K, Thorley‐Lawson DA. Demonstration of the Burkitt's lymphoma Epstein‐Barr virus phenotype in dividing latently infected memory cells in vivo. Proceedings of the National Academy of Sciences. 2004;101(1):239‐44. 57. Münz C. Latency and lytic replication in Epstein–Barr virus‐associated oncogenesis. Nature Reviews Microbiology. 2019;17(11):691‐700. 58. Hadinoto V, Shapiro M, Sun CC, Thorley‐Lawson DA. The dynamics of EBV shedding implicate a central role for epithelial cells in amplifying viral output. PLoS Pathog. 2009;5(7):e1000496.

CHAPTER ONE INTRODUCTION

59. Handel AE, Williamson AJ, Disanto G, Handunnetthi L, Giovannoni G, Ramagopalan SV. An updated meta‐analysis of risk of multiple sclerosis following infectious mononucleosis. PloS one. 2010;5(9):e12496. 60. Thacker EL, Mirzaei F, Ascherio A. Infectious mononucleosis and risk for multiple sclerosis: a meta‐analysis. Annals of neurology. 2006;59(3):499‐503. 61. Nielsen T, Rostgaard K, Askling J, Steffensen R, Oturai A, Jersild C, et al. Effects of infectious mononucleosis and HLA‐DRB1* 15 in multiple sclerosis. Multiple sclerosis. 2009. 62. Munger K, Levin L, O’Reilly E, Falk K, Ascherio A. Anti‐Epstein–Barr virus antibodies as serological markers of multiple sclerosis: a prospective study among United States military personnel. Multiple Sclerosis Journal. 2011;17(10):1185‐93. 63. Ascherio A, Munger KL, Lennette ET, Spiegelman D, Hernán MA, Olek MJ, et al. Epstein‐Barr virus antibodies and risk of multiple sclerosis: a prospective study. Jama. 2001;286(24):3083‐8. 64. Levin LI, Munger KL, O'reilly EJ, Falk KI, Ascherio A. Primary infection with the epstein‐barr virus and risk of multiple sclerosis. Annals of neurology. 2010;67(6):824‐30. 65. Ascherio A, Munger KL. Environmental risk factors for multiple sclerosis. Part I: the role of infection. Annals of neurology. 2007;61(4):288‐99. 66. Pohl D, Krone B, Rostasy K, Kahler E, Brunner E, Lehnert M, et al. High seroprevalence of Epstein‐Barr virus in children with multiple sclerosis. Neurology. 2006;67(11):2063‐5. 67. Banwell B, Krupp L, Kennedy J, Tellier R, Tenembaum S, Ness J, et al. Clinical features and viral serologies in children with multiple sclerosis: a multinational observational study. The Lancet Neurology. 2007;6(9):773‐81. 68. Lünemann J, Huppke P, Roberts S, Brück W, Gärtner J, Münz C. Broadened and elevated humoral immune response to EBNA1 in pediatric multiple sclerosis. Neurology. 2008;71(13):1033‐5. 69. Wandinger K‐P, Jabs W, Siekhaus A, Bubel S, Trillenberg P, Wagner H‐J, et al. Association between clinical disease activity and Epstein–Barr virus reactivation in MS. Neurology. 2000;55(2):178‐84. 70. Lünemann JD, Tintoré M, Messmer B, Strowig T, Rovira Á, Perkal H, et al. Elevated Epstein– Barr virus‐encoded nuclear antigen‐1 immune responses predict conversion to multiple sclerosis. Annals of neurology. 2010;67(2):159‐69. 71. Farrell RA, Antony D, Wall GR, Clark DA, Fisniku L, Swanton J, et al. Humoral immune response to EBV in multiple sclerosis is associated with disease activity on MRI. Neurology. 2009;73(1):32‐8. 72. Afrasiabi A, Parnell GP, Fewings N, Schibeci SD, Basuki MA, Chandramohan R, et al. Evidence from genome wide association studies implicates reduced control of Epstein‐Barr virus infection in multiple sclerosis susceptibility. Genome Medicine. 2019;11(1):26. 73. Alter M, Kahana E, Loewenson R. Migration and risk of multiple sclerosis. Neurology. 1978;28(11):1089‐. 74. Ahlgren C, Lycke J, Odén A, Andersen O. High risk of MS in Iranian immigrants in Gothenburg, Sweden. Multiple sclerosis journal. 2010;16(9):1079‐82. 75. Ahlgren C, Odén A, Lycke J. A nationwide survey of the prevalence of multiple sclerosis in immigrant populations of Sweden. Multiple Sclerosis Journal. 2012;18(8):1099‐107. 76. Kurtzke JF, Beebe GW, Norman JE. Epidemiology of multiple sclerosis in US veterans III. Migration and the risk of MIS. Neurology. 1985;35(5):672‐. 77. Gale CR, Martyn CN. Migrant studies in multiple sclerosis. Progress in neurobiology. 1995;47(4‐5):425‐48. 78. Heath Jr CW, Brodsky AL, Potolsky AI. Infectious mononucleosis in a general population. American journal of epidemiology. 1972;95(1):46‐52. 79. Holick MF, Chen TC, Lu Z, Sauter E. Vitamin D and Skin Physiology: AD‐Lightful Story. Journal of Bone and Mineral Research. 2007;22(S2):V28‐V33. 80. Lips P. Relative value of 25(OH)D and 1,25(OH)2D measurements. J Bone Miner Res. 2007;22(11):1668‐71.

CHAPTER ONE INTRODUCTION

81. Lips P, Netelenbos J, Jongen M, Van Ginkel F, Althuis A, Van Schaik C, et al. Histomorphometric profile and vitamin D status in patients with femoral neck fracture. Metabolic Bone Disease and Related Research. 1982;4(2):85‐93. 82. Christakos S, Dhawan P, Verstuyf A, Verlinden L, Carmeliet G. Vitamin D: metabolism, molecular mechanism of action, and pleiotropic effects. Physiological reviews. 2015;96(1):365‐408. 83. Ramagopalan SV, Heger A, Berlanga AJ, Maugeri NJ, Lincoln MR, Burrell A, et al. A ChIP‐seq defined genome‐wide map of vitamin D receptor binding: associations with disease and evolution. Genome research. 2010;20(10):1352‐60. 84. Chen S, Sims GP, Chen XX, Gu YY, Chen S, Lipsky PE. Modulatory effects of 1, 25‐ dihydroxyvitamin D3 on human B cell differentiation. The Journal of Immunology. 2007;179(3):1634‐ 47. 85. Hewison M, Freeman L, Hughes SV, Evans KN, Bland R, Eliopoulos AG, et al. Differential regulation of vitamin D receptor and its ligand in human monocyte‐derived dendritic cells. The Journal of Immunology. 2003;170(11):5382‐90. 86. Monkawa T, Yoshida T, Hayashi M, Saruta T. Identification of 25‐hydroxyvitamin D3 1&agr;‐ hydroxylase gene expression in macrophages. Kidney international. 2000;58(2):559‐68. 87. Sigmundsdottir H, Pan J, Debes GF, Alt C, Habtezion A, Soler D, et al. DCs metabolize sunlight‐induced vitamin D3 to'program'T cell attraction to the epidermal chemokine CCL27. Nature immunology. 2007;8(3):285‐93. 88. Berer A, Stöckl J, Majdic O, Wagner T, Kollars M, Lechner K, et al. 1, 25‐Dihydroxyvitamin D 3 inhibits dendritic cell differentiation and maturation in vitro. Experimental hematology. 2000;28(5):575‐83. 89. Penna G, Adorini L. 1α, 25‐dihydroxyvitamin D3 inhibits differentiation, maturation, activation, and survival of dendritic cells leading to impaired alloreactive T cell activation. The Journal of Immunology. 2000;164(5):2405‐11. 90. Correale J, Ysrraelit MC, Gaitán MI. Immunomodulatory effects of Vitamin D in multiple sclerosis. Brain. 2009:awp033. 91. Jeffery LE, Burke F, Mura M, Zheng Y, Qureshi OS, Hewison M, et al. 1, 25‐Dihydroxyvitamin D3 and IL‐2 combine to inhibit T cell production of inflammatory cytokines and promote development of regulatory T cells expressing CTLA‐4 and FoxP3. The Journal of Immunology. 2009;183(9):5458‐67. 92. Lacey DL, Axelrod J, Chappel J, Kahn A, Teitelbaum S. Vitamin D affects proliferation of a murine T helper cell clone. The Journal of Immunology. 1987;138(6):1680‐6. 93. Rigby W, Stacy T, Fanger MW. Inhibition of T lymphocyte mitogenesis by 1, 25‐ dihydroxyvitamin D3 (calcitriol). Journal of Clinical Investigation. 1984;74(4):1451. 94. Barrat FJ, Cua DJ, Boonstra A, Richards DF, Crain C, Savelkoul HF, et al. In vitro generation of interleukin 10–producing regulatory CD4+ T cells is induced by immunosuppressive drugs and inhibited by T helper type 1 (Th1)–and Th2‐inducing cytokines. The Journal of experimental medicine. 2002;195(5):603‐16. 95. Jaenisch R, Bird A. Epigenetic regulation of gene expression: how the genome integrates intrinsic and environmental signals. Nature genetics. 2003;33(3s):245. 96. Yan MS‐C, Matouk CC, Marsden PA. Epigenetics of the vascular endothelium. Journal of Applied Physiology. 2010;109(3):916‐26. 97. Ozata DM, Gainetdinov I, Zoch A, O’Carroll D, Zamore PD. PIWI‐interacting RNAs: small RNAs with big functions. Nature Reviews Genetics. 2019;20(2):89‐108. 98. Aravin AA, Sachidanandam R, Bourc'his D, Schaefer C, Pezic D, Toth KF, et al. A piRNA pathway primed by individual transposons is linked to de novo DNA methylation in mice. Molecular cell. 2008;31(6):785‐99. 99. Pezic D, Manakov SA, Sachidanandam R, Aravin AA. piRNA pathway targets active LINE1 elements to establish the repressive H3K9me3 mark in germ cells. Genes & development. 2014;28(13):1410‐28.

CHAPTER ONE INTRODUCTION

100. Verdel A, Jia S, Gerber S, Sugiyama T, Gygi S, Grewal SI, et al. RNAi‐mediated targeting of heterochromatin by the RITS complex. Science. 2004;303(5658):672‐6. 101. Lam JK, Chow MY, Zhang Y, Leung SW. siRNA versus miRNA as therapeutics for gene silencing. Molecular Therapy‐Nucleic Acids. 2015;4:e252. 102. Quinn JJ, Chang HY. Unique features of long non‐coding RNA biogenesis and function. Nature Reviews Genetics. 2016;17(1):47. 103. Galupa R, Heard E. X‐chromosome inactivation: new insights into cis and trans regulation. Current opinion in genetics & development. 2015;31:57‐66. 104. Engreitz JM, Ollikainen N, Guttman M. Long non‐coding RNAs: spatial amplifiers that control nuclear structure and gene expression. Nature Reviews Molecular Cell Biology. 2016;17(12):756. 105. Cedar H, Bergman Y. Linking DNA methylation and histone modification: patterns and paradigms. Nature Reviews Genetics. 2009;10(5):295. 106. Chen T, Dent SY. Chromatin modifiers and remodellers: regulators of cellular differentiation. Nature Reviews Genetics. 2014;15(2):93‐106. 107. Chen P‐Y, Feng S, Joo JWJ, Jacobsen SE, Pellegrini M. A comparative analysis of DNA methylation across human embryonic stem cell lines. Genome Biology. 2011;12(7):R62. 108. Wu M, Rinchik EM, Wilkinson E, Johnson DK. Inherited somatic mosaicism caused by an intracisternal A particle insertion in the mouse tyrosinase gene. Proceedings of the National Academy of Sciences. 1997;94(3):890‐4. 109. Ukai H, Ishii‐Oba H, Ukai‐Tadenuma M, Ogiu T, Tsuji H. Formation of an active form of the interleukin‐2/15 receptor β‐chain by insertion of the intracisternal A particle in a radiation‐induced mouse thymic lymphoma and its role in tumorigenesis. Molecular Carcinogenesis: Published in cooperation with the University of Texas MD Anderson Cancer Center. 2003;37(2):110‐9. 110. Wade PA. Methyl CpG‐binding proteins and transcriptional repression. Bioessays. 2001;23(12):1131‐7. 111. Jones PA. Functions of DNA methylation: islands, start sites, gene bodies and beyond. Nature Reviews Genetics. 2012;13(7):484. 112. Antequera F, Bird A. Number of CpG islands and genes in human and mouse. Proceedings of the National Academy of Sciences. 1993;90(24):11995‐9. 113. Deaton AM, Bird A. CpG islands and the regulation of transcription. Genes & development. 2011;25(10):1010‐22. 114. Guo H, Zhu P, Yan L, Li R, Hu B, Lian Y, et al. The DNA methylation landscape of human early embryos. Nature. 2014;511(7511):606‐10. 115. Jones MJ, Goodman SJ, Kobor MS. DNA methylation and healthy human aging. Aging cell. 2015;14(6):924‐32. 116. Bell JT, Pai AA, Pickrell JK, Gaffney DJ, Pique‐Regi R, Degner JF, et al. DNA methylation patterns associate with genetic and gene expression variation in HapMap cell lines. Genome Biology. 2011;12(1):R10. 117. Bock C, Paulsen M, Tierling S, Mikeska T, Lengauer T, Walter J. CpG island methylation in human lymphocytes is highly correlated with DNA sequence, repeats, and predicted DNA structure. PLoS genetics. 2006;2(3):e26. 118. Schermelleh L, Haemmer A, Spada F, Rösing N, Meilinger D, Rothbauer U, et al. Dynamics of Dnmt1 interaction with the replication machinery and its role in postreplicative maintenance of DNA methylation. Nucleic acids research. 2007;35(13):4301‐12. 119. Chedin F. The DNMT3 family of mammalian de novo DNA methyltransferases. Progress in molecular biology and translational science. 101: Elsevier; 2011. p. 255‐85. 120. Niculescu MD, Zeisel SH. Diet, Methyl Donors and DNA Methylation: Interactions between Dietary Folate, Methionine and Choline. The Journal of Nutrition. 2002;132(8):2333S‐5S. 121. Kriaucionis S, Heintz N. The nuclear DNA base 5‐hydroxymethylcytosine is present in Purkinje neurons and the brain. Science. 2009;324(5929):929‐30.

CHAPTER ONE INTRODUCTION

122. Tahiliani M, Koh KP, Shen Y, Pastor WA, Bandukwala H, Brudno Y, et al. Conversion of 5‐ methylcytosine to 5‐hydroxymethylcytosine in mammalian DNA by MLL partner TET1. Science. 2009;324(5929):930‐5. 123. He Y‐F, Li B‐Z, Li Z, Liu P, Wang Y, Tang Q, et al. Tet‐mediated formation of 5‐ carboxylcytosine and its excision by TDG in mammalian DNA. Science. 2011;333(6047):1303‐7. 124. Ito S, Shen L, Dai Q, Wu SC, Collins LB, Swenberg JA, et al. Tet proteins can convert 5‐ methylcytosine to 5‐formylcytosine and 5‐carboxylcytosine. Science. 2011;333(6047):1300‐3. 125. Pfaffeneder T, Hackner B, Truß M, Münzel M, Müller M, Deiml CA, et al. The discovery of 5‐ formylcytosine in embryonic stem cell DNA. Angewandte Chemie International Edition. 2011;50(31):7008‐12. 126. Wu X, Zhang Y. TET‐mediated active DNA demethylation: mechanism, function and beyond. Nature Reviews Genetics. 2017;18(9):517. 127. An J, Rao A, Ko M. TET family dioxygenases and DNA demethylation in stem cells and cancers. Experimental & molecular medicine. 2017;49(4):e323‐e. 128. Guenther MG, Levine SS, Boyer LA, Jaenisch R, Young RA. A chromatin landmark and transcription initiation at most promoters in human cells. Cell. 2007;130(1):77‐88. 129. Ooi SK, Qiu C, Bernstein E, Li K, Jia D, Yang Z, et al. DNMT3L connects unmethylated lysine 4 of histone H3 to de novo methylation of DNA. Nature. 2007;448(7154):714. 130. Van Soom A. Epigenetics and the periconception environment in ruminants. Proceedings of the Belgian Royal Academies of Medicine. 2013;2:1‐23. 131. Nan X, Meehan RR, Bird A. Dissection of the methyl‐CpG binding domain from the chromosomal protein MeCP2. Nucleic acids research. 1993;21(21):4886‐92. 132. Ohki I, Shimotake N, Fujita N, Jee J‐G, Ikegami T, Nakao M, et al. Solution structure of the methyl‐CpG binding domain of human MBD1 in complex with methylated DNA. Cell. 2001;105(4):487‐97. 133. Du Q, Luu P‐L, Stirzaker C, Clark SJ. Methyl‐CpG‐binding domain proteins: readers of the epigenome. Epigenomics. 2015;7(6):1051‐73. 134. Edwards JR, Yarychkivska O, Boulard M, Bestor TH. DNA methylation and DNA methyltransferases. Epigenetics & chromatin. 2017;10(1):23. 135. Smallwood SA, Kelsey G. De novo DNA methylation: a germ cell perspective. Trends in Genetics. 2012;28(1):33‐42. 136. Lipka DB, Wang Q, Cabezas‐Wallscheid N, Klimmeck D, Weichenhan D, Herrmann C, et al. Identification of DNA methylation changes at cis‐regulatory elements during early steps of HSC differentiation using tagmentation‐based whole genome bisulfite sequencing. Cell Cycle. 2014;13(22):3476‐87. 137. Ji H, Ehrlich LI, Seita J, Murakami P, Doi A, Lindau P, et al. Comprehensive methylome map of lineage commitment from haematopoietic progenitors. Nature. 2010;467(7313):338‐42. 138. Farlik M, Halbritter F, Muller F, Choudry FA, Ebert P, Klughammer J, et al. DNA Methylation Dynamics of Human Hematopoietic Stem Cell Differentiation. Cell Stem Cell. 2016;19(6):808‐22. 139. Hodges E, Molaro A, Dos Santos CO, Thekkat P, Song Q, Uren PJ, et al. Directional DNA methylation changes and complex intermediate states accompany lineage specificity in the adult hematopoietic compartment. Mol Cell. 2011;44(1):17‐28. 140. Bock C, Beerman I, Lien WH, Smith ZD, Gu H, Boyle P, et al. DNA methylation dynamics during in vivo differentiation of blood and skin stem cells. Mol Cell. 2012;47(4):633‐47. 141. Armstrong DA, Lesseur C, Conradt E, Lester BM, Marsit CJ. Global and gene‐specific DNA methylation across multiple tissues in early infancy: implications for children's health research. The FASEB Journal. 2014;28(5):2088‐97. 142. North M, Jones M, MacIsaac J, Morin A, Steacy L, Gregor A, et al. Blood and nasal epigenetics correlate with allergic rhinitis symptom development in the environmental exposure unit. Allergy. 2018;73(1):196‐205.

CHAPTER ONE INTRODUCTION

143. Baccarelli A, Wright RO, Bollati V, Tarantini L, Litonjua AA, Suh HH, et al. Rapid DNA methylation changes after exposure to traffic particles. American journal of respiratory and critical care medicine. 2009;179(7):572‐8. 144. Bollati V, Schwartz J, Wright R, Litonjua A, Tarantini L, Suh H, et al. Decline in genomic DNA methylation through aging in a cohort of elderly subjects. Mechanisms of ageing and development. 2009;130(4):234‐9. 145. Madrigano J, Baccarelli AA, Mittleman MA, Sparrow D, Vokonas PS, Tarantini L, et al. Aging and epigenetics: longitudinal changes in gene‐specific DNA methylation. Epigenetics. 2012;7(1):63‐ 70. 146. Florath I, Butterbach K, Müller H, Bewerunge‐Hudler M, Brenner H. Cross‐sectional and longitudinal changes in DNA methylation with age: an epigenome‐wide analysis revealing over 60 novel age‐associated CpG sites. Human molecular genetics. 2013;23(5):1186‐201. 147. Heindel JJ, Vandenberg LN. Developmental origins of health and disease: a paradigm for understanding disease etiology and prevention. Current opinion in pediatrics. 2015;27(2):248. 148. Alisch RS, Barwick BG, Chopra P, Myrick LK, Satten GA, Conneely KN, et al. Age‐associated DNA methylation in pediatric populations. Genome research. 2012:gr. 125187.111. 149. Yong W‐S, Hsu F‐M, Chen P‐Y. Profiling genome‐wide DNA methylation. Epigenetics & Chromatin. 2016;9(1):26. 150. Frommer M, McDonald LE, Millar DS, Collis CM, Watt F, Grigg GW, et al. A genomic sequencing protocol that yields a positive display of 5‐methylcytosine residues in individual DNA strands. Proceedings of the National Academy of Sciences. 1992;89(5):1827‐31. 151. Gaunt TR, Shihab HA, Hemani G, Min JL, Woodward G, Lyttleton O, et al. Systematic identification of genetic influences on methylation across the human life course. Genome Biology. 2016;17(1):61. 152. Heim C, Binder EB. Current research trends in early life stress and depression: Review of human studies on sensitive periods, gene–environment interactions, and epigenetics. Experimental neurology. 2012;233(1):102‐11. 153. Barker DJ. The fetal and infant origins of adult disease. BMJ: British Medical Journal. 1990;301(6761):1111. 154. Calkins K, Devaskar SU. Fetal origins of adult disease. Current problems in pediatric and adolescent health care. 2011;41(6):158‐76. 155. Cavalli G, Heard E. Advances in epigenetics link genetics to the environment and disease. Nature. 2019;571(7766):489‐99. 156. Bakulski KM, Dou J, Lin N, London SJ, Colacino JA. DNA methylation signature of smoking in lung cancer is enriched for exposure signatures in newborn and adult blood. Sci Rep. 2019;9(1):4576. 157. Barcelona V, Huang Y, Brown K, Liu J, Zhao W, Yu M, et al. Novel DNA methylation sites associated with cigarette smoking among African Americans. Epigenetics. 2019;14(4):383‐91. 158. Emerce E, Ghosh M, Oner D, Duca RC, Vanoirbeek J, Bekaert B, et al. Carbon Nanotube‐ and Asbestos‐Induced DNA and RNA Methylation Changes in Bronchial Epithelial Cells. Chem Res Toxicol. 2019;32(5):850‐60. 159. Guarrera S, Viberti C, Cugliari G, Allione A, Casalone E, Betti M, et al. Peripheral Blood DNA Methylation as Potential Biomarker of Malignant Pleural Mesothelioma in Asbestos‐Exposed Subjects. J Thorac Oncol. 2019;14(3):527‐39. 160. Kettunen E, Hernandez‐Vargas H, Cros MP, Durand G, Le Calvez‐Kelm F, Stuopelyte K, et al. Asbestos‐associated genome‐wide DNA methylation changes in lung cancer. Int J Cancer. 2017;141(10):2014‐29. 161. Bollati V, Baccarelli A, Hou L, Bonzini M, Fustinoni S, Cavallo D, et al. Changes in DNA methylation patterns in subjects exposed to low‐dose benzene. Cancer Res. 2007;67(3):876‐80. 162. Fenga C, Gangemi S, Costa C. Benzene exposure is associated with epigenetic changes (Review). Mol Med Rep. 2016;13(4):3401‐5.

CHAPTER ONE INTRODUCTION

163. Ren J, Cui JP, Luo M, Liu H, Hao P, Wang X, et al. The prevalence and persistence of aberrant promoter DNA methylation in benzene‐exposed Chinese workers. PLoS One. 2019;14(8):e0220500. 164. Ding R, Jin Y, Liu X, Zhu Z, Zhang Y, Wang T, et al. Characteristics of DNA methylation changes induced by traffic‐related air pollution. Mutat Res Genet Toxicol Environ Mutagen. 2016;796:46‐53. 165. Gondalia R, Baldassari A, Holliday KM, Justice AE, Mendez‐Giraldez R, Stewart JD, et al. Methylome‐wide association study provides evidence of particulate matter air pollution‐associated DNA methylation. Environ Int. 2019:104723. 166. Gao X, Jia M, Zhang Y, Breitling LP, Brenner H. DNA methylation changes of whole blood cells in response to active smoking exposure in adults: a systematic review of DNA methylation studies. Clinical epigenetics. 2015;7(1):113. 167. Tiffon C. The Impact of Nutrition and Environmental Epigenetics on Human Health and Disease. Int J Mol Sci. 2018;19(11). 168. Lillycrop KA, Phillips ES, Jackson AA, Hanson MA, Burdge GC. Dietary protein restriction of pregnant rats induces and folic acid supplementation prevents epigenetic modification of hepatic gene expression in the offspring. The Journal of nutrition. 2005;135(6):1382‐6. 169. Hahn O, Grönke S, Stubbs TM, Ficz G, Hendrich O, Krueger F, et al. Dietary restriction protects from age‐associated DNA methylation and induces epigenetic reprogramming of lipid metabolism. Genome biology. 2017;18(1):56. 170. Weisbeck A, Jansen RJ. Nutrients and the Pancreas: An Epigenetic Perspective. Nutrients. 2017;9(3). 171. Larsson SC, Giovannucci E, Wolk A. Folate intake, MTHFR polymorphisms, and risk of esophageal, gastric, and pancreatic cancer: a meta‐analysis. Gastroenterology. 2006;131(4):1271‐83. 172. Heijmans BT, Tobi EW, Stein AD, Putter H, Blauw GJ, Susser ES, et al. Persistent epigenetic differences associated with prenatal exposure to famine in humans. Proc Natl Acad Sci U S A. 2008;105(44):17046‐9. 173. Xue J, Schoenrock SA, Valdar W, Tarantino LM, Ideraabdullah FY. Maternal vitamin D depletion alters DNA methylation at imprinted loci in multiple generations. Clinical epigenetics. 2016;8(1):107. 174. Xue J, Gharaibeh RZ, Pietryk EW, Brouwer C, Tarantino LM, Valdar W, et al. Impact of vitamin D depletion during development on mouse sperm DNA methylation. Epigenetics. 2018;13(9):959‐74. 175. Zhang H, Chu X, Huang Y, Li G, Wang Y, Li Y, et al. Maternal vitamin D deficiency during pregnancy results in insulin resistance in rat offspring, which is associated with inflammation and Ikappabalpha methylation. Diabetologia. 2014;57(10):2165‐72. 176. Wen J, Hong Q, Wang X, Zhu L, Wu T, Xu P, et al. The effect of maternal vitamin D deficiency during pregnancy on body fat and adipogenesis in rat offspring. Scientific reports. 2018;8(1):1‐8. 177. Meems LM, Mahmud H, Buikema H, Tost J, Michel S, Takens J, et al. Parental vitamin D deficiency during pregnancy is associated with increased blood pressure in offspring via Panx1 hypermethylation. American Journal of Physiology‐Heart and Circulatory Physiology. 2016;311(6):H1459‐H69. 178. Neelon SB, White AJ, Vidal AC, Schildkraut JM, Murtha AP, Murphy SK, et al. Maternal vitamin D, DNA methylation at imprint regulatory regions and offspring weight at birth, 1 year and 3 years. International Journal of Obesity. 2018;42(4):587‐93. 179. Suderman M, Stene LC, Bohlin J, Page C, Holvik K, Parr CL, et al. 25‐Hydroxyvitamin D in pregnancy and genome wide cord blood DNA methylation in two pregnancy cohorts (MoBa and ALSPAC). The Journal of steroid biochemistry and molecular biology. 2016;159:102‐9. 180. Mozhui K, Smith AK, Tylavsky FA. Ancestry dependent DNA methylation and influence of maternal nutrition. PloS one. 2015;10(3). 181. Anderson CM, Gillespie SL, Thiele DK, Ralph JL, Ohm JE. Effects of Maternal Vitamin D Supplementation on the Maternal and Infant Epigenome. Breastfeed Med. 2018;13(5):371‐80.

CHAPTER ONE INTRODUCTION

182. Curtis EM, Krstic N, Cook E, D'angelo S, Crozier SR, Moon RJ, et al. Gestational vitamin D supplementation leads to reduced perinatal RXRA DNA methylation: results from the MAVIDOS trial. Journal of Bone and Mineral Research. 2019;34(2):231‐40. 183. Zeitelhofer M, Adzemovic MZ, Gomez‐Cabrero D, Bergman P, Hochmeister S, N'diaye M, et al. Functional genomics analysis of vitamin D effects on CD4+ T cells in vivo in experimental autoimmune encephalomyelitis. Proceedings of the National Academy of Sciences. 2017;114(9):E1678‐E87. 184. Moore JR, Hubler SL, Nelson CD, Nashold FE, Spanier JA, Hayes CE. 1, 25‐Dihydroxyvitamin D3 increases the methionine cycle, CD4+ T cell DNA methylation and Helios+ Foxp3+ T regulatory cells to reverse autoimmune neurodegenerative disease. Journal of neuroimmunology. 2018;324:100‐14. 185. Zhu H, Wang X, Shi H, Su S, Harshfield GA, Gutin B, et al. A genome‐wide methylation study of severe vitamin D deficiency in African American adolescents. The Journal of pediatrics. 2013;162(5):1004‐9. e1. 186. Zhu H, Bhagatwala J, Huang Y, Pollock NK, Parikh S, Raed A, et al. Race/ethnicity‐specific association of vitamin D and global DNA methylation: cross‐sectional and interventional findings. PloS one. 2016;11(4). 187. Junge KM, Bauer T, Geissler S, Hirche F, Thürmann L, Bauer M, et al. Increased vitamin D levels at birth and in early infancy increase offspring allergy risk—evidence for involvement of epigenetic mechanisms. Journal of Allergy and Clinical Immunology. 2016;137(2):610‐3. 188. Valencia RAC, Martino DJ, Saffery R, Ellis JA. In vitro exposure of human blood mononuclear cells to active vitamin D does not induce substantial change to DNA methylation on a genome‐scale. The Journal of steroid biochemistry and molecular biology. 2014;141:144‐9. 189. Booth D, Ding N, Parnell G, Shahijanian F, Coulter S, Schibeci S, et al. Cistromic and genetic evidence that the vitamin D receptor mediates susceptibility to latitude‐dependent autoimmune diseases. Genes & Immunity. 2016;17(4):213‐9. 190. Baubec T, Colombo DF, Wirbelauer C, Schmidt J, Burger L, Krebs AR, et al. Genomic profiling of DNA methyltransferases reveals a role for DNMT3B in genic methylation. Nature. 2015;520(7546):243‐7. 191. Weinberg DN, Papillon‐Cavanagh S, Chen H, Yue Y, Chen X, Rajagopalan KN, et al. The histone mark H3K36me2 recruits DNMT3A and shapes the intergenic DNA methylation landscape. Nature. 2019;573(7773):281‐6. 192. Carlberg C. Molecular endocrinology of vitamin D on the epigenome level. Molecular and cellular endocrinology. 2017;453:14‐21. 193. Schröder PC, Li J, Wong GW, Schaub B. The rural–urban enigma of allergy: What can we learn from studies around the world? Pediatric Allergy and Immunology. 2015;26(2):95‐102. 194. Schaub B, Liu J, Höppler S, Schleich I, Huehn J, Olek S, et al. Maternal farm exposure modulates neonatal immune mechanisms through regulatory T cells. Journal of Allergy and Clinical Immunology. 2009;123(4):774‐82. e5. 195. Lluis A, Depner M, Gaugler B, Saas P, Casaca VI, Raedler D, et al. Increased regulatory T‐cell numbers are associated with farm milk exposure and lower atopic sensitization and asthma in childhood. Journal of allergy and clinical immunology. 2014;133(2):551‐9. e10. 196. Schlums H, Cichocki F, Tesi B, Theorell J, Beziat V, Holmes TD, et al. Cytomegalovirus infection drives adaptive epigenetic diversification of NK cells with altered signaling and effector function. Immunity. 2015;42(3):443‐56. 197. Pacis A, Tailleux L, Morin AM, Lambourne J, MacIsaac JL, Yotova V, et al. Bacterial infection remodels the DNA methylation landscape of human dendritic cells. Genome research. 2015;25(12):1801‐11. 198. Richetto J, Massart R, Weber‐Stadlbauer U, Szyf M, Riva MA, Meyer U. Genome‐wide DNA methylation changes in a mouse model of infection‐mediated neurodevelopmental disorders. Biological psychiatry. 2017;81(3):265‐76.

CHAPTER ONE INTRODUCTION

199. Taniguchi I, Iwaya C, Ohnaka K, Shibata H, Yamamoto K. Genome‐wide DNA methylation analysis reveals hypomethylation in the low‐CpG promoter regions in lymphoblastoid cell lines. Human genomics. 2017;11(1):8. 200. Sun YV, Turner ST, Smith JA, Hammond PI, Lazarus A, Van De Rostyne JL, et al. Comparison of the DNA methylation profiles of human peripheral blood cells and transformed B‐lymphocytes. Human genetics. 2010;127(6):651‐8. 201. Brennan EP, Ehrich M, Brazil DP, Crean JK, Murphy M, Sadlier DM, et al. Comparative analysis of DNA methylation profiles in peripheral blood leukocytes versus lymphoblastoid cell lines. Epigenetics. 2009;4(3):159‐64. 202. Leonard S, Wei W, Anderton J, Vockerodt M, Rowe M, Murray PG, et al. Epigenetic and Transcriptional Changes Which Follow Epstein‐Barr Virus Infection of Germinal Center B Cells and Their Relevance to the Pathogenesis of Hodgkin's Lymphoma. Journal of Virology. 2011;85(18):9568‐ 77. 203. Hernando H, Islam AB, Rodríguez‐Ubreva J, Forne I, Ciudad L, Imhof A, et al. Epstein–Barr virus‐mediated transformation of B cells induces global chromatin changes independent to the acquisition of proliferation. Nucleic acids research. 2013;42(1):249‐63. 204. Hansen KD, Sabunciyan S, Langmead B, Nagy N, Curley R, Klein G, et al. Large‐scale hypomethylated blocks associated with Epstein‐Barr virus‐induced B‐cell immortalization. Genome research. 2013:gr. 157743.113.

CHAPTER TWO

REGULATION OF THE METHYLOME IN

DIFFERENTIATION FROM ADULT STEM CELLS MAY

UNDERPIN VITAMIN D RISK IN MULTIPLE SCLEROSIS

CHAPTER TWO STEM CELL METHYLOME IN DIFFERENTIATED CELLS

Regulation of the methylome in differentiation from adult stem cells may underpin vitamin D risk in MS

*Lawrence T. C. Ong1,2, Grant P. Parnell1, Kelly Veale1, Graeme J. Stewart1,2, Christopher Liddle1, David R. Booth1

1Centre for Immunology and Allergy Research, Westmead Institute for Medical Research, The University of Sydney, 176 Hawkesbury Rd, Westmead, New South Wales 2145, Australia

2Department of Clinical Immunology and Allergy, Westmead Hospital, Cnr Darcy and Hawkesbury Rds, Westmead, New South Wales 2145, Australia

*Corresponding author

Contact Details

Lawrence T. C. Ong – [email protected]

Grant P. Parnell – [email protected]

Kelly Veale – [email protected]

Graeme J. Stewart – [email protected]

Christopher Liddle – [email protected]

David R. Booth – [email protected]

This manuscript has been submitted for peer review in the journal Genes and Immunity

Keywords – DNA methylation, recapitulation, epigenetics, haematopoiesis

CHAPTER TWO STEM CELL METHYLOME IN DIFFERENTIATED CELLS

ABSTRACT

Background: Multiple lines of evidence indicate Multiple Sclerosis (MS) is affected by vitamin D. This effect may be mediated by methylation in immune cell progenitors.

Objective: To determine 1) if haematopoietic stem cell methylation constrains methylation in daughter cells and is variable between individuals, and 2) the interaction of methylation with the vitamin D receptor binding sites.

Methods: We interrogated genomic methylation levels from matching purified CD34+ haematopoietic stem cells and progeny CD14+ monocytes and CD56+ NK cells from 11 individuals using modified reduced representation bisulfite sequencing. Differential methylation of Vitamin D Receptor binding sites and MS risk genes was assessed from this and using pyrosequencing for the vitamin D regulated MS risk gene ZMIZ1.

Results: Although DNA methylation states at CpG islands and other sites are almost entirely recapitulated between progenitor and progeny immune cells, significant variation was detected at some regions between cell subsets and individuals; including around the MS risk genes HLA DRB1 and the vitamin D repressor NCOR2. Methylation of the vitamin D responsive MS risk gene ZMIZ1 was associated with risk SNP and disease.

Conclusion: DNA methylation settings in adult haematopoietic stem cells may contribute to individual variation in vitamin D responses in immune cells.

CHAPTER TWO STEM CELL METHYLOME IN DIFFERENTIATED CELLS

INTRODUCTION

Many of the most common autoimmune diseases show a latitude dependent prevalence, pointing to the importance of this environmental factor in pathogenesis. These latitude dependent autoimmune diseases (LDADs) include Multiple Sclerosis (MS), Type 1 Diabetes, Rheumatoid Arthritis, Crohn’s Disease, Systemic Lupus Erythematosus and Psoriasis (1) . The effect can be very strong, e.g. MS is 7 times more common in Tasmania than Northern Queensland (2), suggesting the therapeutic benefit mediated by the latitude effect may be very strong. Vitamin D is produced when skin is exposed to sunlight, dependent on latitude. It controls expression of many genes through transcription factor activation and plays an important role in immune tolerance (3). Vitamin D deficiency increases MS risk and genes that control vitamin D metabolism are risk factors in LDADs, so that vitamin D‐ influenced immune pathways are thought to underpin the latitude effect(3). However, recent meta‐ analyses have shown that oral supplementation with an inactive form of vitamin D (cholecalciferol), does not ameliorate disease (4, 5). The molecular basis for the putative protective effect of vitamin D needs to be established to improve its utility in the clinic.

To activate the vitamin D receptor (VDR) cholecalciferol needs to be hydroxylated twice: first in the liver by CYP2R1, then in various tissues, including immune cells, by CYP27B1. It is catabolised by CYP24A1. CYP2R1, CYP27B1, CYP24A1 and VDR are all genetic risk loci for MS (6‐9). CYP27B1 and CYP24A1 are most highly expressed in the mononuclear phagocytic cells, including monocytes and dendritic cells (10). From our ChIPseq studies, the VDR peaks in these cells are enriched near LDAD risk genes (11). The MS risk genes with VDR peaks include CYP24A1, and the transcription factors ZMIZ1 and IRF8. These MS risk genes interact with vitamin D to induce a tolerogenic phenotype in mononuclear phagocytic cells (12) in turn inducing a more tolerant immune cell state generally, and likely reducing the risk of development of autoimmunity. In particular, genetic variation in ZMIZ1 is associated with many LDADs (13), and we have recently shown that the risk SNP for ZMIZ1 is near a VDR binding site 11, near a locus known to be variably methylated between individuals (14).

From twin and family studies, the proportion of immune cell types and immune cell transcriptomes are under both genetic and environmental control, with the latter effect being much stronger (15‐ 17). From longitudinal studies, gene expression is much more variable between individuals than within individuals (18). The stability of gene expression over time for individuals suggests it is under

CHAPTER TWO STEM CELL METHYLOME IN DIFFERENTIATED CELLS

tight homeostatic control, by as yet uncharacterised mechanisms. This homeostatic control may define differences in tolerance states between individuals, and differences in responses to therapies. The relative stability of DNA methylation and its potential to be modified by environmental factors makes it a candidate mechanism for mediating homeostatic control and so disease risk in LDADs. Alterations in DNA methylation may result in changes in chromatin conformation and in the affinity of transcription factors for target promoters, thereby altering gene transcription, including for the VDR (19). DNA methylation may be heritable and fixed (for example at imprinted loci); heritable but change in the process of cell differentiation (20) and with aging (21); or non‐heritable, changing in response to environmental factors. In addition, genetic differences between individuals can influence methylation states (22, 23). Vitamin D may regulate methylation through protein:protein interactions with DNA methyltransferases (24), through being regulated by methylation differences (25), and by its own effect on transcription being dependent on promoter methylation of its target genes (19) . Global methylation in response to vitamin D has been shown to increase in mice, and is associated with reduced experimental autoimmune encephalomyelitis, the mouse model of MS (26).

The risk of developing MS is likely more dependent on sun exposure in adolescence than adulthood (27). Variation in childhood sun exposure may affect stem cell methylation, which then sets adult immune cell states. In support of this, Ng et al (28) showed that transplanted bone marrow of mice exposed to UV conferred protection against inflammation in unexposed mice. Further, from studies of DNA methylation through various stages and lineages of normal haematopoietic cell development it has been established that methylation changes occur early in haematopoietic cell differentiation (29, 30), in a progressive(29) and lineage specific fashion (30‐32), and frequently associate with lineage associated transcription factors and binding sites (33). Methylation changes also tend to be modest in magnitude (33). Most of these studies have compared subset specific methylation across different individuals, decreasing the power to detect methylation differences. We focus here on the methylation states and changes within individuals (linear cohort) to determine if CD34 settings could set regulation of MS risk genes in progeny immune cells. This could in turn affect MS disease susceptibility (Fig 1).

Our driving hypothesis is that vitamin D haematopoietic stem cell (HSC, CD34+) methylation states are set in childhood, and transmitted to progeny cells, with individual variation contributing to MS risk. To test if this was feasible we profiled the methylome of HSC derived Natural Killer cells (NK,

CHAPTER TWO STEM CELL METHYLOME IN DIFFERENTIATED CELLS

CD56+) and monocytes (CD14+) from eleven individuals (linear cohort), determined the extent to which methylation state is recapitulated between progenitor and progeny cells in a linear cohort, and if individual variation might mediate differences in vitamin D function, especially through MS risk loci.

MATERIALS AND METHODS

Cell isolation from a linear cohort

Approximately 100 ml of whole blood was drawn from each of 13 healthy individuals into tubes containing EDTA. Peripheral blood mononuclear cells (PBMCs) were then obtained by density gradient centrifugation and divided into multiple aliquots (1x107 PBMCs for CD14+ and CD56+, the remainder for CD34+). Parallel positive selection was performed using CD14 MicroBeads, CD34 MicroBead Kit Ultrapure and CD56 MicroBeads (Miltenyi Biotec, Bergisch Gladbach, Germany) for each respective cell subset with the autoMACS Pro Separator (Miltenyi Biotec, Bergisch Gladbach, Germany) as per manufacturer’s instructions. An aliquot of purified cells was set aside for assessment of purity by flow cytometry. Samples used for downstream analysis demonstrated mean purities of 96% for CD14+, 93% for CD34+ and 96% for CD56+ cells (see Supplemental Data for further details). Purified cells were subjected to two subsequent washes with DPBS before storage at ‐80 oC.

DNA isolation and modified reduced representation bisulfite sequencing (mRRBS)

DNA was extracted using the QIAamp DNA micro kit (Qiagen, Hilden, Germany) as per manufacturer’s instructions. Next generation sequencing libraries were prepared using the gel‐free mRRBS method as previously described (33). One‐hundred base pair paired‐end sequencing was conducted on an Illumina HiSeq 2500 using PhiX spike‐in to counteract low sequence diversity of bisulfite modified sequences

Data QC & alignment

The analysis pipeline is outlined in Supplemental Data. Briefly, the quality of raw sequences was ascertained using FastQC v0.10.1 (34). Adapter trimming was then carried out using Trimmomatic v0.36 (35) with ILLUMINACLIP (seed mismatches = 2, palindrome clip = 30, simple clip threshold = 3)

CHAPTER TWO STEM CELL METHYLOME IN DIFFERENTIATED CELLS

and MAXINFO (target length = 40, strictness = 0.5) options. Reads were aligned to the hg19 genome using the Wildcard Alignment Tool (WALT) v1.0 with default settings and the output .sam files converted to .mr files using to‐mr for further analysis.

Methylation analysis

Downstream methylation analysis was performed using Methpipe v3.4.3 (36). The specific modules used for methylation analyses included bsrate (estimation of bisulfite conversion rates), levels (coverage and methylation level statistics), methcounts (methylation calls, using the ‐n option for CpG context cytosines only) and symmetric‐cpgs (for extracting and merging symmetric methylation levels on both strands using methcounts data). Roimethstat (with ‐P and ‐v options) was used to determine regional methylation levels in conjunction with prespecified annotations, namely CpG islands (UCSC genome browser table browser > assembly: GRCH37/hg19 > group: Regulation > track: CpG Islands) and promoter regions as defined by the Ensembl regulatory build (37). More specifically, this module determines the proportion of methylated CpG reads to total CpG reads within a predefined interval. The mean methylation value for each CpG island (CGI) was then calculated by summing the methylation values for a subset and then dividing by the number of individuals for which a methylation value was able to be calculated. Subset specific methylation was compared by determining the difference in methylation between subsets for a given CGI and CGI with differences >0.25 were considered to be differentially methylated. Multidimensional scaling was performed using BRB Array tools v4.5.1 (38) and heatmap generation and unsupervised hierarchical clustering analysis was performed using Morpheus (39) Proximal and/or overlapping genes or genomic annotations were determined using Bedtools (v2.25.0) closest and intersect respectively.

Permutation testing

The permutation test was performed by taking a matrix (m +n) x o where m refers to 11 individual methylation values for subset a and n refers to the methylation values for the same 11 individuals for subset b. o refers to the CpG islands interrogated. Values in each column were then permuted randomly (across both a and b subsets or m+n columns) and the mean difference of each group of the 11 methylation values calculated. The number of differentially methylated CGI at the 0.25 level were then calculated. The m+n columns were permuted 100 times

CHAPTER TWO STEM CELL METHYLOME IN DIFFERENTIATED CELLS

Processing of public methylomes

Genome wide methylomes of buccal mucosa (40), CD14+, CD34+ and CD56+ cells were downloaded as .bw file and converted to .wig format using the bigWigToWig module of UCSC genome browser tools (v348). The file was subsequently converted to .bed format using Bedops (v2.4.30) wig2bed command (with the ‐‐zero‐indexed option). .bed files were reformatted to .meth files (input for the roimethstat program) by using the Unix awk command to remove column 3 of the .bed file and replacing it with the strand, adding “CpG” (to provide the cytosine context) to column 4 and the value “1” to column 5. CpG island methylation could then be extracted as per Methylation Analysis above.

Comparison of mRRBS with WGBS data

The number of observed and expected overlapping CGI was compared using the χ2 goodness of fit test. The expected number of overlapping CGI by chance was determined by the following formula:

𝑒𝑥𝑝𝑒𝑐𝑡𝑒𝑑 𝑜𝑣𝑒𝑟𝑙𝑎𝑝𝑝𝑖𝑛𝑔 𝐶𝐺𝐼 𝑑𝑖𝑓𝑓 𝑚𝑒𝑡ℎ 𝐶𝐺𝐼 𝑚𝑅𝑅𝐵𝑆 𝑑𝑖𝑓𝑓 𝑚𝑒𝑡ℎ 𝐶𝐺𝐼 𝑊𝐺𝐵𝑆 𝑥 𝑥 𝑡𝑜𝑡𝑎𝑙 𝐶𝐺𝐼 𝑚𝑅𝑅𝐵𝑆 𝑡𝑜𝑡𝑎𝑙 𝐶𝐺𝐼 𝑚𝑅𝑅𝐵𝑆 𝑡𝑜𝑡𝑎𝑙 𝐶𝐺𝐼 𝑊𝐺𝐵𝑆

Assessment of methylation at VDR binding peak loci

The DNA methylation state of 1kb windows centred on CD14+ VDR peaks (11) were derived for each of the 11 individuals from CD14+ and CD34+ subsets using the roimethstat program. Peaks were further analysed only if they contained at least 100 CpG reads per region, per subset, per subject. To determine whether the methylation in these windows was recapitulated between CD34+ and CD14+ cells, regional methylation values had to be correlated by Pearson correlation (p<0.05) and demonstrate a methylation difference of <5% or be not statistically significant by two tailed t‐test at p<0.05. SFP1 and CEBPA/B motifs were downloaded from http://homer.ucsd.edu/homer/data/motifs/homer.KnownMotifs.hg19.170917.bed.gz.

ZMIZ1 methylation, gene expression and MS risk genotype

Peripheral blood was collected from a different cohort of 51 healthy controls and 46 patients with relapsing remitting MS. DNA was isolated from whole blood using Qiagen QIAamp DNA mini kit

CHAPTER TWO STEM CELL METHYLOME IN DIFFERENTIATED CELLS

(Qiagen, Hilden, Germany) as per manufacturer’s instructions. Each sample was genotyped for the ZMIZ1 SNPs rs1250551 using the TaqMan SNP genotyping (Life Technologies, Carlsbad, US). ZMIZ1 gene expression was assessed using a TaqMan gene expression assay (Life Technologies, Carlsbad, US) for ZMIZ1, Hs01119362_m1, with GAPDH (Hs02758991_g1) as a reference gene. ZMIZ1 methylation of seven CpGs in at Chr10:80999771‐80999842 was determined using Pyromark pyrosequencing (Qiagen, Hilden, Germany). This region was chosen as methylation of residues within had been previously found to be associated with genotype at rs12250551 (14).

RESULTS

Of the thirteen individuals assayed, data for two individuals (both males) was excluded from further analysis due to inadequate bisulfite conversion rates (<99%). The remaining 11 individuals (7 female, 4 male) ranged in age from 28‐59 years (mean 36.7, SD 11.6). After adapter and quality trimming, there was an average paired end read count of 13 million reads per subset, per individual. On average, 75% of these reads mapped to the hg19 genome. The mean bisulfite conversion rate for the 11 individuals was 99.4%. The proportion of genome wide CpGs covered across all samples (SD) was 14.7% (2.3). Mean depth of covered CpGs (SD) by subset was 15.0x (4.9) for CD14+, 15.6x (6.4) for CD34+ and 17.9x (3.7) for CD56. The total number of observed CpG reads by subset were calculated in a CGI wise fashion. We observed >89% of CGI had at least 20 CpG reads and 72‐76% of CGI had at least 200 CpG reads (see Supplemental Data for further details)

Methylation is mostly recapitulated across both progeny cell subsets

After excluding sex, mitochondrial and haploid chromosomes (see Supplemental Data), there were a total of 26641 CpG islands annotated in the hg19 genome. There was methylation data for 25138 CGIs across CD14+ subsets, 25105 CGI across CD34+ subsets, and 25152 CGI across CD56+ subsets. The level of DNA methylation at a particular CGI was measured as a proportion, where 0 was completely unmethylated and 1 was completely methylated. We found that the vast number of CGIs were either unmethylated or almost completely methylated, with relatively few displaying intermediate methylation (Figure 2a). Utilising a cut‐off of 0.25 to distinguish differentially methylated CGI between subsets, we found 184 of 24934 (0.74%) of covered CGI to be differentially methylated between CD14+ vs CD34+ (p<0.01, permutation test), 171 of 24958 (0.69%) of covered CGI to be differentially methylated between CD34+ vs CD56+ (p<0.01, permutation test) and 232 of 24993 (0.93%) of covered CGI to be differentially methylated between CD14+ vs CD56+ (p<0.01,

CHAPTER TWO STEM CELL METHYLOME IN DIFFERENTIATED CELLS

permutation test; Figure 2b). In total, there were 395 unique CGI that were differentially methylated between any of the three subsets, representing less than 1.5% of interrogated CGIs.

The methylation levels across the three peripheral blood subsets were then compared to a publicly available buccal mucosal cell methylation dataset (GSE45529) (40). There were between 11 and 16‐ fold more CGIs that were differentially methylated at the 0.25 level when comparing peripheral blood subsets to buccal mucosa than when comparing the peripheral blood subsets to each other (see Figure 2c). This was equivalent to differential methylation at 10.5% of covered CGI. Our data is consistent with the findings of Lowe et al (40), that CpG methylation distinguishes blood cells from buccal mucosa.

Differentially methylated CGI identify cell subsets

To test if the 395 differentially methylated CGI derived from 11 individuals, were randomly distributed across the subsets, we performed hierarchical clustering using methylation data from these 395 CGI across each of the three peripheral blood subsets. Using one minus Pearson correlation and complete linkage, this analysis produced three distinct clusters corresponding to each of the three cell subsets (Figure 3a). This was confirmed with multidimensional scaling analysis, which demonstrated discrete clustering across the three subsets (Figure 3b), indicating the differences were non‐random.

Methylation of CD34+/CD14+/CD56+ cells is similar to peripheral blood cells but not buccal cells

To further validate our differentially methylated CGIs and determine if the methylation pattern we observed was similar to that observed in other studies, we interrogated the whole genome bisulfite sequencing (WGBS) methylomes of blood subsets from Methbase for CD14+ cells, CD34+ cells and CD56+ cells (41). CpG island methylation levels were extracted and compared between subsets. The number of overlapping, differentially methylated CpG islands between subsets determined by the WGBS method and mRRBS are noted in Table 1. In all cases, the number of observed overlaps between mRRBS and WGBS methylomes was significantly higher than expected by chance, with enrichment of 7‐10 fold. This suggests that our method is robustly detecting differentially methylated CGIs also detected by another method.

CHAPTER TWO STEM CELL METHYLOME IN DIFFERENTIATED CELLS

Methylation was higher in progeny cells than their progenitors

Relative methylation between cell subsets of sites that were determined to be differentially methylated at the 0.25 level were examined. Comparison of differentially methylated CGI in CD34+ vs CD14+ showed that CD34+ cells were more methylated than CD14+ at 86 CGI (47% of differentially methylated sites) and that CD14+ were more methylated than CD34+ at 98 CGI (53% of differentially methylated sites; Figure 4a left). Comparison of CD34+ and CD56+ cells however, showed CD56+ cells to be more methylated than CD34+ at 114 CGI (67% of differentially methylated sites) and CD34+ to be more methylated than CD56+ at 57 CGI (33% of differentially methylated sites; Figure 4a right).

Of the 184 differentially methylated CGIs in CD34+ vs CD14+ and the 171 differentially methylated CGI in CD34+ vs CD56+, 60 CGIs were common to both comparisons (Figure 3b). The number of overlapping CGI was much greater than the 1.2 differentially methylated CGI expected by chance using χ2 test (p=5.5x10‐14, df=1). Further analysis of these sites showed that in the vast majority of these CGIs, CD14+ and CD56+ methylation was greater than CD34+ methylation. Specifically, CD34+ methylation was lower than CD56+ methylation in 44/60 CGI (p=1.3x10‐4, sign test) and lower than CD14+ methylation in 46/60 CGI (p=1.5x10‐5, sign test).

Methylation varied most at intragenic regions

In general, CGI methylation at key regulatory elements in the genome are thought to be associated with marked changes in gene expression, such as those at transcription start sites and gene promoters. For example, methylation in promoter CGIs in normal cells is usually restricted to genes subjected to long term silencing such as those at imprinted genes (42). Utilising the Ensembl regulatory build(37), the overlap of these elements with 395 differentially methylated CGIs was determined. Of these 31 overlapped CTCF binding sites, 23 open chromatin regions, 6 predicted enhancer regions, 24 promoter regions, 107 predicted promoter flanking regions and 20 to transcription factor binding sites. Twenty‐two CGI overlapped with multiple annotations, whilst 206 CGI did not overlap with any regulatory build annotations. Figure 4c shows the number of overlapping elements with CGIs determined to be differentially methylated.

CHAPTER TWO STEM CELL METHYLOME IN DIFFERENTIATED CELLS

We further examined the potential functional consequences of differential methylation by classifying CGI overlaps according to whether they were intergenic, intragenic or promoter associated. Most of the differentially methylated CGI occurred in intragenic regions, with a minority occurring at promoters or intergenic regions (see Table 2).

Methylation was not correlated with gene expression

Using a publicly available gene expression database(43), cell subset specific expression of genes overlapping the 395 differentially methylated CGI was extracted (the closest gene was used when CGI were located in intergenic regions). Using a two‐fold difference in expression as the cut‐off between any two subsets, we found relatively few genes corresponding to differentially methylated CGI to be differentially expressed in pairwise subset comparisons. For example, CD34+ vs CD14+ genes were only differentially expressed across 29 regions, CD34+ vs CD56+ genes across 46 regions and CD14+ vs CD56+ across 46 regions (see Supplemental Data). Interestingly, very few differentially methylated promoters were associated with differentially expressed genes.

Methylation displayed little variation between individuals

To determine the magnitude of variation between individuals in CGI methylation, we first removed CGIs for which there was missing data for greater than 50% of subjects. For the remaining CGIs, we calculated the variance across individuals. Across each of the three subsets, the methylation state of most CGI displayed little variation between individuals. Within the top 500 most variable CGIs, the methylation variance range was as follows, CD14+ – 0.04 to 0.27, CD34+ – 0.06 to 0.27, CD56+ – 0.04 to 0.27. We also examined the variability in CGI methylation based on DNA methylation state. In all subsets, the greatest variability occurred at intermediate mean methylation states (Figure 5). Only one MS risk gene, HLA DRB1, was identified as differentially methylated between individuals. The rank of methylation in the CD34+ cells was recapitulated in the CD14+ cells (R=0.8, p=0.005).

MS risk gene VDR peaks were in unmethylated regions in all cell subsets and individuals

Earlier we had identified 1895 VDR peaks in monocytes (11) , of which 1593 were also VDR peaks in dendritic cells. These monocyte VDR peaks were virtually all in regions of open chromatin. Those regions covered in our dataset were mostly in unmethylated promoter regions (72%), and also in unmethylated open chromatin intragenic (18%) and intergenic regions (10%). VDR peaks near MS

CHAPTER TWO STEM CELL METHYLOME IN DIFFERENTIATED CELLS

risk genes are of particular interest, as their regulation by vitamin D are the most likely to be relevant to disease pathogenesis that is mediated by ultraviolet light exposure, and to transmit risk from childhood exposure to adults. VDR peaks in monocytes are overrepresented within 5kb of LDAD risk SNPs, compared to all SNPs (11), including at ZMIZ1 and CYP24A1 loci. The number of identified MS risk SNPs has nearly doubled since 2016, with >200 risk loci now identified. Eleven are within 5kb of VDR peaks in monocytes. We interrogated the methylation within 1kb of these VDR peaks, filtering CpG positions for a minimum of 100 reads across all 11 individuals. Of the 90 peaks passing filtering, 65 VDR peaks were in promoter regions, 16 intragenic, and 9 intergenic. In all but 8 cases, the CpGs were almost completely unmethylated (<10% methylation). Three of these regions methylated >10% were methylated to this extent in both CD14+ and CD34+ cells. In one gene, NCOR2, the intronic CpG methylation was significantly reduced in CD14+ (55%) compared to CD34+ cells (75%, p<0.001, paired samples t‐test) across the 11 individuals. For the unmethylated CpG positions, although less than 1% methylated across the 1kb window, we detected small but significant variation in methylation between CD34+ and CD14+ for around 10% of the VDR peak loci (see Supplemental Data).

Pioneer transcription factors are those capable of initiating chromatin unfolding and enabling access for other transcription factors. VDR itself is not known as a pioneer factor, but the three most common transcription factor binding sites in VDR ChIPseq peaks are VDR, and two pioneer transcription factors, SFP1 and CEBP, suggesting they may form a complex with VDR. SFP1 and CEBP recognition sequences are particularly common in VDR peaks without VDR recognition sequences (11) . As such the binding of the VDR/pioneer factor complexes to CD14+ DNA may be affected by MS risk SNPs and DNA methylation. However, we found no overrepresentation of these sites in the recapitulated or differential VDR methylation sites interrogated (p=0.13 SFP1, p=0.23 CEBP, χ2 test).

Methylation of a ZMIZ1 CpG island is associated with disease, risk SNP genotype but not gene expression in blood

The ZMIZ1 MS risk SNP rs1250551 is associated with variation in methylation (mQTL) (14) for CpGs around the nearest CpG island to the VDR peak in ZMIZ1. ZMIZ1 expression is stable over time, affected by vitamin D, and low in MS (13). To determine if methylation at these CpGs was also associated with disease and expression, we assayed the methylation for this CpG region using pyrosequencing for 51 controls and 46 MS (Supplementary table for cohort details). In this cohort

CHAPTER TWO STEM CELL METHYLOME IN DIFFERENTIATED CELLS

ZMIZ1 expression was low in MS and correlated with risk SNP genotype (Fig 6). There were 7 CpGs in the pyrosequenced region. Methylation state across these 7 was correlated with each other. In all cases the methylation level was also associated with rs1250551 genotype. Methylation level at position 2 was lower in MS compared to controls (see Figure 6). Methylation was not correlated with gene expression.

DISCUSSION

This study sought to determine whether the DNA methylation signature of CD34+ HSCs might regulate the vitamin D response of progeny immune cells. Consistent with our hypothesis, we found that most methylation in CD34+ was recapitulated in CD14+ and CD56+ cells in every member of our linear cohort. Methylation was highly variable between subsets and individuals for some genes, defining lineage differences, and likely contributing to lineage differentiation. Rank analysis across the linear cohort indicated differential methylation around vitamin D associated MS risk genes HLA DRB1 and NCOR2 may have been genetically controlled. Finally, we confirmed that the ZMIZ1 MS risk SNP was an mQTL and showed here that the methylation state was associated with disease. Overall, these data are consistent with vitamin D effects on latitude dependent autoimmune diseases, including MS, being mediated by methylation state in CD34+ cells, and providing a mechanism for childhood sun exposure affecting adult disease risk.

Genomic methylation and vitamin D effect on immune cells

The recapitulation of methylation state in progeny cells is consistent with work of others using different cohort types and techniques to measure methylation (29, 33, 44, 45) . Also, as found by others (46, 47), methylation was higher in progeny cells in all individuals than their progenitors, consistent with the idea that increasing differentiation is associated with increasing methylation, and suggesting that methylation may be tractable. However, Chavez‐Valencia et al (48) have shown that culture of CD14+ cells in vitamin D changes their gene expression and phenotype, but not their methylation state. There are no immunological markers validated as indicating vitamin D response in vivo. The absence of detectable changes to expression and phenotype in vivo on supplementation may indicate that sustained benefit from supplementation is limited by methylation state, set by CD34+ cell methylation, and leading to homeostatic control of expression of VDR regulated genes. So despite the strong epidemiological, genetic and cellular evidence that vitamin D is protective in LDADs, particularly MS, supplementation with cholecalciferol is not effective in vivo (4) .

CHAPTER TWO STEM CELL METHYLOME IN DIFFERENTIATED CELLS

Vitamin D regulated MS risk genes and methylation: HLA DRB1*1501

The immunological response to vitamin D may be dependent on methylation in CD34+ cells and progeny, and the genes regulated by this methylation that are of most relevance to pathogenesis in MS are the MS risk genes regulated by vitamin D. The major MS risk allele HLA DRB1*1501 is regulated by vitamin D (49). We show here that methylation of CpG clusters from this gene are highly variable between individuals. The rank of its methylation in CD14+ for the 11 individuals was the same as its rank in CD34+ cells, so that CD34+ methylation was constraining CD14+ methylation. Hypomethylation of clusters of CpGs have previously been identified in CD4+ T cells from MS compared to controls (50) , but not CD8+ (51) or B cells (52) ; or in CD4+ and CD8+ T cells (53). However, HLA DRB1 is an antigen presenting gene, functional in dendritic and B cells rather than CD4+ or CD8+ T cells. A possible explanation for the hypomethylation in MS in CD4+ and CD8+ T cells is that the variable methylation in CD34+ cells sets methylation in all progeny cells, but has its functional significance in dendritic and B cells. This would suggest that the CD34+ cells from people with MS have hypomethylated HLA DRB1.

Methylation at VDR binding sites near other MS risk genes: NCOR2

If VDR regulation was affected by CD34+ methylation state, we would expect that VDR binding sites in monocytes would fall in regions differentially methylated between individuals, and recapitulated between CD34+ and CD14+ cells. However, these sites were mainly in regions which were almost completely unmethylated. Only minor variation in methylation between subsets and individuals was detected, except for risk gene, NCOR2. The methylation percentage decreased in CD14+ cells compared to CD34+, but still reflected the rank of CD34+ methylation for the 11 individuals. This gene represses function of hormone receptors, including the vitamin D receptor. An intronic CpG island for this gene was the most highly methylated of VDR binding sites examined. This methylation was variable between individuals, and methylation in CD14+s reflected methylation levels in CD34+s. This gene has many isoforms, and has generally low expression in immune cells. The high methylation of the intron might indicate that its gene expression/isoform usage is tightly controlled to regulate VDR inhibition. This gene is therefore a good candidate for mediating a variable vitamin D response between individuals. Nucleic acid or other inhibition of NCOR2 might improve vitamin D response.

CHAPTER TWO STEM CELL METHYLOME IN DIFFERENTIATED CELLS

Disease and risk SNP associated methylation in ZMIZ1

Using very large cohorts, others have shown that although small, differences in methylation may be associated with disease (54), as we found here for ZMIZ1. ZMIZ1 is a risk locus for at least 11 LDADs (13). Its expression is highest in monocytes and tolerogenic DCs, and lowest in inflammatory DCs. High expression would be expected to be protective, and indeed its expression is higher in controls than MS. Previously we have shown in multiple cohorts that expression is lower for the risk than the protective allele, and response to vitamin D is higher for the protective allele (12). We confirm that expression is lower in this MS cohort, and lower for the risk allele. Gaunt et al. (14) have shown that the ZMIZ1 MS risk SNP is an mQTL for the nearest CpG to the risk SNP, which is collocated with a VDR peak. Here we show that the methylation level at this CpG island is associated with disease, but not gene expression in blood. Any association with expression may be context dependent, or need a larger cohort for statistical power to detect it. The implication of risk SNP and disease association is that methylation of this VDR regulated gene affects disease pathogenesis, presumably through effects on transcription factor binding.

Limitations of the study

Although our method of methylation interrogation provides distinct advantages with regard to biological replicates and sequencing depth, it should be noted that methylation outside CGI and promoter regions was not analysed. These regions may potentially show more numerous or perhaps larger methylation differences not detected in CGI and promoters. Secondly, sensitivity to differences in methylation would have been limited by our cohort size and read depth. Thirdly, the association between differential methylation observed here and gene expression was based on public gene expression data rather than that of our longitudinal cohort. This would likely have been more informative if based on the cells we had examined for methylation. Fourthly, low levels of contamination from other cell types may also have affected our estimation of DNA methylation across subsets, although the rates of contamination were generally low.

CONCLUSIONS

In summary, we have profiled the methylome of peripheral blood cell subsets and found that it is almost entirely recapitulated between progenitor and progeny. The individual variability in CGI methylation is consistent with the concept that a DNA methylation setting is present in

CHAPTER TWO STEM CELL METHYLOME IN DIFFERENTIATED CELLS

haematopoietic cells that is variable between individuals and transmitted from progenitor to progeny cells. This setting may contribute to variability in susceptibility to autoimmune disease, including through vitamin D mediated effects on tolerance. Future work may define if and how vitamin D sets CD34 methylation to in turn set immune responsiveness, and how to manipulate this setting for therapeutic benefit.

Ethics approval and consent to participate

This study received ethics approval from the Western Sydney Local Health District Human Research Ethics Committee (HREC2002/9/3.6(1425)). Informed, written consent was obtained from subjects prior to involvement in this study.

Availability of data and material

The datasets supporting the conclusions of this article are available to reviewers in the GEO database, and can be accessed via https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE114254 by entering the token enwzyamidrclruj into the box. The data will be made public once the study has been published.

Conflict of interest statement

The authors declare that they have no competing interests.

Funding

LO was supported by a National Health and Medical Research Council (NHMRC), Trish MS Foundation and MS Research Australia co‐funded postgraduate scholarship. GP was supported by a MS Research Australia Postdoctoral Fellowship and a MS Research Australia/ JDRF Australia/ Macquarie Group Foundation Postdoctoral Fellowship. DB was supported by a NHMRC Senior Research Fellowship and MSRA Project Grant

CHAPTER TWO STEM CELL METHYLOME IN DIFFERENTIATED CELLS

Acknowledgements

The authors would like to thank Ellis Patrick for invaluable biostatistical advice, Nicole Fewings and Stephen Schibeci for assistance with practical aspects of this study and volunteers from the Westmead Institute for Medical Research who kindly donated samples for this project. Flow cytometry was performed at the Flow Cytometry Core Facility that is supported by Westmead Research Hub, Cancer Institute NSW and NHMRC. Bioinformatic analysis was supported by Sydney Informatics Hub, funded by the University of Sydney.

CHAPTER TWO STEM CELL METHYLOME IN DIFFERENTIATED CELLS

REFERENCES

1. Wacker M, Holick MF. Sunlight and Vitamin D: A global perspective for health. Dermato‐ endocrinology. 2013;5(1):51‐108. 2. Miller D, Hammond S, McLeod J, Purdie G, Skegg D. Multiple sclerosis in Australia and New Zealand: are the determinants genetic or environmental? Journal of Neurology, Neurosurgery & Psychiatry. 1990;53(10):903‐5. 3. Dankers W, Colin EM, van Hamburg JP, Lubberts E. Vitamin D in autoimmunity: molecular mechanisms and therapeutic potential. Frontiers in immunology. 2017;7:697. 4. McLaughlin L, Clarke L, Khalilidehkordi E, Butzkueven H, Taylor B, Broadley SA. Vitamin D for the treatment of multiple sclerosis: a meta‐analysis. J Neurol. 2018;265(12):2893‐905. 5. Zheng C, He L, Liu L, Zhu J, Jin T. The efficacy of vitamin D in multiple sclerosis: A meta‐ analysis. Multiple sclerosis and related disorders. 2018;23:56‐61. 6. International Multiple Sclerosis Genetics Consortium. Electronic address ccye, International Multiple Sclerosis Genetics C. Low‐Frequency and Rare‐Coding Variation Contributes to Multiple Sclerosis Risk. Cell. 2018;175(6):1679‐87 e7. 7. Beecham AH, IMSGC. Analysis of immune‐related loci identifies 48 new susceptibility variants for multiple sclerosis. Nat Genet. 2013;45(11):1353‐60. 8. International Multiple Sclerosis Genetics C, Wellcome Trust Case Control C, Sawcer S, Hellenthal G, Pirinen M, Spencer CC, et al. Genetic risk and a primary role for cell‐mediated immune mechanisms in multiple sclerosis. Nature. 2011;476(7359):214‐9. 9. Consortium IMSG. Multiple sclerosis genomic map implicates peripheral immune cells and microglia in susceptibility. Science. 2019;365(6460). 10. Shahijanian F, Parnell GP, McKay FC, Gatt PN, Shojoei M, O'connor KS, et al. The CYP27B1 variant associated with an increased risk of autoimmune disease is underexpressed in tolerizing dendritic cells. Human molecular genetics. 2013;23(6):1425‐34. 11. Booth D, Ding N, Parnell G, Shahijanian F, Coulter S, Schibeci S, et al. Cistromic and genetic evidence that the vitamin D receptor mediates susceptibility to latitude‐dependent autoimmune diseases. Genes and immunity. 2016;17(4):213. 12. Parnell GP, Schibeci SD, Fewings NL, Afrasiabi A, Law SP, Samaranayake S, et al. The latitude‐ dependent autoimmune disease risk genes ZMIZ1 and IRF8 regulate mononuclear phagocytic cell differentiation in response to vitamin D. Human molecular genetics. 2018;28(2):269‐78. 13. Fewings NL, McKay FC, Parnell GP, Schibeci SD, Edwards J, Basuki MA, et al. The autoimmune risk gene ZMIZ1 is a vitamin D responsive marker of a molecular phenotype of multiple sclerosis. Journal of Autoimmunity. 2017;78:57‐69. 14. Gaunt TR, Shihab HA, Hemani G, Min JL, Woodward G, Lyttleton O, et al. Systematic identification of genetic influences on methylation across the human life course. Genome Biology. 2016;17(1):61. 15. Orru V, Steri M, Sole G, Sidore C, Virdis F, Dei M, et al. Genetic variants regulating immune cell levels in health and disease. Cell. 2013;155(1):242‐56. 16. Brodin P, Jojic V, Gao T, Bhattacharya S, Angel CJ, Furman D, et al. Variation in the human immune system is largely driven by non‐heritable influences. Cell. 2015;160(1‐2):37‐47. 17. Roederer M, Quaye L, Mangino M, Beddall MH, Mahnke Y, Chattopadhyay P, et al. The genetic architecture of the human immune system: a bioresource for autoimmunity and disease pathogenesis. Cell. 2015;161(2):387‐403. 18. Carr EJ, Dooley J, Garcia‐Perez JE, Lagou V, Lee JC, Wouters C, et al. The cellular composition of the human immune system is shaped by age and cohabitation. Nat Immunol. 2016;17(4):461‐8. 19. Carlberg C. Molecular endocrinology of vitamin D on the epigenome level. Molecular and cellular endocrinology. 2017;453:14‐21. 20. Guo H, Zhu P, Yan L, Li R, Hu B, Lian Y, et al. The DNA methylation landscape of human early embryos. Nature. 2014;511(7511):606‐10.

CHAPTER TWO STEM CELL METHYLOME IN DIFFERENTIATED CELLS

21. Jones MJ, Goodman SJ, Kobor MS. DNA methylation and healthy human aging. Aging cell. 2015;14(6):924‐32. 22. Bell JT, Pai AA, Pickrell JK, Gaffney DJ, Pique‐Regi R, Degner JF, et al. DNA methylation patterns associate with genetic and gene expression variation in HapMap cell lines. Genome biology. 2011;12(1):R10. 23. Bock C, Paulsen M, Tierling S, Mikeska T, Lengauer T, Walter J. CpG island methylation in human lymphocytes is highly correlated with DNA sequence, repeats, and predicted DNA structure. PLoS genetics. 2006;2(3):e26. 24. Takeyama K‐i, Kato S. The vitamin D3 1alpha‐hydroxylase gene and its regulation by active vitamin D3. Bioscience, biotechnology, and biochemistry. 2011;75(2):208‐13. 25. Wang M, Kong W, He B, Li Z, Song H, Shi P, et al. Vitamin D and the promoter methylation of its metabolic pathway genes in association with the risk and prognosis of tuberculosis. Clin Epigenetics. 2018;10(1):118‐. 26. Moore JR, Hubler SL, Nelson CD, Nashold FE, Spanier JA, Hayes CE. 1,25‐Dihydroxyvitamin D3 increases the methionine cycle, CD4(+) T cell DNA methylation and Helios(+)Foxp3(+) T regulatory cells to reverse autoimmune neurodegenerative disease. J Neuroimmunol. 2018;324:100‐14. 27. Olsson T, Barcellos LF, Alfredsson L. Interactions between genetic, lifestyle and environmental risk factors for multiple sclerosis. Nature Reviews Neurology. 2017;13(1):25. 28. Ng RL, Scott NM, Strickland DH, Gorman S, Grimbaldeston MA, Norval M, et al. Altered immunity and dendritic cell activity in the periphery of mice after long‐term engraftment with bone marrow from ultraviolet‐irradiated mice. The Journal of Immunology. 2013:1202786. 29. Lipka DB, Wang Q, Cabezas‐Wallscheid N, Klimmeck D, Weichenhan D, Herrmann C, et al. Identification of DNA methylation changes at cis‐regulatory elements during early steps of HSC differentiation using tagmentation‐based whole genome bisulfite sequencing. Cell Cycle. 2014;13(22):3476‐87. 30. Farlik M, Halbritter F, Muller F, Choudry FA, Ebert P, Klughammer J, et al. DNA Methylation Dynamics of Human Hematopoietic Stem Cell Differentiation. Cell Stem Cell. 2016;19(6):808‐22. 31. Ji H, Ehrlich LI, Seita J, Murakami P, Doi A, Lindau P, et al. Comprehensive methylome map of lineage commitment from haematopoietic progenitors. Nature. 2010;467(7313):338‐42. 32. Hodges E, Molaro A, Dos Santos CO, Thekkat P, Song Q, Uren PJ, et al. Directional DNA methylation changes and complex intermediate states accompany lineage specificity in the adult hematopoietic compartment. Mol Cell. 2011;44(1):17‐28. 33. Bock C, Beerman I, Lien WH, Smith ZD, Gu H, Boyle P, et al. DNA methylation dynamics during in vivo differentiation of blood and skin stem cells. Mol Cell. 2012;47(4):633‐47. 34. Andrews S. FastQC: a quality control tool for high throughput sequence data. 2010. 35. Bolger AM, Lohse M, Usadel B. Trimmomatic: a flexible trimmer for Illumina sequence data. Bioinformatics. 2014;30(15):2114‐20. 36. Song Q, Decato B, Hong EE, Zhou M, Fang F, Qu J, et al. A reference methylome database and analysis pipeline to facilitate integrative and comparative epigenomics. PLoS One. 2013;8(12):e81148. 37. Zerbino DR, Wilder SP, Johnson N, Juettemann T, Flicek PR. The ensembl regulatory build. Genome biology. 2015;16(1):56. 38. Simon R, Lam A, Li M‐C, Ngan M, Menenzes S, Zhao Y. Analysis of gene expression data using BRB‐array tools. Cancer informatics. 2007;3:117693510700300022. 39. Curtis EM, Krstic N, Cook E, D'Angelo S, Crozier SR, Moon RJ, et al. Gestational Vitamin D Supplementation Leads to Reduced Perinatal RXRA DNA Methylation: Results From the MAVIDOS Trial. Journal of bone and mineral research : the official journal of the American Society for Bone and Mineral Research. 2019;34(2):231‐40. 40. Lowe R, Gemma C, Beyan H, Hawa MI, Bazeos A, Leslie RD, et al. Buccals are likely to be a more informative surrogate tissue than blood for epigenome‐wide association studies. Epigenetics. 2013;8(4):445‐54.

CHAPTER TWO STEM CELL METHYLOME IN DIFFERENTIATED CELLS

41. Chen L, Ge B, Casale FP, Vasquez L, Kwan T, Garrido‐Martín D, et al. Genetic drivers of epigenetic and transcriptional variation in human immune cells. Cell. 2016;167(5):1398‐414. e24. 42. Jones PA. Functions of DNA methylation: islands, start sites, gene bodies and beyond. Nature Reviews Genetics. 2012;13(7):484. 43. Su AI, Wiltshire T, Batalov S, Lapp H, Ching KA, Block D, et al. A gene atlas of the mouse and human protein‐encoding transcriptomes. Proceedings of the National Academy of Sciences of the United States of America. 2004;101(16):6062‐7. 44. Farlik M, Halbritter F, Müller F, Choudry FA, Ebert P, Klughammer J, et al. DNA methylation dynamics of human hematopoietic stem cell differentiation. Cell stem cell. 2016;19(6):808‐22. 45. Hodges E, Molaro A, Dos Santos CO, Thekkat P, Song Q, Uren PJ, et al. Directional DNA methylation changes and complex intermediate states accompany lineage specificity in the adult hematopoietic compartment. Molecular cell. 2011;44(1):17‐28. 46. Hogart A, Lichtenberg J, Ajay SS, Anderson S, Center NIHIS, Margulies EH, et al. Genome‐ wide DNA methylation profiles in hematopoietic stem and progenitor cells reveal overrepresentation of ETS transcription factor binding sites. Genome Res. 2012;22(8):1407‐18. 47. Ji H, Ehrlich LI, Seita J, Murakami P, Doi A, Lindau P, et al. A comprehensive methylome map of lineage commitment from hematopoietic progenitors. Nature. 2010;467(7313):338. 48. Valencia RAC, Martino DJ, Saffery R, Ellis JA. In vitro exposure of human blood mononuclear cells to active vitamin D does not induce substantial change to DNA methylation on a genome‐scale. The Journal of steroid biochemistry and molecular biology. 2014;141:144‐9. 49. Ramagopalan SV, Maugeri NJ, Handunnetthi L, Lincoln MR, Orton SM, Dyment DA, et al. Expression of the multiple sclerosis‐associated MHC class II Allele HLA‐DRB1*1501 is regulated by vitamin D. PLoS genetics. 2009;5(2):e1000369. 50. Graves MC, Benton M, Lea R, Boyle M, Tajouri L, Macartney‐Coxson D, et al. Methylation differences at the HLA‐DRB1 locus in CD4+ T‐Cells are associated with multiple sclerosis. Multiple Sclerosis Journal. 2014;20(8):1033‐41. 51. Maltby VE, Graves MC, Lea RA, Benton MC, Sanders KA, Tajouri L, et al. Genome‐wide DNA methylation profiling of CD8+ T cells shows a distinct epigenetic signature to CD4+ T cells in multiple sclerosis patients. Clin Epigenetics. 2015;7(1):118. 52. Maltby VE, Lea RA, Graves MC, Sanders KA, Benton MC, Tajouri L, et al. Genome‐wide DNA methylation changes in CD19+ B cells from relapsing‐remitting multiple sclerosis patients. Scientific reports. 2018;8(1):17418. 53. Rhead B, Brorson IS, Berge T, Adams C, Quach H, Moen SM, et al. Increased DNA methylation of SLFN12 in CD4+ and CD8+ T cells from multiple sclerosis patients. PloS one. 2018;13(10):e0206511. 54. Hannon E, Gorrie‐Stone TJ, Smart MC, Burrage J, Hughes A, Bao Y, et al. Leveraging DNA‐ Methylation Quantitative‐Trait Loci to Characterize the Relationship between Methylomic Variation, Gene Expression, and Complex Traits. The American Journal of Human Genetics. 2018;103(5):654‐ 65. 55. Krzywinski MI, Schein JE, Birol I, Connors J, Gascoyne R, Horsman D, et al. Circos: An information aesthetic for comparative genomics. Genome Research. 2009.

CHAPTER TWO STEM CELL METHYLOME IN DIFFERENTIATED CELLS

FIGURES

Figure 1. The hypothesis underpinning this study. Foetal stem cells acquire DNA methylation in the process of differentiation and maturation into adult stem cells. This process is influenced by UV light and vitamin D which sets tolerance by 1) direct interaction with DNA methyltransferases 2) methylation of genes involved in vitamin D metabolism (including LDAD risk genes) and 3) methylation of the VDR cistrome (especially LDAD risk genes). This setting is recapitulated in progeny cells to lower the risk of LDADs. Adapted from 0337_Hematopoiesis_new by Rice University, licensed under a Creative Commons Attribution 4.0 International License, except where otherwise noted.

CHAPTER TWO STEM CELL METHYLOME IN DIFFERENTIATED CELLS

Figure 2. A) DNA methylation histograms of immune cell subsets and comparator buccal mucosa cell data derived from whole genome bisulfite sequencing data showing a bimodal pattern of DNA methylation across all cell subsets. The X axis represents methylation state (0 = completely unmethylated, 1 = completely methylated). B) Differentially methylated sites by subset comparison, upper plot CD34+ vs CD14+, middle plot CD34+ vs CD56+, lower plot CD14+ vs CD56+ and genomic location. For upper and middle plots, blue circles represent CGI where CD34+ is hypermethylated, relative to its comparator subset and red circles represent CD34+ hypomethylation relative to its comparator subset. For the lower plot, blue circles represent CD14+ hypermethylation and red circles represent CD14+ hypomethylation relative to CD56+. Figures were generated using Circos(55). C) A comparison of the number of differentially methylated CGIs between peripheral blood subsets and between peripheral blood subsets and buccal mucosa. For buccal mucosa methylation, extant WGBS data(40) was derived from 14 individuals to provide average methylation at base pair resolution. CGI level methylation was calculated as an average of cytosine residues within a CGI.

CHAPTER TWO STEM CELL METHYLOME IN DIFFERENTIATED CELLS

Figure 3. A) Dendrogram and associated heatmap generated using unsupervised hierarchical cluster analysis with one minus pearson correlation and complete linkage. CGI methylation from differentially methylated CGIs across all three comparisons was used. The cluster analysis shows discrete clustering of peripheral blood subsets. The three clusters from left to right are CD56+, CD34+ and CD14+. B) Multidimensional scaling analysis using methylation values from 395 CGIs that were differentially methylated between at least two subsets. The first three components covered 52% of total variation.

CHAPTER TWO STEM CELL METHYLOME IN DIFFERENTIATED CELLS

Figure 4. A) Relative methylation of CGI determined to be differentially methylated between CD34+ and CD14+ (left) and CD34+ and CD56+ (right) peripheral blood subsets. A greater proportion of differentially methylated CGI display higher methylation in CD56+ than CD34+ cells. B) Venn diagram showing the overlap of CGI found to be differentially methylated between CD34+ vs CD14+ and CD34+ vs CD56+ subsets. C) Number of 395 differentially methylated CGIs overlapping with genomic regulatory elements.

CHAPTER TWO STEM CELL METHYLOME IN DIFFERENTIATED CELLS

Figure 5. Individual differences in DNA methylation by subset. The first column shows a histogram indicating the number of CGIs (Y axis) with a specified variance (X axis) for CGIs with methylation data for all 11 individuals for A) CD14+, B) CD34+ and C) CD56+ subsets. The rightmost column plots display variance versus mean DNA methylation state. Variance amongst individuals appears to be greatest at CGI with intermediate mean methylation states.

CHAPTER TWO STEM CELL METHYLOME IN DIFFERENTIATED CELLS

Figure 6. A) Genomic features associated with ZMIZ1, including overlapping CpG islands, MS risk locus rs1250551 and VDR peaks. The pyrosequenced region is displayed at high magnification. B) DNA methylation by genotype at the pyrosequenced region. Each position was differentially methylated, including position 6, which was previously shown to be in mQTL with rs1250551. C) Only position 2 was found to be differentially methylated between MS and healthy controls. D) ZMIZ1 expression is decreased in MS E) ZMIZ1 gene expression by genotype and disease state.

CHAPTER TWO STEM CELL METHYLOME IN DIFFERENTIATED CELLS

TABLES

Table 1. Overlapping differentially methylated CGI by source and subset. WGBS data derived from 1 individual per subset; mRRBS derived from n=11.

Comparison Differentially Differentially Overlapping P (χ2, df=1) methylated methylated WGBS mRRBS CD34+ vs CD14+ 337 184 19 3.13x10‐4 CD34+ vs CD56+ 418 171 22 1.04x10‐4 CD14+ vs CD56+ 278 232 31 7.89x10‐7

CHAPTER TWO STEM CELL METHYLOME IN DIFFERENTIATED CELLS

Table 2. Number of gene/intergenic regions associated with differentially methylated CGI by subset

CD34+ vs CD14+ CD34+ vs CD56+ CD14+ vs CD56+ Intergenic 46 (22%) 28 (14%) 41 (16%) Intragenic 137 (67%) 138 (71%) 188 (71%) Promoter 23 (11%) 28 (14%) 35 (13%) Total 206 194 264

CHAPTER TWO STEM CELL METHYLOME IN DIFFERENTIATED CELLS

CO‐AUTHOR CONTRIBUTIONS

LO, GP and DB planned the experiment, analysed the data and prepared the manuscript. LO conducted all genomic methylation experiments. KV analysed the ZMIZ1 pyrosequencing data. GS and CL assisted in preparation of the manuscript. All authors read and approved the final manuscript.

Grant Parnell ______

Kelly Veale ______

Graeme Stewart ______

Christopher Liddle ______

David Booth ______

CHAPTER THREE

AGE‐DEPENDENT VDR PEAK DNA METHYLATION AS A

MECHANISM FOR LATITUDE‐DEPENDENT MS RISK

CHAPTER THREE VDR PEAK DNA METHYLATION IN MS RISK

Age‐dependent VDR peak DNA methylation as a mechanism for latitude‐dependent MS risk

*Lawrence T C Ong1,2, Stephen D Schibeci1, Nicole L Fewings1, David R Booth1, Grant P Parnell1

1Centre for Immunology and Allergy Research, Westmead Institute for Medical Research, The

University of Sydney, 176 Hawkesbury Rd, Westmead, New South Wales 2145, Australia

2Department of Immunology, Westmead Hospital, Cnr Darcy and Hawkesbury Rds,

Westmead, New South Wales 2145, Australia

*Corresponding author

Contact Details

Lawrence T C Ong – [email protected]

This manuscript has been submitted for peer review in the journal Clinical Epigenetics.

Keywords – DNA methylation, calcitriol, epigenetics, vitamin D, myeloid, VDR binding site

CHAPTER THREE VDR PEAK DNA METHYLATION IN MS RISK

ABSTRACT

Background: The mechanisms linking UV radiation and vitamin D exposure to the risk of acquiring the latitude and critical period dependent autoimmune disease, multiple sclerosis, is unclear. We examined the effect of vitamin D on DNA methylation as well as DNA methylation at vitamin D receptor binding sites in adult and paediatric myeloid cells.

Results: Very few DNA methylation changes occurred in adult and paediatric cells treated with calcitriol. However, several VDR binding sites across the genome demonstrated increased DNA methylation in cells of adult origin. Genes associated with these VDR binding sites were enriched for intracellular signalling and cell activation pathways, suggesting that age‐dependent potential for myeloid cell differentiation and adaptive immune system regulation may be encoded for by DNA methylation.

Conclusions: These results suggest vitamin D exposure at critical periods in immune system development may contribute to the well characterised latitude related differences in autoimmune disease incidence.

CHAPTER THREE VDR PEAK DNA METHYLATION IN MS RISK

BACKGROUND

The prevalence of autoimmune diseases such as multiple sclerosis (MS), type 1 diabetes mellitus, rheumatoid arthritis and atopic diseases such as asthma follow a latitude gradient, with increasing prevalence at latitudes more distant from the equator (1, 2). Ultraviolet light exposure and hence serum vitamin D levels are known to correlate with latitude, yet a precise mechanism linking vitamin D to immune disease remains elusive. DNA methylation, an important epigenetic mark, has been posited as a potential link between environmental exposures and disease due to its susceptibility to environmental change (3) and relative stability over time (4).

Some latitude dependent diseases such as MS also demonstrate a critical period, where risk factors such as latitude of residence appear to exert their influence during childhood and adolescence (5‐7). This critical period is perhaps underpinned by age‐related susceptibility to alterations in DNA methylation. DNA methylation changes have been detected in leukocyte development at key histone modifiers, chromatin remodellers and immune susceptibility loci within the first five years of life (8). DNA methylation changes proceed more rapidly in normal childhood development, with changes in peripheral blood occurring at a three to four‐fold higher rate compared with adults (9). Prenatal susceptibility to environmental insults such as famine, are also highly influenced by gestational age, resulting in persistent DNA methylation changes into adulthood (10, 11).

Vitamin D exerts its genomic effects through binding the vitamin D receptor (VDR). Binding of the active form of vitamin D, calcitriol, results in heterodimerisation of the VDR with the retinoid X receptor (RXR). This heterodimer binds regions of DNA known as vitamin D response elements (VDREs), which lie in the promoter regions of vitamin D responsive genes and lead to subsequent upregulation or suppression of DNA transcription. DNA methylation at VDREs may therefore interfere with the effects of calcitriol on transcriptional regulation.

Whilst the genomic effects of vitamin D have been well characterised through its interactions via the VDR‐RXR complex, its effects on DNA methylation have only been partially characterised. A study of human ex vivo leukocytes found vitamin D3 supplementation during and shortly after pregnancy led to mixed DNA methylation changes in mothers and infants. In infants, methylation loss was associated with genes regulating apoptosis and antigen presentation (12). In another study, global

CHAPTER THREE VDR PEAK DNA METHYLATION IN MS RISK

leukocyte DNA methylation increased in a dose‐dependent manner with vitamin D3 administration (13).

There have been few studies on the effect of vitamin D on the methylome of specific immune cell subsets. A study of cholecalciferol supplementation and mouse CD4+ T cells in experimental autoimmune encephalomyelitis (EAE; a mouse model of MS), showed global decreases in DNA methylation. This was associated with changes in the expression of enzymes involved in the establishment and maintenance of DNA methylation marks, with concomitant decreases in CD4+ T cell proliferation and differentiation into inflammatory Th1 and Th17 subsets (14). Another study found increases in Helios+ Foxp3+ T regulatory cells with 1,25(OH)2Vitamin D3 (calcitriol) supplementation that were associated with amelioration of EAE, with an increase rather than decrease in global DNA methylation (15).

More MS risk genes are predominantly expressed in mononuclear phagocytic cells than any other cell subset (16). These cells are likely to be important in the pathogenesis of MS through their regulation of immune cell differentiation, via mechanisms such as antigen presentation and expression of key vitamin D associated MS risk genes (16, 17). An epigenome wide study of vitamin D treatment on the human monocyte cell line, THP‐1, found marked changes in chromatin accessibility due to vitamin D with maximal chromatin opening after 24 hours (18). Despite this, ex vivo mononuclear cells cultured with vitamin D for up to 120 hours did not show any differentially methylated CpGs despite extensive changes in gene expression (19). The authors suggested donor age may have affected DNA methylation plasticity, however other factors including duration of culture, use of terminally differentiated cells and heterogeneous cell population, may also have contributed to the apparent lack of effect on DNA methylation.

We therefore hypothesised that differentiating haematopoietic progenitors into monocyte/macrophage lineage cells in the presence of sustained calcitriol exposure would result in age‐dependent DNA methylation changes. Thus, we sought to determine whether immune cell DNA methylation is affected by exposure to calcitriol. Secondly, because the multiple sclerosis latitude gradient appears to be mediated by a critical period, we sought to determine whether calcitriol‐ related DNA methylation changes vary with age. Because the genomic effects of vitamin D are

CHAPTER THREE VDR PEAK DNA METHYLATION IN MS RISK

mediated by its receptor, we also investigated the DNA methylation changes at corresponding binding sites.

RESULTS

Calcitriol results in changes in cell number, morphology and phenotype in cell culture

We conducted preliminary cell culture with varying calcitriol concentrations to determine effects on cell morphology and immunophenotype. Using CD34+ haematopoietic progenitor cells originating from an adult subject, at day 22, we noted a marked decrease in cell number and a less activated immunophenotype with higher concentrations of calcitriol (Figure 1). This preliminary experiment confirmed that a physiological concentration of calcitriol (0.1nM) was sufficient to elicit phenotypic changes in cultured cells. In the final experiment that cultured cells from two adult and two paediatric subjects, overall CD14+ percentage as a subset of CD45+ cells, was greater in cells of paediatric origin than those of adult origin with mean CD14+ proportion 78.9% vs 11.7% (p<4.99x10‐ 4, two‐tailed t‐test).

DNA methylation varies more due to age and individual differences than due to calcitriol

Whole genome bisulfite sequencing reads, alignment statistics, bisulfite conversion rates and coverage rates are detailed in Additional file 1. On average, 96% of genome wide CpGs were covered at an average depth of 16x. Genome wide DNA methylation did not vary to a large extent by age or calcitriol exposure. The average proportion of methylated reads was 0.83 for each of the four sample categories (adult + vitamin D, paediatric + vitamin D; see Figure 2A). Multidimensional scaling analysis of methylation values by sample found little difference in DNA methylation secondary to calcitriol. Differences due to calcitriol exposure were generally much smaller than those due to individual differences or age (Figure 2B).

At an individual CpG level there were few that differed in methylation state following calcitriol exposure. In cells of adult origin, there were 382 differentially methylated CpGs (FDR<0.05) corresponding to 29 autosomal and 11 mitochondrial genes/gene promoters. Amongst cells of paediatric origin, there were 37 differentially methylated CpGs corresponding to 2 genes. None of the differentially methylated CpGs overlapped between adult and paediatric samples or with VDR peaks. Of all adult differentially methylated CpGs, a subset mapped to the promoter region of one

CHAPTER THREE VDR PEAK DNA METHYLATION IN MS RISK

MS risk gene, PAPD7. Details of differentially methylated CpGs can be found in Additional file 2 and 3.

DNA methylation varies markedly by donor age at myeloid VDR peaks

VDR binding sites are another mechanism by which calcitriol may exert effects on gene expression. There were marked differences in the distribution of DNA methylation between samples of adult and paediatric origin at myeloid VDR peaks, which were not apparent at other transcription factor binding sites or regulatory regions (Figure 3). There was overall lower DNA methylation at myeloid VDR peaks in cells of paediatric origin (p<2.2x10‐16, Wilcoxon Rank Sum Test). RADmeth(20) was used to call differentially methylated CpGs between samples of adult and paediatric origin regardless of exposure to calcitriol. There were 26134 differentially methylated CpGs corresponding to 7244 VDR peaks (52% of all myeloid VDR peaks) and 2973 genes (Additional file 4). In comparison, analysis of CD14+ transcription factor binding sites (TFBS) yielded 7125 differentially methylated CpGs corresponding to 1896 TFBS (22% of annotated CD14+ TFBS). In comparison to TFBS, differential methylation was proportionally greater at VDR peaks than TFBS (χ2=3983, p<1x10‐5). A statistical overrepresentation test(21) (Panther GO‐slim annotation version 14.1, released March 12, 2019) found many immune and intracellular signalling ontologies to be enriched amongst genes corresponding to differentially methylated myeloid VDR peaks (Figure 4B).

The genomic annotations overlapping differentially methylated myeloid VDR peaks were then determined. Of interest, were CD14+ Ensembl regulatory build annotations and hg19 CpG island/shore annotations. Five prime and 3’ CpG island shores were designated as 2000bp upstream and downstream of the corresponding hg19 CpG island respectively. Fifty‐two percent of myeloid VDR peaks contained differentially methylated CpGs, with most of these being hypomethylated in cells of paediatric origin relative to cells of adult origin. The peaks overlapped predominantly with promoter regions, promoter flanking regions and CTCF binding sites, although not in the expected proportion in comparison to all myeloid VDR peaks (χ2=300.8, p<2.2x10‐16, df=8), suggestive of enrichment for specific genomic annotations (Figure 4C). The overlap of 352 CpGs previously identified as markers of biological age(22) with differentially methylated VDR peaks was ascertained to determine whether differential methylation could be attributed to the cumulative effects of epigenetic maintenance. None of the differentially methylated VDR peaks contained a “clock” CpG. Ninety of the differentially methylated VDR peaks overlapped with non‐HLA multiple sclerosis risk

CHAPTER THREE VDR PEAK DNA METHYLATION IN MS RISK

genes (Figure 4D), underlining the potential importance of DNA methylation differences in this latitude dependent autoimmune disease.

Transcriptomic effects of calcitriol vary by age

Transcriptomic analysis by RNA‐seq identified 183 downregulated and 154 upregulated genes amongst adult cells treated with calcitriol compared to no calcitriol, using a fold‐change threshold of two. Amongst cells of paediatric origin, there were 167 downregulated and 158 upregulated genes due to calcitriol (Figure 5A). Overall, only 75 differentially expressed genes overlapped between cells of adult and paediatric origin (Figure 5B), 57 in the same direction with calcitriol exposure and 18 in opposite directions (Additional file 6). A statistical overrepresentation test did not yield any statistically significant GO terms associated with any of the differentially expressed gene sets. None of the differentially expressed genes secondary to calcitriol were differentially methylated.

Age‐dependent transcriptomic effects are greater than calcitriol dependent effects

A greater number of differentially expressed genes were observed between adult and paediatric cells, independent of calcitriol exposure. We found 1002 genes were downregulated and 1252 upregulated in cells of paediatric compared to adult origin. Of these genes, 509 overlapped with differentially methylated myeloid VDR peaks (p<9.38x10‐6, hypergeometric test; Figure 5C & 6; co‐ location with genomic annotations is noted in Additional file 7). There was a negative correlation between expression fold‐change and methylation difference at CD14+ annotated promoters, consistent with the known relationship between DNA methylation at promoter regions and gene expression (Figure 5D). Of the differentially expressed genes, those underexpressed in cells of paediatric origin were enriched for biological processes relating to inflammation, intracellular signalling and metal ion homeostasis. Those overexpressed in cells of paediatric origin were associated with cellular replication and cell cycle processes (Figure 5E).

Of the 509 overlapping differentially expressed/methylated genes, 28 overlapped with 265 non‐HLA MS risk genes (p=1.82x10‐6, hypergeometric test). Two hundred and seventy‐six of these overlapping genes underexpressed in paediatric cells were enriched for the GO terms “cell surface receptor signalling”, “cell migration”, “intracellular signal transduction” and “protein phosphorylation”. The

CHAPTER THREE VDR PEAK DNA METHYLATION IN MS RISK

remaining 233 overexpressed genes were not enriched for any biological process terms (see Additional file 7).

DISCUSSION

This study examined potential interactions between calcitriol and DNA methylation in myeloid cells, to identify mechanisms underlying critical periods in the development of latitude dependent autoimmune diseases such as MS. Calcitriol addition resulted in marked morphologic and phenotypic effects, however, DNA methylation changes were relatively minor in comparison. Vitamin D independent DNA methylation changes differed between cells of paediatric and adult origin, especially at myeloid VDR peaks. MS risk genes were prominent among differentially methylated VDR peaks. Gene expression changes due to calcitriol, were marked, and differed strikingly between adult and paediatric cells, but only ~22% of differentially expressed genes were common to both conditions. The changes in VDR peak methylation due to age and calcitriol may be sufficient to drive the profoundly different transcriptomes between paediatric and adult derived mononuclear phagocytic cells, with evidence of MS risk gene involvement.

The calcitriol bound VDR‐RXR complex binds to VDREs and participates in transcriptional regulation. Therefore, DNA methylation changes at these regions are likely to have important functional consequences. We found 52% of myeloid VDR peaks were differentially methylated between cells of adult and paediatric origin. Differential methylation was enriched above background CD14+ TFBS, providing support for VDR specificity of age‐dependent changes. Most of the differentially methylated peaks also displayed decreased methylation in cells of paediatric origin. At 17% of these sites, there was concomitant differential gene expression, suggesting an immediate functional effect of these methylation differences within a subset of myeloid VDR peaks. Biological processes associated with underexpressed genes in paediatric cells were predominantly associated with inflammation, intracellular signalling and response to divalent cations, whereas those associated with overexpressed genes were predominantly associated with cellular proliferation.

Genes in cis with most differentially methylated VDR peaks were associated with biological processes important in inflammation and cellular differentiation, but not with changes in gene expression. VDR binding sites proximal to genes encoding PI3K subunits (including PIK3R1/3/6,

CHAPTER THREE VDR PEAK DNA METHYLATION IN MS RISK

PIK3CG, PIK3C2B) were differentially methylated between cells of adult and paediatric origin. The PI3K molecular pathway plays a role in myeloid cell differentiation (23, 24), monocyte antimycobacterial activity (25) and promotion of macrophage differentiation (26). Genes relating to the MAPK cascade were also enriched amongst differentially methylated VDR peak genes. The MAPK cascade is involved in significant crosstalk with the PI3K/AKT pathway (27) and has pleiotropic effects in monocytes/macrophages depending on the triggering stimulus and cell type. These effects include differentiation and activation(28). Together, the differential methylation of genes relating to both the PI3K and MAPK pathways suggest that the differing potential for vitamin D related myeloid cell differentiation between adult and paediatric cells may be encoded by DNA methylation.

Genes belonging to ontologies relating to the regulation of the adaptive immune response and regulation of T cell proliferation were also associated with differential methylation between adult and paediatric cells. This suggests that myeloid cells are differentially primed to influence the adaptive immune system in childhood compared with adulthood.

Consistent with previous work on vitamin D supplementation in mononuclear cells (19), DNA methylation in our myeloid cells appeared to be relatively insensitive to the effects of calcitriol. This occurred despite differentiation from haematopoietic progenitor cells from paediatric donors, which would presumably demonstrate greater DNA methylation plasticity in response to environmental stimuli. In contrast, the previously documented effect of vitamin D on DNA methylation in mouse CD4+ T cells was much more prominent(14). Many effects of vitamin D on human monocytes/macrophages may be mediated by epigenetic marks other than DNA methylation (18).

How age‐dependent methylomic differences at myeloid VDR binding sites confer long term risk for latitude dependent diseases is unclear. Monocytes typically persist in the circulation for up to ~1 week (29), and would be an unlikely substrate for DNA methylation dependent risk unless they migrate to peripheral sites. These tissue resident macrophages are known to persist for much longer periods (months to years) (30), perhaps transmitting early life influenced phenotypic changes that either predispose or limit propensity to autoimmunity in later life (Figure 7). Another possibility is that VDR agonism (or lack thereof) during early life leads to persistent changes in VDR binding site methylation and later susceptibility to VDR effector activities.

CHAPTER THREE VDR PEAK DNA METHYLATION IN MS RISK

One potential reason for finding minimal DNA methylation changes with vitamin D could be related to the duration of the cell cultures. Increasing passage number is associated with increases in DNA methylation (22), potentially distorting or obscuring the effects of vitamin D or age on DNA methylation. Due to the lack of CD14+ cells at earlier stages of culture, we were also unable to determine temporal effects on DNA methylation with vitamin D exposure. It has been previously shown that chromatin accessibility due to calcitriol exposure peaks at 24 hours and virtually returns to baseline levels after 48 hours (18). It could be argued however, that DNA methylation changes are less likely to occur within these time frames in comparison to histone modifications and non‐coding RNAs. Finally, whether the phenomenon of age‐related differential methylation at VDR binding sites occurs in vivo requires further investigation.

Future studies will need to ascertain the robustness of our present findings across a greater number of biological replicates. This study also raises questions regarding age‐dependent VDR methylation in other cell lineages as well as haematopoietic progenitor cells, and whether VDR methylation settings might be transmitted to progeny cells. The model of altered tissue macrophage phenotype might also be amenable to study by comparison of their phenotype/function in MS with normal individuals, for example in co‐culture.

CONCLUSION

Whilst vitamin D has minor effects on the myeloid methylome, age‐dependent differences in VDR peak DNA methylation suggest vitamin D exposure at critical periods in immune system development may contribute to well characterised latitude related differences in autoimmune disease risk.

METHODS

Cell isolation

Adult peripheral blood was extracted by injecting approximately 50ml of Dulbecco’s PBS (DPBS) with 10% acid citrate dextrose (Sigma Aldrich) into leukocyte reduction system (LRS) chambers discarded following plateletpheresis from two male subjects and allowing the chamber to drain under gravity. Umbilical cord blood units obtained from one male and one female donor, were diluted in a 1:1 ratio with DPBS. Mononuclear cells from both adult and paediatric samples were then obtained by density gradient centrifugation (Ficoll‐Paque PLUS, GE Healthcare). Positive selection of CD34+

CHAPTER THREE VDR PEAK DNA METHYLATION IN MS RISK

haematopoietic progenitor cells was then performed using CD34 MicroBead Kit Ultrapure Microbeads (Miltenyi Biotec) at a volume of 50µl per 108 cells, with an autoMACS Pro Separator (Miltenyi Biotec) as per manufacturer’s instructions. Purified cells were plated at a density of 5x104 cells/ml of media.

Cell culture

CD14+ cells were cultured from haematopoietic cell precursors (CD34+) using a previously published protocol with minor modifications(31). The culture cocktail contained X‐VIVO10 (Lonza), Albumex (Seqirus) 0.05%, SCF (Peprotech) 200ng/ml, GM‐CSF (Peprotech) 0.03ug/ml, M‐CSF (premium grade; Miltenyi Biotec) 5000U/ml, IL‐6 (Peprotech) 10ng/ml, FLT3 ligand (Peprotech) 50ng/ml and

Gentamicin (Sigma Aldrich) 50µg/ml. Calcitriol (1,25(OH)2Vitamin D3; BioGems) was added at a

o physiological concentration of 0.1nM. Cells were incubated at 37 C with 5% CO2 for one week before replating at a density of 1x105 cells/ml of media. Media was then changed every third day by demi‐ depletion. Cells were harvested on day 21.

Harvested cells were centrifuged at 300xg for 5 minutes before the supernatant was removed. The cells were resuspended in 1ml of chilled PBS, stained with 1µl of LIVE/DEAD Fixable Aqua Dead Cell Stain (Life Technologies) and incubated on ice for 30 minutes. The cells were then washed and stained with the following antibody cocktail: CD45 BUV395 (BD Horizon), CD14 BV421 (BD), CD16 BV650 (BD), HLA‐DR FITC (BD Pharmingen), CD34 PE (BD Pharmingen), CD11b APC (BD). FACS sorting was carried out on a BD Influx and CD14+ cells subjected to two washes with DPBS at 1000xg before storage at ‐80oC.

Whole genome bisulfite sequencing

DNA was extracted using QIAamp DNA extraction kit (Qiagen) as per manufacturer’s instructions. Whole genome bisulfite sequencing libraries were generated with the Accel‐NGS Methyl‐seq DNA Library Kit (Swift Biosciences) and sequenced on a HiSeq X10 (Illumina) in 150bp PE mode with PhiX spike‐in to counteract low sequence diversity.

CHAPTER THREE VDR PEAK DNA METHYLATION IN MS RISK

Data QC, alignment and processing

The quality of raw sequences was ascertained using FastQC (32). Quality trimming was carried out using Trim galore (33) in paired end mode with the following parameters ‐quality 20, ‐‐ three_prime_clip_R1 5, ‐‐clip_R2 15. Reads were aligned to the hg19 genome using the Wildcard Alignment Tool (WALT) with default settings and the output .sam files converted to .mr files using to‐ mr before further processing with Methpipe (34). Firstly, duplicate removal was performed using duplicate‐remover, followed by estimation of bisulfite conversion rates and coverage/methylation level statistics using bsrate and levels respectively. Methylation calls were made using methcounts (using the ‐n option for CpG context cytosines only) before symmetric‐cpgs was used to extract and merge methylation data from both strands. Regional methylation analysis was performed using the roimethstat module (with ‐P and ‐v options), to determine methylation state within a prespecified region of interest.

Differential methylation analysis

RADmeth (20) was utilised for differential methylation analysis. The effects of calcitriol and age were considered separately by comparing the effects of calcitriol amongst cells of adult and paediatric origin separately. To further examine the specific effects of vitamin D, myeloid vitamin D receptor (VDR) peaks (35) and a 500bp region up and downstream were also examined. DNA methylation at these sites was compared to experimentally validated CD14+ transcription factor binding sites (TFBS) (36) (+500bp). Overlap between differentially methylated regions and genes/genomic annotations was determined using Bedtools (37) intersect, with closest being used to determine the nearest gene to differentially methylated myeloid VDR peaks.

RNA sequencing

Due to low cell number, RNA‐seq was not performed on cells cultured from one of the adult subjects. RNA was extracted from other cultured CD14+ cells with the RNeasy Mini Kit (Qiagen) as per manufacturer’s instructions. Sequencing libraries were generated with the QIAseq Stranded mRNA Select Kit (Qiagen) and sequenced on the Novaseq 6000 (Illumina) using 100bp SE mode. The quality of raw sequences was ascertained using FastQC (32). Fifteen base pairs were trimmed from the start of each read using Trimmomatic (38) before alignment with TopHat2 (39). Assignment of aligned reads to genes was performed using featureCounts (40). Quantile normalised RPKM values

CHAPTER THREE VDR PEAK DNA METHYLATION IN MS RISK

were utilised to determine fold‐change differences. Genes with aggregate read counts <100 or belonging to sex chromosomes were excluded from further analysis, resulting in 12492 genes. Only non‐zero numerators or denominators were kept for fold‐change and downstream calculations.

Declarations

Ethics approval and consent to participate

This study received ethics approval from the Western Sydney Local Health District Human Research Ethics Committee (HREC2002/9/3.6(1425) & (5366) AU RED LNR/17/WMEAD/447).

Consent for publication

Not applicable

Availability of data and materials

The datasets used and/or analysed during the current study are available from the corresponding author on reasonable request.

Competing interests

The authors declare that they have no competing interests.

Funding

This study was supported by a Multiple Sclerosis Research Australia Incubator Grant. LO received support from a co‐funded NHMRC/Multiple Sclerosis Research Australia/Trish MS Foundation scholarship and a NSW Health Pathology Postgraduate Fellowship

Acknowledgements

The authors would like to acknowledge Australian Red Cross Blood Services, the Sydney Cord Blood Bank and donors for providing samples for this study. The authors also acknowledge Prof David

CHAPTER THREE VDR PEAK DNA METHYLATION IN MS RISK

Brown for his invaluable comments on the manuscript. Flow cytometry was performed at the Flow Cytometry Core Facility supported by the Westmead Research Hub, Cancer Institute NSW and NHMRC. Bioinformatic analysis was supported by Sydney Informatics Hub, funded by the University of Sydney.

CHAPTER THREE VDR PEAK DNA METHYLATION IN MS RISK

REFERENCES

1. Wacker M, Holick MF. Sunlight and Vitamin D: A global perspective for health. Dermato‐ endocrinology. 2013;5(1):51‐108. 2. Osborne NJ, Ukoumunne OC, Wake M, Allen KJ. Prevalence of eczema and food allergy is associated with latitude in Australia. Journal of allergy and clinical immunology. 2012;129(3):865‐7. 3. Heim C, Binder EB. Current research trends in early life stress and depression: Review of human studies on sensitive periods, gene–environment interactions, and epigenetics. Experimental neurology. 2012;233(1):102‐11. 4. Cedar H, Bergman Y. Linking DNA methylation and histone modification: patterns and paradigms. Nature Reviews Genetics. 2009;10(5):295. 5. Ahlgren C, Lycke J, Odén A, Andersen O. High risk of MS in Iranian immigrants in Gothenburg, Sweden. Multiple sclerosis journal. 2010;16(9):1079‐82. 6. Gale CR, Martyn CN. Migrant studies in multiple sclerosis. Progress in neurobiology. 1995;47(4‐5):425‐48. 7. Ahlgren C, Odén A, Lycke J. A nationwide survey of the prevalence of multiple sclerosis in immigrant populations of Sweden. Multiple Sclerosis Journal. 2012;18(8):1099‐107. 8. Acevedo N, Reinius LE, Vitezic M, Fortino V, Söderhäll C, Honkanen H, et al. Age‐associated DNA methylation changes in immune genes, histone modifiers and chromatin remodeling factors within 5 years after birth in human blood leukocytes. Clinical epigenetics. 2015;7(1):34. 9. Alisch RS, Barwick BG, Chopra P, Myrick LK, Satten GA, Conneely KN, et al. Age‐associated DNA methylation in pediatric populations. Genome research. 2012:gr. 125187.111. 10. Tobi EW, Lumey L, Talens RP, Kremer D, Putter H, Stein AD, et al. DNA methylation differences after exposure to prenatal famine are common and timing‐and sex‐specific. Human molecular genetics. 2009;18(21):4046‐53. 11. Tobi EW, Slieker RC, Stein AD, Suchiman HED, Slagboom PE, van Zwet EW, et al. Early gestation as the critical time‐window for changes in the prenatal environment to affect the adult human blood methylome. International journal of epidemiology. 2015;44(4):1211‐23. 12. Anderson CM, Gillespie SL, Thiele DK, Ralph JL, Ohm JE. Effects of Maternal Vitamin D Supplementation on the Maternal and Infant Epigenome. Breastfeed Med. 2018;13(5):371‐80. 13. Zhu H, Bhagatwala J, Huang Y, Pollock NK, Parikh S, Raed A, et al. Race/ethnicity‐specific association of vitamin D and global DNA methylation: cross‐sectional and interventional findings. PloS one. 2016;11(4):e0152849. 14. Zeitelhofer M, Adzemovic MZ, Gomez‐Cabrero D, Bergman P, Hochmeister S, N'Diaye M, et al. Functional genomics analysis of vitamin D effects on CD4+ T cells in vivo in experimental autoimmune encephalomyelitis. Proceedings of the National Academy of Sciences of the United States of America. 2017;114(9):E1678‐e87. 15. Moore JR, Hubler SL, Nelson CD, Nashold FE, Spanier JA, Hayes CE. 1,25‐Dihydroxyvitamin D3 increases the methionine cycle, CD4(+) T cell DNA methylation and Helios(+)Foxp3(+) T regulatory cells to reverse autoimmune neurodegenerative disease. J Neuroimmunol. 2018;324:100‐14. 16. Parnell GP, Booth DR. The multiple sclerosis (MS) genetic risk factors indicate both acquired and innate immune cell subsets contribute to MS pathogenesis and identify novel therapeutic opportunities. Frontiers in immunology. 2017;8:425. 17. Shahijanian F, Parnell GP, McKay FC, Gatt PN, Shojoei M, O'Connor KS, et al. The CYP27B1 variant associated with an increased risk of autoimmune disease is underexpressed in tolerizing dendritic cells. Human molecular genetics. 2014;23(6):1425‐34. 18. Seuter S, Neme A, Carlberg C. Epigenome‐wide effects of vitamin D and their impact on the transcriptome of human monocytes involve CTCF. Nucleic acids research. 2016;44(9):4090‐104. 19. Chavez Valencia RA, Martino DJ, Saffery R, Ellis JA. In vitro exposure of human blood mononuclear cells to active vitamin D does not induce substantial change to DNA methylation on a genome‐scale. The Journal of steroid biochemistry and molecular biology. 2014;141:144‐9.

CHAPTER THREE VDR PEAK DNA METHYLATION IN MS RISK

20. Dolzhenko E, Smith AD. Using beta‐binomial regression for high‐precision differential methylation analysis in multifactor whole‐genome bisulfite sequencing experiments. BMC bioinformatics. 2014;15(1):215. 21. Mi H, Huang X, Muruganujan A, Tang H, Mills C, Kang D, et al. PANTHER version 11: expanded annotation data from Gene Ontology and Reactome pathways, and data analysis tool enhancements. Nucleic acids research. 2016;45(D1):D183‐D9. 22. Horvath S. DNA methylation age of human tissues and cell types. Genome biology. 2013;14(10):3156. 23. Hmama Z, Nandan D, Sly L, Knutson KL, Herrera‐Velit P, Reiner NE. 1α, 25‐dihydroxyvitamin D3–induced myeloid cell differentiation is regulated by a vitamin D receptor–phosphatidylinositol 3‐ kinase signaling complex. The Journal of experimental medicine. 1999;190(11):1583‐94. 24. Neri LM, Marchisio M, Colamussi ML, Bertagnolo V. Monocytic differentiation of HL‐60 cells is characterized by the nuclear translocation of phosphatidylinositol 3‐kinase and of definite phosphatidylinositol‐specific phospholipase C isoforms. Biochemical and biophysical research communications. 1999;259(2):314‐20. 25. Sly LM, Lopez M, Nauseef WM, Reiner NE. 1α, 25‐Dihydroxyvitamin D3‐induced monocyte antimycobacterial activity is regulated by phosphatidylinositol 3‐kinase and mediated by the NADPH‐ dependent phagocyte oxidase. Journal of Biological Chemistry. 2001;276(38):35482‐93. 26. Liu Q, Ning W, Dantzer R, Freund GG, Kelley KW. Activation of protein kinase C‐ζ and phosphatidylinositol 3′‐kinase and promotion of macrophage differentiation by insulin‐like growth factor‐I. The Journal of Immunology. 1998;160(3):1393‐401. 27. Aksamitiene E, Kiyatkin A, Kholodenko BN. Cross‐talk between mitogenic Ras/MAPK and survival PI3K/Akt pathways: a fine balance. Portland Press Ltd.; 2012. 28. Rao KMK. MAP kinase activation in macrophages. Journal of leukocyte biology. 2001;69(1):3‐ 10. 29. Patel AA, Zhang Y, Fullerton JN, Boelen L, Rongvaux A, Maini AA, et al. The fate and lifespan of human monocyte subsets in steady state and systemic inflammation. Journal of Experimental Medicine. 2017;214(7):1913‐23. 30. Parihar A, Eubank TD, Doseff AI. Monocytes and macrophages regulate immunity through dynamic networks of survival and cell death. J Innate Immun. 2010;2(3):204‐15. 31. Way KJ, Dinh H, Keene MR, White KE, Clanchy FI, Lusby P, et al. The generation and properties of human macrophage populations from hemopoietic stem cells. Journal of leukocyte biology. 2009;85(5):766‐78. 32. Andrews S. FastQC: a quality control tool for high throughput sequence data. 2010. 2017. 33. Krueger F. Trim galore. A wrapper tool around Cutadapt and FastQC to consistently apply quality and adapter trimming to FastQ files. 2015. 34. Song Q, Decato B, Hong EE, Zhou M, Fang F, Qu J, et al. A reference methylome database and analysis pipeline to facilitate integrative and comparative epigenomics. PloS one. 2013;8(12):e81148. 35. Booth D, Ding N, Parnell G, Shahijanian F, Coulter S, Schibeci S, et al. Cistromic and genetic evidence that the vitamin D receptor mediates susceptibility to latitude‐dependent autoimmune diseases. Genes and immunity. 2016;17(4):213. 36. Zerbino DR, Wilder SP, Johnson N, Juettemann T, Flicek PR. The ensembl regulatory build. Genome biology. 2015;16(1):56. 37. Quinlan AR. BEDTools: the Swiss‐army tool for genome feature analysis. Current protocols in bioinformatics. 2014;47(1):11.2. 1‐.2. 34. 38. Bolger AM, Lohse M, Usadel B. Trimmomatic: a flexible trimmer for Illumina sequence data. Bioinformatics. 2014;30(15):2114‐20. 39. Kim D, Pertea G, Trapnell C, Pimentel H, Kelley R, Salzberg SL. TopHat2: accurate alignment of transcriptomes in the presence of insertions, deletions and gene fusions. Genome biology. 2013;14(4):R36.

CHAPTER THREE VDR PEAK DNA METHYLATION IN MS RISK

40. Liao Y, Smyth GK, Shi W. featureCounts: an efficient general purpose program for assigning sequence reads to genomic features. Bioinformatics. 2014;30(7):923‐30. 41. Taylor C. 3 CELLS fvcc104. OpenStax CNX 3 Nov 2019 [Available from: http://cnx.org/contents/21a91101‐9df0‐4826‐bb9b‐[email protected].

CHAPTER THREE VDR PEAK DNA METHYLATION IN MS RISK

FIGURES

Figure 1. Light microscopy and flow cytometric characterisation of cultured cells at day 22. Cultured cells of adult origin with media containing 0nM calcitriol (left), 0.1nM calcitriol (centre) and 50nM calcitriol (right). There were marked morphologic and immunophenotypic changes, with overall decrease in cell number, fewer fusiform shaped cells, greater CD14+ proportion and decreases in HLA‐DR and CD16 expression (not shown) at higher calcitriol concentrations.

CHAPTER THREE VDR PEAK DNA METHYLATION IN MS RISK

Figure 2. Global DNA methylation by age and calcitriol status. A) Frequency histograms of 1kb tile, genome wide DNA methylation, showing similar distribution of DNA methylation between conditions. B) Multidimensional scaling analysis of CpG wise methylation values demonstrating only minor differences in DNA methylation with the addition of calcitriol. A – adult, P – paediatric, N – no calcitriol, D – with calcitriol, 1 or 2 refer to adult or paediatric subject 1 or 2.

CHAPTER THREE VDR PEAK DNA METHYLATION IN MS RISK

Figure 3. Distribution of DNA methylation by genomic feature. A) Violin plots of DNA methylation across various genomic features by cell origin and culture condition. Myeloid VDR peaks demonstrated DNA methylation that was skewed towards lower methylation levels in cells of paediatric origin in comparison to those of adult origin. The effects of calcitriol on DNA methylation distribution were not evident. B) Methylation difference between cells of paediatric and adult origin at myeloid VDR peaks and transcription factor binding sites, showing skewing towards paediatric

CHAPTER THREE VDR PEAK DNA METHYLATION IN MS RISK

hypomethylation at myeloid VDR peaks, but not at transcription factor binding sites. A – adult, P – paediatric, N – no calcitriol, D – with calcitriol, 1 or 2 refer to adult or paediatric subject 1 or 2.

100

CHAPTER THREE VDR PEAK DNA METHYLATION IN MS RISK

Figure 4. Characteristics of differentially methylated VDR peaks. A) Breakdown of VDR peaks based on differential methylation status and overlap with regions of interest (regulatory regions, CpG islands and island shores; left) and differentially methylated VDR peaks overlapping with regions of interest (right). The majority of regulatory regions demonstrated hypomethylation in cells of paediatric origin. B) Overrepresented GO terms (FDR <0.05) associated with differentially methylated VDR peaks. C) Breakdown of differentially methylated myeloid VDR peaks and corresponding annotation overlaps compared with all annotated VDR peaks. D) Overlap of currently known non‐HLA MS risk genes and their overlap with differentially methylated myeloid VDR peaks. PTK – protein tyrosine kinase, PI3P – phosphatidylinositol‐3‐phosphate, CGI – CpG island, PFR – promoter flanking region, TFBS – transcription factor binding site.

101

CHAPTER THREE VDR PEAK DNA METHYLATION IN MS RISK

Figure 5. Differentially expressed genes, associated gene ontologies and DNA methylation overlap. A) Number of genes demonstrating changes in expression (>2 fold) due to calcitriol in cells of adult and paediatric origin. B) A minority of common genes are differentially expressed in response to calcitriol amongst cells of adult and paediatric origin. C) Overlap between differentially expressed genes and differentially methylated myeloid VDR peaks/genes when comparing cells of adult and paediatric origin. D) Scatter plot of overlapping sites from C) corresponding to annotated promoter regions. There was a significant negative correlation between methylation difference (paediatric – adult) and log fold‐change (paediatric/adult). E) Overrepresented GO biological process terms (FDR<0.05) of differentially expressed sites from C) with two‐fold decreased expression in cells of

102

CHAPTER THREE VDR PEAK DNA METHYLATION IN MS RISK

paediatric origin (left) or two‐fold increase in cells of paediatric origin (right). DE – differentially expressed

103

CHAPTER THREE VDR PEAK DNA METHYLATION IN MS RISK

Figure 6. An example of differential methylation at myeloid VDR peaks overlapping with MS risk genes. In cells of paediatric origin, DNA methylation was increased at IRF8 (left) and decreased at TNIP3 (right) relative to cells of adult origin. Both genes were also differentially expressed between cells of adult and paediatric origin (see Additional file 7). Black arrows denote differentially methylated regions. Red tracks – paediatric, blue tracks – adult.

104

CHAPTER THREE VDR PEAK DNA METHYLATION IN MS RISK

Figure 7. A potential mechanism for the development of autoimmune disease risk dependent on early life vitamin D exposure. Decreased VDR binding site methylation in early life increases phenotypic plasticity and susceptibility to vitamin D exposure. Because tissue macrophages persist for months to years, phenotypic settings resulting from vitamin D exposure in early life may lead to a tolerogenic or autoimmune propensity in later life. Macrophage images adapted from (41).

105

CHAPTER THREE VDR PEAK DNA METHYLATION IN MS RISK

CO‐AUTHOR CONTRIBUTIONS

LO, GP and DB devised the experiments. NF and SS assisted in planning and analysis of cell culture and flow cytometric experiments. LO conducted the experiments, analyses and prepared the manuscript. GP performed RNA‐seq and assisted in data analysis. All authors read and approved the final manuscript.

Stephen Schibeci ______

Nicole Fewings ______

David Booth ______

Grant Parnell ______

106

CHAPTER FOUR

TRANSCRIBED B LYMPHOCYTE GENES AND MULTIPLE

SCLEROSIS RISK GENES ARE UNDERREPRESENTED IN

EPSTEIN‐BARR VIRUS HYPOMETHYLATED REGIONS

CHAPTER FOUR EBV HYPOMETHYLATION IN MS RISK GENES

Transcribed B lymphocyte genes and multiple sclerosis risk genes are underrepresented in Epstein‐

Barr virus hypomethylated regions

*Lawrence T. C. Ong1,2, Grant P. Parnell1, Ali Afrasiabi1, Graeme J. Stewart1,2, Sanjay Swaminathan1,2,

David R. Booth1

1 Centre for Immunology and Allergy Research, Westmead Institute for Medical Research, The

University of Sydney, 176 Hawkesbury Rd, Westmead, NSW 2145, Australia

2 Department of Clinical Immunology and Allergy, Westmead Hospital, Cnr Darcy and Hawkesbury

Rds, Westmead, NSW 2145, Australia

Corresponding author: Lawrence T. C. Ong – [email protected]

This manuscript has been published:

Ong, LTC, Parnell, GP, Afrasiabi, A, Stewart, GJ, Swaminathan, S, Booth DR. Transcribed B lymphocyte genes and multiple sclerosis risk genes are underrepresented in Epstein‐Barr Virus hypomethylated regions. Genes and Immunity (2019); 16, pp. 1‐9

108

CHAPTER FOUR EBV HYPOMETHYLATION IN MS RISK GENES

ABSTRACT

Epstein‐Barr Virus (EBV) infection appears to be necessary for the development of Multiple Sclerosis

(MS), although the specific mechanisms are unknown. More than 200 single nucleotide polymorphisms (SNPs) are known to be associated with the risk of developing MS. About a quarter of these are also highly associated with proximal gene expression in B cells infected with EBV

(lymphoblastoid cell lines ‐ LCLs). The DNA of LCLs is hypomethylated compared to both uninfected and activated B cells. Since methylation can affect gene expression, and so cell differentiation and immune evasion, we hypothesised that EBV‐driven hypomethylation may affect the interaction between EBV infection and MS. We interrogated an existing data set comprising three individuals with whole genome bisulfite sequencing data from EBV transformed B cells and CD40 activated B cells. DNA methylation surrounding MS risk SNPs associated with gene expression in LCLs (LCLeQTL) was less likely to be hypomethylated than randomly selected chromosomal regions. Differential methylation was independent of genomic features such as promoter regions, but genes preferentially expressed in EBV infected B cells, including the LCLeQTL genes, were underrepresented in the hypomethylated regions. Our data does not indicate MS genetic risk is affected by EBV hypomethylation.

109

CHAPTER FOUR EBV HYPOMETHYLATION IN MS RISK GENES

INTRODUCTION

Epstein‐Barr Virus (EBV) is a gamma herpesvirus that has been implicated as an aetiological factor in several autoimmune and malignant conditions. Although most people have life‐long infection with this virus, infection is usually under sufficient immunological control that there are no pathogenic consequences. In immunocompromised individuals, or those with certain mutations, EBV infection and proliferation can cause significant immune perturbation and death (1). Suboptimal control of

EBV infection may underpin the development of autoimmune disease and malignant conditions.

Multiple sclerosis (MS) is an autoimmune and neurodegenerative condition that affects the central nervous system, causing demyelination and progressive disability. Although most healthy individuals will become seropositive for EBV (80‐90%), almost all MS affected individuals are seropositive (2).

This suggests that EBV infection is necessary, but not sufficient for the development of MS and that interactions with other risk mechanisms might be necessary for disease. At present, over 200 single nucleotide polymorphisms (SNPs) have been found to be associated with increased susceptibility to

MS (3), and the vast proportion of the genes proximal to these SNPs are associated with immune cell pathways, and are associated with altered gene expression in immune cells of the blood (4). Many of the MS risk SNPs have a stronger or different association with gene expression in EBV infected B cells than blood (5). The altered expression of risk genes appears in turn to affect infected B cell functions, including cell proliferation, and immune system evasion. Genetic variation due to somatic variation in genes affecting these functions has also been identified in nasopharyngeal carcinomas caused by EBV (6). The MS risk SNPs may affect gene expression through altering transcription factor control of gene expression, through direct mechanisms such as affecting binding motifs, or through indirect mechanisms, through interaction with other gene regulatory processes.

110

CHAPTER FOUR EBV HYPOMETHYLATION IN MS RISK GENES

One of the potential indirect mechanisms is through the effects of EBV infection on host DNA methylation. DNA methylation is a relatively stable epigenetic modification that is associated with changes in gene transcription, particularly at transcription start sites and promoter regions. An interaction between risk polymorphisms and EBV methylation may lead to abnormal activation, proliferation or persistence, and so disease.

EBV infection is known to cause widespread changes in genomic DNA methylation. Genome wide studies of lymphoblastoid cell lines (EBV transformed B lymphocyte lines) have mostly found DNA methylation to be decreased relative to comparator subsets such as whole blood cells (7, 8) and peripheral blood lymphocytes or leukocytes (9) perhaps due to decreased DNMT1 expression (10).

Studies comparing the DNA methylation of B cells and LCLs are more robust at detecting changes due specifically to EBV infection, as they are not confounded by cell lineage specific methylation differences. Although these studies are limited (10‐12), they have found LCLs demonstrate hypomethylation of promoter regions corresponding to B cell biological pathways compared to resting B cells (12). In addition, EBV infection tends to increase DNA methylation in high CpG content promoters and decrease DNA methylation in low CpG content promoters (10). Finally, EBV infection has been found to result in hypomethylation of over two‐thirds of the entire genome (11).

We therefore hypothesised that EBV‐driven hypomethylation may affect the interaction between

EBV infection and MS. Specifically, we predicted that if hypomethylation contributes to the dysregulation of infected B cells leading to MS, MS risk loci would be over‐represented in these hypomethylated regions, especially the subset of these risk loci known to correlate with gene expression in LCLs.

111

CHAPTER FOUR EBV HYPOMETHYLATION IN MS RISK GENES

MATERIALS AND METHODS

Datasets, alignment and methylation Calling

We utilised extant whole genome bisulfite sequencing data from LCLs, CD40L activated B cells and resting B cells from three individuals(11). Resting B cells have very different transcriptomes compared to LCLs and activated B cells. Therefore, unless otherwise specified, methylation of CD40L activated B cells was compared to LCLs to distinguish between the effects of proliferation due to virus and B cell activation. The quality of raw sequences was ascertained using FastQC v0.11.7(13).

Adapter trimming was then carried out using Trimmomatic v0.36 (14) with HEADCROP (set to 3) and

MAXINFO (target length = 40, strictness = 0.5) options. Reads were aligned to the hg19 genome using the Wildcard Alignment Tool (WALT) v1.0. Modules from Methpipe v3.4.3(15) were then used to process aligned files. Firstly, .sam files from the alignment were converted to .mr format using the to‐mr module. Duplicate removal was performed using duplicate‐remover, followed by estimation of bisulfite conversion rates using bsrate. Methylation calls were performed using methcounts with the

‐n option, to limit calls to CpG context cytosines only. Methcounts data were then merged using merge‐methcounts for biological replicates within each condition i.e. LCL, CD40L activated B cells and resting B cells. Coverage data was determined using the levels module. Because CpG methylation is most commonly symmetric, the symmetric‐cpg module was then used to merge methylation data from both strands prior to further analysis. Regional methylation analysis was performed using the roimethstat module (with ‐P and ‐v options), which determines the average methylation state within a prespecified region of interest.

Regions of interest

We performed tiling analysis using 1kb tiles centred upon MS risk loci identified previously by the

International MS Genetics Consortium as being associated with increased disease risk (16)

112

CHAPTER FOUR EBV HYPOMETHYLATION IN MS RISK GENES

(MSGWAS). We also chose a subset of these risk loci which have been previously found to be associated with gene expression in LCLs (5) (LCLeQTL). We compared the number of differentially methylated regions with regions centred upon an unbiased list of single nucleotide polymorphism from the NHGRI‐EBI Catalog of published genome‐wide association studies (17) (GWAS Catalog; downloaded 8 September 2018). Whole genome 1kb tiles were derived using Bedtools (v2.25.0) makewindows with ‐w set to 1000, based on hg19 coordinates. Significant variant‐gene association lists were downloaded from https://storage.googleapis.com/gtex_analysis_v7/single_tissue_eqtl_data/GTEx_Analysis_v7_eQTL.t ar.gz and the loci with the most significant associations (nominal p value <10‐20) extracted for whole blood (Whole_Blood.v7.signif_variant_gene_pairs.txt.gz) and LCLs (Cells_EBV‐ transformed_lymphocytes.v7.signif_variant_gene_pairs.txt.gz). Promoter regions were defined by the Ensembl regulatory build (18) and downloaded from ftp://ftp.ensembl.org/pub/grch37/current/regulation/homo_sapiens/. Exon and intron coordinates were derived from UCSC Genome Browser Table Browser > group: Genes and Gene Predictions > track: UCSC Genes > table: knownGene. Regions of the genome not covered by promoters, introns or exons were considered to be intergenic.

Differential methylation

Differentially methylated regions were determined by taking subset specific methylation values for a region of interest (as determined by roimethstat) and subtracting them from each other. An absolute methylation difference of >0.2 was used as the threshold for calling a differentially methylated region (DMR). Bedtools v2.25.0 was used for analysis of genomic features (19) and statistical analyses were performed using R statistical software and GraphPad Prism 8.0.0. Sex, mitochondrial and haploid chromosomes were excluded from analyses.

113

CHAPTER FOUR EBV HYPOMETHYLATION IN MS RISK GENES

Proximal genes

Genes proximal to an MS risk locus, the midpoint of a 1kb tile or a region of interest were determined by using the closest module of Bedtools v2.25.0.

RNA sequencing & analysis

To correlate EBV induced DNA methylation alterations with gene expression, we performed global gene expression profiling by RNA‐seq on paired resting B cells and LCL samples from five individuals as previously described (5). The raw and processed sequencing data generated are available from the NCBI Gene Expression Omnibus (GEO) under accession number GSE126379. Log2 of normalised relative gene expression levels (logFC) was correlated with the methylation difference between LCL and resting B cells at exons, introns and promoters proximal to these genes. Where more than one gene was proximal to an annotation, one gene was selected at random as the proximal gene.

RESULTS

Bisulfite conversion rates and coverage data following sequence alignment and duplicate removal can be found in Supplementary Information 1 and 2. In summary, the mean bisulfite conversion rate was 98.3% for LCLs, 98.4% for activated B cells and 96.9% for resting B cells. Mean coverage was 5.7x for LCLs, 5.8x for activated B cells and 5.5x for resting B cells. We used a methylation difference of

>0.2 as the threshold for identifying differentially methylated regions (DMRs).

LCLs demonstrate widespread hypomethylation relative to activated B cells

To confirm widespread hypomethylation found previously by Hansen et al.(11), we performed tiling analysis using 1kb tiles across the genome as DNA methylation has previously been found to

114

CHAPTER FOUR EBV HYPOMETHYLATION IN MS RISK GENES

correlate across this scale (20, 21). We found 45.9% of tiles to be hypomethylated in LCLs and 0.07% to be hypermethylated in LCLs. This was consistent with a left‐shift in the distribution of methylation values in LCLs (see Figure 1A).

Differentially methylated regions are underrepresented at immune and B cell loci

There was a profound underrepresentation of differentially methylated regions when comparing

MSGWAS and LCLeQTL loci with tiles generated from GWAS Catalog SNPs and genome wide 1kb tiles

(Table 1). To determine whether the observed number of DMRs were significantly different from chance, we randomly sampled 34 and 199 tiles from the GWAS Catalog loci 100 times and found the minimum number of DMRs to be 7 (20.59%) and 63 (31.66%) respectively, suggesting the results obtained would occur less than 1 in 100 times by chance alone. Hansen et al (11) indicated previously that genes within hypomethylated blocks tended to be hypervariable with regard to expression. Using the same normalised LCL expression data (22), we found nine genes associated with differentially methylated LCLeQTL loci, three of which had expression level data. Of these, only one gene demonstrated hypervariable expression per Hansen et al (11).

To determine the source of this differential methylation, we compared methylation values between

LCLeQTL, MSGWAS and GWAS catalog lists for LCLs only or CD40L activated B cells only. We found only CD40L activated B cell methylation to be significantly different when comparing immune/B cell loci with the GWAS catalog list (Table 2 & Figure 1B), suggesting that the lack of DMRs is due to the lower methylation state LCLeQTL and MSGWAS loci in CD40L activated B cells rather than LCLs.

To confirm that immune loci were specifically underrepresented in EBV demethylation, we sought to determine the number of DMRs using 1kb tiles centred on loci most highly correlated with whole

115

CHAPTER FOUR EBV HYPOMETHYLATION IN MS RISK GENES

blood gene expression (nominal p‐value <10‐20) derived from the Genotype Tissue Expression (GTEx)

Project (23). We found 15.17% of the 73771 tiles with data to be differentially methylated. We repeated the analysis with 1kb tiles centred upon loci most highly correlated with LCL gene expression (nominal p‐value <10‐20) from the GTEx project. In this instance, 9477 tiles contained data, with 15.02% of these demonstrating differential methylation.

We next sought to determine whether genes associated with DMRs across the genome were associated with biological gene ontology processes. Using a statistical overrepresentation test with default settings (24) (annotation version 14.1, released March 12 2019), we found genes proximal to

DMRs demonstrated an underrepresentation of B cell and immune activation ontologies (Table 3), consistent with the underrepresentation of DMRs amongst immune and B cell loci (see

Supplementary Information 3). The only biological process overrepresented relative to background was multicellular organismal process (GO:0032501), which was not specific to immune or B cell processes.

The underrepresentation of DMRs also occurs at MHC regions

HLA haplotypes have previously been shown to have the largest genetic contribution to MS risk. In addition, differential methylation at the MHC region in CD4+ T cells has been found to be associated with MS (25, 26). We therefore interrogated the MHC region in LCLs and CD40L activated B cells by examining methylation in 1kb windows centred upon SNPs in LD with MS HLA risk alleles (27). We found DMRs to be underrepresented, with 3/17 (17.65%) windows hypomethylated (see

Supplementary Information 4).

116

CHAPTER FOUR EBV HYPOMETHYLATION IN MS RISK GENES

The underrepresentation of DMRs is not due specifically to B cell activation

To determine whether the underrepresentation of immune and B cell loci amongst DMRs was due to the activation state of both LCLs and CD40L activated B cells, we compared methylation between

LCLs and resting B cells from the same individuals. Once again, there was a statistically significant underrepresentation of DMRs amongst LCLeQTL (p<0.01; minimum from random sampling of GWAS

Catalog – 8 DMRs) and MSGWAS loci (p<0.01; minimum from random sampling of GWAS Catalog –

103 DMRs) (Table 4). To determine whether the DMRs occurred at similar regions in both LCL vs

CD40L activated B cells (LvA) and LCL vs resting B cells (LvR), we determined the proportion of overlapping DMRs between the comparisons on a genome wide basis. Almost all LvA DMRs (93.31%) overlapped with LvR DMRs, whereas 85.84% of LvR DMRs overlapped with LvA DMRs (Figure 2). We also performed a statistical overrepresentation test on LvR DMRs across genome wide 1kb tiles.

There was a large overlap between GO terms belonging to both LvR and LvA lists (see Supplementary

Information 5). Overall, this suggests that the underrepresentation of DMRs amongst immune and B cell loci between LCLs and CD40L activated B cells is not due primarily to the activation status of these subsets.

The underrepresentation of DMRs is not due to differential distribution of B cell/immune cell loci in specific gene annotations

The underrepresentation of DMRs may be due to the differential distribution of these regions amongst various genomic annotations. For example, if specific gene annotations are associated with lower or higher methylation states, this may bias the likelihood of DMRs if loci predominantly overlap with these annotations. Thus, we sought to determine the overlap of loci in each list with genomic annotations, comprising promoters, exons and introns and intergenic regions, not otherwise covered by other annotations (Figure 3). There was no difference in distribution of loci between LCLeQTL and GWAS Catalog lists by Fisher’s exact test (p=0.14), however the distribution

117

CHAPTER FOUR EBV HYPOMETHYLATION IN MS RISK GENES

was significantly different between MSGWAS and GWAS Catalog lists (p=5.90x10‐8, χ2=36.49, df=3).

Finally, a comparison of the midpoint of whole genome 1kb tiles and the GWAS Catalog list showed a significant difference between observed and expected numbers of loci corresponding to each annotation (p<2.20x10‐16, χ2=1651.20, df=3).

Given the differences in expected proportions between MSGWAS, whole genome 1kb tiles and

GWAS Catalog loci, we examined the effect of EBV infection on DNA methylation for particular gene annotations. Interestingly, promoter methylation was minimally affected by EBV infection, whereas other gene annotations demonstrated marked hypomethylation due to EBV. Despite this, the relatively small proportion of loci overlapping with promoter regions would not account for the paucity of DMRs in immune and B cell loci. The difference between the proportion of promoters amongst LCLeQTL/MSGWAS loci versus GWAS Catalog loci was approximately 6%. Even assuming that these 6% were DMRs, the probability of obtaining 7 and 46 DMRs for LCLeQTL and MSGWAS loci respectively would be 0.01 and <0.01 respectively.

Promoter regions are not hypomethylated in LCLs

Methylation of promoter regions is thought to be closely related to gene expression. Interestingly, despite large areas of hypomethylation in LCLs, promoter regions only appeared to be minimally affected by EBV transformation (see Figure 3B). Of 13983 promoters with data, only 99 were differentially methylated (0.71%). A statistical overrepresentation test (24) did not show any statistically significant gene ontology associations.

118

CHAPTER FOUR EBV HYPOMETHYLATION IN MS RISK GENES

Gene expression correlates weakly with differential methylation in an annotation specific manner

Having found that the majority of differentially methylated tiles overlap between LvR and LvA conditions, we compared the expression of resting B cells and LCLs from a separate RNA‐seq dataset, to determine the extent to which differential expression occurs and to what extent it is correlated with differential methylation. This was done by comparing log transformed relative expression as a function of methylation difference between resting B cells and LCLs by gene annotation i.e. exon, intron and promoter regions (Figure 4). Of 14740 genes with expression level data, 8515 (57.77%) demonstrated at least single log‐fold changes in expression level (LCL > resting B cells – 2731, LCL < resting B cells 5784). In comparing expression and differential methylation, for exons and introns, points tended to cluster at x>0, consistent with the effect of EBV infection on DNA methylation. For promoters, points clustered around x=0, as expected due to the relatively small effect of EBV on

DNA methylation at promoters. The correlation between methylation difference and expression was weakly positive for exons and introns and was not present at all for promoters. We also determined the number of differentially methylated regions corresponding to high (>single log‐fold) or low or absent (< single log‐fold) expression differences between subsets (Table 5). Regions with high levels of differential expression were more likely to be differentially methylated at the 0.2 level than those with low or no expression differences between subsets (χ2 test; exons p<0.0001; introns p<0.0001; promoters p=0.0391).

DISCUSSION

This study interrogated whole genome bisulfite sequencing methylomes to determine whether EBV mediated dysregulation of DNA methylation occurs at MS risk loci. Consistent with previous studies, we found widespread dysregulation across the entire genome, with large areas of hypomethylation in LCLs relative to other subsets. Given the persistence and rapid proliferation of LCLs, we expected

B cell/immune activation loci to demonstrate differential methylation to a greater extent than other

119

CHAPTER FOUR EBV HYPOMETHYLATION IN MS RISK GENES

regions. Surprisingly, we found the opposite to be true, that hypomethylation in LCLs at B cell/immune loci occur much less frequently than other regions of the methylome. Further analysis showed that this was due in part to constitutive B cell utilisation of these regions, with relative hypomethylation in CD40L activated B cells compared to other regions. This effect was independent of B cell activation state and distribution of loci amongst gene annotations.

Given that EBV hypomethylation affects 2.18 Gb and including one‐third of genes, we would expect by chance that >65 MS risk genes would be in the hypomethylated regions, but only 35 were. Also, of the 47 MS risk SNPs we had earlier identified as associated with gene expression in LCL (5), only 6 were in a hypomethylated region. Many of these were not coding genes, and the others did not have obvious functions related to EBV infection, compared to the other risk genes (Supplementary

Information 6). In addition, a minority of SNPs associated with MHC related MS risk alleles were in hypomethylated regions. This may indicate that the hypomethylation trait is more related to the generation of tumours and immortalization than to the effect of EBV on autoimmune risk. Hansen et al (11) had established that hypomethylated blocks, and the genes contained within them, overwhelmingly correspond to those seen in cancer, with an overlap of 1.72 GB. In a study of over

2000 tissues, most from tumours, Perez et al (28) concluded that hypomethylation was shared across tumour types, independent of tumour tissue origin, and distinct from the hypomethylation associated with aging. Although many of the genes required for methylation are encoded in heterochromatic regions which are hypomethylated in EBV, suggesting their reduced expression may reduce effective methylation on proliferation, the hypomethylation in tumour and LCLs has little overlap with proliferation (11, 12). The shared hypomethylation between tumours and LCLs appears to be related to chromatin modification, especially due the inhibiting histone modification

H3K9me3 (28) . Hansen et al (11) have suggested the hypomethylation might favour immortalization

120

CHAPTER FOUR EBV HYPOMETHYLATION IN MS RISK GENES

by generating hypervariable gene expression, enabling selection of cell lines avoiding immune detection and those with apoptosis avoiding transcriptional programs.

A limitation of this study is that we cannot know which MS risk loci, if any, are risk loci because they affect host response to EBV. We also cannot be sure if hypomethylation is different in LCLs from people with MS, especially given that different methylation has been detected in some regions of the uninfected immune cell genome in MS compared to controls (25, 26, 29). A larger sample size of individuals may reveal more diversity in differentially methylated regions. Further, the interaction between methylation and transcription is complex, and risk SNPs may interact with the methylation machinery independently of co‐location with differentially methylated regions.

Although our data does not indicate a link between MS risk genes and EBV hypomethylation, it is notable that others have suggested that the EBV hypomethylation is associated with immortalisation

(see above). EBV immortalisation of forbidden clones targeting myelin, especially through lymph node processes, has been considered a likely mechanism driving the association of EBV with MS (30).

Genetic and environmental factors controlling this immortalisation may drive difference in risk between individuals. The genetic component could be due to germline variation or somatic variation. Future work targeting variation between individuals and the process of EBV hypomethylation may identify genetic variation important in the contribution of EBV infection to MS pathogenesis.

Acknowledgements

LO was supported by a National Health and Medical Research Council (NHMRC), Trish MS

Foundation and MS Research Australia co‐funded postgraduate scholarship. GP was supported by a

121

CHAPTER FOUR EBV HYPOMETHYLATION IN MS RISK GENES

MS Research Australia Postdoctoral Fellowship and a MS Research Australia/ JDRF Australia/

Macquarie Group Foundation Postdoctoral Fellowship. DB was supported by a NHMRC Senior

Research Fellowship and MSRA Project Grant. Bioinformatic analysis was supported by Sydney

Informatics Hub, funded by the University of Sydney.

Competing interests

No competing interests to declare.

122

CHAPTER FOUR EBV HYPOMETHYLATION IN MS RISK GENES

REFERENCES

1. Young LS, Yap LF, Murray PG. Epstein‐Barr virus: more than 50 years old and still providing surprises. Nat Rev Cancer. 2016;16(12):789‐802. 2. Levin LI, Munger KL, O'reilly EJ, Falk KI, Ascherio A. Primary infection with the epstein‐barr virus and risk of multiple sclerosis. Annals of neurology. 2010;67(6):824‐30. 3. International Multiple Sclerosis Genetics Consortium. Electronic address ccye, International Multiple Sclerosis Genetics C. Low‐Frequency and Rare‐Coding Variation Contributes to Multiple Sclerosis Risk. Cell. 2018;175(6):1679‐87 e7. 4. James T, Linden M, Morikawa H, Fernandes SJ, Ruhrmann S, Huss M, et al. Impact of genetic risk loci for multiple sclerosis on expression of proximal genes in patients. Hum Mol Genet. 2018;27(5):912‐28. 5. Afrasiabi A, Parnell GP, Fewings N, Schibeci SD, Basuki MA, Chandramohan R, et al. Evidence from genome wide association studies implicates reduced control of Epstein‐Barr virus infection in multiple sclerosis susceptibility. Genome Med. 2019;11(1):26. 6. Li YY, Chung GT, Lui VW, To KF, Ma BB, Chow C, et al. Exome and genome sequencing of nasopharynx cancer identifies NF‐kappaB pathway activating mutations. Nat Commun. 2017;8:14121. 7. Taniguchi I, Iwaya C, Ohnaka K, Shibata H, Yamamoto K. Genome‐wide DNA methylation analysis reveals hypomethylation in the low‐CpG promoter regions in lymphoblastoid cell lines. Human genomics. 2017;11(1):8. 8. Sun YV, Turner ST, Smith JA, Hammond PI, Lazarus A, Van De Rostyne JL, et al. Comparison of the DNA methylation profiles of human peripheral blood cells and transformed B‐lymphocytes. Human genetics. 2010;127(6):651‐8. 9. Brennan EP, Ehrich M, Brazil DP, Crean JK, Murphy M, Sadlier DM, et al. Comparative analysis of DNA methylation profiles in peripheral blood leukocytes versus lymphoblastoid cell lines. Epigenetics. 2009;4(3):159‐64. 10. Leonard S, Wei W, Anderton J, Vockerodt M, Rowe M, Murray PG, et al. Epigenetic and Transcriptional Changes Which Follow Epstein‐Barr Virus Infection of Germinal Center B Cells and Their Relevance to the Pathogenesis of Hodgkin's Lymphoma. Journal of Virology. 2011;85(18):9568‐ 77. 11. Hansen KD, Sabunciyan S, Langmead B, Nagy N, Curley R, Klein G, et al. Large‐scale hypomethylated blocks associated with Epstein‐Barr virus‐induced B‐cell immortalization. Genome research. 2013:gr. 157743.113. 12. Hernando H, Shannon‐Lowe C, Islam AB, Al‐Shahrour F, Rodríguez‐Ubreva J, Rodríguez‐ Cortez VC, et al. The B cell transcription program mediates hypomethylation and overexpression of key genes in Epstein‐Barr virus‐associated proliferative conversion. Genome biology. 2013;14(1):R3‐ R. 13. Andrews S. FastQC: a quality control tool for high throughput sequence data. 2010. 14. Bolger AM, Lohse M, Usadel B. Trimmomatic: a flexible trimmer for Illumina sequence data. Bioinformatics. 2014;30(15):2114‐20. 15. Song Q, Decato B, Hong EE, Zhou M, Fang F, Qu J, et al. A reference methylome database and analysis pipeline to facilitate integrative and comparative epigenomics. PLoS One. 2013;8(12):e81148. 16. Patsopoulos N, Baranzini SE, Santaniello A, Shoostari P, Cotsapas C, Wong G, et al. The Multiple Sclerosis Genomic Map: Role of peripheral immune cells and resident microglia in susceptibility. BioRxiv. 2017:143933. 17. MacArthur J, Bowler E, Cerezo M, Gil L, Hall P, Hastings E, et al. The new NHGRI‐EBI Catalog of published genome‐wide association studies (GWAS Catalog). Nucleic acids research. 2016;45(D1):D896‐D901.

123

CHAPTER FOUR EBV HYPOMETHYLATION IN MS RISK GENES

18. Zerbino DR, Wilder SP, Johnson N, Juettemann T, Flicek PR. The Ensembl Regulatory Build. Genome Biology. 2015;16(1):56. 19. Quinlan AR. BEDTools: the Swiss‐army tool for genome feature analysis. Current protocols in bioinformatics. 2014;47(1):11.2. 1‐.2. 34. 20. Eckhardt F, Lewin J, Cortese R, Rakyan VK, Attwood J, Burger M, et al. DNA methylation profiling of human chromosomes 6, 20 and 22. Nature genetics. 2006;38(12):1378‐85. 21. Ball MP, Li JB, Gao Y, Lee J‐H, LeProust EM, Park I‐H, et al. Targeted and genome‐scale strategies reveal gene‐body methylation signatures in human cells. Nature biotechnology. 2009;27(4):361‐8. 22. Choy E, Yelensky R, Bonakdar S, Plenge RM, Saxena R, De Jager PL, et al. Genetic analysis of human traits in vitro: drug response and gene expression in lymphoblastoid cell lines. PLoS genetics. 2008;4(11):e1000287. 23. Lonsdale J, Thomas J, Salvatore M, Phillips R, Lo E, Shad S, et al. The genotype‐tissue expression (GTEx) project. Nature genetics. 2013;45(6):580. 24. Muruganujan A, Mills C, Kang D, Tang H, Huang X, Mi H, et al. PANTHER version 11: expanded annotation data from Gene Ontology and Reactome pathways, and data analysis tool enhancements. Nucleic Acids Research. 2016;45(D1):D183‐D9. 25. Maltby VE, Lea RA, Sanders KA, White N, Benton MC, Scott RJ, et al. Differential methylation at MHC in CD4+ T cells is associated with multiple sclerosis independently of HLA‐DRB1. Clinical epigenetics. 2017;9(1):71. 26. Kular L, Liu Y, Ruhrmann S, Zheleznyakova G, Marabita F, Gomez‐Cabrero D, et al. DNA methylation as a mediator of HLA‐DRB1* 15: 01 and a protective variant in multiple sclerosis. Nature communications. 2018;9(1):2397. 27. Moutsianas L, Jostins L, Beecham AH, Dilthey AT, Xifara DK, Ban M, et al. Class II HLA interactions modulate genetic risk for multiple sclerosis. Nature genetics. 2015;47(10):1107. 28. Perez RF, Tejedor JR, Bayon GF, Fernandez AF, Fraga MF. Distinct chromatin signatures of DNA hypomethylation in aging and cancer. Aging Cell. 2018;17(3):e12744. 29. Ewing E, Kular L, Fernandes SJ, Karathanasis N, Lagani V, Ruhrmann S, et al. Combining evidence from four immune cell types identifies DNA methylation patterns that implicate functionally distinct pathways during Multiple Sclerosis progression. EBioMedicine. 2019. 30. Pender MP. The essential role of Epstein‐Barr virus in the pathogenesis of multiple sclerosis. Neuroscientist. 2011;17(4):351‐67.

124

CHAPTER FOUR EBV HYPOMETHYLATION IN MS RISK GENES

TABLES

Table 1. Differentially methylated regions obtained when comparing LCLs to CD40L activated B cells

Number of 1kb Tiles With DMRs as % of

List Loci/SNPs Data DMRs Tiles With Data

LCLeQTL 35 34 6 17.65

MSGWAS 201 199 34 17.09

GWAS Catalog 51899 51151 20191 39.47

Genome Wide 2881045 2648736 1216588 45.93

Table 2. Summary statistics of 1kb tile methylation

LCLeQTL MSGWAS GWAS Catalog

Activated LCL Activated LCL Activated LCL

Maximum 1.00 1.00 1.00 1.00 1.00 1.00

Minimum 0.02 0.02 0.01 0.01 0.00 0.00

Median 0.80 0.68 0.85 0.74 0.88 0.68

First Quartile 0.42 0.27 0.61 0.40 0.79 0.52

Third Quartile 0.93 0.89 0.93 0.86 0.93 0.83

IQR 0.51 0.62 0.31 0.46 0.14 0.31

125

CHAPTER FOUR EBV HYPOMETHYLATION IN MS RISK GENES

Table 3. Top 10 biological processes associated with genes proximal to genome wide DMRs (for full listing see Supplementary Information 3)

Fold

Biological Process Enrichment Raw P‐Value FDR

Positive regulation of lymphocyte activation

(GO:0051251) 0.06 7.40E‐17 1.33E‐13

B cell receptor signaling pathway

(GO:0050853) 0.13 1.70E‐14 1.53E‐11

Phagocytosis (GO:0006909) 0.29 5.61E‐10 3.36E‐07

Regulation of lymphocyte activation

(GO:0051249) 0.29 8.67E‐10 3.89E‐07

Regulation of leukocyte activation

(GO:0002694) 0.31 2.71E‐09 9.72E‐07

Antigen receptor‐mediated signaling pathway

(GO:0050851) 0.34 6.22E‐09 1.86E‐06

Regulation of cell activation (GO:0050865) 0.32 7.86E‐09 2.02E‐06

Immune response (GO:0006955) 0.58 1.71E‐08 3.84E‐06

B cell activation (GO:0042113) 0.35 3.01E‐08 6.00E‐06

Defense response to bacterium (GO:0042742) 0.38 8.60E‐08 1.54E‐05

126

CHAPTER FOUR EBV HYPOMETHYLATION IN MS RISK GENES

Table 4. Differentially methylated regions obtained when comparing LCLs to resting B cells

Number of 1kb Tiles With DMRs As % of List DMRs Loci/SNPs Data Tiles With Data

LCLeQTL 35 34 7 20.59

MSGWAS 201 199 44 22.11

GWAS Catalog 51899 51144 22250 43.50

Genome Wide 2881045 2648736 1321140 49.93

Table 5. Rates of differential expression by expression level and annotation

Exons Introns Promoters

< One log‐fold Difference in Expression 61482 64846 4900

% Differentially Methylated 7.16% 6.74% 0.67%

> One log‐fold Difference in Expression 78155 82312 5318

% Differentially Methylated 14.51% 17.80% 1.05%

127

CHAPTER FOUR EBV HYPOMETHYLATION IN MS RISK GENES

FIGURES

128

CHAPTER FOUR EBV HYPOMETHYLATION IN MS RISK GENES

Figure 1. A) distribution of methylation values of 1kb tiles across the CD40L activated B cell and LCL methylome. B) Boxplot of DNA methylation by loci list and subset. Only CD40L activated B cell tiles demonstrated significant differences when compared to regions centred on GWAS Catalog SNPS, suggesting that LCLeQTL and MSGWAS loci are less methylated even prior to EBV transformation.

129

CHAPTER FOUR EBV HYPOMETHYLATION IN MS RISK GENES

Figure 2. Overlap of genome wide DMRs for LvA and LvR comparisons based on 1kb tiles

130

CHAPTER FOUR EBV HYPOMETHYLATION IN MS RISK GENES

Figure 3. A) Breakdown of loci overlapping with gene annotations and B) violin plots of DNA methylation by gene annotation and cell subset.

131

CHAPTER FOUR EBV HYPOMETHYLATION IN MS RISK GENES

Figure 4. Scatter plots comparing gene expression as a function of DNA methylation differences between LCLs and resting B cells.

132

CHAPTER FOUR EBV HYPOMETHYLATION IN MS RISK GENES

CO‐AUTHOR CONTRIBUTIONS

LO has full access to the data in this study and takes responsibility for the integrity of the data and accuracy of the data analysis. LO, DB and SS conceived and designed the experiments based on extant DNA methylation data. LO performed the data analysis. Data analysis was reviewed by GP.

AA, GS, GP, SS and DB contributed to the revision of the manuscript. All authors read and approved the manuscript.

Grant Parnell ______

Ali Afrasiabi ______

Graeme Stewart ______

Sanjay Swaminathan ______

David Booth ______

133

CHAPTER FIVE

SUMMARY AND CONCLUSIONS

CHAPTER FIVE SUMMARY AND CONCLUSIONS

5.1 SUMMARY OF FINDINGS

5.1.1 Recapitulation of haematopoietic progenitor DNA methylation in progeny cells

The presence of a critical period in MS risk suggests the presence of a mechanism capable of transmitting “memory” for early life exposures. If DNA methylation is a mechanism for transmitting this risk, one would expect the presence of a DNA methylation signature that is present in adult haematopoietic progenitors to be recapitulated in progeny immune cells. In addition, for DNA methylation related risk to occur, one would also expect individual variation at these recapitulated sites. Thus, it was hypothesised that DNA methylation setting in CD34+ haematopoietic progenitors would be recapitulated in CD14+ monocytes and CD56+ NK cells. Secondly, it was hypothesised that individual differences in DNA methylation would be detectable amongst these subsets.

The study presented in Chapter Two showed that DNA methylation at CpG islands was almost completely recapitulated using a DNA methylation proportion cut‐off of 0.25, with less than 1.5% of all interrogated CGIs being differentially methylated at this level between any of the three cell subsets. Many of these differentially methylated sites were likely involved in the differentiation of cell subsets or cell subset specific function. There was little difference in DNA methylation between individuals at these sites, with greatest variability occurring at regions displaying intermediate mean methylation states. Despite this, HLA‐DRB1, the MS risk gene with by far the highest odds ratio for risk, was identified as being one of the genes demonstrating interindividual variability in DNA methylation. In addition, the relative DNA methylation rank between individuals was recapitulated at HLA‐DRB1 from CD34+ to CD14+ cells.

135

CHAPTER FIVE SUMMARY AND CONCLUSIONS

5.1.1.1 Significance of these findings

At the interrogated CpG islands, DNA methylation was recapitulated between progenitor and progeny cells. This is consistent with the idea that a DNA methylation signature (potentially influenced by early environmental exposures) is transmitted from haematopoietic progenitors to progeny cells. That there were sites that were differentially methylated between individuals, is also consistent with the hypothesis that varying environmental exposures may perhaps be linked to differences in DNA methylation and thus be a substrate for disease risk as in the case of MS.

5.1.1.2 Study limitations

A relatively high threshold was used to call regions that were differentially methylated. A lower threshold would have resulted in the identification of a greater number of differentially methylated regions. However, the significance of smaller differences is unclear with regards to effects on gene expression or phenotypic change. Studying a larger number of samples would have provided more statistical power to detect differences between cell subsets and between individuals, if present.

Reduced representation bisulfite sequencing provided high coverage of limited regions of the genome, particularly at CpG rich sites such as CpG islands and promoter regions. However, data on other parts of the genome that may have demonstrated more numerous differences in DNA methylation were limited due to the restriction enzyme and fragment size selection strategy used.

In addition, only control subjects were studied and dysregulation of the methylome associated with disease was not ascertained. This could have provided further insight as to the regions of the methylome where a DNA methylation setting might be found. Other immune cell subsets contributing to MS pathogenesis, including through the effect of vitamin D, may have different methylation patterns to those examined. Finally, because adult subjects were only examined at one

136

CHAPTER FIVE SUMMARY AND CONCLUSIONS

point in time, it remains to be shown that the DNA methylation signature set by early life environmental exposures is maintained throughout the lifespan.

5.1.2 Effects of calcitriol on DNA methylation in myeloid cells

Because DNA methylation is relatively stable over time (1), and susceptible to environmental influences (2), the effect of vitamin D on DNA methylation was investigated as a possible mechanism mediating latitude dependent autoimmune disease risk, such as in MS. Given that the genomic effects of vitamin D are mediated through the vitamin D receptor, the difference in DNA methylation at VDR binding sites was also investigated to determine whether methylation at these regions might mediate age‐dependent effects of vitamin D in autoimmune disease risk. A further motivation for examining the effect of vitamin D on DNA methylation was the heterogeneity of results due to variability in experimental factors such as cell type/heterogeneity, type of vitamin D supplemented/measured and methylation assays utilised.

The study reported in Chapter Three found that calcitriol had very limited, but different effects on

DNA methylation in adult and paediatric myeloid cells. This occurred despite several genes being differentially expressed due to calcitriol. Interestingly, the differentially expressed genes varied greatly between cells of adult and paediatric origin with only 75 differentially expressed genes overlapping between groups (337 adult and 325 paediatric genes differentially expressed in total).

Despite this, 52% of myeloid VDR peaks (DC1, DC2, CD14+ monocytes) (3) were differentially methylated between adult and paediatric cells, with the majority of these overlapping with regulatory regions such as gene promoters, promoter flanking regions and CTCF binding sites. The latter suggests their importance in transcriptional regulation, and the former that vitamin D regulation of genes in these cells is age‐dependent. Furthermore, genes proximal to 90 differentially

137

CHAPTER FIVE SUMMARY AND CONCLUSIONS

methylated VDR peaks were also MS risk genes, indicating a possible role of this age‐dependent difference in DNA methylation contributing to the effect of vitamin D on MS pathogenesis.

Consistent with this, differentially methylated VDR peaks were also associated with pathways associated with regulation of the adaptive immune response and intracellular signalling pathways associated with myeloid cell differentiation.

5.1.2.1 Significance of these findings

The study found that the myeloid cell methylome was relatively insensitive to the effects of calcitriol.

This was consistent with a previous in vitro study of calcitriol on mononuclear cells (4). However, the age‐dependent differences at VDR peaks was an unexpected finding. As discussed in Chapter 1, DNA methylation at a transcription factor binding site can inhibit transcription factor binding or lead to recruitment of MBD and chromatin remodellers to alter transcription factor binding. Therefore, differences in DNA methylation at VDR peaks might point to an increased susceptibility of paediatric myeloid cells to the effects of vitamin D. A very high proportion of MS risk genes were proximal to these differentially methylated peaks. Further, because of the associated enrichment of VDR peak genes with adaptive immune response and myeloid cell differentiation pathways, exposure to vitamin D in early life likely contributes to development along these pathways, toward a more or less tolerogenic phenotype. This is a plausible candidate mechanism for the critical period dependent effects of the latitude gradient on MS and autoimmune disease risk. It may also explain why vitamin

D supplementation in MS patients has not provided evidence of clear benefit to date.

5.1.2.2 Study limitations

Whole genome bisulfite sequencing on four subjects allowed the methylomes of these subjects to be interrogated at significant depth, increasing power to detect differences between subjects. The

138

CHAPTER FIVE SUMMARY AND CONCLUSIONS

differential methylation at VDR peaks need to be confirmed amongst a larger number of biological replicates, especially in the context of large interindividual differences in DNA methylation. This could perhaps be achieved using more targeted sequencing approaches such as pyrosequencing, or targeted genome wide approaches specific for VDR peaks. The study was also conducted on ex vivo cultured cells, and validation would also be important in primary CD14+ monocytes.

The effects of vitamin D on DNA methylation may not have been detected for a number of reasons.

Firstly, the culture occurred over three weeks and it is possible that any DNA methylation changes may have occurred earlier in the culture period. Cultured cells have also been shown to increase their DNA methylation with increasing passage number (5), potentially distorting or obscuring the effects of vitamin D or age on DNA methylation. The timing of addition of calcitriol, and its concentration may also be important in mediating differences in methylation between adult and paediatric cells.

5.1.3 DNA methylation of LCLs at MS risk genes

The almost complete seropositivity for EBV seen in MS‐affected individuals (6) suggests that EBV is necessary, but not sufficient, for the development of MS. Although there are several possible mechanisms for EBV mediated immune dysregulation in the pathogenesis of MS, the age‐dependent effects of EBV mediated MS risk (as in infectious mononucleosis), also make DNA methylation a candidate for mediating this risk. EBV is also known to cause widespread disruption of DNA methylation with DNA hypomethylation noted at over two‐thirds of the entire LCL genome (7).

Therefore, it was hypothesised that dysregulation of DNA methylation would occur preferentially at

MS risk loci, given that these regions are likely to be key mediators of risk.

139

CHAPTER FIVE SUMMARY AND CONCLUSIONS

The study reported in Chapter Four found that dysregulation of the methylome was less likely to occur at MS risk and B cell loci compared with other regions of the LCL genome, perhaps due to the decreased methylation of these loci in activated B cells compared with an unbiased list of SNPs

(GWAS Catalog). Gene ontology analysis showed that genes proximal to genome wide differentially methylated regions were underrepresented for pathways related to lymphocyte and B cell activation as well as immune cell functions. It was also found that promoter regions (whose methylation is most correlated with gene expression), was less likely to be altered compared to other regions e.g. intergenic, exonic, intronic. Despite this and overrepresentation of promoter regions amongst MS risk and an unbiased list of SNPs, the number of DMRs in the worst case remained less than that due to chance alone. Overall, these results suggested that the relative lack of hypomethylation at immune and B cell loci was consistent with constitutive utilisation of these regions by B cells/LCLs.

5.1.3.1 Significance of these findings

The results of this study argue against a specific role of DNA hypomethylation at MS risk loci contributing to MS risk. This does not rule out a role for widespread EBV mediated hypomethylation in MS risk however. Widespread DNA hypomethylation for example, is also seen in various human malignancies (8, 9), and it is possible that this contributes to viral immortalisation in EBV infected B cells in vivo. This is particularly relevant if immortalised clones are autoreactive and capable of CNS targeting.

5.1.3.2 Study limitations

The study presented in Chapter Four only examined normal individuals and it is unclear whether the effects found in this study will be replicated in those with MS, especially when baseline immune cell methylation may already differ between MS and controls (10‐12). In addition, the study was only

140

CHAPTER FIVE SUMMARY AND CONCLUSIONS

based on three healthy controls and a larger cohort should be examined to provide confirmation of these results.

5.2 FUTURE DIRECTIONS

The study from Chapter Two found stable DNA methylation at CGIs between subsets, but future studies should interrogate areas of the methylome not included in this analysis. To further substantiate the findings from the chapter, other cell types should also be included in subsequent investigations. As mentioned earlier, the longitudinal stability of recapitulated regions would also need to be established for these regions to be truly candidates for transmitting the effects of early life influences.

Although the study from Chapter Three failed to find anticipated changes in DNA methylation with calcitriol exposure, factors such as duration of treatment need to be explored further, especially in light of the known epigenetic changes that typically occur within 24 hours of treatment (13).Time course experiments would be useful in helping to elucidate this effect, although the use of CD34+ progenitors would need to be carefully considered given the lack of CD14+ numbers early on in culture.

Perhaps the best test of functional differences in VDR peak methylation between adult and paediatric cells would be to examine the effects of vitamin D and non‐vitamin D supplemented

CD14+ cells in co‐culture with CD4+ T cells to determine whether the predicted phenotypic effects arise. Given the individual differences in DNA methylation found in Chapter Three, the effect of vitamin D on DNA methylation in other cell subsets (e.g. CD4+ T cells, especially given the changes

141

CHAPTER FIVE SUMMARY AND CONCLUSIONS

seen in animal models (14, 15)) would also be worthwhile interrogating. The phenomenon of differential VDR peak methylation also requires confirmation in primary cells.

With regard to EBV and DNA methylation in MS risk, the significance of widespread DNA demethylation is unclear and would be better studied in the context of other epigenetic modifications. The pattern of LCL methylation in MS patients (both pre and post diagnosis), would help to determine whether differential susceptibility to EBV induced DNA methylation differences might contribute to disease risk.

This thesis has only considered two of the many environmental risk factors that could affect a critical period in the development of MS. It raises questions about DNA methylation and epigenetic alterations associated with other environmental risk factors such as shift work, adolescent obesity and organic solvent exposure. Although it is unlikely that they would affect VDR peaks, an understanding of the DNA methylation changes attributable to these factors might provide details as to common immunological pathways, cell subset specificity and genomic regions where targeted treatments might be employed. In addition, studies on these factors might provide further resolution on the exact timing and characteristics of susceptibility periods to these exposures.

In the long term, insights gained into the field of environmental epigenetics are needed to provide clear strategies for risk mitigation and better targeted measures in the prevention and treatment of

MS. Epigenetic marks have already been exploited as drug targets, such as in cancer, due to widespread dysregulation in malignant cells. For example, the DNA methylation inhibitor 5’‐ azacytidine is used in the treatment of different myeloid neoplasms. HDAC inhibitors (e.g. vorinostat, romidepsin) are used in the treatment of haematological and solid organ malignancies. The use of

142

CHAPTER FIVE SUMMARY AND CONCLUSIONS

these drugs in immune diseases is not as advanced, but shows promise (16). If the spatiotemporal features of key epigenetic susceptibility loci are mapped in MS, therapies targeting DNA methylation might be employed in disease treatment and prevention. Although current therapies act on the epigenome in a relatively untargeted fashion, future treatments capable of methylating or demethylating specific loci may allow artificial regulation of target genes as has already been demonstrated in mice using a d‐Cas9 system (17).

5.3 CONCLUSIONS

The intersection of environment, epigenetics and disease risk as an area of study in multiple sclerosis is in its infancy. This thesis has established the potential for DNA methylation as a mediator environmentally driven risk and generated several potential avenues for exploration. The limits currently imposed by technology and its cost, will likely be rapidly overcome with the ever decreasing cost of next generation sequencing, allowing multi‐omics approaches to provide new and more complete insights into the field.

143

CHAPTER FIVE SUMMARY AND CONCLUSIONS

5.4 REFERENCES

1. Cedar H, Bergman Y. Linking DNA methylation and histone modification: patterns and paradigms. Nature Reviews Genetics. 2009;10(5):295. 2. Heim C, Binder EB. Current research trends in early life stress and depression: Review of human studies on sensitive periods, gene–environment interactions, and epigenetics. Experimental neurology. 2012;233(1):102‐11. 3. Booth D, Ding N, Parnell G, Shahijanian F, Coulter S, Schibeci S, et al. Cistromic and genetic evidence that the vitamin D receptor mediates susceptibility to latitude‐dependent autoimmune diseases. Genes and immunity. 2016;17(4):213. 4. Chavez Valencia RA, Martino DJ, Saffery R, Ellis JA. In vitro exposure of human blood mononuclear cells to active vitamin D does not induce substantial change to DNA methylation on a genome‐scale. The Journal of steroid biochemistry and molecular biology. 2014;141:144‐9. 5. Horvath S. DNA methylation age of human tissues and cell types. Genome biology. 2013;14(10):3156. 6. Levin LI, Munger KL, O'reilly EJ, Falk KI, Ascherio A. Primary infection with the epstein‐barr virus and risk of multiple sclerosis. Annals of neurology. 2010;67(6):824‐30. 7. Hansen KD, Sabunciyan S, Langmead B, Nagy N, Curley R, Klein G, et al. Large‐scale hypomethylated blocks associated with Epstein‐Barr virus‐induced B‐cell immortalization. Genome research. 2013:gr. 157743.113. 8. Hansen KD, Timp W, Bravo HC, Sabunciyan S, Langmead B, McDonald OG, et al. Increased methylation variation in epigenetic domains across cancer types. Nature genetics. 2011;43(8):768. 9. Berman BP, Weisenberger DJ, Aman JF, Hinoue T, Ramjan Z, Liu Y, et al. Regions of focal DNA hypermethylation and long‐range hypomethylation in colorectal cancer coincide with nuclear lamina–associated domains. Nature genetics. 2012;44(1):40. 10. Maltby VE, Lea RA, Sanders KA, White N, Benton MC, Scott RJ, et al. Differential methylation at MHC in CD4+ T cells is associated with multiple sclerosis independently of HLA‐DRB1. Clinical epigenetics. 2017;9(1):71. 11. Kular L, Liu Y, Ruhrmann S, Zheleznyakova G, Marabita F, Gomez‐Cabrero D, et al. DNA methylation as a mediator of HLA‐DRB1* 15: 01 and a protective variant in multiple sclerosis. Nature communications. 2018;9(1):2397. 12. Moutsianas L, Jostins L, Beecham AH, Dilthey AT, Xifara DK, Ban M, et al. Class II HLA interactions modulate genetic risk for multiple sclerosis. Nature genetics. 2015;47(10):1107. 13. Seuter S, Neme A, Carlberg C. Epigenome‐wide effects of vitamin D and their impact on the transcriptome of human monocytes involve CTCF. Nucleic acids research. 2016;44(9):4090‐104. 14. Zeitelhofer M, Adzemovic MZ, Gomez‐Cabrero D, Bergman P, Hochmeister S, N'diaye M, et al. Functional genomics analysis of vitamin D effects on CD4+ T cells in vivo in experimental autoimmune encephalomyelitis. Proceedings of the National Academy of Sciences. 2017;114(9):E1678‐E87. 15. Moore JR, Hubler SL, Nelson CD, Nashold FE, Spanier JA, Hayes CE. 1, 25‐Dihydroxyvitamin D3 increases the methionine cycle, CD4+ T cell DNA methylation and Helios+ Foxp3+ T regulatory cells to reverse autoimmune neurodegenerative disease. Journal of neuroimmunology. 2018;324:100‐14. 16. Hull EE, Montgomery MR, Leyva KJ. HDAC inhibitors as epigenetic regulators of the immune system: impacts on cancer therapy and inflammatory diseases. BioMed research international. 2016;2016. 17. Liu XS, Wu H, Ji X, Stelzer Y, Wu X, Czauderna S, et al. Editing DNA Methylation in the Mammalian Genome. Cell. 2016;167(1):233‐47.e17.

144

APPENDICES

APPENDIX ONE

SUPPLEMENTARY MATERIAL FOR CHAPTER TWO

APPENDIX ONE SUPPLEMENTARY MATERIAL FOR CHAPTER TWO

Supplemental Table 1. Differentially methylated CGI, closest gene and differential expression for CD34+ vs CD14+ subsets

Differentially Chromosome Start End Closest Gene CGI Overlap Class Expressed (>2 fold) chr12 125599306 125599625 AACS intragenic chr7 48494361 48494753 ABCA13 intragenic chr16 89118801 89119152 ACSF3 intergenic chr10 1415976 1416249 ADARB2 intragenic chr10 1585174 1585421 ADARB2 intragenic chr10 1585174 1585421 ADARB2‐AS1 intragenic chr12 5936876 5937085 ANO2 intragenic chr19 17443965 17444633 ANO8 promoter chr19 42412375 42412584 ARHGEF1 intergenic chr19 950020 950370 ARID3A intragenic chr1 1389555 1389889 ATAD3C intragenic chr19 835272 835524 AZU1 intergenic Y chr8 145494447 145494712 BOP1 intragenic chr6 31867691 31867957 C2 promoter chr1 181767493 181767893 CACNA1E intragenic chr1 19665010 19665276 CAPZB intragenic chr20 32232170 32232403 CBFA2T2 intragenic chr1 1643593 1643811 CDK11A intragenic chr1 1643593 1643811 CDK11B intragenic chr3 48677529 48677877 CELSR3 intragenic chr4 2431798 2432328 CFAP99 intragenic chr11 890191 890506 CHID1 intragenic Y chr16 1495039 1495366 CLCN7 intragenic chr1 9790292 9790704 CLSTN1 intragenic chr13 111109139 111109652 COL4A2 intragenic chr13 111109139 111109652 COL4A2‐AS2 intragenic chr9 137660513 137660742 COL5A1 intragenic chr19 1947721 1948042 CSNK1G2 intragenic chr18 77552401 77552603 CTDP1 intergenic Y chr12 56329518 56329790 DGKA intragenic chr14 102050814 102051025 DIO3 intergenic chr15 41228442 41228872 DLL4 intragenic chr19 55673460 55673663 DNAAF3 intragenic chr5 13810126 13810333 DNAH5 intragenic chr22 16192654 16193098 DUXAP8 promoter chr19 48244150 48244693 EHD2 intragenic chr8 623372 623710 ERICH1 intragenic Y chr2 239009072 239009361 ESPNL intragenic chr22 21505626 21506073 FAM230B intergenic chr9 41222254 41222736 FAM74A1 intergenic chr12 133022649 133022863 FBRSL1 intergenic chr12 133140304 133140634 FBRSL1 intragenic chr9 79629865 79630084 FOXB2 intergenic chr7 4806860 4807086 FOXK1 intragenic chr1 41889270 41889626 FOXO6 intergenic chr12 130623142 130623492 FZD10‐AS1 intergenic chr4 871276 871630 GAK intragenic chr19 2511392 2511782 GNG7 intragenic chr19 2643287 2643633 GNG7 intragenic chr22 47017355 47017654 GRAMD4 intragenic chr19 48908176 48908597 GRIN2D intragenic chr11 67052394 67053110 GRK2 promoter chr12 96389405 96389675 HAL intragenic chr15 20711403 20711823 HERC2P3 promoter

147

APPENDIX ONE SUPPLEMENTARY MATERIAL FOR CHAPTER TWO

chr16 32199219 32199676 HERC2P4 promoter chr7 139416286 139416522 HIPK2 intragenic chr19 12880574 12880888 HOOK2 intragenic chr7 154862680 154863245 HTR5A intragenic chr7 154862680 154863245 HTR5A‐AS1 intragenic chr7 150901550 150901978 IQCA1L intragenic chr5 3962286 3962512 IRX1 intergenic chr1 949329 949851 ISG15 promoter Y chr9 5042797 5043270 JAK2 intragenic chr18 77585913 77586856 KCNG2 intergenic chr19 5119220 5119445 KDM4B intragenic chr22 45598721 45599080 KIAA0930 intragenic Y chr11 33680219 33680428 KIAA1549L intragenic chr1 245498883 245499218 KIF26B intragenic chr9 110479034 110479420 KLF4 intergenic Y chr10 134971717 134971945 KNDC1 intergenic chr3 42977830 42978090 KRBOX1 promoter chr3 42977830 42978090 KRBOX1‐AS1 promoter chr1 31246010 31246280 LAPTM5 intergenic Y chr18 6920591 6920874 LINC00668 intergenic chr7 158816092 158816409 LINC00689 intragenic chr16 86411154 86411457 LINC00917 intergenic chr12 122235106 122235310 LINC01089 promoter chr2 132152766 132153044 LINC01120 intergenic chr6 72298274 72298528 LINC01626 intergenic chr21 45575451 45575833 LINC01678 intergenic chr20 59542508 59542739 LINC01718 intergenic chr2 496823 497038 LINC01874 intergenic chr3 195564747 195565017 LINC01983 intragenic chr14 101908781 101909022 LINC02314 intragenic chr16 967241 967619 LMF1 intragenic chr7 62809609 62809812 LOC100287704 intragenic chr7 62809609 62809812 LOC100287834 intragenic chr22 21695101 21695548 LOC100996335 intergenic chr2 91911039 91911371 LOC101927050 intergenic chr16 32199219 32199676 LOC102723753 promoter chr7 62809609 62809812 LOC102724738 intragenic chr14 106033298 106033520 LOC105370697 intergenic chr6 168613132 168613346 LOC105378137 intergenic chr17 28903507 28903832 LOC107133515 intragenic chr8 58172076 58173863 LOC286177 intragenic chr1 43353947 43354170 LOC339539 intragenic chr16 32489789 32490078 LOC390705 intergenic chr5 1663730 1664340 LOC728613 intergenic chr19 2323912 2324127 LSM7 intragenic Y chr7 1967664 1967994 MAD1L1 intragenic chr4 1304768 1305114 MAEA intragenic chr17 81100253 81101228 METRNL intergenic Y chr16 4730262 4730487 MGRN1 intragenic chr16 4731561 4731835 MGRN1 intragenic chr16 4732219 4732487 MGRN1 intragenic chr1 17006742 17007151 MIR3675 intergenic chr7 63222279 63222688 MIR4283‐1 intergenic chr17 34093571 34093836 MMP28 intragenic chr17 56355207 56355502 MPO intragenic chr17 56356374 56357013 MPO intragenic chr5 176734599 176734944 MXD3 promoter chr11 78285405 78285995 NARS2 promoter chr1 149149250 149149505 NBPF25P intergenic chr18 77284291 77285544 NFATC1 intragenic

148

APPENDIX ONE SUPPLEMENTARY MATERIAL FOR CHAPTER TWO

chr19 50059574 50060175 NOSIP intragenic chr9 139417115 139417518 NOTCH1 intragenic chr10 135203101 135203316 PAOX intragenic chr21 47288455 47288722 PCBP3 intragenic chr5 140736821 140737058 PCDHGA1 intragenic chr5 140751298 140751883 PCDHGA1 intragenic chr5 140736821 140737058 PCDHGA2 intragenic Y chr5 140751298 140751883 PCDHGA2 intragenic Y chr5 140736821 140737058 PCDHGA3 intragenic chr5 140751298 140751883 PCDHGA3 intragenic chr5 140736821 140737058 PCDHGA4 intragenic Y chr5 140751298 140751883 PCDHGA4 intragenic Y chr5 140751298 140751883 PCDHGA5 intragenic chr5 140736821 140737058 PCDHGB1 intragenic Y chr5 140751298 140751883 PCDHGB1 intragenic Y chr5 140751298 140751883 PCDHGB2 intragenic Y chr5 140751298 140751883 PCDHGB3 intragenic Y chr6 165747763 165747970 PDE10A intragenic chr16 335829 336034 PDIA2 intragenic chr21 45730220 45730444 PFKL intragenic chr1 9775708 9775997 PIK3CD intragenic chr16 15012708 15013276 PKD1P3‐NPIPA1 intragenic chr16 16451347 16451960 PKD1P4‐NPIPA8 intragenic PKD1P5‐ chr16 18487934 18488547 LOC105376752 intragenic chr8 145003589 145004145 PLEC intragenic Y chr8 145018815 145019214 PLEC promoter Y chr19 1526462 1526681 PLK5 intragenic chr19 45885786 45885999 PPP1R13L intragenic chr10 72360255 72360605 PRF1 intragenic chr1 2082314 2082529 PRKCZ intragenic chr19 47181683 47181893 PRKD2 intragenic chr9 132437924 132438156 PRRX2 intragenic Y chr19 836533 836899 PRTN3 intergenic Y chr19 847898 848129 PRTN3 intragenic Y chr20 62168432 62168684 PTK6 promoter chr11 67203319 67203662 PTPRCAP intragenic Y chr7 157503469 157504144 PTPRN2 intragenic chr1 29587087 29587412 PTPRU promoter chr21 46268896 46269263 PTTG1IP intergenic chr19 8464627 8465001 RAB11B intragenic chr2 130750104 130750424 RAB6C intergenic Y chr2 132110849 132111169 RAB6D intergenic chr13 114807632 114807834 RASA3 intragenic chr19 10434024 10434248 RAVER1 intragenic chr20 55982631 55983074 RBM38 intragenic chr2 88125153 88125434 RGPD1 intragenic chr2 88125153 88125434 RGPD2 intragenic chr20 19955536 19956034 RIN2 intragenic chr3 77147167 77147399 ROBO2 intragenic chr17 152117 152438 RPH3AL intragenic chr19 54711106 54711375 RPS9 promoter chr18 76552775 76553014 SALL3 intergenic chr9 136566573 136566896 SARDH intragenic chr19 1155020 1155305 SBNO2 intragenic chr7 83378885 83379153 SEMA3E intergenic chr19 4543407 4544578 SEMA6B intragenic chr6 134497536 134497756 SGK1 promoter chr11 64329520 64329825 SLC22A11 intragenic chr8 145641915 145642118 SLC39A4 intragenic chr17 1478584 1479214 SLC43A2 intragenic

149

APPENDIX ONE SUPPLEMENTARY MATERIAL FOR CHAPTER TWO

chr17 1492086 1492322 SLC43A2 intragenic chr8 142219197 142219445 SLC45A4 intragenic chr8 142221016 142221482 SLC45A4 intragenic chr4 681313 681514 SLC49A3 promoter chr5 1232352 1232594 SLC6A18 intragenic chr14 23291078 23291337 SLC7A7 intergenic Y chr16 3639092 3639306 SLX4 intragenic chr19 51107425 51107742 SNAR‐F intergenic chr4 132669279 132669602 SNHG27 intergenic chr2 1286393 1286642 SNTG2 intragenic chr4 7666019 7666347 SORCS2 intragenic chr13 112610658 112611433 SOX1‐OT intergenic chr9 65602613 65602872 SPATA31A5 intergenic chr5 171482573 171482777 STK10 intragenic Y chr6 158507719 158508126 SYNJ2 intragenic chr4 940613 941057 TMEM175 intragenic chr16 33140199 33140659 TP53TG3 intergenic chr16 32264531 32265444 TP53TG3D promoter chr13 20134266 20134485 TPTE2 intragenic chr8 143407246 143408048 TSNARE1 intragenic chr19 4950670 4950940 UHRF1 intragenic Y chr19 30392613 30392952 URI1 intergenic Y chr9 136758461 136758902 VAV2 intragenic chr6 31867691 31867957 ZBTB12 promoter chr19 4059917 4060131 ZBTB7A promoter chr5 755873 756254 ZDHHC11B intragenic chr14 69256676 69257036 ZFP36L1 promoter chr16 88559820 88560089 ZFPM1 intragenic chr3 147115764 147116421 ZIC4 intragenic chr19 57306682 57307032 ZIM2 intragenic chr19 57306682 57307032 ZIM2‐AS1 intragenic chr16 88502475 88502684 ZNF469 intragenic

150

APPENDIX ONE SUPPLEMENTARY MATERIAL FOR CHAPTER TWO

Supplemental Table 2. Differentially methylated CGI, closest gene and differential expression for CD34+ vs CD56+ subsets

Differentially Expressed (>2 Chromosome Start End Closest Gene CGI Overlap Class fold) chr17 47296969 47297314 ABI3 promoter Y chr19 39217530 39217857 ACTN4 intragenic Y chr2 236773883 236774084 AGAP1 intragenic Y chr21 45713509 45713813 AIRE intragenic chr19 17443965 17444633 ANO8 promoter chr7 144052115 144052650 ARHGEF5 promoter chr19 947547 947981 ARID3A intragenic chr13 113540350 113540558 ATP11A intragenic chr7 99016967 99017289 ATP5MF‐PTCD1 intragenic chr5 17275369 17275638 BASP1 intragenic Y chr8 145494447 145494712 BOP1 intragenic chr7 99016967 99017289 BUD31 intragenic chr17 40191787 40192450 C17orf113 intragenic chr17 263083 263412 C17orf97 promoter chr17 36997448 36997661 C17orf98 promoter chr22 37578232 37578660 C1QTNF6 intragenic chr1 181767493 181767893 CACNA1E intragenic chr1 19665010 19665276 CAPZB intragenic Y chr13 111331166 111331373 CARS2 intragenic chr16 89033529 89033786 CBFA2T3 intragenic Y chr9 70645539 70645863 CBWD5 intergenic chr11 6291338 6291558 CCKBR intragenic chr1 1634252 1634690 CDK11A intragenic chr1 1643593 1643811 CDK11A intragenic chr1 1634252 1634690 CDK11B intragenic chr1 1643593 1643811 CDK11B intragenic chr9 21989773 21990108 CDKN2A intragenic chr9 22005887 22006229 CDKN2B intragenic chr9 22005887 22006229 CDKN2B‐AS1 intragenic chr22 47081750 47082045 CERK intragenic chr11 890191 890506 CHID1 intragenic Y chr11 62690974 62691488 CHRM1 intergenic chr7 73752819 73753343 CLIP2 intragenic Y chr13 111109139 111109652 COL4A2 intragenic chr13 111109139 111109652 COL4A2‐AS2 intragenic chr17 80202799 80203289 CSNK1D intragenic Y chr4 1233872 1234420 CTBP1 intragenic chr9 90439605 90440092 CTSLP8 intergenic chr6 39869579 39869800 DAAM2 intragenic chr2 3833815 3834318 DCDC2C intragenic chr4 961347 962155 DGKQ intragenic chr4 963804 964115 DGKQ intragenic chr15 41228442 41228872 DLL4 intragenic chr5 13810126 13810333 DNAH5 intragenic chr8 21769694 21770080 DOK2 promoter Y chr7 76129395 76129621 DTX2 intragenic DTX2P1‐UPK3BP1‐ chr7 76629263 76629489 PMS2P11 intragenic chr22 16192654 16193098 DUXAP8 promoter chr19 48244150 48244693 EHD2 intragenic Y chr19 10226164 10226702 EIF3G intragenic Y chr9 130706912 130707273 FAM102A intragenic

151

APPENDIX ONE SUPPLEMENTARY MATERIAL FOR CHAPTER TWO

chr22 21505626 21506073 FAM230B intergenic chr9 41222254 41222736 FAM74A1 intergenic chr12 133022649 133022863 FBRSL1 intergenic chr19 40382281 40382821 FCGBP intragenic chr19 17877468 17877777 FCHO1 intragenic chr8 21905461 21905757 FGF17 intragenic chr9 79629865 79630084 FOXB2 intergenic chr1 47899125 47899398 FOXD2‐AS1 intragenic chr1 47899661 47900385 FOXD2‐AS1 intragenic chr1 41889270 41889626 FOXO6 intergenic chr4 871276 871630 GAK intragenic chr19 2511392 2511782 GNG7 intragenic chr19 48908176 48908597 GRIN2D intragenic chr16 85676291 85676592 GSE1 intragenic Y chr1 9324202 9324439 H6PD intragenic chr10 94452200 94452428 HHEX promoter Y chr14 100141601 100142051 HHIPL1 intragenic chr11 315739 316539 IFITM1 intergenic Y chr11 330785 331110 IFITM3 intergenic Y chr11 2176839 2177043 INS‐IGF2 intragenic chr7 150901550 150901978 IQCA1L intragenic chr1 949329 949851 ISG15 promoter Y chr12 53591254 53591767 ITGB7 intragenic Y chr9 5042797 5043270 JAK2 intragenic chr6 15517282 15517557 JARID2 intragenic chr19 44278273 44278777 KCNN4 intragenic Y chr19 5074591 5074814 KDM4B intragenic chr11 33680219 33680428 KIAA1549L intragenic chr1 245498883 245499218 KIF26B intragenic chr17 51834650 51834900 KIF2B intergenic chr15 31689500 31689707 KLF13 intragenic Y chr9 110479034 110479420 KLF4 intergenic chr10 3823789 3824017 KLF6 promoter chr3 42977830 42978090 KRBOX1 promoter chr3 42977830 42978090 KRBOX1‐AS1 promoter chr9 139640640 139640885 LCN6 intragenic chr18 13641584 13642415 LDLRAD4 intragenic chr2 110705052 110705960 LIMS3‐LOC440895 intragenic chr18 6920591 6920874 LINC00668 intergenic chr2 132152766 132153044 LINC01120 intergenic chr1 47899125 47899398 LINC01389 intragenic chr1 47899661 47900385 LINC01389 intragenic chr20 61660249 61660881 LINC01749 intragenic chr5 176759071 176759361 LMAN2 intragenic Y chr9 139640640 139640885 LOC100128593 intragenic chr16 88590204 88590410 LOC100128882 intragenic chr2 110705052 110705960 LOC100288570 intragenic chr14 106033298 106033520 LOC105370697 intergenic chr17 263083 263412 LOC105371430 promoter chr3 195358440 195358686 LOC105374297 intergenic chr10 3598475 3598930 LOC105376360 intragenic chr17 28903507 28903832 LOC107133515 intragenic chr8 58172076 58173863 LOC286177 intragenic chr1 43353947 43354170 LOC339539 intragenic chr17 18992044 18992633 LOC388436 intergenic chr2 110705052 110705960 LOC440895 intragenic chr5 43037259 43037520 LOC648987 promoter chr22 25755941 25756164 LRP5L intragenic

152

APPENDIX ONE SUPPLEMENTARY MATERIAL FOR CHAPTER TWO

chr7 1967664 1967994 MAD1L1 intragenic Y chr1 26097904 26098107 MAN1C1 intragenic chr19 18260252 18260525 MAST3 intragenic Y chr6 41620950 41621262 MDFI intragenic chr1 17006742 17007151 MIR3675 intergenic chr17 34093571 34093836 MMP28 intragenic chr7 45002111 45002845 MYO1G intragenic Y chr18 3067297 3067533 MYOM1 intragenic chr12 125002880 125003101 NCOR2 intragenic chr12 125003217 125003482 NCOR2 intragenic chr20 50108795 50109456 NFATC2 intragenic chr19 55450822 55451128 NLRP7 intragenic chr2 157184389 157184632 NR4A2 intragenic chr9 140347840 140348320 NSMF promoter chr7 144052115 144052650 OR2A1‐AS1 promoter chr5 140751298 140751883 PCDHGA1 intragenic Y chr5 140751298 140751883 PCDHGA2 intragenic Y chr5 140751298 140751883 PCDHGA3 intragenic chr5 140751298 140751883 PCDHGA4 intragenic Y chr5 140751298 140751883 PCDHGA5 intragenic chr5 140751298 140751883 PCDHGB1 intragenic Y chr5 140751298 140751883 PCDHGB2 intragenic Y chr5 140751298 140751883 PCDHGB3 intragenic Y chr20 44574821 44575022 PCIF1 intragenic Y chr21 47845595 47845864 PCNT intragenic Y chr16 15223723 15224262 PDXDC1 intragenic chr1 912869 913153 PERM1 intragenic chr21 45730220 45730444 PFKL intragenic chr11 118505508 118505741 PHLDB1 intragenic chr10 3180174 3180578 PITRM1 intragenic chr16 15012708 15013276 PKD1P3‐NPIPA1 intragenic chr16 16451347 16451960 PKD1P4‐NPIPA8 intragenic chr16 15223723 15224262 PKD1P6‐NPIPP1 intragenic chr8 145018815 145019214 PLEC promoter chr19 1526462 1526681 PLK5 intragenic chr15 89876390 89876956 POLG promoter chr19 45885786 45885999 PPP1R13L intragenic chr10 72360255 72360605 PRF1 intragenic Y chr8 48743841 48744177 PRKDC intragenic Y chr5 139138875 139139242 PSD2‐AS1 intergenic chr2 113956343 113957042 PSD4 promoter Y chr7 99016967 99017289 PTCD1 intragenic chr1 29587087 29587412 PTPRU promoter chr21 46268896 46269263 PTTG1IP intergenic chr17 17603583 17604147 RAI1 intragenic chr13 114868894 114869836 RASA3 intragenic Y chr13 114875198 114875418 RASA3 intragenic Y chr19 10434024 10434248 RAVER1 intragenic chr20 55982631 55983074 RBM38 intragenic Y chr9 94662565 94662818 ROR2 intragenic chr9 137252115 137252451 RXRA intragenic Y chr18 76552775 76553014 SALL3 intergenic chr7 92672789 92673016 SAMD9 intergenic Y chr7 83378885 83379153 SEMA3E intergenic chr11 75277366 75278032 SERPINH1 promoter chr10 105420685 105421076 SH3PXD2A promoter chr19 51189824 51190109 SHANK1 intragenic chr5 1108802 1109051 SLC12A7 promoter Y

153

APPENDIX ONE SUPPLEMENTARY MATERIAL FOR CHAPTER TWO

chr17 80195043 80195866 SLC16A3 intragenic Y chr8 145641915 145642118 SLC39A4 intragenic chr17 1478047 1478259 SLC43A2 intragenic chr8 142219197 142219445 SLC45A4 intragenic chr8 142221016 142221482 SLC45A4 intragenic chr14 23291078 23291337 SLC7A7 intergenic Y chr4 132669279 132669602 SNHG27 intergenic chr11 47399788 47400006 SPI1 promoter Y chr6 158507719 158508126 SYNJ2 intragenic chr7 100218511 100218723 TFR2 intragenic Y chr11 1947531 1947958 TNNT3 intragenic chr11 1958934 1959247 TNNT3 intragenic chr16 32264531 32265444 TP53TG3D promoter chr13 20134266 20134485 TPTE2 intragenic chr8 27145131 27145601 TRIM35 intragenic chr5 180622178 180622658 TRIM7 intragenic chr21 45789090 45789373 TRPM2 intragenic chr17 56401728 56402343 TSPOAP1 promoter chr22 46685379 46685796 TTC38 intragenic chr9 136758461 136758902 VAV2 intragenic chr2 98340397 98340959 ZAP70 intragenic Y chr19 4059917 4060131 ZBTB7A promoter chr14 69256676 69257036 ZFP36L1 promoter chr9 115835264 115835927 ZFP37 intergenic chr16 88590204 88590410 ZFPM1 intragenic chr7 63498112 63498314 ZNF727 intergenic chr17 3907378 3907703 ZZEF1 intergenic

154

APPENDIX ONE SUPPLEMENTARY MATERIAL FOR CHAPTER TWO

Supplemental Table 3. Differentially methylated CGI, closest gene and differential expression for CD14+ vs CD56+ subsets

Differentially Chromosome Start End Closest Gene CGI Overlap Class Expressed (>2 fold) chr12 125599306 125599625 AACS intragenic chr9 139909878 139910267 ABCA2 intragenic Y chr19 39217530 39217857 ACTN4 intragenic chr1 27901660 27902688 AHDC1 intragenic chr21 45713509 45713813 AIRE intragenic chr7 2774444 2774655 AMZ1 intragenic chr12 5936876 5937085 ANO2 intragenic chr7 4832111 4832481 AP5Z1 intragenic chr7 144052115 144052650 ARHGEF5 promoter chr19 947547 947981 ARID3A intragenic chr19 950020 950370 ARID3A intragenic chr2 9514723 9514995 ASAP2 intragenic chr13 113540350 113540558 ATP11A intragenic chr7 70096263 70096501 AUTS2 intragenic chr19 835272 835524 AZU1 intergenic chr17 79380359 79380723 BAHCC1 intragenic chr15 91362445 91362658 BLM intergenic Y chr1 92414718 92414986 BRDT intragenic chr17 263083 263412 C17orf97 promoter chr17 37331425 37331629 CACNB1 intragenic chr13 111331166 111331373 CARS2 intragenic chr20 32232170 32232403 CBFA2T2 intragenic Y chr16 89033529 89033786 CBFA2T3 intragenic Y chr16 89034060 89034293 CBFA2T3 intragenic Y chr9 21989773 21990108 CDKN2A intragenic chr9 22005887 22006229 CDKN2B intragenic chr9 22005887 22006229 CDKN2B‐AS1 intragenic chr22 47081750 47082045 CERK intragenic Y chr4 2431798 2432328 CFAP99 intragenic chr1 9790292 9790704 CLSTN1 intragenic Y chr13 100310204 100310446 CLYBL intragenic chr17 40838849 40839062 CNTNAP1 intragenic chr10 93805512 93805845 CPEB3 intergenic chr4 1233872 1234420 CTBP1 intragenic Y chr18 77552401 77552603 CTDP1 intergenic Y chr18 77560088 77560292 CTDP1 intergenic Y chr19 49132603 49134291 DBP intragenic chr4 961347 962155 DGKQ intragenic chr4 963804 964115 DGKQ intragenic chr13 50707585 50708019 DLEU1 intragenic chr8 21914275 21914525 DMTN intragenic chr19 55673460 55673663 DNAAF3 intragenic chr8 21769694 21770080 DOK2 promoter chr12 113515164 113515970 DTX1 intragenic chr19 48244150 48244693 EHD2 intragenic chr19 10226164 10226702 EIF3G intragenic chr19 852855 853495 ELANE intragenic chr8 615143 615352 ERICH1 intragenic Y chr8 623372 623710 ERICH1 intragenic Y chr2 239009072 239009361 ESPNL intragenic chr11 128419198 128419513 ETS1 intragenic Y chr9 140268713 140268993 EXD3 intragenic

155

APPENDIX ONE SUPPLEMENTARY MATERIAL FOR CHAPTER TWO

chr9 130706912 130707273 FAM102A intragenic chr12 133022649 133022863 FBRSL1 intergenic chr19 40382281 40382821 FCGBP intragenic chr19 17877468 17877777 FCHO1 intragenic Y chr19 4785250 4785490 FEM1A intergenic chr8 21905461 21905757 FGF17 intragenic chr11 128694084 128694688 FLI1 intergenic Y chr5 43000343 43000791 FLJ32255 intergenic chr9 132652350 132652715 FNBP1 promoter Y chr1 47899125 47899398 FOXD2‐AS1 intragenic chr1 47899661 47900385 FOXD2‐AS1 intragenic chr19 3532433 3532666 FZR1 intragenic chr1 230415342 230415689 GALNT2 intragenic chr7 2774444 2774655 GNA12 intragenic chr19 2643287 2643633 GNG7 intragenic chr17 72442927 72443194 GPRC5C intragenic chr2 11750766 11751072 GREB1 intragenic chr11 67052394 67053110 GRK2 promoter chr7 73894815 73895110 GTF2IRD1 intragenic chr12 96389405 96389675 HAL intragenic chr19 611291 611579 HCN2 intragenic chr15 20711403 20711823 HERC2P3 promoter chr17 80393470 80393752 HEXDC intragenic chr10 94452200 94452428 HHEX promoter Y chr7 139416286 139416522 HIPK2 intragenic chr21 38352856 38353274 HLCS intragenic chr13 31019860 31020137 HMGB1 intergenic chr17 46641534 46642110 HOXB3 intragenic chr17 46654053 46654369 HOXB3 promoter chr17 46654053 46654369 HOXB4 promoter chr8 38831592 38832353 HTRA4 intragenic chr12 6657727 6657973 IFFO1 promoter chr11 315739 316539 IFITM1 intergenic Y chr11 330785 331110 IFITM3 intergenic chr2 242663547 242663872 ING5 intragenic chr5 3962286 3962512 IRX1 intergenic chr16 30485382 30485721 ITGAL intragenic Y chr17 73726238 73726526 ITGB4 intragenic chr12 53591254 53591767 ITGB7 intragenic Y chr18 77585913 77586856 KCNG2 intergenic chr17 61611260 61611586 KCNH6 intragenic chr12 121890223 121890510 KDM2B intragenic Y chr19 5074591 5074814 KDM4B intragenic chr14 104645553 104645762 KIF26A intragenic chr15 31689500 31689707 KLF13 intragenic chr10 3823789 3824017 KLF6 promoter chr1 31246010 31246280 LAPTM5 intergenic chr18 13641584 13642415 LDLRAD4 intragenic LIMS3‐ chr2 110705052 110705960 LOC440895 intragenic chr12 122235106 122235310 LINC01089 promoter chr1 47899125 47899398 LINC01389 intragenic chr1 47899661 47900385 LINC01389 intragenic chr21 45575451 45575833 LINC01678 intergenic chr20 59542508 59542739 LINC01718 intergenic chr14 101908781 101909022 LINC02314 intragenic chr4 6659271 6659787 LINC02482 intergenic chr5 176759071 176759361 LMAN2 intragenic

156

APPENDIX ONE SUPPLEMENTARY MATERIAL FOR CHAPTER TWO

chr16 88590204 88590410 LOC100128882 intragenic chr2 110705052 110705960 LOC100288570 intragenic chr22 21695101 21695548 LOC100996335 intergenic chr2 91911039 91911371 LOC101927050 intergenic chr11 94883335 94883565 LOC101929295 intergenic chr17 263083 263412 LOC105371430 promoter chr3 195358440 195358686 LOC105374297 intergenic chr15 37179516 37179782 LOC145845 intergenic chr2 110705052 110705960 LOC440895 intragenic chr21 46714984 46715206 LOC642852 intragenic chr5 1663730 1664340 LOC728613 intergenic chr5 1489799 1490071 LPCAT1 intragenic Y chr19 2323912 2324127 LSM7 intragenic chr11 1892037 1892556 LSP1 intragenic chr4 1304768 1305114 MAEA intragenic chr1 26097904 26098107 MAN1C1 intragenic chr19 12758975 12759210 MAN2B1 intragenic Y chr19 18260252 18260525 MAST3 intragenic chr18 74824149 74824414 MBP intragenic Y chr6 41620950 41621262 MDFI intragenic chr16 4730262 4730487 MGRN1 intragenic chr16 4732219 4732487 MGRN1 intragenic chr16 4732830 4733184 MGRN1 intragenic chr17 56355207 56355502 MPO intragenic chr17 56356374 56357013 MPO intragenic chr11 1986015 1986362 MRPL23 intergenic Y chr11 1253903 1254494 MUC5B intragenic chr5 176734599 176734944 MXD3 promoter chr18 3067297 3067533 MYOM1 intragenic chr11 78285405 78285995 NARS2 promoter chr15 20988422 20989666 NBEAP1 intergenic chr12 124941314 124941584 NCOR2 intragenic chr12 125002880 125003101 NCOR2 intragenic chr12 125003217 125003482 NCOR2 intragenic chr18 77284291 77285544 NFATC1 intragenic chr20 50108795 50109456 NFATC2 intragenic chr19 3369477 3369913 NFIC promoter chr9 139417115 139417518 NOTCH1 intragenic chr9 35791584 35791924 NPR2 intergenic chr2 157184389 157184632 NR4A2 intragenic Y chr2 157185557 157186355 NR4A2 intragenic Y chr9 102590742 102591303 NR4A3 intragenic chr7 144052115 144052650 OR2A1‐AS1 promoter chr19 10224442 10225138 P2RY11 intragenic chr19 727956 728177 PALM intragenic chr10 135203101 135203316 PAOX intragenic chr5 140736821 140737058 PCDHGA1 intragenic chr5 140751298 140751883 PCDHGA1 intragenic chr5 140755254 140755941 PCDHGA1 intragenic chr5 140736821 140737058 PCDHGA2 intragenic chr5 140751298 140751883 PCDHGA2 intragenic chr5 140755254 140755941 PCDHGA2 intragenic chr5 140736821 140737058 PCDHGA3 intragenic chr5 140751298 140751883 PCDHGA3 intragenic chr5 140755254 140755941 PCDHGA3 intragenic chr5 140736821 140737058 PCDHGA4 intragenic chr5 140751298 140751883 PCDHGA4 intragenic chr5 140755254 140755941 PCDHGA4 intragenic

157

APPENDIX ONE SUPPLEMENTARY MATERIAL FOR CHAPTER TWO

chr5 140751298 140751883 PCDHGA5 intragenic chr5 140755254 140755941 PCDHGA5 intragenic chr5 140755254 140755941 PCDHGA6 intragenic chr5 140736821 140737058 PCDHGB1 intragenic chr5 140751298 140751883 PCDHGB1 intragenic chr5 140755254 140755941 PCDHGB1 intragenic chr5 140751298 140751883 PCDHGB2 intragenic chr5 140755254 140755941 PCDHGB2 intragenic chr5 140751298 140751883 PCDHGB3 intragenic chr5 140755254 140755941 PCDHGB3 intragenic chr21 47845595 47845864 PCNT intragenic Y chr6 165747763 165747970 PDE10A intragenic chr16 335829 336034 PDIA2 intragenic chr8 22560949 22561271 PEBP4 intergenic chr17 17465135 17465397 PEMT intragenic chr1 912869 913153 PERM1 intragenic chr11 118505508 118505741 PHLDB1 intragenic chr19 3697837 3698378 PIP5K1C promoter chr17 17109893 17110296 PLD6 intergenic chr8 145003589 145004145 PLEC intragenic Y chr8 145024449 145025179 PLEC promoter Y chr7 72413249 72413793 POM121 intragenic chr19 10224442 10225138 PPAN‐P2RY11 intragenic chr10 72360255 72360605 PRF1 intragenic Y chr1 2082314 2082529 PRKCZ intragenic chr19 836533 836899 PRTN3 intergenic chr19 847898 848129 PRTN3 intragenic chr2 113956343 113957042 PSD4 promoter chr20 62168432 62168684 PTK6 promoter chr11 67203319 67203662 PTPRCAP intragenic Y chr19 8464627 8465001 RAB11B intragenic chr9 139715663 139716441 RABL6 intragenic chr17 17603583 17604147 RAI1 intragenic chr13 114868894 114869836 RASA3 intragenic Y chr13 114875198 114875418 RASA3 intragenic Y chr20 55982631 55983074 RBM38 intragenic chr11 63679403 63679638 RCOR2 intragenic chr4 3374733 3374998 RGS12 intragenic chr20 19955536 19956034 RIN2 intragenic Y chr19 54711106 54711375 RPS9 promoter chr9 137252115 137252451 RXRA intragenic Y chr19 1155020 1155305 SBNO2 intragenic chr12 125223520 125223750 SCARB1 intergenic chr12 125299474 125299874 SCARB1 intragenic chr10 102279162 102279730 SEC31B intragenic chr19 4543407 4544578 SEMA6B intragenic chr6 134497536 134497756 SGK1 promoter Y chr10 105420685 105421076 SH3PXD2A promoter chr19 51189824 51190109 SHANK1 intragenic chr5 1108802 1109051 SLC12A7 promoter Y chr7 100463758 100464146 SLC12A9 promoter Y chr12 129280256 129280486 SLC15A4 intragenic Y chr17 80195043 80195866 SLC16A3 intragenic Y chr20 61590816 61591210 SLC17A9 intragenic chr14 100795072 100795376 SLC25A47 intragenic chr11 57267006 57267316 SLC43A1 promoter chr17 1478584 1479214 SLC43A2 intragenic chr17 1492086 1492322 SLC43A2 intragenic

158

APPENDIX ONE SUPPLEMENTARY MATERIAL FOR CHAPTER TWO

chr8 142219197 142219445 SLC45A4 intragenic chr4 681313 681514 SLC49A3 promoter chr19 33726654 33726946 SLC7A10 intergenic chr16 3639092 3639306 SLX4 intragenic chr1 156261199 156261425 SMG5 promoter chr2 1286393 1286642 SNTG2 intragenic chr13 112610658 112611433 SOX1‐OT intergenic chr19 49132603 49134291 SPHK2 intragenic chr11 47399788 47400006 SPI1 promoter Y chr1 54821990 54822333 SSBP3 intragenic chr6 158507719 158508126 SYNJ2 intragenic chr12 117484305 117484627 TESC intragenic Y chr7 100218511 100218723 TFR2 intragenic chr13 114145965 114146172 TMCO3 promoter chr2 220412341 220412678 TMEM198 intragenic chr1 156261199 156261425 TMEM79 promoter chr19 6659886 6660094 TNFSF14 intergenic chr11 1947531 1947958 TNNT3 intragenic chr11 1958934 1959247 TNNT3 intragenic chr11 1299275 1299478 TOLLIP intragenic chr13 19964791 19965017 TPTE2 intergenic chr11 89713070 89713801 TRIM51EP intergenic chr8 143407246 143408048 TSNARE1 intragenic chr17 56401728 56402343 TSPOAP1 promoter Y chr22 46685379 46685796 TTC38 intragenic Y chr22 50657480 50657937 TUBGCP6 intragenic chr19 4950670 4950940 UHRF1 intragenic chr19 17759189 17759410 UNC13A intragenic chr19 30392613 30392952 URI1 intergenic Y chr10 88295205 88295592 WAPL intergenic chr1 228075377 228075750 WNT9A intergenic chr14 69256676 69257036 ZFP36L1 promoter chr9 115835264 115835927 ZFP37 intergenic chr16 88536936 88537375 ZFPM1 intragenic chr16 88559820 88560089 ZFPM1 intragenic chr16 88590204 88590410 ZFPM1 intragenic chr16 31075084 31075756 ZNF668 intragenic chr4 206377 206892 ZNF876P promoter chr17 3907378 3907703 ZZEF1 intergenic

159

APPENDIX ONE SUPPLEMENTARY MATERIAL FOR CHAPTER TWO

Supplemental Table 4. Chromosomes omitted from bioinformatic analysis

chr1_gl000191_random chrUn_gl000215 chr1_gl000192_random chrUn_gl000216 chr11_gl000202_random chrUn_gl000217 chr17_ctg5_hap1 chrUn_gl000218 chr17_gl000204_random chrUn_gl000219 chr17_gl000205_random chrUn_gl000220 chr4_ctg9_hap1 chrUn_gl000221 chr4_gl000193_random chrUn_gl000222 chr4_gl000194_random chrUn_gl000223 chr6_apd_hap1 chrUn_gl000224 chr6_cox_hap2 chrUn_gl000225 chr6_dbb_hap3 chrUn_gl000228 chr6_mann_hap4 chrUn_gl000229 chr6_mcf_hap5 chrUn_gl000231 chr6_qbl_hap6 chrUn_gl000235 chr6_ssto_hap7 chrUn_gl000236 chr8_gl000197_random chrUn_gl000237 chr9_gl000199_random chrUn_gl000240 chr9_gl000200_random chrUn_gl000241 chr9_gl000201_random chrUn_gl000242 chrUn_gl000211 chrUn_gl000243 chrUn_gl000212 chrX chrUn_gl000213 chrY chrUn_gl000214

160

APPENDIX ONE SUPPLEMENTARY MATERIAL FOR CHAPTER TWO

Supplemental Table 5. Purity of cell subsets following positive selection

Subject Subset Cell Count Purity (%) CeLi CD14+ 7.02E+05 96 CeLi CD34+ 4.68E+04 95 CeLi CD56+ 7.80E+04 99 DaBo CD14+ 8.20E+05 95 DaBo CD34+ 5.10E+04 98 DaBo CD56+ 3.80E+05 97 DeDe CD14+ 7.61E+05 99 DeDe CD34+ 1.72E+04 88 DeDe CD56+ 3.51E+05 98 EuLe CD14+ 8.78E+05 96 EuLe CD34+ 4.29E+04 91 EuLe CD56+ 7.41E+05 96 GrPa CD14+ 7.41E+05 97 GrPa CD34+ 8.58E+04 93 GrPa CD56+ 3.71E+05 98 JiWu CD14+ 3.90E+05 94 JiWu CD34+ 5.46E+04 87 JiWu CD56+ 4.49E+05 93 JuDj CD14+ 2.93E+05 97 JuDj CD34+ 1.56E+04 94 JuDj CD56+ 1.17E+05 96 NiFe CD14+ 8.29E+05 96 NiFe CD34+ 5.85E+04 98 NiFe CD56+ 3.80E+05 99 PrGa CD14+ 6.14E+05 98 PrGa CD34+ 1.95E+04 96 PrGa CD56+ 8.78E+04 96 ScRe CD14+ 5.85E+05 94 ScRe CD34+ 4.68E+04 91 ScRe CD56+ 4.88E+05 92 StSc CD14+ 4.29E+05 98 StSc CD34+ 7.80E+03 95 StSc CD56+ 8.78E+04 94

161

APPENDIX ONE SUPPLEMENTARY MATERIAL FOR CHAPTER TWO

Supplemental Table 6. Bisulfite conversion rates, coverage data and alignment statistics

Bisulfite Fraction of Mean Depth of Total Read Uniquely Ambiguous Subject Subset Conversion Symmetric CpGs Covered Symmetric Pairs Mapped Pairs Mapped Rate Covered CpGs CeLi CD14+ 0.99 0.15 19.14 15076009 10918659 868471 CeLi CD34+ 0.99 0.14 9.16 7494635 4572520 339620 CeLi CD56+ 0.99 0.15 26.77 26592159 18177834 1760606 DaBo CD14+ 1.00 0.14 14.33 11726529 8599867 643764 DaBo CD34+ 1.00 0.08 14.08 6984406 4746737 398960 DaBo CD56+ 1.00 0.15 15.70 14423006 10671618 842362 DeDe CD14+ 1.00 0.17 14.86 13926375 9618900 824316 DeDe CD34+ 1.00 0.14 22.36 14759463 10495381 738680 DeDe CD56+ 1.00 0.17 14.96 13932937 9454423 879915 EuLe CD14+ 1.00 0.15 10.21 8459675 5798207 506782 EuLe CD34+ 1.00 0.14 8.58 6500707 4344434 370920 EuLe CD56+ 0.99 0.16 19.66 18723696 13746724 1069891 GrPa CD14+ 1.00 0.17 18.61 18505357 12178776 1039607 GrPa CD34+ 1.00 0.16 20.26 17243639 11819321 1059543 GrPa CD56+ 1.00 0.18 22.15 21943392 14898743 1240129 JiWu CD14+ 1.00 0.15 15.86 14790093 10324835 1021899 JiWu CD34+ 1.00 0.10 9.51 6158375 4227041 331999 JiWu CD56+ 1.00 0.15 17.50 16460359 11569190 1056965 JuDj CD14+ 0.99 0.15 19.42 15644476 10872686 946321 JuDj CD34+ 1.00 0.12 21.11 12533709 8883704 731052 JuDj CD56+ 1.00 0.13 18.08 12815697 9197762 836397 NiFe CD14+ 1.00 0.12 11.88 8721201 5765138 527184 NiFe CD34+ 1.00 0.10 26.18 12350711 8854835 588782 NiFe CD56+ 0.99 0.16 14.20 13392231 9467413 834140 PrGa CD14+ 1.00 0.16 9.50 8678264 5745857 518678 PrGa CD34+ 1.00 0.14 13.81 10711780 7095184 657199 PrGa CD56+ 1.00 0.16 15.48 12634376 8320695 748797 ScRe CD14+ 1.00 0.15 8.05 6158471 4338486 373699 ScRe CD34+ 1.00 0.17 18.65 16026757 11122795 977667 ScRe CD56+ 0.99 0.15 16.15 15493305 10425854 1038832 StSc CD14+ 1.00 0.18 23.56 23118833 16205189 1332747 StSc CD34+ 1.00 0.14 8.03 5977149 4085740 361837 StSc CD56+ 1.00 0.17 16.17 14459470 10004622 833224

162

APPENDIX ONE SUPPLEMENTARY MATERIAL FOR CHAPTER TWO

Supplemental Table 7. Average CpG reads by cell subset and CpG island

Subset >2 >4 >10 >20 >100 >200 >1000 >5000 CD14+ 24780 24585 24260 23867 21803 19736 8931 173 CD14+ (%) 93 92 91 90 82 74 34 1 CD34+ 24702 24479 24075 23601 21353 19279 9769 271 CD34+ (%) 93 92 90 89 80 72 37 1 CD56+ 24837 24674 24363 23993 22187 20355 10098 276 CD56+ (%) 93 93 91 90 83 76 38 1 Mean proportion (all 0.93 0.92 0.91 0.89 0.82 0.74 0.36 0.01 subsets)

30000

CD14+ 25000 CD34+ 20000 CD56+

15000

Number of CGI 10000

5000

0 >2 >4 >10 >20 >100 >200 >1000 >5000 Number of CpG Reads Within CGI

163

APPENDIX ONE SUPPLEMENTARY MATERIAL FOR CHAPTER TWO

Supplemental Table 8. DNA methylation levels for filtered VDR peaks for CD14+ and CD34+

Methylation Mean CD14+ Standard Mean CD34+ Standard Difference (CD14+ ‐ Chromosome Start End Annotation Methylation Deviation Methylation Deviation CD34+) P Value Closest Gene chr1 3712646 3713646 promoter‐TSS 0.00 0.00 0.00 0.00 0.00 0.21 LRRC47 chr1 23345314 23346314 promoter‐TSS 0.00 0.00 0.00 0.00 0.00 0.20 KDM1A chr1 24017689 24018689 promoter‐TSS 0.07 0.02 0.10 0.04 ‐0.02 0.14 RPL11 chr1 28285564 28286564 promoter‐TSS 0.05 0.03 0.07 0.04 ‐0.02 0.16 SMPDL3B/XKR8 chr1 38455321 38456321 promoter‐TSS 0.03 0.02 0.03 0.02 0.00 0.98 SF3A3 chr1 84971784 84972784 promoter‐TSS 0.00 0.00 0.00 0.00 0.00 0.52 GNG5/SPATA1 chr1 204485043 204486043 promoter‐TSS 0.00 0.00 0.00 0.00 0.00 0.84 MDM4 chr1 246729171 246730171 promoter‐TSS 0.00 0.00 0.00 0.00 0.00 0.27 TFB2M/CNST chr10 101418115 101419115 promoter‐TSS 0.00 0.00 0.00 0.01 0.00 0.90 ENTPD7 chr11 30344115 30345115 promoter‐TSS 0.00 0.00 0.00 0.00 0.00 0.89 ARL14EP chr11 57297812 57298812 promoter‐TSS 0.00 0.00 0.00 0.01 0.00 0.99 TIMM10 chr11 61582437 61583437 promoter‐TSS 0.00 0.00 0.01 0.00 0.00 0.10 FADS1/MIR1908 chr11 62438563 62439563 promoter‐TSS 0.00 0.00 0.00 0.00 0.00 0.23 LBHD1/UQCC3 chr11 64646421 64647421 promoter‐TSS 0.00 0.00 0.01 0.00 0.00 0.03 EHD1 intron (NM_032871, chr11 73087721 73088721 intron 1 of 10) 0.01 0.01 0.01 0.01 0.00 0.68 RELT chr12 12162805 12163805 Intergenic 0.00 0.00 0.01 0.02 ‐0.01 0.11 BCL2L14 chr12 25538905 25539905 Intergenic 0.00 0.00 0.00 0.00 0.00 0.60 LMNTD1 intron (NM_006009, chr12 49581557 49582557 intron 1 of 3) 0.00 0.00 0.00 0.00 0.00 0.51 TUBA1A 5' UTR (NM_182729, chr12 104680306 104681306 exon 1 of 15) 0.00 0.00 0.00 0.00 0.00 0.90 TXNRD1 chr12 110906703 110907703 promoter‐TSS 0.01 0.01 0.02 0.02 ‐0.01 0.07 FAM216A intron (NM_001077261, chr12 124911539 124912539 intron 12 of 47) 0.55 0.05 0.75 0.04 ‐0.20 0.00 NCOR2 intron (NM_001077261, chr12 124942182 124943182 intron 7 of 47) 0.10 0.05 0.14 0.06 ‐0.04 0.03 NCOR2

164

APPENDIX ONE SUPPLEMENTARY MATERIAL FOR CHAPTER TWO

chr13 44452706 44453706 promoter‐TSS 0.00 0.00 0.00 0.00 0.00 0.94 CCDC122/LACC1 intron (NM_018283, chr13 48611701 48612701 intron 1 of 2) 0.00 0.00 0.00 0.00 0.00 0.05 NUDT15 intron (NM_020529, chr14 35872568 35873568 intron 1 of 5) 0.00 0.00 0.00 0.00 0.00 0.50 NFKBIA chr14 35873679 35874679 promoter‐TSS 0.00 0.00 0.01 0.01 0.00 0.07 NFKBIA chr14 73524621 73525621 promoter‐TSS 0.00 0.00 0.00 0.01 0.00 0.87 RBM25 chr14 74769274 74770274 promoter‐TSS 0.14 0.05 0.14 0.12 0.00 0.96 ABCD4 chr14 75229411 75230411 promoter‐TSS 0.00 0.00 0.00 0.00 0.00 0.38 YLPM1 chr14 92505904 92506904 promoter‐TSS 0.00 0.00 0.00 0.00 0.00 0.77 TRIP11 chr15 40225545 40226545 promoter‐TSS 0.10 0.05 0.13 0.11 ‐0.03 0.34 EIF2AK4 chr15 63795033 63796033 Intergenic 0.01 0.01 0.01 0.01 0.00 0.38 USP3 intron (NM_006305, chr15 69087366 69088366 intron 1 of 6) 0.01 0.01 0.10 0.07 ‐0.09 0.00 ANP32A chr15 82553851 82554851 promoter‐TSS 0.00 0.01 0.00 0.01 0.00 0.54 EFL1 chr15 98503011 98504011 promoter‐TSS 0.00 0.00 0.01 0.00 0.00 0.18 ARRDC4 chr16 14726215 14727215 promoter‐TSS 0.00 0.00 0.00 0.00 0.00 0.83 BFAR chr16 21964047 21965047 promoter‐TSS 0.00 0.00 0.00 0.00 0.00 0.04 UQCRC2 chr16 69373094 69374094 promoter‐TSS 0.00 0.00 0.00 0.00 0.00 0.21 COG8/NIP7 chr16 85604175 85605175 Intergenic 0.08 0.02 0.21 0.08 ‐0.13 0.00 GSE1 intron (NM_001142864, chr16 88837113 88838113 intron 1 of 50) 0.08 0.02 0.13 0.06 ‐0.05 0.04 PIEZO1 chr17 8198532 8199532 promoter‐TSS 0.02 0.02 0.01 0.01 0.01 0.01 SLC25A35 chr17 19550841 19551841 promoter‐TSS 0.02 0.01 0.02 0.01 ‐0.01 0.29 ALDH3A2 chr17 25659085 25660085 Intergenic 0.00 0.00 0.01 0.01 ‐0.01 0.19 WSB1 chr17 33905191 33906191 promoter‐TSS 0.00 0.00 0.00 0.00 0.00 0.69 PEX12 chr17 46985162 46986162 promoter‐TSS 0.00 0.00 0.00 0.00 0.00 0.90 LOC105371814/UBE2Z chr17 74349727 74350727 promoter‐TSS 0.00 0.00 0.00 0.00 0.00 0.83 PRPSAP1 chr18 32869826 32870826 promoter‐TSS 0.00 0.00 0.00 0.00 0.00 0.56 ZSCAN30/ZNF271P chr18 61604024 61605024 Intergenic 0.08 0.03 0.32 0.05 ‐0.24 0.00 SERPINB10 intron (NM_052847, chr19 2546376 2547376 intron 3 of 4) 0.51 0.16 0.63 0.16 ‐0.11 0.00 GNG7

165

APPENDIX ONE SUPPLEMENTARY MATERIAL FOR CHAPTER TWO

chr19 4457368 4458368 promoter‐TSS 0.00 0.00 0.00 0.00 0.00 0.99 UBXN6 intron (NM_001012753, chr19 12075861 12076861 intron 1 of 3) 0.00 0.00 0.00 0.01 0.00 0.25 ZNF763 chr19 39935674 39936674 promoter‐TSS 0.00 0.00 0.00 0.00 0.00 0.94 SUPT5H chr19 41304234 41305234 promoter‐TSS 0.00 0.00 0.00 0.00 0.00 0.57 RAB4B‐EGLN2/EGLN2 chr19 42387900 42388900 promoter‐TSS 0.00 0.00 0.00 0.00 0.00 0.38 ARHGEF1 chr19 49122096 49123096 promoter‐TSS 0.00 0.00 0.00 0.00 0.00 0.46 RPL18/SPHK2 chr2 3622285 3623285 promoter‐TSS 0.01 0.01 0.02 0.02 ‐0.01 0.10 RPS7 intron (NR_038432, chr2 7570774 7571774 intron 4 of 5) 0.01 0.00 0.01 0.01 ‐0.01 0.05 LOC100506274 chr2 73460809 73461809 promoter‐TSS 0.00 0.00 0.00 0.00 0.00 0.86 CCT7 chr20 43991222 43992222 promoter‐TSS 0.00 0.00 0.00 0.00 0.00 0.58 SYS1‐DBNDD2/SYS1 chr20 49126112 49127112 promoter‐TSS 0.00 0.00 0.00 0.01 0.00 0.24 PTPN1 chr20 52790122 52791122 promoter‐TSS 0.03 0.02 0.03 0.02 0.00 0.84 CYP24A1 chr22 17700128 17701128 promoter‐TSS 0.10 0.08 0.06 0.07 0.03 0.22 ADA2 chr22 36924899 36925899 promoter‐TSS 0.00 0.00 0.00 0.00 0.00 0.75 EIF3D chr3 14692615 14693615 promoter‐TSS 0.01 0.01 0.00 0.00 0.00 0.29 CCDC174 chr3 42053954 42054954 Intergenic 0.00 0.00 0.01 0.01 0.00 0.11 ULK4 ZNF660‐ ZNF197/ZNF197‐ chr3 44665969 44666969 promoter‐TSS 0.00 0.00 0.00 0.00 0.00 0.82 AS1/ZNF197 chr3 47421915 47422915 promoter‐TSS 0.00 0.00 0.00 0.00 0.00 0.45 PTPN23 chr3 114866196 114867196 promoter‐TSS 0.00 0.01 0.00 0.00 0.00 0.38 ZBTB20 chr3 121264392 121265392 promoter‐TSS 0.00 0.00 0.00 0.00 0.00 0.39 POLQ chr3 133291589 133292589 promoter‐TSS 0.00 0.00 0.00 0.00 0.00 0.79 CDV3 chr4 141294336 141295336 promoter‐TSS 0.00 0.00 0.00 0.00 0.00 0.35 SCOC/SCOC‐AS1 non‐coding (NR_104027, exon 2 SMARCA5/SMARCA5‐ chr4 144434263 144435263 of 2) 0.00 0.00 0.00 0.00 0.00 0.68 AS1 chr5 34915200 34916200 promoter‐TSS 0.00 0.00 0.00 0.00 0.00 0.19 RAD1/BRIX1 chr5 43040514 43041514 promoter‐TSS 0.00 0.00 0.00 0.00 0.00 0.96 LOC648987 chr5 137910886 137911886 promoter‐TSS 0.07 0.06 0.02 0.02 0.06 0.02 HSPA9

166

APPENDIX ONE SUPPLEMENTARY MATERIAL FOR CHAPTER TWO

chr5 154237612 154238612 promoter‐TSS 0.00 0.00 0.00 0.00 0.00 0.24 CNOT8 chr5 180649644 180650644 promoter‐TSS 0.00 0.01 0.00 0.00 0.00 0.59 TRIM41 chr5 180649996 180650996 promoter‐TSS 0.00 0.01 0.00 0.00 0.00 0.74 TRIM41 intron (NM_003913, chr6 4021204 4022204 intron 1 of 14) 0.00 0.00 0.00 0.00 0.00 0.74 PRPF4B chr6 28109155 28110155 promoter‐TSS 0.00 0.00 0.00 0.00 0.00 0.23 ZKSCAN8 chr6 28367039 28368039 promoter‐TSS 0.02 0.02 0.03 0.03 ‐0.02 0.15 ZSCAN12 chr6 33128703 33129703 Intergenic 0.01 0.00 0.01 0.00 0.00 0.77 COL11A2 chr6 106773122 106774122 promoter‐TSS 0.01 0.01 0.00 0.00 0.00 0.21 ATG5 chr6 107780334 107781334 promoter‐TSS 0.00 0.00 0.01 0.01 0.00 0.10 PDSS2 chr6 134273800 134274800 promoter‐TSS 0.00 0.00 0.00 0.00 0.00 0.98 TBPL1 chr6 170125017 170126017 Intergenic 0.03 0.01 0.06 0.05 ‐0.04 0.02 PHF10 intron (NM_181552, chr7 101499679 101500679 intron 1 of 23) 0.00 0.00 0.00 0.00 0.00 0.34 CUX1 intron (NM_012257, chr7 106809844 106810844 intron 1 of 10) 0.00 0.00 0.00 0.00 0.00 0.56 HBP1 chr7 140714333 140715333 promoter‐TSS 0.00 0.00 0.00 0.00 0.00 0.55 MRPS33 chr9 140095063 140096063 promoter‐TSS 0.00 0.00 0.00 0.00 0.00 0.26 TPRN

167

APPENDIX ONE SUPPLEMENTARY MATERIAL FOR CHAPTER TWO

Supplemental Data 9. Bioinformatic pipeline used in this study

168

APPENDIX ONE SUPPLEMENTARY MATERIAL FOR CHAPTER TWO

Supplemental Table 10. Participant characteristics for ZMIZ1 pyrosequencing cohort

MS Control Number 46 51 Female (%) 33 (72) 25 (49) Age range (yrs) 23‐67 18‐72 Mean age 39.5 46

169

APPENDIX TWO

SUPPLEMENTARY MATERIAL FOR CHAPTER THREE

APPENDIX TWO SUPPLEMENTARY MATERIAL FOR CHAPTER THREE

Additional file 1 – Cell harvest, WGBS alignment, bisulfite conversion and coverage statistics

Percentage Percentage Adult CD14+CD45+ Paediatric CD14+CD45+ AN1 2.3 PN1 72.9 AD1 2.7 PD1 58.4 AN2 18.0 PN2 91.0 AD2 24.0 PD2 93.4 Mean 11.7 Mean 78.9

A = adult, P = paediatric, N = no calcitriol, D = with calcitriol

171

APPENDIX TWO SUPPLEMENTARY MATERIAL FOR CHAPTER THREE

Ambiguous Total Proportion Bisulfite Total Read Unique Mapped Mapped Proportion Symmetric of CpGs Conversion Sample Pairs Mapped Pairs Pairs Ambiguous % Pairs Mapped CpG Depth Covered Rate (%) AN1 292152728 215555656 8028846 2.75 223584502 0.77 14.7 0.96 99.4 AD1 337195352 245453525 9286648 2.75 254740173 0.76 16.6 0.96 99.5 AN2 331620166 232035956 9345366 2.82 241381322 0.73 15.8 0.96 99.5 AD2 354982701 247138417 9586601 2.70 256725018 0.72 17.2 0.96 99.3 PN1 344145545 255498958 10383908 3.02 265882866 0.77 16.2 0.96 99.3 PD1 355229613 261918760 10307644 2.90 272226404 0.77 16.8 0.96 99.3 PN2 306949998 222088613 8510116 2.77 230598729 0.75 14.9 0.96 99.3 PD2 306411056 217247901 7762532 2.53 225010433 0.73 15.8 0.96 99.4

A = adult P = paediatric

N = no calcitriol D = calcitriol

1 or 2 = adult or paediatric sample

172

APPENDIX TWO SUPPLEMENTARY MATERIAL FOR CHAPTER THREE

Additional file 2 – Genome wide paediatric differentially methylated CpGs

No Calcitriol Calcitriol

FDR Difference (No Raw P Modified P Corrected P Total Methylated Proportion Methylated Proportion Calcitriol ‐ Chromosome Position Value Value Value Reads Reads Methylated Total Reads Reads Methylated Calcitriol) Gene chr2 210073653 0.16 1.64E‐08 0.02 32 25 0.78 32 29 0.91 ‐0.13 Intergenic (5' of MAP2) chr2 210073671 0.06 2.76E‐09 0.00 30 21 0.70 28 25 0.89 ‐0.19 Intergenic (5' of MAP2) chr2 210073682 0.07 2.76E‐09 0.00 27 20 0.74 18 17 0.94 ‐0.20 Intergenic (5' of MAP2) chr2 210073692 0.03 1.15E‐09 0.00 26 18 0.69 19 18 0.95 ‐0.26 Intergenic (5' of MAP2) chr2 210073708 0.02 1.19E‐09 0.00 20 9 0.45 13 11 0.85 ‐0.40 Intergenic (5' of MAP2) chr2 210073773 0.00 1.29E‐10 0.00 15 15 1.00 14 6 0.43 0.57 Intergenic (5' of MAP2) chr2 210073783 0.00 1.29E‐10 0.00 15 15 1.00 17 9 0.53 0.47 Intergenic (5' of MAP2) chr2 210073792 0.01 1.17E‐11 0.00 15 14 0.93 17 8 0.47 0.46 Intergenic (5' of MAP2) chr2 210073813 0.00 1.17E‐11 0.00 18 18 1.00 27 16 0.59 0.41 Intergenic (5' of MAP2) chr2 210073823 0.00 6.35E‐11 0.00 19 19 1.00 27 16 0.59 0.41 Intergenic (5' of MAP2) chr2 210073829 0.03 1.32E‐10 0.00 21 19 0.90 26 16 0.62 0.29 Intergenic (5' of MAP2) chr2 210073831 0.00 1.32E‐10 0.00 21 21 1.00 26 15 0.58 0.42 Intergenic (5' of MAP2) chr2 210073834 0.01 1.32E‐10 0.00 22 21 0.95 27 16 0.59 0.36 Intergenic (5' of MAP2) chr2 210073839 0.01 1.32E‐10 0.00 22 21 0.95 27 16 0.59 0.36 Intergenic (5' of MAP2) chr2 210073845 0.02 1.26E‐10 0.00 24 22 0.92 28 17 0.61 0.31 Intergenic (5' of MAP2) chr2 210073856 0.04 6.36E‐10 0.00 25 23 0.92 26 18 0.69 0.23 Intergenic (5' of MAP2) chr2 210073863 0.02 6.36E‐10 0.00 21 19 0.90 26 16 0.62 0.29 Intergenic (5' of MAP2) chr2 210073891 0.04 6.15E‐09 0.01 22 20 0.91 27 18 0.67 0.24 Intergenic (5' of MAP2) chr2 210073903 0.18 1.15E‐08 0.01 21 16 0.76 26 15 0.58 0.18 Intergenic (5' of MAP2) chr2 210073926 0.10 6.64E‐08 0.05 22 18 0.82 25 15 0.60 0.22 Intergenic (5' of MAP2) chr2 210073932 0.36 6.64E‐08 0.05 21 15 0.71 24 14 0.58 0.13 Intergenic (5' of MAP2) chr20 57430056 0.05 3.45E‐08 0.03 40 13 0.33 37 23 0.62 ‐0.30 GNAS chr20 57430062 0.03 2.85E‐08 0.03 40 13 0.33 35 23 0.66 ‐0.33 GNAS chr20 57430065 0.02 3.38E‐08 0.03 37 11 0.30 34 22 0.65 ‐0.35 GNAS

173

APPENDIX TWO SUPPLEMENTARY MATERIAL FOR CHAPTER THREE

chr20 57430077 0.01 3.38E‐08 0.03 34 9 0.26 35 23 0.66 ‐0.39 GNAS chr20 57430087 0.00 5.17E‐08 0.04 33 8 0.24 36 24 0.67 ‐0.42 GNAS chr5 172110655 0.11 7.63E‐08 0.05 25 17 0.68 25 10 0.40 0.28 NEURL1B chr5 172110658 0.22 7.63E‐08 0.05 24 14 0.58 25 8 0.32 0.26 NEURL1B chr5 172110661 0.19 7.63E‐08 0.05 25 14 0.56 24 9 0.38 0.19 NEURL1B chr5 172110664 0.19 7.08E‐08 0.05 24 16 0.67 23 11 0.48 0.19 NEURL1B chr5 172110677 0.04 7.48E‐08 0.05 31 16 0.52 28 7 0.25 0.27 NEURL1B chr5 172110684 0.05 5.88E‐08 0.05 34 24 0.71 28 13 0.46 0.24 NEURL1B chr5 172110707 0.19 3.58E‐08 0.03 36 15 0.42 27 7 0.26 0.16 NEURL1B chr5 172110710 0.12 3.78E‐08 0.03 37 21 0.57 27 10 0.37 0.20 NEURL1B chr5 172110721 0.07 2.81E‐08 0.03 34 18 0.53 27 8 0.30 0.23 NEURL1B chr5 172110725 1.00 7.09E‐08 0.05 36 16 0.44 27 12 0.44 0.00 NEURL1B chr5 172110729 0.01 4.90E‐08 0.04 36 25 0.69 26 9 0.35 0.35 NEURL1B

174

APPENDIX TWO SUPPLEMENTARY MATERIAL FOR CHAPTER THREE

Additional file 3 – Genome wide adult differentially methylated CpGs

SEE XLSX FILE – Appendix 2_Additional file 3.xlsx

Additional file 4 – Differentially methylated VDR myeloid peaks

SEE XLSX FILE – Appendix 2_Additional file 4.xlsx

175

APPENDIX TWO SUPPLEMENTARY MATERIAL FOR CHAPTER THREE

Additional file 5 – MS risk genes overlapping differentially methylated VDR peaks

ADO C10orf55 CSF2RB FAM69A JARID2 MARCH1 PLEK RREB1 TNFAIP8 AFF1 CARD11 CTLA4 FOXP1 KPNB1 MERTK PLXNC1 SH2B3 TNIP3 ANKRD55 CBLB CTSH GABARAPL3 LOC100130476 MIR1208 PTGER4 SLAMF7 UBASH3B ATXN1 CD226 CXCR4 GRB2 LOC152225 MIR548AN PVR SLC9A8 VMP1 B4GALT5 CD28 CXCR5 IFNGR2 LOC285626 MTHFR PVT1 SP140 WWOX BATF CD48 CYP24A1 IL2RA LPP NCF4 RAD51B SPRED2 ZBTB38 BATF3 CD58 DLEU1 IL7R LPP‐AS2 NR1D1 RASGRF1 STAT3 ZC3HAV1 BCAS1 CD86 ELMO1 IRF5 LY86 NSL1 RCOR1 TAGAP ZFP36L2 BCL6 CLEC16A EOMES IRF8 MAF PDE4A RGS1 TBX6 ZMIZ1 BCL9L CNR2 ETS1 JAK1 MAP3K14 PLEC RMI2 TET2 ZNF365

176

APPENDIX TWO SUPPLEMENTARY MATERIAL FOR CHAPTER THREE

Additional file 6 – Differentially expressed genes, vitamin D vs no vitamin D

SEE XLSX FILE – Appendix 2_Additional file 6.xlsx

Additional file 7 – Differentially expressed genes, adult vs paediatric

SEE XLSX FILE – Appendix 2_Additional file 7.xlsx

177

APPENDIX THREE

SUPPLEMENTARY MATERIAL FOR CHAPTER FOUR

APPENDIX THREE SUPPLEMENTARY MATERIAL FOR CHAPTER FOUR

Supplementary Information 1. Bisulfite conversion rates, coverage and alignment statistics by subject and cell subset

Uniquely Proportion Mean Depth Bisulfite Total Read Ambiguous Subject Subset Mapped of CpGs of Covered Conversion Rate Pairs Mapped Pairs Pairs Covered CpGs

B1 Activated 0.984 140860942 93489539 2551583 0.912 5.16 B1 LCL 0.978 178588206 121319039 3257426 0.933 6.92 B1 Resting 0.981 195911086 115429244 3468242 0.916 6.12 B2 Activated 0.983 204213232 113261160 3787137 0.916 5.63 B2 LCL 0.986 214421911 125935885 4340271 0.922 6.27 B2 Resting 0.977 147760733 107018786 3304048 0.871 4.51 B3 Activated 0.986 213890590 119459177 3512096 0.925 6.51 B3 LCL 0.986 140983322 102563247 2705258 0.837 3.88 B3 Resting 0.95 206558449 113549628 3319035 0.912 5.87

179

APPENDIX THREE SUPPLEMENTARY MATERIAL FOR CHAPTER FOUR

Supplementary Information 2. Coverage statistics for merged methcounts data by subset

Subset Proportion of CpGs Covered Mean Depth of Covered CpGs Activated 0.97 16.58 LCL 0.97 16.17 Resting 0.97 15.59

180

APPENDIX THREE SUPPLEMENTARY MATERIAL FOR CHAPTER FOUR

Supplementary Information 3. GO process terms associated with differentially methylated 1kb tiles for CD40L activated B cells and LCLs

UNDERREPRESENTED PROCESSES

Fold Raw P‐ Biological Process Enrichment Value FDR Positive regulation of lymphocyte activation (GO:0051251) 0.06 7.40E‐17 1.33E‐13 B cell receptor signaling pathway (GO:0050853) 0.13 1.70E‐14 1.53E‐11

Phagocytosis (GO:0006909) 0.29 5.61E‐10 3.36E‐07 Regulation of lymphocyte activation (GO:0051249) 0.29 8.67E‐10 3.89E‐07 Regulation of leukocyte activation (GO:0002694) 0.31 2.71E‐09 9.72E‐07 Antigen receptor‐mediated signaling pathway (GO:0050851) 0.34 6.22E‐09 1.86E‐06

Regulation of cell activation (GO:0050865) 0.32 7.86E‐09 2.02E‐06

Immune response (GO:0006955) 0.58 1.71E‐08 3.84E‐06

B cell activation (GO:0042113) 0.35 3.01E‐08 6.00E‐06 Defense response to bacterium (GO:0042742) 0.38 8.60E‐08 1.54E‐05

Response to bacterium (GO:0009617) 0.38 1.12E‐07 1.55E‐05

Response to other organism (GO:0051707) 0.38 1.12E‐07 1.67E‐05

Response to biotic stimulus (GO:0009607) 0.39 1.38E‐07 1.77E‐05 Response to external biotic stimulus (GO:0043207) 0.38 1.12E‐07 1.83E‐05 Immune response‐regulating cell surface receptor signaling pathway (GO:0002768) 0.42 2.08E‐07 2.33E‐05 Immune response‐activating cell surface receptor signaling pathway (GO:0002429) 0.42 2.08E‐07 2.49E‐05

Adaptive immune response (GO:0002250) 0.45 2.15E‐06 2.27E‐04

181

APPENDIX THREE SUPPLEMENTARY MATERIAL FOR CHAPTER FOUR

Immune effector process (GO:0002252) 0.56 2.70E‐05 2.55E‐03

Lymphocyte activation (GO:0046649) 0.52 2.66E‐05 2.65E‐03

Humoral immune response (GO:0006959) 0.52 3.24E‐05 2.77E‐03

Leukocyte activation (GO:0045321) 0.56 6.56E‐05 5.36E‐03 Vesicle budding from membrane (GO:0006900) 0.62 1.31E‐04 9.77E‐03

Membrane invagination (GO:0010324) 0.62 1.31E‐04 1.02E‐02

Innate immune response (GO:0045087) 0.6 1.57E‐04 1.13E‐02

Protein metabolic process (GO:0019538) 0.74 5.41E‐04 3.74E‐02

Primary metabolic process (GO:0044238) 0.79 7.13E‐04 4.74E‐02

OVERREPRESENTED PROCESSES

Fold Raw P‐ Biological Process Enrichment Value FDR Multicellular organismal process (GO:0032501) 1.18 3.21E‐05 2.88E‐03

Analysed using PANTHER Go slim version 14.1 released 12 March 2019

182

APPENDIX THREE SUPPLEMENTARY MATERIAL FOR CHAPTER FOUR

Supplementary Information 4. SNPs in LD with MS MHC risk alleles and differential methylation status in LCLs vs CD40L activated B cells

SNP In LD HLA Allele Associated With Risk Differentially With MS Allele Methylated HLA A*02:01 rs2523822 N HLA A*02:01 rs4713274 N HLA A*02:01 rs9295825 N HLA B*44:02 rs9266773 Y HLA B*55:01 rs3819284 N HLA B*55:01 rs3093547 N HLA DQA1*01:01 rs13193645 N HLA DQB1*03:01 rs2858312 N HLA DQB1*03:02 rs3957146 Y HLA DQB1*03:02 rs3998159 N HLA DRB1*03:01 rs2854275 Y HLA DRB1*08:01 rs7775055 N HLA DRB1*08:01 rs4713586 N HLA DRB1*15:01 rs3135391 N HLA DRB1*15:01 rs3135388 N HLA DRB1*15:01 rs3129889 N HLA DRB1*15:01 rs9271366 N

183

APPENDIX THREE SUPPLEMENTARY MATERIAL FOR CHAPTER FOUR

Supplementary Information 5. GO process terms associated with differentially methylated 1kb tiles for resting B cells and LCLs

UNDERREPRESENTED PROCESSES

Biological Process Fold Enrichment Raw P‐Value FDR

Positive regulation of lymphocyte activation (GO:0051251) 0.06 2.92x10‐17 5.25x10‐14

B cell receptor signaling pathway (GO:0042113) 0.14 2.87x10‐14 2.58x10‐11

Regulation of lymphocyte activation (GO:0051249) 0.28 3.10x10‐10 1.85x10‐7

Phagocytosis (GO:0006909) 0.30 4.18x10‐10 1.88x10‐7

Regulation of leukocyte activation (GO:0002694) 0.30 9.86x10‐10 3.54x10‐7

Regulation of cell activation (GO:0050865) 0.32 2.94x10‐9 8.80x10‐7

Immune response (GO:0006955) 0.58 9.10x10‐9 2.33x10‐6

Antigen receptor‐mediated signaling (GO:0050851) 0.35 1.31x10‐8 2.95x10‐6

B cell activation (GO:0042113) 0.35 3.42x10‐8 6.83x10‐6

Defense response to bacterium (GO:0042742) 0.37 3.56x10‐8 6.39x10‐6

Response to external biotic stimulus (GO:0043207) 0.38 8.38x10‐8 1.37x10‐5

Response to other organism (GO:0051707) 0.38 8.38x10‐8 1.25x10‐5

Response to bacterium (GO:0009617) 0.38 8.38x10‐8 1.16x10‐5

184

APPENDIX THREE SUPPLEMENTARY MATERIAL FOR CHAPTER FOUR

Response to biotic stimulus (GO:0009607) 0.39 1.05x10‐7 1.35x10‐5 Immune response‐activating cell surface receptor signaling pathway (GO:0002429) 0.42 1.57x10‐7 1.88x10‐5 Immune response‐regulating cell surface receptor signaling pathway (GO:0002768) 0.42 1.57x10‐7 1.76x10‐5

Adaptive immune response (GO:0002250) 0.44 7.27x10‐7 7.68x10‐5

Lymphocyte activation (GO:0046649) 0.51 1.41x10‐5 1.40x10‐3

Immune effector process (GO:0002252) 0.55 1.62x10‐5 1.53x10‐3

Humoral Immune response (GO:0006959) 0.51 1.67x10‐5 1.50x10‐3

Leukocyte activation (GO:0045321) 0.54 3.58x10‐5 3.06x10‐3

Innate immune response (GO:0045087) 0.59 9.84x10‐5 8.03x10‐3

Membrane invagination (GO:0010324) 0.64 3.28x10‐4 2.45x10‐2

Vesicle budding from membrane (GO:0006900) 0.64 3.25x10‐4 2.36x10‐2

OVERREPRESENTED PROCESSES

Biological Process Fold Enrichment Raw P‐Value FDR Multicellular organismal process (GO:0032501) 1.16 1.83x10‐4 1.43x10‐2

Analysed using PANTHER version 14.1, released 12 March 2019

185

APPENDIX THREE SUPPLEMENTARY MATERIAL FOR CHAPTER FOUR

Supplementary Information 6. Genes associated with differentially methylated loci in MSGWAS and LCLeQTL lists

MSGWAS Associated Gene Function B4GALT5 Encodes type II membrane‐bound glycoprotein. The function of the ezyme encoded by this gene is unclear BACH2 Transcriptional regulator that plays an important role in coordinating transcription activation and repression by MAFK BCAS1 Candidate oncogene. Required for myelination BCL6 Sequence specific repressor of transcription mainly required for germinal centre formation and antibody affinity maturation. C1orf52 CD28 Essential for T‐cell proliferation and survival, cytokine production and Th2 development Component of the MITRAC (mitochondrial translation regulation assembly intermediate of cytochrome c oxidase complex (complex) that CMC1 regulates cytochrome c oxidase assembly CSF2RB Common beta chain of the high affinity receptor for IL‐3, IL‐5 and CSF CTLA4 Member of immunoglobulin superfamily, encodes protein that transmits inhibitory signal to T cells Chemokine receptor expressed in mature B‐cells and Burkitt's lymphoma. Binds to B‐lymphocyte chemoattractant and is involved in B‐ CXCR5 cell migration into B‐cell follicles of spleen and Peyer patches CYP24A1 Mitochondrial protein initiating degradation of 1,25dihydroxyvitamin D3. DDX6 Encodes RNA helicase protein found in P‐bodies and stress granules, functioning in translation suppression and mRNA degradation. DLEU1 May act as a tumor suppressor Transcription factor crucial for embryonic development of mesoderm and CNS in vertebrates. May also be necessary for the EOMES differentiation of CD8+ T cells ETS1 Functions as either transcriptional activator or repressor of numerous genes ETV7 Transcription factor predominantly expressed in haematopoietic tissues EXOC6 A component of a multiprotein complex required for exocytosis FAM76B NEDD8 specific protease activity Encodes Fc receptor‐like glycoprotein. Has four extracellular C2‐type Ig domains, transmembrane domain and cytoplasmic domain with FCRL2 inhibitor and activation motifs. May have regulatory role in normal and neoplastic B cell development. Encodes Fc receptor‐like glycoprotein. Contains both activating and inhibitory motifs in its cytoplasmic domain. Promotes TLR9‐induced FCRL3 B cell proliferation, activation and survival but inhibits antibody production and suppresses plasma cell differentiation. FRMD6 FERM domain containing 6 FUCA2 Encodes alpha‐L‐fucosidase 2, which catalyses the hydrolysis of alpha‐1,6‐linked fucose

186

APPENDIX THREE SUPPLEMENTARY MATERIAL FOR CHAPTER FOUR

Encodes one of the gamma subunits of guanine nucleotide‐binding proteins (G proteins), involved as a modulator/transducer of GNG2 intracellular signalling HHEX Member of the homeobox family of transcription factors. May play a role in haematopoieitc differentiation. IGSF23 Immunoglobulin superfamily 23, encodes protein with one immunoglobulin domain. IL12A‐AS1 RNA gene Binds to IFN‐stimulated response elements to regulate expression of genes stimulated by type I interferons. Specifically binds to IRF8 regulatory region upstream of type I IFN and IFN‐inducible MHC class I genes. JAK1 Janus kinase 1, phosphorylates STAT proteins and plays key role in type I and II IFN signal transduction LBH Transcriptional activator which may act in mitogen‐activated protein kinase signalling pathway LINC01082 RNA gene LINC01146 RNA gene LOC100129203 RNA gene affiliated with ncRNA class LOC101929282 RNA gene affiliated with ncRNA class LOC285626 RNA gene affiliated with ncRNA class LOC285740 LPP‐AS2 RNA gene Transcription factor that can function as an activator or repressor. Activates expression of IL4 in Th2 cells, increases T cell susceptibility MAF to apoptosis. Represses transcription of CD13 promoter in early stages of myelopoiesis. Activates HECT domain‐containing E3 ubiquitin‐protein ligases. Prevents chronic Th mediated inflammation, restricts proinflammatory NDFIP1 cytokine production in Th17 cells. Limits cytokine signalling and expansion of effector Th2 cells. PRR5L Associates and regulates mTORC2 complex, regulates cellular survival and cytoskeletal organisation. Controls cell migration. PVR Transmembrane glycoprotein mediating NK cell adhesion and triggering NK cell effector functions. Binds CD96 and CD226 on NK cells. RAD51B Involved in homologous recombination repair pathway of dsDNA breaks during DNA replication or induced by DNA damaging agents. RBM43 RNA binding motif 43 Forms heterodimeric core‐binding factor with CBFB that binds a number of different enhancers and promoters including TCR enhancers, RUNX3 LCK, IL3 and GM‐CSF promoters. May be involved in cellular proliferation and/or differentiation SLC9A8 Integral transmembrane proteins involved in cellular exchange of Na+ and H+ ions SPRY4 Inhibitor of receptor transduced MAPK signalling pathway SYDE2 GTPase activator for the Rho‐type GTPases by converting them to an inactive GDP‐bound state Ligand for TNFRSF14, which is a herpesvirus entry mediator. Stimulates the proliferation of T cells and triggers apoptosis of various TNFSF14 tumor cells.

187

APPENDIX THREE SUPPLEMENTARY MATERIAL FOR CHAPTER FOUR

Participates in signal transduction of CD40, regulating pathways leading ot activation of NFkB and MAPK. Plays central role in B cell survival. Required for isotype switching from IgM to IgG. EBV encoded latent membrane protein 1 (LMP1), can interact with this and TRAF3 several other members of the TRAF family WWOX Tumor suppressor gene. Able to induce apoptosis. Plays a role in TGFB1 signaling and TGFB1 mediated cell death Zinc finger RNA binding protein that provides pathway for attenuation of protein synthesis through promoting removal or deadenylation ZFP36L1 of poly(A) tail of AU‐rich element‐containing mRNA transcripts

NB. There are more genes than tested regions due to multiple genes being imputed in the MSGWAS studies. Italicised genes are also in the LCLeQTL list

188

APPENDIX FOUR

VITAMIN D AND DNA METHYLATION IN

DEVELOPMENT, AGING AND DISEASE

(MANUSCRIPT IN PREPARATION)

APPENDIX FOUR REVIEW OF VITAMIN D AND DNA METHYLATION

Vitamin D and DNA methylation in development, aging and disease

*Lawrence T. C. Ong1,2, David R. Booth1, Grant P. Parnell1

1Centre for Immunology and Allergy Research, Westmead Institute for Medical Research, The

University of Sydney, 176 Hawkesbury Rd, Westmead, New South Wales 2145, Australia

2Department of Immunology, Westmead Hospital, Cnr Darcy and Hawkesbury Rds,

Westmead, New South Wales 2145, Australia

*Corresponding author

Contact Details

Lawrence T C Ong – [email protected]

Keywords – vitamin D, calcitriol, calcidiol, DNA methylation, epigenetics

190

APPENDIX FOUR REVIEW OF VITAMIN D AND DNA METHYLATION

ABSTRACT

DNA methylation is increasingly being recognised as a mechanism through which environmental exposures confer disease risk. Several studies have examined the association between vitamin D and changes in DNA methylation in areas as diverse as human and animal development, genomic stability, chronic disease risk and malignancy. In many cases, they have demonstrated clear associations between vitamin D and DNA methylation in candidate disease pathways. Despite this, a clear understanding of the mechanisms by which these factors interact is unclear. This paper reviews our current understanding of the effects of vitamin D on DNA methylation. In light of current knowledge in the field, the potential mechanisms mediating vitamin D effects on DNA methylation are discussed, as are the limiting factors and future avenues for research into this exciting area.

191

APPENDIX FOUR REVIEW OF VITAMIN D AND DNA METHYLATION

OVERVIEW

Introduction

Vitamin D and DNA methylation – A background

Vitamin D: Metabolism and genomic effects

DNA Methylation

Examining the association between vitamin D and DNA methylation

Developmental studies

Immune disease

Epigenetic aging

Genomic instability

Cancer

Furthering the field of vitamin D and DNA methylation research

Methodological issues

DNA methylation assessment

Intervention

Heterogeneity of cells studied

How does vitamin D regulate DNA methylation?

Future research

Conclusion

192

APPENDIX FOUR REVIEW OF VITAMIN D AND DNA METHYLATION

INTRODUCTION

Vitamin D has pleiotropic effects, with important roles in physiologic homeostasis, disease pathogenesis and organism development. Vitamin D is also known to modulate risk in many chronic diseases including osteoporosis, cardiovascular disease, diabetes and certain cancers. In some of these conditions, the mechanisms underlying disease risk are well understood, however in many, this is less clear. An example of this is multiple sclerosis (MS), whose incidence increases at latitudes further from the equator, perhaps under the influence of serum vitamin D levels. This latitude dependent risk is conferred during childhood and adolescence (1‐5), however once an individual contracts the disease, vitamin D supplementation is ineffective in its treatment (6, 7). A candidate mechanism for the prolonged effect of vitamin D in latitude‐dependent autoimmune disease risk is

DNA methylation. Due to its age‐related plasticity (8) and relative stability over time, DNA methylation might also explain early life or life‐long impacts of nutrition on development, chronic disease and cancer.

DNA methylation is thought to be an important interface between environment and disease (9, 10), and interest is increasing in regard to the effects of vitamin D on this important epigenetic mark.

Understanding how vitamin D affects the methylome will provide clarity and precision regarding the use of vitamin D in the treatment and prevention of disease. This paper will discuss vitamin D metabolism and its actions on the genome, before introducing the field of epigenetics and DNA methylation. We then review the existing evidence on vitamin D and its effects on DNA methylation in animal models, humans and malignant cells. The review will then discuss some of the shortfalls of the present research, potential mechanisms by which vitamin D might exert its effects on the methylome, and finally, directions for future studies in the area.

193

APPENDIX FOUR REVIEW OF VITAMIN D AND DNA METHYLATION

VITAMIN D AND DNA METHYLATION – A BACKGROUND

Vitamin D: Metabolism and genomic effects

Very little vitamin D is contained in foods, such that humans derive ~90% of their vitamin D requirements from exposure to sunlight (11). UVB radiation (a constituent of sunlight), is involved in the conversion of 7‐dehydrocholesterol into previtamin D3, which then isomerises to vitamin D3

(cholecalciferol). Vitamin D3 is released from the cellular plasma membrane and binds to vitamin D binding protein (DBP) in the circulation. It is then hydroxylated in the liver by 25‐hydroxylase

(CYP2R1) to 25‐hydroxyvitamin D3 (25(OH)D3; calcidiol). This prohormone is converted by the kidney through the action of 1α‐hydroxylase (CYP27B1) into the biologically active 1,25(OH)2D3 (calcitriol; see Figure 1). Human cells express different levels of vitamin D activating and inactivating enzymes, and therefore demonstrate varying ability to utilise vitamin D or calcidiol (12).

Calcitriol exerts its genomic effects through binding the vitamin D receptor (VDR), which is comprised of an N‐terminal DNA binding domain and C‐terminal ligand‐binding domain. Binding of calcitriol to the ligand‐binding domain results in heterodimerisation of the VDR with the retinoid X receptor (RXR). This heterodimer binds regions of DNA known as vitamin D response elements

(VDREs), which lie in the promoter regions of vitamin D responsive genes, resulting in upregulation or suppression of DNA transcription.

DNA methylation

DNA methylation involves the covalent addition of a methyl group to cytosine residues, typically in the context of a guanine nucleotide i.e. a CpG dinucleotide. Along with other epigenetic modifications (e.g. histone marks and non‐coding RNAs) it is involved in regulating gene transcription. DNA methylation is important in multiple physiologic functions, such as random X

194

APPENDIX FOUR REVIEW OF VITAMIN D AND DNA METHYLATION

inactivation in females, and genomic imprinting. It is also thought to play a role in stabilising the human genome by preventing the unwanted transposition of transposon derived repetitive elements (e.g. LINE‐1) which would otherwise result in significant genomic disruption and mutagenesis (13, 14). These sequences are relatively CpG rich, and uniformly methylated to prevent this from occurring (15).

Different enzymes are involved in the maintenance of existing DNA methylation marks and de novo

DNA methylation. During mitosis, DNA methyltransferase 1 (DNMT1) maintains DNA methylation by replicating the methylation pattern of hemimethylated DNA. DNMT3A, DNMT3B and DNMT3L participate in de novo DNA methylation (16). DNA demethylation occurs either in a passive fashion, when DNA replication is not accompanied by maintenance of DNA methylation, or in an active fashion, mediated by the family of ten‐eleven‐translocase (TET) enzymes.

Changes in DNA methylation may be associated with gene expression through alterations in the conformational state of DNA. Generally, increased DNA methylation is associated with compaction of chromatin filaments, reduced transcription factor access and repression of gene expression at that region. This is particularly true in the case of gene promoters and transcription start sites (17).

However, the opposite is thought to occur in gene bodies and increased methylation at these sites is generally thought to be associated with increased transcription (18).

195

APPENDIX FOUR REVIEW OF VITAMIN D AND DNA METHYLATION

repression domain which can recruit other factors that condense chromatin such as histone modifiers (21) (Figure 2A). A secondary mechanism which is probably less important in DNA methylation related transcriptional repression, is the inhibition of transcription factor binding (22)

(Figure 2B). The remainder of this review will focus on the relationship between vitamin D and DNA methylation. The interaction between vitamin D and other epigenetic marks has already been reviewed elsewhere (23, 24).

EXAMINING THE ASSOCIATION BETWEEN VITAMIN D AND DNA METHYLATION

Developmental studies

Animal models

Mouse and rat models have been used to study the effects of prenatal vitamin D deficiency. Vitamin

D affects DNA methylation at a number of imprinted genes. Imprinting is the process by which genes are silenced depending on their parent of origin. This typically occurs by DNA methylation at imprinting control regions. Xue et al. (25) found tissue specific changes in DNA methylation due to vitamin D deficiency at key imprinted genes that varied by generation and parent. In sperm for example, DNA methylation at IG‐DMR was lower in two generations of males (G1 and G2), following exposure to a low vitamin D diet in the grandparental generation (G0). This suggests that the effect of vitamin D may persist through multiple generations. Interestingly, this finding was not borne out in somatic tissues, although different loci appeared to be differentially methylated in G1 and G2 generations despite the absence of exposure to low vitamin D diet in G2. A subsequent study by Xue and colleagues (26) used targeted next generation sequencing to interrogate a greater number of

CpGs across the methylome of mouse sperm. Compared to those exposed to low vitamin D prenatally, controls had 15827 differentially methylated CpGs with predominantly higher DNA

196

APPENDIX FOUR REVIEW OF VITAMIN D AND DNA METHYLATION

methylation levels. The differentially methylated CpGs occurred at genes related to development and metabolism such as cadherin, Wnt and PDGF.

Maternal vitamin D deficiency during pregnancy has been linked to DNA methylation changes and physiological dysregulation such as in insulin resistance, body fat and blood pressure. The Panx1 promoter was found to be hypermethylated in the kidney of vitamin D depleted rats (but not the aorta or left ventricle) corresponding to elevated blood pressure and impaired endothelial relaxation of large blood vessels (27). The Panx1 gene encodes for a protein involved in vascular relaxation.

Metabolic dysregulation and higher body weight in offspring of vitamin D deficient rats was associated with changes in adipocyte DNA methylation at 608 promoters, and 204 CpG islands. Of the 305 corresponding genes, 101 were also differentially expressed. The Vldlr gene, encoding a lipoprotein receptor important in cholesterol and triglyceride metabolism, was found to be hypermethylated and lowly expressed, whilst Hif1α (involved in energy metabolism) was demethylated and highly expressed in offspring of vitamin D deficient mothers (28). Finally, hepatocytes from offspring of vitamin D deficient rats demonstrated global hypermethylation and elevated IκBα methylation compared to controls. This was associated with insulin resistance and higher inflammatory cytokine concentrations (29). Overall, these studies suggest that in utero vitamin D exposure may predispose to metabolic disease, perhaps in part mediated by vitamin D. A summary of the abovementioned animal studies is presented in Table 1.

Humans

Observational studies in general, have found few associations between vitamin D and DNA methylation. Neelon et al. (30) using cord blood leukocytes, found no association between maternal plasma calcidiol levels and DNA methylation at nine genomically imprinted sites. Similarly, mid

197

APPENDIX FOUR REVIEW OF VITAMIN D AND DNA METHYLATION

pregnancy calcidiol levels also showed no association with DNA methylation in cord blood using an array assaying more than 450,000 genomic CpGs (31). Mozhui et al. (32) in another observational study demonstrated similar findings using an array‐based methylation method, but extended their analysis to identify coordinated gene networks whose methylation changed in response to vitamin

D. DNA methylation of one gene module involved in responding to the presence of organic substances correlated negatively with calcidiol concentrations. Another module associated with immune responses showed a significant race‐by‐calcidiol interaction.

Although observational studies are useful, randomised controlled human trials provide greater precision in the examination of vitamin D effects on DNA methylation. A trial of high dose cholecalciferol in the late pregnancy and early postpartum period showed variable effects in mothers and newborns. Leukocyte DNA methylation differed between vitamin D and control subjects at <0.01% of assayed sites. Differential methylation occurred in both directions. At four to six weeks postpartum, mothers had methylation gain/loss at 200/102 CpGs, whilst children had gain/loss at 217/213 CpGs. Interestingly, genes associated with methylation loss in children were involved in regulation of apoptosis and antigen presentation (33). In another study, umbilical cord tissue RXRA DNA methylation was found to be reduced in those mothers supplemented with vitamin

D, with a mean difference of ~2%. The authors of this study proposed RXRA methylation as a surrogate for the action of vitamin D on other tissues following birth (34). Human studies on vitamin

D and DNA methylation are summarised in Table 2.

198

APPENDIX FOUR REVIEW OF VITAMIN D AND DNA METHYLATION

Immune disease

Animal models

As discussed earlier, vitamin D is thought to be involved in the pathogenesis of several immune diseases, especially those with a latitude dependent risk profile. Two studies have examined the effect of vitamin D on CD4+ T cells in an animal model of multiple sclerosis known as experimental autoimmune encephalomyelitis (EAE). Zeitelhofer et al. (35) found lower global DNA methylation and predominantly lower regional DNA methylation with cholecalciferol supplementation in mice with EAE. These differentially methylated regions (DMRs) were associated with 413 genes, which the authors attributed to lower expression of DNA methyltransferases and histone modifiers. At a nominal significance value, the authors also found several DMRs associated with gene expression changes including those at key T cell co‐receptors and intracellular activation molecules. Moore et al.

(36) using a different method to determine global CD4+ T cell DNA methylation, found two‐fold increased levels in CD4+ T cells cultured with calcitriol for 3 days compared to those that were cultured in the absence of calcitriol. The absolute change in global DNA methylation was ~3.5%. They attributed this difference to increased expression of Bhmt1 that ultimately led to increased availability of substrate for DNA methylation. The differences in global DNA methylation between the two studies may perhaps be related to the different assays used.

Cholecalciferol supplementation in pregnant rats was associated with lower DNA methylation in offspring at the IFNG gene (37). There was a concurrent increase in IFNγ, decrease in overall DNMT activity and increase in Th1:Th2 ratio. This may provide a mechanism for vitamin D mediated risk in allergic disease.

199

APPENDIX FOUR REVIEW OF VITAMIN D AND DNA METHYLATION

Humans

An observational study comparing 11 vitamin D deficient boys with age matched controls found only

2 CpG sites (in a panel of >27000) to be differentially methylated in leukocytes (38). These sites corresponded to microtubule associated protein MAPRE2 (37% lower in vitamin D deficient), and

DIO3 (24% higher in vitamin D deficient), that has important functions in the maturation and function of the thyroid axis. A randomised controlled trial found cholecalciferol supplementation in adolescents over 16 weeks to be associated with dose‐dependent increases in global leukocyte DNA methylation (39). These changes correlated with serum calcidiol levels. Junge et al. (40) examined cord blood of 3 children with high and low serum calcidiol levels. They found 508 DMRs, of which

311 were hypomethylated in those with high calcidiol levels. The authors found the methylation of the TSLP enhancer, to be correlated in a separate validation cohort, but only in those with high serum calcidiol levels, perhaps demonstrating an association between elevated serum vitamin D and predisposition to allergic disease. Another study found a CpG associated with MICAL3 to be hypomethylated in vitamin D deficient children (41). MICAL3 is involved in reactive oxygen species production and was hypothesised by the authors to be involved in pathogenesis of atopic dermatitis.

In adults aged 50‐75, there were no associations found between serum calcidiol levels and leukocyte

DNA methylation after adjusting for confounders (42). Another study conducted on similarly aged subjects found vitamin D intake as assessed by food questionnaire to be negatively correlated with

ABCA1 (involved in cholesterol metabolism) promoter DNA methylation in leukocytes (43). Chavez‐

Valencia et al. (44) did not find any statistically significant differentially methylated CpGs in adult peripheral blood mononuclear cells cultured with calcitriol for 120 hours. They speculated that this may have been related to decreased methylomic plasticity (as the cells were of adult origin) and masking of methylation changes in a heterogeneous cell population. To address this, Ong et al. (45) conducted whole genome bisulfite sequencing on CD34+ haematopoieitic progenitors from adults

200

APPENDIX FOUR REVIEW OF VITAMIN D AND DNA METHYLATION

and cord blood. The effect of calcitriol was minor, with very few regions found to be differentially methylated in cells of adult or paediatric origin. Interestingly, however DNA methylation at VDR binding sites was markedly lower in paediatric cells, perhaps providing a substrate for susceptibility to vitamin D in early life.

Epigenetic aging

Horvath’s seminal study on the prediction of chronological age using CpG methylation at specific loci, led to the use of DNA methylation as a biomarker of biological aging (46). Changes in DNA methylation at specific CpGs with aging were thought by Horvath to represent the cumulative effects of epigenetic maintenance. His discovery paved the way for a new field of epigenetic inquiry examining the relationship between disease states, environmental exposures and their ability to modulate an individuals’ biological age.

One such study examined the effects of vitamin D on the “Horvath clock”. This study found increasing serum calcidiol levels and cholecalciferol supplementation to be associated with lower epigenetic age (47). Another study examined the association between vitamin D status, an epigenetic mortality risk score based on 10 CpGs and all‐cause mortality in 1467 individuals aged between 50 and 75. Although they found no association between serum calcidiol levels and epigenetic mortality risk score, they found both to be independent predictors of mortality that predicted mortality even more strongly in combination (48).

Different epigenetic clocks have been developed to predict gestational age of neonates. Chen et al.

(49) found high dose cholecalciferol supplementation during pregnancy to be associated with lower epigenetic gestational age acceleration (GAA) amongst an African American subgroup. GAA in adults

201

APPENDIX FOUR REVIEW OF VITAMIN D AND DNA METHYLATION

has been associated with negative health outcomes, although the long‐term outcomes for neonates with unfavourable GAA is unclear.

Genomic stability

Retrotransposons are remnants of viral genomes that have co‐evolved with the human genome.

They are usually heavily methylated to silence their expression and prevent unwanted translocation, genomic instability and carcinogenesis. Three studies have examined the impact of vitamin D on

LINE‐1 methylation in human blood (50‐52), and a further two on human tissues (adipocytes (53), rectal mucosa (54)). The studies on peripheral blood did not find an association between vitamin D and LINE‐1 methylation. In tissue however, there was an association between higher LINE‐1 methylation and higher calcidiol levels, in adipocytes from colorectal cancer patients but not healthy controls (53). In older adults, rectal mucosa LINE‐1 methylation showed a positive association with serum calcidiol levels, consistent with the known protective effect of vitamin D against colorectal cancer (55).

Cancer

Few studies have been examined the effects of vitamin D on DNA methylation in malignant cells.

Most studies have provided in vitro data, allowing greater control of vitamin D exposures. Studies on breast cancer cell lines have found an association between calcitriol supplementation and decreased

CDH1 promoter (56) and PDLIM2 promoter methylation (57) and an inverse relationship between cholecalciferol and PTEN promoter methylation (58). CDH1 encodes E‐cadherin, a marker of breast cancer cell differentiation, whilst PDLIM2 encodes an adaptor molecule with tumour suppressor activity. PTEN (phosphatase and tensin homologue), is a tumour suppressor gene. Induction of these genes with calcitriol or cholecalciferol, may result in anti‐tumour effects in breast cancer.

202

APPENDIX FOUR REVIEW OF VITAMIN D AND DNA METHYLATION

In a malignant prostate cancer cell line, calcitriol triggered recruitment of NCOR1, which was associated with increases in the transcriptionally repressive histone mark H3K9me2 and a decrease in the transcriptionally permissive mark H3K9ac (59). Concurrently, there was also a reduction in the expression of VDR target genes involved in cell cycle regulation (CDKN1A and GADD45A) and mixed

DNA methylation changes at CDKN1A. These findings potentially explain the differential anti‐ proliferative effects of vitamin D on normal and malignant cells. Another study examining vitamin D effects on prostate cancer cell lines (LNCaP) found long term calcitriol exposure induced hypomethylation of genes linked to the mTOR signalling pathway, potentially accounting for their resistance to the antiproliferative effects of vitamin D (60).

Finally, an observational study found DNA methylation of DKK1 and WNT5A promoters in colorectal cancer to be negatively associated with vitamin D intake (61). The products of these genes participate in inhibition of the Wnt signalling pathway and thus malignant proliferation. A summary of studies examining malignant cells is presented in Table 3.

FURTHERING THE FIELD OF VITAMIN D AND DNA METHYLATION RESEARCH

Methodological Issues

Although it may seem difficult to reconcile the varying effect or lack of effect of vitamin D on the methylome, several methodological differences likely contribute to this heterogeneity. These will be discussed in turn.

DNA methylation assessment

DNA methylation assays provide information on different aspects of DNA methylation. Growing knowledge on the function of DNA methylation suggest the location of these changes are critical in

203

APPENDIX FOUR REVIEW OF VITAMIN D AND DNA METHYLATION

determining their function. To this extent, global differences in DNA methylation provide little information on the functional impacts of altered DNA methylation, and base resolution profiling is likely to be more informative. Array‐based DNA methylation, whilst providing base resolution information provides less breadth of information than whole genome bisulfite sequencing (WGBS) that is now considered to be the gold standard for DNA methylation analysis. The use of WGBS is limited by its cost, especially in large cohorts, and is probably limited to hypothesis‐free “discovery” studies at present. Where a target region is known however, regional, base resolution techniques such as mass array or pyrosequencing provide significant depth and information at target CpGs.

Thus, the heterogeneity of vitamin D effects in this review may in part be accounted for by the different CpG residues captured in the assays used. The characteristics of DNA methylation assays in this review are presented in Table 4.

Intervention

The form of vitamin D utilised or measured in the studies might also account for the variable effects on DNA methylation. The genomic effects of vitamin D can be attributed to the effect of its active metabolite, calcitriol, and it is likely that its methylomic effects are also mediated by calcitriol.

Therefore, in vitro studies of cells unable to metabolise calcidiol or cholecalciferol into active calcitriol, may not manifest DNA methylation changes. In vivo/ex vivo studies supplementing inactive vitamin D, would require UV radiation to ensure adequate calcitriol levels in serum. Another factor affecting methylomic susceptibility to vitamin D might be timing and duration of supplementation.

An epigenome wide study of the effects of vitamin D on the human monocyte cell line, THP‐1, found marked changes in chromatin accessibility due to vitamin D with maximal chromatin opening after

24 hours, that closed to almost baseline levels after 48 hours (62). Consistent with this, Lai et al found marked DNA methylation changes in prostate cancer cells that became resistant to the antiproliferative effects of calcitriol following long term exposure (60).

204

APPENDIX FOUR REVIEW OF VITAMIN D AND DNA METHYLATION

Heterogeneity of cells studied

DNA methylation is known to differ between cell types. Cells of the haematopoietic lineage for example, demonstrate greater similarity in DNA methylation to each other than to cells of non‐ haematopoieitic origin e.g. skin cells or connective tissue (63). Therefore, bulk DNA methylation of heterogeneous populations may not provide an accurate reflection of vitamin D effects on cell specific methylomes. In addition, the response to vitamin D is likely to vary between different cells due to cell‐specific variation in vitamin D receptor peaks (24). The age of the host cell is also likely to contribute to heterogeneity in the assessment of vitamin D related methylation change. Previous research has identified windows of methylomic susceptibility to environment, especially in early life

(64).

How does vitamin D regulate DNA methylation?

Different mechanisms have been posited to account for the effect of vitamin D on DNA methylation.

It is likely that DNA methylation is susceptible to a combination of these, along with other factors including environmental exposures and cell type.

Vitamin D may alter expression of genes that directly or indirectly affect DNA methylation. As discussed earlier, Bhmt1 expression was increased in CD4+ T cells of mice supplemented with cholecalciferol (36). This enzyme catalyses the generation of methionine from homocysteine.

Methionine is converted to S‐adenosylmethionine, which is a methyl‐donor utilised by DNA methyltransferases in DNA methylation.

205

APPENDIX FOUR REVIEW OF VITAMIN D AND DNA METHYLATION

Decreased DNA methyltransferase expression was found by Zeitelhofer and colleagues (35) in CD4+

T cells of cholecalciferol supplemented mice. Similarly, Lai et al. found decreased expression of

DNMT1 and DNMT3B in calcitriol treated prostate cancer cells (60). In DNMT3B, they accounted for this effect in two ways i) calcitriol interfering with transcription factor binding to the DNMT3B promoter, that was reversed by removal of distal VDREs from the DNMT3B promoter region and ii) via miR‐98‐5p upregulation, that is known to target the DNMT3B 3’UTR. Although the above mechanisms might account for global hypomethylation with vitamin D treatment, it does not explain the bidirectional, locus‐specific DNA methylation changes seen in other studies. Evidence exists for the effect of vitamin D on other DNA methylation altering enzymes, with VDR binding sites previously found proximal to DNMT1, DNMT3, TET2 and TET3 (65). Specific targeting mechanisms would then be required to localise active and passive alterations of DNA methylation at genic regions.

Vitamin D may influence DNA methylation indirectly through other epigenetic mechanisms. For example, targeting of DNMT3A and DNMT3B appears to be dependent on methylation of H3K36 (66,

67) which may be triggered by VDR dependent alterations in chromatin (e.g. through interactions with nuclear proteins and altering transcription of chromatin modifying/remodelling genes; reviewed in (24)).

The mechanism of vitamin D effects on DNA methylation might be further explored through an understanding of other micronutrients and their impacts on DNA methylation. Vitamin E has anti‐ oxidative and anti‐inflammatory effects. NF‐κB is a pro‐inflammatory transcription factor that is activated by reactive oxygen species (ROS). Given that NF‐κB is known to stimulate Dnmt1 expression (68), inhibition of ROS by vitamin E may potentially alter DNA methylation via this mechanism.

206

APPENDIX FOUR REVIEW OF VITAMIN D AND DNA METHYLATION

FUTURE RESEARCH

The disparate findings relating to the methylomic effects of vitamin D attests to the complexity of epigenetic responses to environmental stimuli. Future studies should take into account the cell specific nature of responses to vitamin D, the specific form of vitamin D used, duration of exposure and assay method. This is likely to be aided by improvements in the technology available to assay

DNA methylation and its associated costs. Interindividual differences in DNA methylation in response to vitamin D would also be important to consider. A more complete knowledge of the effects on

DNA methylation is likely to be gained by concurrent examination of other epigenetic marks and their coordinated responses to vitamin D. Understanding these various factors will allow personalised and targeted use of vitamin D in the prevention and treatment of chronic disease.

CONCLUSIONS

Vitamin D has important effects on the methylome, that likely mediates risk in several chronic diseases. The research to date suggests specific effects that vary by organism, cell type and duration of supplementation/exposure. The mechanisms by which vitamin D acts on the methylome are yet to be fully elucidated, but studies to date indicate the importance of targeted use of vitamin D in the treatment and prevention of disease.

Competing interests

The authors declare that they have no competing interests.

207

APPENDIX FOUR REVIEW OF VITAMIN D AND DNA METHYLATION

Funding

LO received support from a co‐funded NHMRC/Multiple Sclerosis Research Australia/Trish MS

Foundation scholarship and NSW Health Pathology Postgraduate Fellowship.

208

APPENDIX FOUR REVIEW OF VITAMIN D AND DNA METHYLATION

REFERENCES

1. Alter M, Kahana E, Loewenson R. Migration and risk of multiple sclerosis. Neurology. 1978;28(11):1089‐. 2. Ahlgren C, Lycke J, Odén A, Andersen O. High risk of MS in Iranian immigrants in Gothenburg, Sweden. Multiple sclerosis journal. 2010;16(9):1079‐82. 3. Ahlgren C, Odén A, Lycke J. A nationwide survey of the prevalence of multiple sclerosis in immigrant populations of Sweden. Multiple Sclerosis Journal. 2012;18(8):1099‐107. 4. Kurtzke JF, Beebe GW, Norman JE. Epidemiology of multiple sclerosis in US veterans III. Migration and the risk of MIS. Neurology. 1985;35(5):672‐. 5. Gale CR, Martyn CN. Migrant studies in multiple sclerosis. Progress in neurobiology. 1995;47(4‐5):425‐48. 6. Jagannath VA, Filippini G, Di Pietrantonj C, Asokan GV, Robak EW, Whamond L, et al. Vitamin D for the management of multiple sclerosis. Cochrane Database of Systematic Reviews. 2018(9). 7. McLaughlin L, Clarke L, Khalilidehkordi E, Butzkueven H, Taylor B, Broadley SA. Vitamin D for the treatment of multiple sclerosis: a meta‐analysis. Journal of Neurology. 2018;265(12):2893‐905. 8. Alisch RS, Barwick BG, Chopra P, Myrick LK, Satten GA, Conneely KN, et al. Age‐associated DNA methylation in pediatric populations. Genome research. 2012:gr. 125187.111. 9. Calkins K, Devaskar SU. Fetal origins of adult disease. Current problems in pediatric and adolescent health care. 2011;41(6):158‐76. 10. Heindel JJ, Vandenberg LN. Developmental origins of health and disease: a paradigm for understanding disease etiology and prevention. Current opinion in pediatrics. 2015;27(2):248. 11. Holick MF, Chen TC, Lu Z, Sauter E. Vitamin D and Skin Physiology: AD‐Lightful Story. Journal of Bone and Mineral Research. 2007;22(S2):V28‐V33. 12. Shahijanian F, Parnell GP, McKay FC, Gatt PN, Shojoei M, O'Connor KS, et al. The CYP27B1 variant associated with an increased risk of autoimmune disease is underexpressed in tolerizing dendritic cells. Human molecular genetics. 2014;23(6):1425‐34. 13. Wu M, Rinchik EM, Wilkinson E, Johnson DK. Inherited somatic mosaicism caused by an intracisternal A particle insertion in the mouse tyrosinase gene. Proceedings of the National Academy of Sciences. 1997;94(3):890‐4. 14. Ukai H, Ishii‐Oba H, Ukai‐Tadenuma M, Ogiu T, Tsuji H. Formation of an active form of the interleukin‐2/15 receptor β‐chain by insertion of the intracisternal A particle in a radiation‐induced mouse thymic lymphoma and its role in tumorigenesis. Molecular Carcinogenesis: Published in cooperation with the University of Texas MD Anderson Cancer Center. 2003;37(2):110‐9. 15. Wade PA. Methyl CpG‐binding proteins and transcriptional repression. Bioessays. 2001;23(12):1131‐7. 16. Chedin F. The DNMT3 family of mammalian de novo DNA methyltransferases. Progress in molecular biology and translational science. 101: Elsevier; 2011. p. 255‐85. 17. Bell JT, Pai AA, Pickrell JK, Gaffney DJ, Pique‐Regi R, Degner JF, et al. DNA methylation patterns associate with genetic and gene expression variation in HapMap cell lines. Genome Biology. 2011;12(1):R10. 18. Jones PA. Functions of DNA methylation: islands, start sites, gene bodies and beyond. Nature Reviews Genetics. 2012;13(7):484. 19. Nan X, Meehan RR, Bird A. Dissection of the methyl‐CpG binding domain from the chromosomal protein MeCP2. Nucleic acids research. 1993;21(21):4886‐92. 20. Ohki I, Shimotake N, Fujita N, Jee J‐G, Ikegami T, Nakao M, et al. Solution structure of the methyl‐CpG binding domain of human MBD1 in complex with methylated DNA. Cell. 2001;105(4):487‐97. 21. Du Q, Luu P‐L, Stirzaker C, Clark SJ. Methyl‐CpG‐binding domain proteins: readers of the epigenome. Epigenomics. 2015;7(6):1051‐73.

209

APPENDIX FOUR REVIEW OF VITAMIN D AND DNA METHYLATION

22. Chen P‐Y, Feng S, Joo JWJ, Jacobsen SE, Pellegrini M. A comparative analysis of DNA methylation across human embryonic stem cell lines. Genome Biology. 2011;12(7):R62. 23. Fetahu IS, Hobaus J, Kallay E. Vitamin D and the epigenome. Front Physiol. 2014;5:164. 24. Carlberg C. Molecular endocrinology of vitamin D on the epigenome level. Molecular and cellular endocrinology. 2017;453:14‐21. 25. Xue J, Schoenrock SA, Valdar W, Tarantino LM, Ideraabdullah FY. Maternal vitamin D depletion alters DNA methylation at imprinted loci in multiple generations. Clinical epigenetics. 2016;8(1):107. 26. Xue J, Gharaibeh RZ, Pietryk EW, Brouwer C, Tarantino LM, Valdar W, et al. Impact of vitamin D depletion during development on mouse sperm DNA methylation. Epigenetics. 2018;13(9):959‐74. 27. Meems LM, Mahmud H, Buikema H, Tost J, Michel S, Takens J, et al. Parental vitamin D deficiency during pregnancy is associated with increased blood pressure in offspring via Panx1 hypermethylation. American Journal of Physiology‐Heart and Circulatory Physiology. 2016;311(6):H1459‐H69. 28. Wen J, Hong Q, Wang X, Zhu L, Wu T, Xu P, et al. The effect of maternal vitamin D deficiency during pregnancy on body fat and adipogenesis in rat offspring. Scientific reports. 2018;8(1):1‐8. 29. Zhang H, Chu X, Huang Y, Li G, Wang Y, Li Y, et al. Maternal vitamin D deficiency during pregnancy results in insulin resistance in rat offspring, which is associated with inflammation and Ikappabalpha methylation. Diabetologia. 2014;57(10):2165‐72. 30. Neelon SB, White AJ, Vidal AC, Schildkraut JM, Murtha AP, Murphy SK, et al. Maternal vitamin D, DNA methylation at imprint regulatory regions and offspring weight at birth, 1 year and 3 years. International Journal of Obesity. 2018;42(4):587‐93. 31. Suderman M, Stene LC, Bohlin J, Page C, Holvik K, Parr CL, et al. 25‐Hydroxyvitamin D in pregnancy and genome wide cord blood DNA methylation in two pregnancy cohorts (MoBa and ALSPAC). The Journal of steroid biochemistry and molecular biology. 2016;159:102‐9. 32. Mozhui K, Smith AK, Tylavsky FA. Ancestry dependent DNA methylation and influence of maternal nutrition. PloS one. 2015;10(3). 33. Anderson CM, Gillespie SL, Thiele DK, Ralph JL, Ohm JE. Effects of Maternal Vitamin D Supplementation on the Maternal and Infant Epigenome. Breastfeed Med. 2018;13(5):371‐80. 34. Curtis EM, Krstic N, Cook E, D'angelo S, Crozier SR, Moon RJ, et al. Gestational vitamin D supplementation leads to reduced perinatal RXRA DNA methylation: results from the MAVIDOS trial. Journal of Bone and Mineral Research. 2019;34(2):231‐40. 35. Zeitelhofer M, Adzemovic MZ, Gomez‐Cabrero D, Bergman P, Hochmeister S, N'diaye M, et al. Functional genomics analysis of vitamin D effects on CD4+ T cells in vivo in experimental autoimmune encephalomyelitis. Proceedings of the National Academy of Sciences. 2017;114(9):E1678‐E87. 36. Moore JR, Hubler SL, Nelson CD, Nashold FE, Spanier JA, Hayes CE. 1, 25‐Dihydroxyvitamin D3 increases the methionine cycle, CD4+ T cell DNA methylation and Helios+ Foxp3+ T regulatory cells to reverse autoimmune neurodegenerative disease. Journal of neuroimmunology. 2018;324:100‐14. 37. Jiao X, Wang L, Wei Z, Liu B, Liu X, Yu X. Vitamin D deficiency during pregnancy affects the function of Th1/Th2 cells and methylation of IFN‐gamma gene in offspring rats. Immunol Lett. 2019;212:98‐105. 38. Zhu H, Wang X, Shi H, Su S, Harshfield GA, Gutin B, et al. A genome‐wide methylation study of severe vitamin D deficiency in African American adolescents. The Journal of pediatrics. 2013;162(5):1004‐9. e1. 39. Zhu H, Bhagatwala J, Huang Y, Pollock NK, Parikh S, Raed A, et al. Race/ethnicity‐specific association of vitamin D and global DNA methylation: cross‐sectional and interventional findings. PloS one. 2016;11(4).

210

APPENDIX FOUR REVIEW OF VITAMIN D AND DNA METHYLATION

40. Junge KM, Bauer T, Geissler S, Hirche F, Thürmann L, Bauer M, et al. Increased vitamin D levels at birth and in early infancy increase offspring allergy risk—evidence for involvement of epigenetic mechanisms. Journal of Allergy and Clinical Immunology. 2016;137(2):610‐3. 41. Cho HJ, Sheen YH, Kang MJ, Lee SH, Lee SY, Yoon J, et al. Prenatal 25‐hydroxyvitamin D deficiency affects development of atopic dermatitis via DNA methylation. J Allergy Clin Immunol. 2019;143(3):1215‐8. 42. Florath I, Schottker B, Butterbach K, Bewerunge‐Hudler M, Brenner H. Epigenome‐wide search for association of serum 25‐hydroxyvitamin D concentration with leukocyte DNA methylation in a large cohort of older men. Epigenomics. 2016;8(4):487‐99. 43. Fujii R, Yamada H, Munetsuna E, Yamazaki M, Ando Y, Mizuno G, et al. Associations between dietary vitamin intake, ABCA1 gene promoter DNA methylation, and lipid profiles in a Japanese population. Am J Clin Nutr. 2019;110(5):1213‐9. 44. Valencia RAC, Martino DJ, Saffery R, Ellis JA. In vitro exposure of human blood mononuclear cells to active vitamin D does not induce substantial change to DNA methylation on a genome‐scale. The Journal of steroid biochemistry and molecular biology. 2014;141:144‐9. 45. Ong L, Schibeci S, Fewings N, Booth D, Parnell G. Age‐dependent VDR peak DNA methylation as a mechanism for latitude‐dependent MS risk. bioRxiv. 2020. 46. Horvath S. DNA methylation age of human tissues and cell types. Genome biology. 2013;14(10):3156. 47. Chen L, Dong Y, Bhagatwala J, Raed A, Huang Y, Zhu H. Effects of vitamin D3 supplementation on epigenetic aging in overweight and obese African Americans with suboptimal vitamin D status: a randomized clinical trial. The Journals of Gerontology: Series A. 2019;74(1):91‐8. 48. Gao X, Zhang Y, Schottker B, Brenner H. Vitamin D status and epigenetic‐based mortality risk score: strong independent and joint prediction of all‐cause mortality in a population‐based cohort study. Clin Epigenetics. 2018;10:84. 49. Chen L, Wagner CL, Dong Y, Wang X, Shary JR, Huang Y, et al. Effects of Maternal Vitamin D3 Supplementation on Offspring Epigenetic Clock of Gestational Age at Birth: A Post‐hoc Analysis of a Randomized Controlled Trial. Epigenetics. 2020:1‐11. 50. Nair‐Shalliker V, Dhillon V, Clements M, Armstrong BK, Fenech M. The association between personal sun exposure, serum vitamin D and global methylation in human lymphocytes in a population of healthy adults in South Australia. Mutation Research/Fundamental and Molecular Mechanisms of Mutagenesis. 2014;765:6‐10. 51. Hübner U, Geisel J, Kirsch SH, Kruse V, Bodis M, Klein C, et al. Effect of 1 year B and D vitamin supplementation on LINE‐1 repetitive element methylation in older subjects. Clinical chemistry and laboratory medicine. 2013;51(3):649‐55. 52. Ong L, Schibeci S, Fewings N, Booth D, Parnell G. LINE‐1 DNA methylation in response aging and vitamin D. bioRxiv. 2020. 53. Castellano‐Castillo D, Morcillo S, Crujeiras AB, Sánchez‐Alcoholado L, Clemente‐Postigo M, Torres E, et al. Association between serum 25‐hydroxyvitamin D and global DNA methylation in visceral adipose tissue from colorectal cancer patients. BMC cancer. 2019;19(1):93. 54. Tapp HS, Commane DM, Bradburn DM, Arasaradnam R, Mathers JC, Johnson IT, et al. Nutritional factors and gender influence age‐related DNA methylation in the human rectal mucosa. Aging cell. 2013;12(1):148‐55. 55. McCullough ML, Zoltick ES, Weinstein SJ, Fedirko V, Wang M, Cook NR, et al. Circulating Vitamin D and Colorectal Cancer Risk: An International Pooling Project of 17 Cohorts. JNCI: Journal of the National Cancer Institute. 2018;111(2):158‐69. 56. Lopes N, Carvalho J, Duraes C, Sousa B, Gomes M, Costa JL, et al. 1Alpha, 25‐ dihydroxyvitamin D3 induces de novo E‐cadherin expression in triple‐negative breast cancer cells by CDH1‐promoter demethylation. Anticancer research. 2012;32(1):249‐57.

211

APPENDIX FOUR REVIEW OF VITAMIN D AND DNA METHYLATION

57. Vanoirbeek E, Eelen G, Verlinden L, Carmeliet G, Mathieu C, Bouillon R, et al. PDLIM2 expression is driven by vitamin D and is involved in the pro‐adhesion, and anti‐migration and‐ invasion activity of vitamin D. Oncogene. 2014;33(15):1904‐11. 58. Stefanska B, Salamé P, Bednarek A, Fabianowska‐Majewska K. Comparative effects of retinoic acid, vitamin D and resveratrol alone and in combination with adenosine analogues on methylation and expression of phosphatase and tensin homologue tumour suppressor gene in breast cancer cells. British Journal of Nutrition. 2012;107(6):781‐90. 59. Doig CL, Singh PK, Dhiman VK, Thorne JL, Battaglia S, Sobolewski M, et al. Recruitment of NCOR1 to VDR target genes is enhanced in prostate cancer cells and associates with altered DNA methylation patterns. Carcinogenesis. 2013;34(2):248‐56. 60. Lai GR, Lee YF, Yan SJ, Ting HJ. Active vitamin D induced gene‐specific hypomethylation in prostate cancer cells developing vitamin D resistance. Am J Physiol Cell Physiol. 2020. 61. Rawson JB, Sun Z, Dicks E, Daftary D, Parfrey PS, Green RC, et al. Vitamin D intake is negatively associated with promoter methylation of the Wnt antagonist gene DKK1 in a large group of colorectal cancer patients. Nutrition and cancer. 2012;64(7):919‐28. 62. Seuter S, Neme A, Carlberg C. Epigenome‐wide effects of vitamin D and their impact on the transcriptome of human monocytes involve CTCF. Nucleic acids research. 2016;44(9):4090‐104. 63. Bock C, Beerman I, Lien WH, Smith ZD, Gu H, Boyle P, et al. DNA methylation dynamics during in vivo differentiation of blood and skin stem cells. Mol Cell. 2012;47(4):633‐47. 64. Cavalli G, Heard E. Advances in epigenetics link genetics to the environment and disease. Nature. 2019;571(7766):489‐99. 65. Booth D, Ding N, Parnell G, Shahijanian F, Coulter S, Schibeci S, et al. Cistromic and genetic evidence that the vitamin D receptor mediates susceptibility to latitude‐dependent autoimmune diseases. Genes & Immunity. 2016;17(4):213‐9. 66. Baubec T, Colombo DF, Wirbelauer C, Schmidt J, Burger L, Krebs AR, et al. Genomic profiling of DNA methyltransferases reveals a role for DNMT3B in genic methylation. Nature. 2015;520(7546):243‐7. 67. Weinberg DN, Papillon‐Cavanagh S, Chen H, Yue Y, Chen X, Rajagopalan KN, et al. The histone mark H3K36me2 recruits DNMT3A and shapes the intergenic DNA methylation landscape. Nature. 2019;573(7773):281‐6. 68. Kim AY, Park YJ, Pan X, Shin KC, Kwak SH, Bassas AF, et al. Obesity‐induced DNA hypermethylation of the adiponectin gene mediates insulin resistance. Nat Commun. 2015;6:7585.

62. Neelon SEB, White AJ, Vidal AC, Schildkraut JM, Murtha AP, Murphy SK, et al. Maternal vitamin D, DNA methylation at imprint regulatory regions, and offspring weight at birth, 1 year, and 3 years. International journal of obesity (2005). 2017. 63. O’Brien KM, Sandler DP, Xu Z, Kinyamu HK, Taylor JA, Weinberg CR. Vitamin D, DNA methylation, and breast cancer. Breast Cancer Research. 2018;20(1):70. 64. Seuter S, Neme A, Carlberg C. Epigenome‐wide effects of vitamin D and their impact on the transcriptome of human monocytes involve CTCF. Nucleic acids research. 2016;44(9):4090‐104. 65. Bock C, Beerman I, Lien WH, Smith ZD, Gu H, Boyle P, et al. DNA methylation dynamics during in vivo differentiation of blood and skin stem cells. Mol Cell. 2012;47(4):633‐47. 66. Cavalli G, Heard E. Advances in epigenetics link genetics to the environment and disease. Nature. 2019;571(7766):489‐99. 67. Booth D, Ding N, Parnell G, Shahijanian F, Coulter S, Schibeci S, et al. Cistromic and genetic evidence that the vitamin D receptor mediates susceptibility to latitude‐dependent autoimmune diseases. Genes & Immunity. 2016;17(4):213‐9. 68. Baubec T, Colombo DF, Wirbelauer C, Schmidt J, Burger L, Krebs AR, et al. Genomic profiling of DNA methyltransferases reveals a role for DNMT3B in genic methylation. Nature. 2015;520(7546):243‐7.

212

APPENDIX FOUR REVIEW OF VITAMIN D AND DNA METHYLATION

69. Weinberg DN, Papillon‐Cavanagh S, Chen H, Yue Y, Chen X, Rajagopalan KN, et al. The histone mark H3K36me2 recruits DNMT3A and shapes the intergenic DNA methylation landscape. Nature. 2019;573(7773):281‐6. 70. Kim AY, Park YJ, Pan X, Shin KC, Kwak SH, Bassas AF, et al. Obesity‐induced DNA hypermethylation of the adiponectin gene mediates insulin resistance. Nat Commun. 2015;6:7585.

213

APPENDIX FOUR REVIEW OF VITAMIN D AND DNA METHYLATION

TABLES

Table 1. Summary of animal studies examining the effects of vitamin D on DNA methylation

Global DNA Vitamin D Local DNA Methylation Species Author Year Methylation Assay Cell Type Methylation with Supplemented/Assayed with ↑Vitamin D ↑Vitamin D

Bisulfite PCR/Sanger Zhang, H. et al. Rat 2014 sequencing/ Calcidiol Hepatocyte Lower Iκbα (first exon) ‐ lower (29) Methylflash

G1/cross 2: IG‐DMR ‐ higher. G2/cross 2: IG‐ Xue, J. et al. Mouse 2016 Pyrosequencing Calcidiol Sperm N/A DMR ‐ higher. Combined (25) G2 cross 1 and cross 2 IG‐ DMR ‐ higher

G1/cross 1: H19/Igf2 Cbs2, Grb10DMR ‐ higher; G1/cross 2: H19/Igf2 Cbs2 ‐ lower. Combined G1 cross 1 and Liver N/A cross 2: H19/Igf2 Cbs2 ‐ higher. G2/cross 2: H19/Igf2 Cbs4 ‐ higher. Combined G2 cross 1 and cross 2: H19/Igf2 Cbs4‐ higher

214

APPENDIX FOUR REVIEW OF VITAMIN D AND DNA METHYLATION

Differentially methylated genes (direction not stated) ‐ Dqx1, Gyltl1b, Meems, L. et Methylation Vitamin D (not Rat 2016 Kidney N/A LOC499742, Otop2, al. (27) array/Pyrosequencing specified)/Calcidiol RGD1562608, Zbtb7b, Capn2, Bcl2, Adh7. Panx1 ‐ lower Zeitelhofer, M. 491 DMRs ‐ 410 lower, 81 Rat 2017 MBD‐seq/CHARM Cholecalciferol CD4+ T cell Lower et al. (35) higher 15827 CpG differentially methylated. 69% of sites higher. Differences Target enriched next localised to Xue, J. et al. Mouse 2018 generation Cholecalciferol Sperm Higher developmental and (26) sequencing metabolic genes with enrichment for Cadherin, Wnt, PDGF and Integrin signaling pathways Moore, J. et al. Mouse 2018 Methylflash Calcitriol/Cholecalciferol CD4+ T cells Higher N/A (36) Differentially methylated CpGs corresponded to 608 promoters, 204 CpG islands and 305 genes. Wen, J. et al. Methylation array/ Vitamin D (not Direction not specified. Rat 2018 Adipocyte N/A (28) Mass array specified)/Calcidiol 141 differentially methylated genes also differentially expressed. Vldlr methylation ‐ lower, Hif1α ‐ higher

215

APPENDIX FOUR REVIEW OF VITAMIN D AND DNA METHYLATION

Jiao, X. et al. Whole Rat 2019 Mass array Cholecalciferol N/A IFNγ ‐ lower (37) blood

CBS ‐ CTCF binding site, CHARM – comprehensive high throughput arrays for relative methylation, DMR – differentially methylated region, MBD – methyl‐ CpG‐binding domain

216

APPENDIX FOUR REVIEW OF VITAMIN D AND DNA METHYLATION

Table 2. A summary of studies examining the effects of vitamin D on human DNA methylation

Global DNA Vitamin D Local DNA Methylation Methylation Study Type Author Year Supplemented/ Cell Type Methylation with Assay with Assayed ↑Vitamin D ↑Vitamin D Zhu, H. et al. Methylation MAPRE2 promoter ‐ Observational 2013 Calcidiol Leukocyte N/A (38) array higher, DIO3 ‐ lower LINE‐1 ‐ higher; Tapp, H. et al. MethylPCR/RT‐ Higher (LINE‐ Observational 2013 Calcidiol Rectal MYOD, PCA1, APC ‐ (54) PCR 1) lower Hubner, U. et Vitamin D (not Whole Interventional 2013 Pyrosequencing N/A LINE‐1 ‐ nil al. (51) specified) blood Nair‐Shalliker, Lymphocyt Observational 2014 Pyrosequencing Calcidiol N/A LINE‐1 ‐ nil V. et al. (50) e Chavez Interventional Methylation Valencia, R. et 2014 Calcitriol PBMC Nil Nil (in vitro) array al. (44) Mozhui, K., Smith, A., Methylation Observational 2015 Calcidiol Leukocyte Nil Nil Tylavsky, F. array (32) Observational/ Zhu, H. et al. 2016 Direct ELISA Calcidiol Leukocyte Higher N/A Interventional (39) Suderman, M. Methylation Observational 2016 Calcidiol Cord blood Nil Nil et al. (31) array Mostly decreased ‐ 311/508 differentially Junge, K. et al. WGBS/Mass Obervational 2016 Calcidiol Cord blood N/A methylated CpGs; (40) array TSLP enhancer ‐ lower

217

APPENDIX FOUR REVIEW OF VITAMIN D AND DNA METHYLATION

Florath, I. et Methylation Observational 2016 Calcidiol Leukocyte N/A Nil al. (42) array Neelon, S. et Observational 2017 Pyrosequencing Calcidiol Leukocyte N/A Nil al. (62) Differentially methylated CpGs: maternal at birth ‐ 76 lower, 89 higher; Anderson, C. Methylation maternal 4‐6 weeks Interventional 2018 Cholecalciferol Leukocyte N/A et al. (33) array post partum ‐ 102 lower, 200 higher. Paediatric 4‐6 weeks post partum ‐ 217 lower, 213 higher 23/198 candidate CpGs differentially methylated; CYP27B1 TSS1500 ‐ lower, GC gene body, RXRA gene body, NADSYN1 O'Brien, K. et Methylation Whole Observational 2018 Calcidiol N/A TSS200, DHCR7 gene al. (63) array blood body, DHCR7 TSS1500, GC 3' UTR ‐ higher, NADSYN1 gene body, DHCR7 3'UTR, RXRA gene body ‐ mixed No association between DNA Gao, X. et al. Methylation Whole Observational 2018 Calcidiol N/A methylation based (48) array blood mortality score (MS) and serum calcidiol

218

APPENDIX FOUR REVIEW OF VITAMIN D AND DNA METHYLATION

Chen, L. et al. Methylation Cholecalciferol/ Horvath methylation Interventional 2019 Leukocyte N/A (47) array Calcidiol age lower

Curtis, E. et al. Umbilical Interventional 2019 Pyrosequencing Cholecalciferol N/A RXRA ‐ lower (34) cord Higher (LINE‐ LINE‐1 ‐ higher 1 in Castellano‐ methylation in colorectal Observational Castillo, D. et 2019 Pyrosquencing Calcidiol Adipocyte colorectal cancer cancer al. (53) patients, but not patients controls only) Cho, H., Methylation MICAL3 ‐ higher, Observational Sheen, Y., 2019 Calcidiol Leukocyte N/A array RSPH10B2 ‐ lower Kang, M. (41) Fujii, R. et al. Vitamin D (not Observational 2019 Pyrosequencing Leukocyte N/A ABCA1 ‐ lower (43) specified) Bohlin and Knight Chen, L. et al. Methylation Interventional 2020 Cholecalciferol Leukocyte N/A methylation age (49) array acceleration lower Paediatric cells: GNAS ‐ higher, NEURL1B ‐ CD14+ Ong, L. et al. lower. Adult cells: Interventional Submitted WGBS Calcitriol myeloid N/A (45) autosomal genes or cells promoters ‐ 25 higher, 12 lower Manuscript CD14+ Ong, L. et al. Interventional in WGBS Calcitriol myeloid N/A LINE‐1 ‐ nil (52) preparation cells ELISA – enzyme linked immunosorbent assay, PBMC – peripheral blood mononuclear cell, TSS1500 – 1500bp upstream of transcription start site, TSS200 – 200bp upstream of transcription start site, WGBS – whole genome bisulfite sequencing

219

APPENDIX FOUR REVIEW OF VITAMIN D AND DNA METHYLATION

Table 3. A summary of studies examining human cancer cell DNA methylation in response to vitamin D

Global DNA Methylation Vitamin D Local DNA Methylation Study Type Author Year Cell Type Methylation with Assay Supplemented/Estimated with ↑Vitamin D ↑Vitamin D

Bisulfite Breast cancer Interventional Lopes, N. et al. 2012 PCR/sanger Calcitriol cell line (MDA‐ N/A CDH1 promoter ‐ lower (in vitro) (56) sequencing MB‐231)

Methylation Breast cancer Interventional Stefanska, B. et sensitive PTEN promoter 2012 Cholecalciferol cell line (MCF‐7, N/A (in vitro) al. (58) restriction methylation ‐ lower MDA‐MB‐231) analysis

DKK1 promoter/CGI ‐ Rawson, J. et al. Methyl/RT‐ Colorectal Observational 2012 Vitamin D (not specified) N/A lower, WNT5A (61) PCR cancer promoter/CGI ‐ lower

Cell, site and duration specific methylation Prostate cancer changes: RWPE‐1 cells Interventional Doig, C. et al. 2013 Mass array Calcitriol cell line (RWPE‐ N/A CDKN1A regulatory (in vitro) (59) 1, PC3) regions ‐ lower, PC3 cells CDKN1A regulatory regions ‐ mixed Bisulfite Interventional Vanoirbeek, E. Breast cancer 2014 PCR/sanger Calcitriol N/A PDLIM2 promoter ‐ lower (in vitro) et al. (57) cell line (MCF‐7) sequencing

220

APPENDIX FOUR REVIEW OF VITAMIN D AND DNA METHYLATION

Bisulfite several genes lower Interventional Prostate cancer Lai, G. et al. (60) 2020 PCR/sanger Calcitriol N/A including those linked to (in vitro) cell line (LNCaP) sequencing mTOR signaling pathway CGI – CpG island

221

APPENDIX FOUR REVIEW OF VITAMIN D AND DNA METHYLATION

Table 4. A summary of DNA methylation assays

Single Base Assesses Global Assesses Local

Method Resolution Methylation Methylation

WGBS Y Y Y

Methylation array Y Y Y

Pyrosequencing Y N Y

Target enriched next generation

sequencing Y N Y

MBD‐Seq N N Y

Direct ELISA N Y N

Methyl/RT‐PCR Y N* Y

Mass array Y N Y

Methylation sensitive restriction

analysis N N Y

Bisulfite PCR/Sanger Y N Y

*repeat elements may be used as a surrogate marker of global DNA methylation

222

APPENDIX FOUR REVIEW OF VITAMIN D AND DNA METHYLATION

FIGURES

Figure 1. The vitamin D metabolic pathway

223

APPENDIX FOUR REVIEW OF VITAMIN D AND DNA METHYLATION

Figure 2. The effects of DNA methylation on gene transcription A) Methyl‐CpG‐binding domain proteins such as MeCP1 bind methylated DNA and recruit transcriptional corepressors that participate in chromatin compaction and silence transcription. B) Methylated CpGs can also inhibit the binding of transcription factors and therefore transcription. MBD – methyl‐CpG‐binding domain protein. Filled circles indicate methylated CpGs, whereas open circles are unmethylated.

224

APPENDIX FIVE

LINE‐1 DNA METHYLATION IN RESPONSE TO AGING

AND VITAMIN D

(MANUSCRIPT IN PREPARATION)

APPENDIX FIVE LINE‐1 DNA METHYLATION, AGING AND VITAMIN D

LINE‐1 DNA methylation in response to aging and vitamin D

*Lawrence T. C. Ong1,2, Stephen D Schibeci1, Nicole L Fewings1, David R Booth1, Grant P Parnell1

1Centre for Immunology and Allergy Research, Westmead Institute for Medical Research, The

University of Sydney, 176 Hawkesbury Rd, Westmead, New South Wales 2145, Australia

2Department of Immunology, Westmead Hospital, Cnr Darcy and Hawkesbury Rds,

Westmead, New South Wales 2145, Australia

*Corresponding author

Contact Details

Lawrence T C Ong – [email protected]

Keywords – retrotransposon, DNA methylation, vitamin D, calcitriol, epigenetics

226

APPENDIX FIVE LINE‐1 DNA METHYLATION, AGING AND VITAMIN D

ABSTRACT

Retrotransposons are genetic elements capable of their own propagation and insertion into the human genome. Because of their mutagenic potential, retrotransposons are heavily suppressed by mechanisms including DNA methylation. Increased age is associated with decreasing DNA methylation of the LINE‐1 retrotransposon and may partially explain the predisposition towards malignancy with advancing age. Vitamin D has been investigated for its effects on DNA methylation at LINE‐1 elements with mixed results. This study hypothesised that LINE‐1 DNA methylation is altered by vitamin D exposure and age. Using whole genome bisulfite sequencing of adult and newborn haematopoietic progenitors cultured with or without vitamin D, DNA methylation at LINE‐1 elements was not found to vary between cells cultured with or without vitamin D both in adults and newborns. In contrast, several LINE‐1 regions were found to be differentially methylated between adults and children, but these were not uniformly hypermethylated in paediatric cells. The results of this study suggest that at least in haematopoietic cells, vitamin D does not appear to affect LINE‐1 methylation.

227

APPENDIX FIVE LINE‐1 DNA METHYLATION, AGING AND VITAMIN D

INTRODUCTION

Transposable elements encompass a diverse group of genetic elements that are capable of replicating and inserting copies of themselves into the genome. Retrotransposons form part of this family, capable of reinsertion through RNA intermediates. The family of retrotransposons includes long terminal repeat elements (LTRs), long interspersed nuclear elements (LINEs) and short interspersed nuclear elements (SINEs). In humans, they form a large proportion of our genetic material for example, LINE‐1 retrotransposons and Alu elements comprise 17‐25% and 11% of the human genome respectively (1, 2). Retrotransposons are generally thought to have deleterious effects due to their ability to induce malignancy through mutagenic insertions. Some inflammatory diseases are also thought to arise from transcription of retrotransposon elements that act as potent promoters for inflammatory genes (e.g. interferon‐responsive genes in Aicardi Goutieres syndrome), or through triggering of innate immunity via pattern recognition receptors (e.g. MDA5, Toll‐like receptors) during reactivation (3, 4). Retrotransposons like LINE‐1 also enable transposition of non‐ autonomous retroelements and the HERV‐W endogenous retrovirus subfamily (5) previously linked to multiple sclerosis.

DNA methylation involves the covalent addition of a methyl group to a cytosine residue. In general,

DNA methylation is associated with compacted chromatin and transcriptional repression.

Retrotransposons are routinely methylated to decrease the risk of unwanted transposition and mutagenesis (6, 7). However, DNA methylation at these sites is also known to decrease with aging

(8‐10), particularly at Alu elements (9), perhaps in part explaining susceptibility to malignancy with age.

228

APPENDIX FIVE LINE‐1 DNA METHYLATION, AGING AND VITAMIN D

Vitamin D is a hormone that has pleiotropic effects, with important roles in physiologic homeostasis, disease pathogenesis and organism development. It is an important mediator of risk in autoimmune and allergic diseases such as multiple sclerosis, type 1 diabetes and atopic dermatitis. In addition, it has been shown to have antiproliferative effects both in healthy and malignant cells (11). Several studies support the notion that the effects of vitamin D might be mediated at least in part through its effects on DNA methylation (see (12) for a review).

Studies examining the effect of vitamin D on retrotransposon DNA methylation have exclusively focused on LINE‐1 DNA methylation. In humans, studies on peripheral blood have not found any changes in LINE‐1 DNA methylation due to vitamin D. In one study, 50 individuals were supplemented with vitamin B and D, with no detectable changes in whole blood LINE‐1 methylation after one year (13). Another study on peripheral blood lymphocytes, did not find any association between serum calcidiol and LINE‐1 methylation (14). Studies on colorectal tissue have found higher serum vitamin D levels to be associated with increased LINE‐1 methylation (15), perhaps explaining its effect on decreasing colorectal cancer risk. A study on visceral adipose tissue found a positive correlation between vitamin D levels and LINE‐1 DNA methylation in colorectal cancer patients, but not controls (16).

In the present study, it was hypothesised that cell culture of haematopoietic progenitor cells with vitamin D, would result in changes in LINE‐1 DNA methylation. Given the known effects of age on

DNA methylation, it was also hypothesised that LINE‐1 DNA methylation would be increased in cells derived from children compared to those from adults.

229

APPENDIX FIVE LINE‐1 DNA METHYLATION, AGING AND VITAMIN D

METHODS

Cell isolation & culture

The methods have been previously described elsewhere (17). In brief, positive selection of CD34+ haematopoietic progenitors was performed on peripheral blood from two healthy adult male platelet donors (27 and 47 years old) and two umbilical cord blood units from one male and one female donor. Purified cells were plated at a density of 5x104 cells/ml of media. The culture cocktail contained X‐VIVO10 (Lonza), Albumex (Seqirus) 0.05%, SCF (Peprotech) 200ng/ml, GM‐CSF

(Peprotech) 0.03ug/ml, M‐CSF (premium grade; Miltenyi Biotec) 5000U/ml, IL‐6 (Peprotech)

10ng/ml, FLT3 ligand (Peprotech) 50ng/ml and Gentamicin (Sigma Aldrich) 50µg/ml. Calcitriol

(1,25(OH)2Vitamin D3; BioGems) was added at a physiological concentration of 0.1nM. Cells were

o 5 incubated at 37 C with 5% CO2 for one week before replating at a density of 1x10 cells/ml of media.

Media was then changed every third day by demi‐depletion. Cells were harvested on day 21 and

FACS sorting was performed to obtain purified CD14+CD45+ mononuclear phagocytes.

Whole genome bisulfite sequencing, data QC, alignment and processing

Whole genome bisulfite sequencing libraries were generated with the Accel‐NGS Methyl‐seq DNA

Library Kit (Swift Biosciences) and sequenced on a HiSeq X10 (Illumina) in 150bp PE mode with PhiX spike‐in to counteract low sequence diversity. The quality of raw sequences was ascertained using

FastQC (18). Quality trimming was carried out using Trim galore (19) in paired end mode with the following parameters ‐quality 20, ‐‐three_prime_clip_R1 5, ‐‐clip_R2 15. Reads were aligned to the hg19 genome using the Wildcard Alignment Tool (WALT) before further processing with Methpipe

(20). Methylation calls were made using methcounts (using the ‐n option for CpG context cytosines only) before symmetric‐cpgs was used to extract and merge methylation data from both strands.

Merge‐methcounts was used to merge locus specific read counts. Regional methylation analysis was

230

APPENDIX FIVE LINE‐1 DNA METHYLATION, AGING AND VITAMIN D

performed using the roimethstat module (with ‐P and ‐v options), to determine methylation state within a prespecified region of interest. Sex chromosomes were omitted from analyses. Unless otherwise specified, analyses were performed on filtered data with an average read count of > 10 reads per covered CpG for the entire region. Comparisons were made on regions where both datasets met the filtering criteria. Global estimates of LINE‐1 DNA methylation were determined by dividing methylated reads by total reads.

Differential methylation and expression analysis

RADmeth (21) was utilised for differential methylation analysis. Unfiltered, CpG wise methylation counts from annotated LINE‐1 regions were used as input. The ‐bins parameter was set to 1:200:1 and a CpG‐wise FDR‐adjusted significance of 0.05 was used to determine differentially methylated sites. The effects of calcitriol and age were considered separately by comparing the effects of calcitriol amongst cells of adult and paediatric origin separately. LINE‐1 annotations were derived from (22). The intersection of LINE‐1 regions, annotated CD14+ promoter regions (23) and hg19 genes was determined with BEDtools intersect and closest (24) respectively. Details of differential expression analysis can be found in (17).

RESULTS

On average, 96% of genome wide CpGs were covered at a depth of 16x. Approximately 65% of all annotated LINE‐1 regions were covered (see Table 1). We had previously shown that across all genome wide CpGs, the average methylation was 0.83 irrespective of age or vitamin D.

Unsurprisingly, LINE‐1 regions demonstrated higher DNA methylation levels than those estimated at all genome wide CpGs and was slightly higher in paediatric than adult cells (0.89 vs 0.88; see Table 2 for descriptive statistics).

231

APPENDIX FIVE LINE‐1 DNA METHYLATION, AGING AND VITAMIN D

There were no differentially methylated CpGs when comparing adult CD14+ with or without calcitriol. Similarly, no CpGs were found to be differentially methylated amongst paediatric CD14+ in the presence of absence of calcitriol. Interestingly, when comparing adult and paediatric data, there were 12450 differentially methylated LINE‐1 CpGs corresponding to 5356 different LINE‐1 regions

(Supplementary Data 1 & 2). Of the differentially methylated CpGs, DNA methylation at 5160 CpGs was lower in cells of paediatric origin and higher in 7290.

Differentially methylated LINE‐1 regions were further examined by determining their overlap with annotated CD14+ promoter regions and their corresponding hg19 genes. This yielded 312 genes

(Supplementary Data 3) of which 46 were also differentially expressed (Supplementary Data 4).

Several of these differentially methylated and differentially expressed genes have immunoregulatory functions.

DISCUSSION

Vitamin D has broad ranging physiological effects, of which a subset may be mediated through DNA methylation. This study explored the effects of vitamin D and age on LINE‐1 DNA methylation.

Consistent with previous studies on haematopoietic cells, there was no effect of vitamin D on LINE‐1 methylation. In contrast, age appeared to be associated with changes in almost 11% of LINE‐1 regions passing filter. The differentially methylated CpGs varied with regard to the direction of differences. Although most loci were more highly methylated in cells of paediatric origin, a large proportion (41%) were less methylated. Thus, LINE‐1 methylation is not only locus specific, but also variable with regard to the direction and magnitude of differences across age.

232

APPENDIX FIVE LINE‐1 DNA METHYLATION, AGING AND VITAMIN D

The results of this study suggest that vitamin D does not affect DNA methylation at LINE‐1 sites in

CD14+ myeloid cells. This was consistent with the two previously identified studies on haematopoietic cells. These results may be reconciled with those examining tissue LINE‐1 methylation, by the fact that DNA methylation in different cell types appears to vary in its susceptibility to vitamin D induced alterations. These results suggest that at least in myeloid cells, vitamin D may not act through LINE‐1 DNA methylation to alter genomic stability or propensity to disease.

With respect to age related differences in LINE‐1 DNA methylation, the overall magnitude of these differences at the global level was small. Previous studies examining age and LINE‐1 methylation have studied adults. To date, the trajectory of global LINE‐1 methylation loss has not been determined and this study suggests that changes may be smaller in childhood. Of particular interest however, was that the direction of differences varied between LINE‐1 regions. This may point to the need for more targeted assessment of LINE‐1 DNA methylation, rather than assuming that changes in their methylation are unidirectional across loci. Age related differences in LINE‐1 DNA methylation also occurred at different gene promoters, a subset of which were associated with changes in gene expression. This may indicate a regulatory role for some of these LINE‐1 regions.

One of the strengths of this study was that we utilised a number of strategies to increase the likelihood of detecting DNA methylation changes. DNA methylation is known to be more plastic in early life (25), and susceptible to environmental stimuli during differentiation (26). The comparison of cells originating from adults and newborns, and differentiation of CD34+ haematopoietic progenitors allowed us to take advantage of these previous findings. Secondly, we examined a single subset of leukocytes, decreasing the likelihood of DNA methylation changes being obscured by

233

APPENDIX FIVE LINE‐1 DNA METHYLATION, AGING AND VITAMIN D

heterogeneous subsets. Finally, by performing an in vitro study, the potential for confounding variables between groups was controlled for compared with the previously reported ex vivo studies.

Perhaps adversely influencing the results of this study were the use of cultured cells, whose DNA methylation has previously been shown to increase with passage number (27). This may have obscured differences in DNA methylation due to vitamin D. In addition, the use of these cultured cells may not reflect DNA methylation states in primary cells. Finally, whole‐genome bisulfite sequencing captured only 65% of LINE‐1 regions and may not be reflective of DNA methylation changes in the remaining 35% of LINE‐1 regions.

The results of this study reiterate the lack of vitamin D effects on LINE‐1 methylation in haematopoietic cells and the likely differential effects of vitamin D in different tissues. Age related differences in LINE‐1 DNA methylation were found, although the relationship is not as clear as that portrayed by global assessments of LINE‐1 DNA methylation. Future studies might consider the rate of change in LINE‐1 DNA methylation across the life span and locus specific methylation to better elucidate the functional consequences of age related differences.

Declarations

Ethics approval and consent to participate

This study received ethics approval from the Western Sydney Local Health District Human Research

Ethics Committee (HREC2002/9/3.6(1425) & (5366) AU RED LNR/17/WMEAD/447).

Consent for publication

Not applicable

234

APPENDIX FIVE LINE‐1 DNA METHYLATION, AGING AND VITAMIN D

Availability of data and materials

The datasets used and/or analysed during the current study are available from the corresponding author on reasonable request.

Competing Interests

The authors declare that they have no competing interests.

Funding

Authors’ Contributions

Acknowledgements

The authors would like to acknowledge Australian Red Cross Blood Services, the Sydney Cord Blood

Bank and donors for providing samples for this study. Flow cytometry was performed at the Flow

Cytometry Core Facility supported by the Westmead Research Hub, Cancer Institute NSW and

235

APPENDIX FIVE LINE‐1 DNA METHYLATION, AGING AND VITAMIN D

NHMRC. Bioinformatic analysis was supported by Sydney Informatics Hub, funded by the University of Sydney.

236

APPENDIX FIVE LINE‐1 DNA METHYLATION, AGING AND VITAMIN D

REFERENCES

1. Kazazian HH, Jr. Mobile elements: drivers of genome evolution. Science. 2004;303(5664):1626‐32. 2. Richardson SR, Doucet AJ, Kopera HC, Moldovan JB, Garcia‐Perez JL, Moran JV. The influence of LINE‐1 and SINE retrotransposons on mammalian genomes. Mobile DNA III. 2015:1165‐208. 3. Ahmad S, Mu X, Yang F, Greenwald E, Park JW, Jacob E, et al. Breaching self‐tolerance to Alu duplex RNA underlies MDA5‐mediated inflammation. Cell. 2018;172(4):797‐810. e13. 4. Perron H, Lang A. The human endogenous retrovirus link between genes and environment in multiple sclerosis and in multifactorial diseases associating neuroinflammation. Clinical reviews in allergy & immunology. 2010;39(1):51‐61. 5. Grandi N, Tramontano E. Type W human endogenous retrovirus (HERV‐W) integrations and their mobilization by L1 machinery: contribution to the human transcriptome and impact on the host physiopathology. Viruses. 2017;9(7):162. 6. Wu M, Rinchik EM, Wilkinson E, Johnson DK. Inherited somatic mosaicism caused by an intracisternal A particle insertion in the mouse tyrosinase gene. Proceedings of the National Academy of Sciences. 1997;94(3):890‐4. 7. Ukai H, Ishii‐Oba H, Ukai‐Tadenuma M, Ogiu T, Tsuji H. Formation of an active form of the interleukin‐2/15 receptor β‐chain by insertion of the intracisternal A particle in a radiation‐induced mouse thymic lymphoma and its role in tumorigenesis. Molecular Carcinogenesis: Published in cooperation with the University of Texas MD Anderson Cancer Center. 2003;37(2):110‐9. 8. Bollati V, Schwartz J, Wright R, Litonjua A, Tarantini L, Suh H, et al. Decline in genomic DNA methylation through aging in a cohort of elderly subjects. Mechanisms of ageing and development. 2009;130(4):234‐9. 9. Jintaridth P, Mutirangura A. Distinctive patterns of age‐dependent hypomethylation in interspersed repetitive sequences. Physiological genomics. 2010;41(2):194‐200. 10. Cho YH, Woo HD, Jang Y, Porter V, Christensen S, Hamilton RF, Jr., et al. The Association of LINE‐1 Hypomethylation with Age and Centromere Positive Micronuclei in Human Lymphocytes. PLOS ONE. 2015;10(7):e0133909. 11. Banerjee P, Chatterjee M. Antiproliferative role of vitamin D and its analogs–a brief overview. Molecular and cellular biochemistry. 2003;253(1‐2):247‐54. 12. Fetahu IS, Hobaus J, Kallay E. Vitamin D and the epigenome. Front Physiol. 2014;5:164. 13. Hübner U, Geisel J, Kirsch SH, Kruse V, Bodis M, Klein C, et al. Effect of 1 year B and D vitamin supplementation on LINE‐1 repetitive element methylation in older subjects. Clinical chemistry and laboratory medicine. 2013;51(3):649‐55. 14. Nair‐Shalliker V, Dhillon V, Clements M, Armstrong BK, Fenech M. The association between personal sun exposure, serum vitamin D and global methylation in human lymphocytes in a population of healthy adults in South Australia. Mutation Research/Fundamental and Molecular Mechanisms of Mutagenesis. 2014;765:6‐10. 15. Tapp HS, Commane DM, Bradburn DM, Arasaradnam R, Mathers JC, Johnson IT, et al. Nutritional factors and gender influence age‐related DNA methylation in the human rectal mucosa. Aging cell. 2013;12(1):148‐55. 16. Castellano‐Castillo D, Morcillo S, Crujeiras AB, Sánchez‐Alcoholado L, Clemente‐Postigo M, Torres E, et al. Association between serum 25‐hydroxyvitamin D and global DNA methylation in visceral adipose tissue from colorectal cancer patients. BMC cancer. 2019;19(1):93. 17. Ong LTC, Schibeci SD, Fewings NL, Booth DR, Parnell GP. Age‐dependent VDR peak DNA methylation as a mechanism for latitude‐dependent MS risk. bioRxiv. 2020. 18. Andrews S. FastQC: a quality control tool for high throughput sequence data. 2010. 2017. 19. Krueger F. Trim galore. A wrapper tool around Cutadapt and FastQC to consistently apply quality and adapter trimming to FastQ files. 2015.

237

APPENDIX FIVE LINE‐1 DNA METHYLATION, AGING AND VITAMIN D

20. Song Q, Decato B, Hong EE, Zhou M, Fang F, Qu J, et al. A reference methylome database and analysis pipeline to facilitate integrative and comparative epigenomics. PloS one. 2013;8(12):e81148. 21. Dolzhenko E, Smith AD. Using beta‐binomial regression for high‐precision differential methylation analysis in multifactor whole‐genome bisulfite sequencing experiments. BMC bioinformatics. 2014;15(1):215. 22. Smit AFA, Hubley R, Green P. RepeatMasker Open‐4.0 2013‐2015 [Available from: http://www.repeatmasker.org. 23. Zerbino DR, Wilder SP, Johnson N, Juettemann T, Flicek PR. The ensembl regulatory build. Genome biology. 2015;16(1):56. 24. Quinlan AR. BEDTools: the Swiss‐army tool for genome feature analysis. Current protocols in bioinformatics. 2014;47(1):11.2. 1‐.2. 34. 25. Alisch RS, Barwick BG, Chopra P, Myrick LK, Satten GA, Conneely KN, et al. Age‐associated DNA methylation in pediatric populations. Genome research. 2012;22(4):623‐32. 26. Heindel JJ, Vandenberg LN. Developmental origins of health and disease: a paradigm for understanding disease etiology and prevention. Current opinion in pediatrics. 2015;27(2):248. 27. Horvath S. DNA methylation age of human tissues and cell types. Genome biology. 2013;14(10):3156.

238

APPENDIX FIVE LINE‐1 DNA METHYLATION, AGING AND VITAMIN D

TABLES

Table 1. Coverage statistics of genome‐wide annotated LINE‐1 regions

Regions With

Total Regions With Average > 10 Reads % of Total Global

Sample Annotated Any Reads Per CpG Annotated Methylation

Adult NoD 876238 573767 566842 0.65 0.88

Adult VitD 876238 573776 567907 0.65 0.88

Paed NoD 876238 573638 566492 0.65 0.89

Paed VitD 876238 573785 567313 0.65 0.89

All Paed 876238 574401 571124 0.65 0.89

All Adult 876238 574437 571313 0.65 0.88

239

APPENDIX FIVE LINE‐1 DNA METHYLATION, AGING AND VITAMIN D

Table 2. Descriptive statistics of filtered LINE‐1 regions

Sample Max Min Median Mean IQR

Adult NoD 1 0 0.92 0.88 0.11

Adult VitD 1 0 0.91 0.88 0.11

Paed NoD 1 0 0.92 0.89 0.11

Paed VitD 1 0 0.93 0.89 0.10

All Paed 1 0 0.92 0.89 0.10

All Adult 1 0 0.91 0.88 0.11

IQR – interquartile range

240

APPENDIX FIVE LINE‐1 DNA METHYLATION, AGING AND VITAMIN D

FIGURES

Figure 1. A) Histograms demonstrating DNA methylation differences by condition and B) violin plots displaying the distribution of DNA methylation values at genome wide LINE‐1 regions. Paed – paediatric. NoD – without vitamin D. VitD – with vitamin D

241

APPENDIX FIVE LINE‐1 DNA METHYLATION, AGING AND VITAMIN D

SUPPLEMENTARY DATA

Supplementary Data 1 – Differentially methylated LINE‐1 CpGs

SEE XLSX FILE – Appendix 5_Supplementary data 1.xlsx

Supplementary Data 2 – LINE‐1 regions containing differentially methylated CpGs

SEE XLSX FILE – Appendix 5_Supplementary data 2.xlsx

Supplementary Data 3 – Genes corresponding to differentially methylated LINE‐1 promoters

SEE XLSX FILE – Appendix 5_Supplementary data 3.xlsx

242

APPENDIX FIVE LINE‐1 DNA METHYLATION, AGING AND VITAMIN D

Supplementary Data 4 – Differentially expressed genes corresponding to differentially methylated

LINE‐1 promoters

ABCG1 DDX60 IFI44 METTL7A SLAMF6 AHRR DHRS9 IFI44L MSR1 SLC12A8 ATP2A3 EHD1 ISOC2 MTUS1 TBC1D4 BIN1 EVL KCNK13 PARP9 TMEM176A C21orf58 FCGBP KCTD7 RABGEF1 TMEM176B C2orf62 FSD1L KIAA0125 SERINC5 VNN3 CD70 GFI1B KLRG1 SIDT1 CD72 HCG11 LIMS1 SIGLEC7 CD86 HERC5 MARCH1 SIGLEC9 CDC45 HNMT MCTP2 SIRPB1

243