Effects of Folic Acid Supplementation on Human Leukocyte DNA Methylation and Gene Expression

James M. A. Elworthy

A thesis submitted for the degree of

Doctor of Philosophy

University of Otago Dunedin New Zealand

March 2011

Abstract

Folic acid and natural folates are B group vitamins obtained exclusively from the diet. Biochemically, folate derivatives participate in the transfer of single carbon units, acting as methyl group donors for numerous cellular processes, including the addition of methyl moieties to genomic CpG dinucleotides, a process termed DNA methylation. This is an important epigenetic mechanism that contributes to the control of gene expression and maintenance of genomic integrity. Low folate intake can result in hypomethylation of DNA, which can disrupt gene expression and promote genetic instability. Genomic hypomethylation is associated with the progression of numerous diseases, including multiple cancers and atherosclerosis. Conversely, limited evidence suggests excessive folate intake may also enhance development of certain carcinomas.

The enzyme methylenetetrahydrofolate reductase (MTHFR) commits folate-bound one- carbon moieties to the DNA methylation pathway. A common enzyme variant of reduced catalytic function may modify the relationship between dietary folate intake, DNA methylation and disease propensity. However, there is a paucity of studies in humans that wholly link these events. Additionally, New Zealand is considering introducing mandatory folic acid fortification to reduce the incidence of neural tube defects. Accordingly, the current study sought to assess changes in DNA methylation as a result of modulated folate intakes in a normal, healthy New Zealand population, which might infer a propensity towards disease.

Peripheral blood was obtained from 93 participants (equally matched by number, age and sex across all MTHFR genotypes) following a thirteen week folic acid avoidance period and a subsequent eleven week folic acid supplementation period (400 µg folic acid/day). Mean plasma folate fell across all MTHFR genotypes by 7 nmol/L (p < 0.001) and increased by 26 nmol/L (p < 0.001) following folic acid avoidance and supplementation, respectively. Total homocysteine, an inverse and intermediate indicator between folate status and DNA methylation, decreased significantly following supplementation (p < 0.001). This was most pronounced in individuals homozygous for

iii

the MTHFR polymorphism with a low folate status (< 12 nmol/L) during the avoidance period.

Despite global hypomethylation being observed under similar conditions by other researchers, no significant change in leukocyte DNA total methylation was found in the current study, either between MTHFR genotypes, or between folate avoidance and supplementation periods. This was measured using high performance liquid chromatography and mass spectrometry methods.

Microarray analysis of leukocyte-derived RNA revealed a small number of genes differentially expressed between genotypes and dietary regimes; however fold change was slight and follow-up analysis indicated none were probable disease candidates. The methylation status of repetitive satellite 2 elements was queried using a methylation- sensitive restriction digest followed by quantitative PCR. Participants homozygous for the MTHFR polymorphism displayed a small but significantly lower level of Sat2 methylation. This difference was not observed following supplementation.

These results suggested a negligible DNA methylation response to modulated folate status and no indication of disease propensity – DNA methylation is particularly immune from medium term perturbations to folate intake in healthy New Zealanders.

Acknowledgements

Well, what has been a protracted and often tortuous ordeal has finally come to an end. Accordingly, this is the part where I get to thank the many people who helped, encouraged or cajoled me to get there.

Firstly, I would like to express my gratitude towards my supervisors: Professor Tony Reeve, who was pivotal in getting this work off the ground and obtaining funding and was a source of ongoing support. Thank you for your constructive critiques and offerings of assistance, drive and focus to counter my waywardness. I am also extremely grateful for the support given by Professor Ian Morison, who has been a great font of inspiration and who salvaged me from a rather unproductive place, giving me the encouragement and support needed to get finished. A very, very big thank you Ian!

I was also very fortunate to have two active committee members. Dr Tim Green and Dr Bernard Venn gave tremendous help and support, endured multiple paper rewrites and gave sage advice and support throughout the project. Thank you both.

Thanks to the members of the Cancer Genetics Laboratory for making the lab an interesting and enjoyable place to work. A special mention is reserved for Jackie Ludgate – thank you for all your assistance, especially the early mornings spent helping me process those bloods! A big thanks too to Les McNoe from the Genomics Facility for his expertise with the arrays.

Likewise, I would like to thank the great people from the Human Nutrition Department who helped make the intervention study run so smoothly.

The vital components in this project were of course the trial participants. Naturally, they are to be thanked here as well. Your contributions constitute the meat of this book!

I am grateful to The National Research Centre for Growth and Development for funding the bulk of this project and providing my PhD stipend, and to the Laurenson Foundation for providing funding for the trial proper.

Finally I would like to thank my family. To my amazing parents who have been so incredibly magnanimous and supportive throughout – you’ve been wonderful. Milly: yes I’ve finally finished my book and have found a job! Thanks for helping keep things in perspective. And to Megan, I truly couldn’t have done it without you. My heartfelt thanks to you for your love and support for so long. Now we can finally put this bloody thing to bed.

Table of Contents

CHAPTER ONE - Introduction

1.1 Folate, the core of one-carbon metabolism ...... 1

1.1.1 A brief history of the discovery and characterisation of folate ...... 3 1.1.2 Sources of folate ...... 4 1.1.3 Recommended dietary intake of folate ...... 5 1.1.4 Folate transport, homeostasis and catabolism ...... 5 1.1.5 Tetrahydrofolate one-carbon acceptance and interconversion ...... 7

1.1.6 Metabolic destinations of folate C1 units ...... 10 1.1.7 Overview of the methionine cycle ...... 11 1.1.8 Methylenetetrahydrofolate reductase (MTHFR) ...... 13

1.2 Folate’s role in the pathogenesis of human disease ...... 15

1.2.1 General effects of folate deficiency and reduced MTHFR efficiency ...... 16 1.2.2 Folate and atherosclerosis ...... 17 1.2.3 Adverse pregnancy outcomes ...... 18 1.2.4 Neuropsychiatric disorders ...... 18 1.2.5 Neural tube defects ...... 20 1.2.6 Folic acid, neural tube defects and public health...... 22 1.2.7 Folate and cancer ...... 24 1.2.7.1 MTHFR C677T as a risk modifier ...... 26 1.2.7.2 Folic acid as a public health measure – potentially deleterious ramifications of supplementation ...... 27

1.3 An introduction to epigenetics ...... 29

1.4 Components of epigenetics ...... 30

1.4.1 Histones ...... 30 1.4.2 MicroRNAs ...... 31 vii

1.5 DNA methylation ...... 32

1.5.1 Introduction ...... 32 1.5.2 5′-Methylcytosine and the CpG dinucleotide ...... 32 1.5.3 Genomic distribution of 5′-methylcytosine ...... 36 1.5.4 CpG islands help determine gene expression state ...... 37

1.6 Concomitant roles of DNA methylation ...... 39

1.6.1 Transposon suppression – DNA methylation as a genetic defence ...... 39 1.6.2 DNA methylation and development ...... 40 1.6.3 X chromosome inactivation ...... 41 1.6.4 Imprinting ...... 41

1.7 Improper DNA methylation culminates as disease ...... 43

1.7.1 Syndromes mediated by errors of DNA methylation, in brief ...... 43 1.7.2 Aberrant DNA methylation in cardiovascular disease and other human pathologies ...... 44 1.7.3 Improper DNA methylation and cancer – a well established example...... 45

1.8 Environmental factors influence (disease) phenotypes through DNA methylation ...... 46

1.8.1 Retrospective studies ...... 47 1.8.2 Experimental studies ...... 48

1.9 Does inappropriate folate status or the MTHFR C677T polymorphism lead to disrupted DNA methylation, altered gene expression and thus disease? .... 50

1.9.1 Evidence of folate as an environmental influence ...... 50 1.9.2 Folate, DNA methylation and neural tube defects ...... 51 1.9.3 Folate, the MTHFR polymorphism, methylation and cancer ...... 52

1.10 Evidence for perturbations in folate metabolism and MTHFR affecting DNA methylation in humans ...... 54

1.11 Aims of the study ...... 55

viii

CHAPTER TWO – Materials & Methods

2.1 Chemicals and materials ...... 59

2.1.1 Enzymes...... 59

2.2 Buffers and solutions ...... 59

2.3 Folic acid supplementation trial ...... 59

2.3.1 Ethical approval ...... 60 2.3.2 Study preparation ...... 60 2.3.3 Participant recruitment and screening ...... 61 2.3.4 Folic acid supplementation trial ...... 62 2.3.5 Compliance ...... 63

2.4 Biochemical assays ...... 63

2.4.1 Folate measurements ...... 64 2.4.2 Homocysteine ...... 64

2.4.3 Vitamin B12 ...... 64 2.4.4 Statistics ...... 64

2.5 DNA preparations ...... 65

2.5.1 DNA extraction from buccal cells ...... 65 2.5.2 DNA extraction from peripheral blood ...... 65 2.5.3 DNA preparation for LC/MS and HPLC ...... 66 2.5.3.1 Removal of ribonucleic acid ...... 66 2.5.3.2 DNA hydrolysis ...... 67

2.6 RNA preparation ...... 67

2.6.1 RNA extraction from peripheral blood leukocytes ...... 67 2.6.2 RNA quantification...... 69 2.6.2.1 RNA gels ...... 69 2.6.2.2 Experion automated electrophoresis system ...... 70

2.7 Assessment of MTHFR genotype status ...... 70

2.7.1 MTHFR C677T polymerase chain reaction (PCR) ...... 70 2.7.2 PCR product digestion ...... 71

2.8 Global DNA Methylation Analysis using LC/MS ...... 72

2.8.1 Equipment ...... 72 2.8.2 System setup ...... 72 2.8.3 Analysis of participant DNA hydrolysates ...... 74

2.9 Global Methylation Analysis utilising HPLC ...... 74

2.10 Microarray Analysis ...... 75

2.10.1 Array design ...... 75 2.10.2 Sample selection for gene expression array studies...... 76 2.10.3 Arrays ...... 76 2.10.4 RNA to cDNA reverse-transcription ...... 76 2.10.4.1 Labeling of cDNA ...... 77 2.10.5 Hybridisation of labeled cDNA ...... 78 2.10.5.1 Slide preparation ...... 78 2.10.5.2 Loading ...... 79 2.10.5.3 Post-hybridisation ...... 79 2.10.6 Scanning of arrays ...... 80 2.10.7 Data normalisation ...... 81 2.10.8 Expression data analysis ...... 81

2.11 Satellite 2 DNA methylation assay ...... 81

CHAPTER THREE – Folic Acid Supplementation Study

3.1 Trial overview and provisos ...... 85

3.1.1 Peripheral blood’s suitability as a source and a proxy for whole-body folate status ...... 85 3.1.2 Compensation for MTHFR allele distribution ...... 87 x

3.1.3 Determinates of study duration ...... 87 3.1.3.1 Peripheral blood cell turnover ...... 88 3.1.4 B vitamin supplementation and diet considerations ...... 89 3.1.5 Blood processing ...... 90

3.2 Results ...... 91

3.2.1 MTHFR polymorphism genotyping...... 91 3.2.2 MTHFR C677T allelic distribution ...... 92 3.2.3 Participant flow and follow up ...... 92 3.2.4 Study participant statistics and compliance ...... 93 3.2.5 Blood processing ...... 95 3.2.6 Plasma folate...... 95 3.2.7 Red cell folate (erythrocyte folate) ...... 96 3.2.8 Homocysteine ...... 97 3.2.9 Additional micronutrient data ...... 97 3.2.10 Plasma homocysteine as a function of folate, MTHFR C677T genotype

and B12 status ...... 99

3.2.11 Sex effect in measured outcomes ...... 101

3.3 Discussion of immediate study outcomes ...... 101

CHAPTER FOUR – Global DNA Methylation Analysis

4.1 Brief overview of existing global DNA methylation assays ...... 105

4.2 Development of LC/MS procedure ...... 108

4.2.1 Principles of mass spectrometry ...... 108 4.2.2 Instrument optimisation ...... 108 4.2.3 Initial assessment of standard separation...... 109 4.2.4 Genomic DNA hydrolysis, separation and species identification ...... 110 4.2.5 Limits of detection, inter-assay variation and standard curves ...... 114

4.2.6 Internal standard considerations ...... 114 4.2.7 Equipment issues ...... 116 4.3 Results of LC/MS assay of total DNA methylation ...... 116

4.4 High performance liquid chromatography ...... 117

4.5 HPLC results ...... 118

4.6 Discussion ...... 121 4.6.1 Results of assay and comparison to data from Friso et al...... 121 4.6.2 Comments on the merits of the LC/MS and HPLC assays ...... 122

CHAPTER FIVE – Microarray Analysis

5.1 Introduction ...... 125

5.1.1 Large scale discrete DNA methylation analysis methods ...... 126 5.1.2 DNA microarrays – gene expression as a proxy for DNA methylation ..... 128

5.2 Experimental design ...... 129

5.2.1 Reversed-dye microarray platform ...... 129 5.2.2 Paired sample non-reference design ...... 130 5.2.3 Additional considerations ...... 131

5.3 RNA quality assessment ...... 131

5.4 Selection of participant samples for microarray analysis ...... 132

5.5 Data filtering, normalisation and collation ...... 133

5.5.1 Accession number streamlining using DAVID and SOURCE ...... 134

5.6 Expression analysis of genes implicated in folate metabolism and DNA methylation ...... 135 5.6.1 Generation of folate- and DNA methylation-centric gene lists ...... 135 5.6.2 Analysis of folate-centric genes using t-tests, SAM and clustering ...... 139 5.6.3 Statistical analysis outcomes ...... 141

xii

5.6.4 Biological relevance of identified DE genes ...... 141 5.6.5 Summary of folate-related differential gene expression ...... 146 5.7 Assessment of complete array dataset ...... 146

5.7.1 Utilisation of genome-wide profiling data ...... 147 5.7.2 SAM analysis and t-tests of complete expression dataset ...... 151 5.7.3 Characteristics of identified genes ...... 151 5.7.4 Interpretation of expression data ...... 154

5.8 Discussion ...... 155

5.8.1 Further comments on identified genes...... 156 5.8.2 Assay limitations – inter-subject variability ...... 157 5.8.3 Genes involved in methylation remodelling – implications for the wider genome? ...... 158

CHAPTER SIX – Satellite 2 Methylation Analysis

6.1 Introduction ...... 159

6.1.1 Characteristics of satellite sequences ...... 159 6.1.2 Aberrant Sat2 methylation and disease ...... 160

6.2 Satellite 2 methylation assay ...... 162

6.3 Results ...... 163

6.4 Discussion regarding Sat2 results ...... 166

CHAPTER SEVEN – Discussion

7.1 Synopsis of the project ...... 169

7.1.1 Intervention trial and micronutrient data ...... 169 7.1.2 Genomic DNA methylation analysis ...... 170 7.1.3 Microarray analysis ...... 171 7.1.4 Satellite 2 methylation analysis ...... 171

xiii

7.1.5 Summary ...... 172

7.2 Limitations of the study ...... 172 7.2.1 Potential confounders of the immediate trial ...... 172 7.2.2 Assay limitations ...... 175 7.2.3 Tissue suitability ...... 175

7.3 Novel aspects of the study and comparisons with associated literature ...... 176

7.3.1 Studies conducted under a background of folic acid ...... 177 7.3.2 Metabolic unit studies versus “free-living” cohort ...... 177 7.3.3 Cohort differences ...... 178 7.3.4 DNA methylation assay methods ...... 178 7.3.5 Comparison to cross sectional studies ...... 179 7.3.6 Literature describing no relationship between folate status and DNA methylation ...... 182 7.3.7 Summary ...... 183

7.4 Brief comment on the folic acid fortification debate ...... 184

7.5 Future research directions ...... 184

7.6 Conclusions ...... 186

APPENDICIES

Appendix A ...... 189 Appendix B ...... 192

REFERENCES

References ...... 199

xiv

List of Figures

Figure 1.1 Chemical structure of folic acid ...... 2

Figure 1.2 Tetrahydrofolate and its derivatives ...... 8

Figure 1.3 Interconversion of single-carbon units carried by tetrahydrofolate ...... 9

Figure 1.4 Biosynthetic fates of single-carbon units in the THF pool, incorporating methionine metabolism and the multiple methyl acceptors of S-adenosylmethionine ...... 12

Figure 1.5 General mechanism of transcription silencing involving DNA methylation ...... 33

Figure 1.6 Attributes of 5′-methylcytosine ...... 35

Figure 1.7 Proposed interrelationship between MTHFR C677T genotype, folate status and DNA methylation, with respect to cancer ...... 53

Figure 3.1 Intervention study procedural overview ...... 86

Figure 3.2 Sites of action of cofactors involved in folate metabolism ...... 89

Figure 3.3 Electrophoretogram of digested PCR amplicons ...... 91

Figure 3.4 Participant flow and follow up ...... 93

Figure 3.5 Plasma folate concentration according to MTHFR C677T genotype at baseline, following folic acid avoidance (week 13) and following folic acid supplementation (week 24) ...... 96

Figure 3.6 Red cell (erythrocyte) folate concentration according to MTHFR C677T genotype at baseline, following folic acid avoidance (week 13) and following folic acid supplementation (week 24) ...... 97

Figure 3.7 Homocysteine concentration according to MTHFR C677T genotype at baseline, following folic acid avoidance (week 13) and following folic acid supplementation (week 24) ...... 98

Figure 4.1 Procedural overview of the global 5′-methylcytosine assay ...... 107

Figure 4.2 Preliminary trace of ion abundance versus time for 25:1 molar mix of cytidine and methylcytidine standards, as separated by HPLC at 0.2 mL/min ...... 110

Figure 4.3 Qualitative MS assessment of hydrolysed DNA following HPLC separation ...... 111

Figure 4.4 Species identification of hydrolysed DNA by MS base peak ...... 112

Figure 4.5 Representative run of hydrolysed genomic DNA ...... 113

Figure 4.6 Standard curves of cytosine and 4% methylcytosine ...... 115

Figure 4.7 Standard curves of cytidine and 5′-methylcytidine as performed by HPLC ...... 118

Figure 4.8 Comparison of replicate variation between LC/MS and HPLC assays ...... 119

Figure 4.9 DNA methylation stratified by post-avoidance plasma folate concentration (folate status), and separated according to MTHFR C677T genotype ...... 120

Figure 4.10 Comparison of global DNA methylation data between Italian and New Zealand populations according to MTHFR genotype and plasma folate status ...... 121

Figure 5.1 Array procedure overview ...... 131

Figure 5.2 Assessment of RNA quality – output of Nanodrop spectrophotometer and Bio-Rad Experion systems ...... 132

Figure 6.1 Satellite 2 methylation according to MTHFR C677T genotype following folic acid avoidance and folic acid supplementation ...... 163

Figure 6.2 Global DNA methylation versus satellite 2 methylation, subdivided by plasma folate status and MTHFR genotype following folic acid avoidance ...... 165

xvi

List of Tables

Table 1.1 Distribution of CpG dinucleotides in the human genome ...... 37

Table 3.1 Age and sex distributions of participants that completed the study ...... 94

Table 3.2 Vitamin capsule consumption by MTHFR C677T genotype...... 95

Table 3.3 Red blood cell folate, plasma folate, cobalamin and homocysteine concentration by MTHFR C677T genotype at baseline, following folate avoidance and repletion ...... 99

Table 4.1 LC/MS assay of global DNA methylcytosine by MTHFR C677T genotype at baseline, following folate avoidance and following repletion ...... 117

Table 4.2 Global DNA methylcytosine by MTHFR C677T genotype at baseline, following folate avoidance and following repletion, according to HPLC ...... 120

Table 5.1 Arrayed participant statistics ...... 133

Table 5.2 Folate- and DNA methylation-centric gene list for microarray data inquiry ...... 137

Table 5.3 Statistically significant differentially expressed genes as determined by Significance Analysis of Microarray ...... 142

Table 5.4 Loci possessing putative methylated promoter CpG islands, as reported by Shen and colleagues, used to probe differentially expressed genes identified in current dataset ...... 148

Table 5.5 Significant DE genes containing methylated CpG islands as identified by SAM ...... 152

Table 5.6 CpG island attributes of genes identified by SAM ...... 153

xvii

xviii

List of Abbreviations

5mC 5-methylcytosine AdoHcy S-adenosylhomocysteine (also abbreviated as SAH) AdoMet S-adenosylmethionine (also abbreviated as SAM)

B2 Riboflavin (vitamin B2)

B6 Pyridoxine (vitamin B6)

B12 Cyanocobalamin (vitamin B12) bp base pairs

C1 one-carbon unit CpG cytosine-guanine dinucleotide CV Coefficient of variation (standard deviation / mean, expressed as a percentage) CyDye Monofunctional NHS-reactive fluorescent dye ddH2O double-distilled water DE differentially expressed; differential expression DEPC diethylpyrocarbonate DFE dietary folate equivalent DHF dihydrofolate EDTA ethylenediaminetetraacetic acid ESI Electrospray ionisation EtBr ethidium bromide FOCM Folate-mediated one-carbon metabolism HPLC high performance liquid chromatography IMP inosine monophosphate kb kilobase LC/MS liquid chromatography coupled mass spectrometry min minutes mRNA messenger ribonucleic acid MS mass spectrometry, mass spectrometer MSQ-PCR Methylation-specific quantitative polymerase chain reaction MTHFR 5,10-methylenetetrahydrofolate reductase

xix

m/z Mass to charge ratio NTD neural tube defect PABA para-aminobenzoic acid (p-aminobenzoic acid) PBL peripheral blood leukocytes PBMC peripheral blood mononuclear cells PCR polymerase chain reaction PSI Pounds per square inch RBC Red blood cells (erythrocytes) RNA ribonucleic acid rpm revolutions per minute RT Retention time SAH S-adenosylhomocysteine (also abbreviated as AdoHcy) SAM S-adenosylmethionine (also abbreviated as AdoMet) SAM Significance analysis of Microarray Sat2 Satellite 2 repetitive DNA SD standard deviation SDS sodium dodecyl sulphate THF tetrahydrofolate Tris 2-amino-2-(hydroxymethyl)-1,3-propandiol UCSC University of California Santa Cruz Genome browser UL tolerable upper intake level v/v volume per volume w/v weight per volume

xxi

xxii

Folate & DNA Methylation: 1 An Introduction .

The vitamin folate has a long history in the scientific literature. The first reports of megaloblastic anaemia, of which folate deficiency is a cause, date back almost two hundred years. Today, folate remains in the spotlight: the vitamin is heralded as an effective agent in the reduction of neural tube defects and is at the centre of one of the biggest public health initiatives in history – mandatory folic acid fortification in the United States. A common theme of the vitamin has always been its prominence in public health, and the multiplicity of potential disorders when dietary intake of the vitamin is inadequate. Indeed, biochemically, folate derivatives play a role in many, varied biochemical processes, leading to a web of interactions among cellular systems. As a consequence, the observed effects of changes in the dietary intake of folate can be numerous or nuanced. Chapter 1 focuses on the role of folate in disease and the interplay between folate and DNA methylation, the latter being an equally intriguing phenomenon in its own right, with purported roles as diverse as X chromosome inactivation and protection of the genome from parasitic genetic elements. Here though, the epigenetic governance of gene expression by DNA methylation, its implication in disease development and folate’s role in the process are of interest and lay the foundation for this thesis.

1.1 Folate, the core of one-carbon metabolism

The term folate encompasses a number of compounds chemically based on the fully oxidised molecule pteroylmonoglutamic acid, commonly known as folic acid (Mann and Truswell, 2007). Folic acid is not typically found in nature, but is the form primarily used in synthetic supplements due to its stability and high bioavailability (Lawrence, 2007). The base molecule consists of a pteridine group, para-aminobenzoic acid (PABA), and glutamic acid (Figure 1.1). Natural folates are invariably reduced at

2 Chapter 1 positions 5-8 of the pterin ring and are termed tetrahydrofolates (THF). Folic acid is reduced to THF enzymatically via dihydrofolate (DHF), with both reactions facilitated by dihydrofolate reductase (Abali et al., 2008). Multiple glutamate groups (typically three to eight) are added to the tail of the folate molecule following cellular uptake (Shane, 1995).

2-Amino-4-oxo-6-methylpterin Para-aminobenzoic acid Glutamates (n = 1-8)

Pteroic acid

Pteroylglutamic acid

Figure 1.1 Chemical structure of folic acid.

Also termed vitamin B9, folates, once metabolised by the body, function as transporter coenzymes responsible for accepting and donating single-carbon moieties within the cell (Stover, 2004). The transfer reactions that folate derivatives facilitate are involved in multiple processes, including DNA and RNA synthesis, through involvement in the generation of the nucleobases adenine, guanine and thymine, and the interconversion of various amino acids (Wagner, 1995). One such conversion is the methylation of homocysteine to methionine, the latter being precursor to S-adenosylmethionine (abbreviated as SAM or alternatively, AdoMet). SAM is required in diverse methylation reactions including that of histones, neurotransmitters, proteins, phospholipids and importantly in the case of this work, methylation of cytosine-guanine dinucleotides – a process termed DNA methylation – which is a central theme of this thesis.

Because of the multiple fates of single carbon units in the folate pool, modulation of folate intake and metabolism can influence these various downstream pathways. Consequently, from a health standpoint, poor dietary folate intake or folate metabolism deficiencies have the potential to materialise as a number of disorders. Interestingly, it was the research into one such disorder that led to the discovery of folic acid.

Introduction 3

1.1.1 A brief history of the discovery and characterisation of folate

The characterisation of folic acid had its beginnings in the 1870s with the classification of megaloblastic anaemia, a disorder in which there is a presence of large, immature and non-functioning red blood cells (referred to as macrocytosis) in the bone marrow (Williams, 1990). This improper cell synthesis occurs as a result of incorrect DNA synthesis, which is due to insufficient folate or vitamin B12, or both – these vitamins being essential for the generation of thymidylate and purines. The initial description of the disorder was made by Hayem in 1877 and by Ehrlich in 1880, who identified the associated disordered neutrophils with their multi-segmented nuclei (Carmel, 2005). However it was not until the 1930s that advances were made into the causes of megaloblastic anaemia.

Briton Lucy Wills, during a sabbatical in Bombay (now Mumbai), India, began investigating macrocytic anaemia in pregnant Indian women (Hoffbrand and Weir, 2001). This anaemia was more prevalent in poorer populations whose diets were high in white bread and polished rice but low in protein, fruit and vegetables. Yeast or the yeast extract Marmite, when added to the diet, prevented such anaemias (Wills, 1931) and additional experiments performed upon rats with dietary anaemia resulted in similar outcomes (Wills and Mehta, 1931). In 1938, Day et al. described a “Vitamin M” which corrected nutritional anaemia in monkeys (Hoffbrand and Weir, 2001) and Stokstad and

Manning reported the discovery of a “Factor S”, or “Vitamin BC”, present in yeast, which cured nutritional anaemia in chickens (Stokstad and Manning, 1938). Wills’s yeast factor, “Vitamin M”, and “Vitamin BC” were in fact the same molecule – folate. The vitamin was isolated from four tons of spinach in 1941, giving rise to its name, folium in Latin meaning “leaf” (Mitchell et al., 1941). Folic acid was chemically synthesised in 1945 by Angier et al. who determined that it consisted of a pteridine ring, p-amino benzoic acid and glutamate, giving rise to its chemical term, pteroylglutamic acid (Angier et al., 1945).

The majority of pathways and enzymes involved in folate metabolism were discovered during the two decades subsequent to folate’s structural elucidation (Hoffbrand and Weir, 2001). Additionally, physiological modes of folate absorption, transport and storage were also elucidated (Halstead, 1980; Shane, 1995), although it is acknowledged

4 Chapter 1 that this work is not complete; particular vitamer absorption properties, for example, are not fully understood (Sanderson et al., 2003). Also, bigger questions pertaining to the causal biochemical mechanisms responsible for many pathologies associated with folate remain unanswered (Fox and Stover, 2008). This aside, information pertinent to the current thesis regarding what is known about folate sources, metabolism, interconversion and catabolism is summarised in sections 1.1.2 to 1.1.8 below.

1.1.2 Sources of folate

By its very definition as a vitamin, folate is not produced de novo in humans. Although components of the final molecule can be synthesised, the ability to covalently link the pteridine ring and p-amino benzoic acid (see Figure 1.1) has been lost. (Ifergan and Assaraf, 2008). Instead, the molecule is obtained from dietary fruit and vegetables and from intestinal microflora capable of folate synthesis (Rong et al., 1991). Folate is most abundant in green leafy vegetables such as spinach, cabbage and broccoli. It is also in foods such as beans, peanuts, yeast extract, avocadoes, bananas, eggs, liver and kidney (Birn, 2006). In New Zealand, vegetables contribute more than a quarter of total folate intake. Cereals and breads are also provide significant contributions – around 10% and 15% respectively. A small proportion of the population (approximately 1%) acquire some folate from supplements. There are no major age-related differences in folate intake (Russell et al., 1999).

In addition to acquiring natural folates from the diet, individuals in many countries are increasingly consuming synthesised folic acid that is being added to refined foods, a process termed fortification (Oakley and Johnston, 2004). In fact, in the United States, fortified foods have become the primary contributor of total folate intake (Institute of Medicine et al., 2000). In this country, fortification may contribute up to 70 μg/day (Russell et al., 1999). The mandatory introduction of folic acid into refined foods in America and elsewhere is primarily a response to the desire to reduce the prevalence of neural tube defects (NTDs). A comprehensive topic in itself, folic acid fortification and NTDs will not be discussed here but instead affords its own section in part 1.2.

Introduction 5

1.1.3 Recommended dietary intake of folate

Natural folates (from food) are estimated to be only 50% bioavailable, whereas folic acid is around 85% when taken with food, and almost 100% when consumed while fasting (Gregory, 1997). Natural folates have lower bioavailability due to the incomplete breakdown of foods during digestion and decreased exposure of free folate to enterocytes. Because of this, the convention has been to express requirements of the vitamin in terms of dietary folate equivalents (DFEs). For example, 0.6 µg of folic acid from fortified food or supplements is the approximate dietary equivalent of 1 µg of food folate. Thus 0.6 µg of folic acid equals 1 DFE (Caudill, 2010). The recommended dietary intake of folate for adults is 400 µg DFEs per day, as specified by the New Zealand Ministry of Health (New Zealand Ministry of Health, 2010). Actual median intake of natural folates in New Zealand has been estimated at 242 μg per day. Prevalence of folate inadequacy is only around 7%, however, and this is regarded as an overestimate (Russell et al., 1999).

Subsets of the population can have varying folate requirements. For example, men may have a higher requirement than women (Sanderson et al., 2003). This is reflected in the higher daily intake in men (in New Zealand, 278 μg/day versus 212 μg/day for women. Perturbations in folate metabolism via compromised enzymes can also modify one’s folate requirements (Guinotte et al., 2003; Hung et al., 2006) – the most comprehensively studied of these being 5,10-methylenetetrahydrofolate reductase (MTHFR), which will be discussed further in section 1.1.8. Additionally, women capable of or planning pregnancies are advised to consume additional folic acid in the form of a supplement at a level of 400 µg per day for at least one month before and three months after conception, due to the vitamin’s importance in fetal development. The US Institute of Medicine quotes the Tolerable Upper Intake Level (UL) for supplements for adults at 1000 µg per day, in addition to consumed natural folates (Institute of Medicine et al., 2000).

1.1.4 Folate transport, homeostasis and catabolism

Folates from the diet are primarily absorbed in the proximal small intestine. Folyl- polyglutamate carboxypeptidase (FOLH1) first deconjugates folate’s glutamate tails to

6 Chapter 1 monoglutamates (Chandler et al., 1986). Absorption is then performed by PCFT/HCP1, a proton-coupled folate symport (Qiu et al., 2006). Both enzymes reside in the apical membrane of enterocytes. The bulk of this active transport takes place at the duodenum and to a lesser extent at the jejunum. The reduced folate carrier (RFC), a folate transporter found in multiple tissues, is expressed throughout the gut, but is less suited to the low pH of the upper small intestine, so contributes little to primary absorption. It is instead thought to contribute to folate uptake in the colon (Matherly and Hou, 2008), where intestinal microflora supplies a significant contribution of the vitamin (Rong et al., 1991). In addition to this active, saturable transport, pharmacological doses of folic acid are absorbed by passive diffusion (Institute of Medicine et al., 2000).

Digested folates, converted mainly to 5-methyl-THF in the enterocytes, are then moved across the basolateral membrane by ABC transporters MRPs 1 and 3 (Assaraf, 2006) and taken via the hepatic portal vein to the liver where they are stored (Matherly and Hou, 2008) before being secreted into the blood or bile. Upon release into the bloodstream, around two thirds of folate becomes bound to carrier proteins, particularly albumin (Institute of Medicine et al., 2000) but also to some high-affinity folate binding proteins. Folate is then taken up into peripheral tissues by either folate binding protein/folate receptors (FR) or the reduced folate carrier. The first category of transporters are unidirectional and membrane bound, and possibly transport folates via receptor-mediated endocytosis (Antony, 1996), whereas reduced folate carriers are carrier-mediated transport systems capable of bidirectional flux (Sirotnak and Tolner, 1999). They are more efficient than folate receptors (Henderson et al., 1988; Kamen and Capdevila, 1986), however the extent to which these two systems contribute to folate uptake is unclear, and the degree of expression of each family varies according to cell type (Sirotnak and Tolner, 1999). Indeed, modulation of receptor expression, activity or localisation is believed to play a part in cellular folate homoeostasis (Jansen et al., 1990; Wong et al., 1998), however the exact biochemical feedback mechanisms by which this occurs have not been ascertained.

Folate turnover and catabolism is also not fully understood but it is a complex process that, in addition to probable expression modulation of the aforementioned receptors, utilises the addition or removal of glutamate to alter cellular folate retention, plus cleavage and inactivation of folate molecule (Suh et al., 2001). Addition of glutamate

Introduction 7 polypeptide tails allows better retention of folate by cells and increases folate’s metabolic activity by raising its affinity for folate-binding enzymes two or three orders of magnitude higher than their equivalent monoglutamates (Schirch and Strong, 1989). Folate binding proteins including enzymes involved in folate processing are present in excess in the cell and as a consequence, most intracellular folate is protein-bound. Glutamate addition is catalysed by folylpolyglutamate synthetase, or FPGS (Moran, 1999). Bound folate is also sterically protected from polyglutamate peptide cleavage by glutamyl hydrolase, which hydrolyses polyglutamates to the monoglutamate form (Wang et al., 1993). The monoglutamate form can be readily taken up (Shane, 1995) but can be effluxed equally easily from the cell (Suh et al., 2001). Bound folates consequently have a much longer intracellular half life than free monoglutamates. Folate can also be oxidatively degraded to yield pterin (6-formyltetrahydropterin) and PABA-glutamate (p-aminobenzoylpolyglutamate) components which are excreted from the cell and subsequently removed from circulation by the kidneys.

1.1.5 Tetrahydrofolate one-carbon acceptance and interconversion

As an enzymatic cofactor, tetrahydrofolate is the recipient of single-carbon moieties in various states of oxidation, these being donated from multiple sources. The modified folates are subsequently able to donate their one-carbon (C1) groups to a variety of biosynthetic pathways – each pathway being dependent on a one-carbon unit from a different and specific THF cofactor form (described further in section 1.1.6).

The amino acids serine and glycine constitute the main sources from which single- carbon units enter the THF pool. In a reaction catalysed by serine hydroxymethyl transferase, serine donates a methylene group (–CH2–) to THF, becoming glycine and 5,10-methylene-THF, respectively. Glycine can be further hydrolysed to donate a methylene moiety to a second THF.

Additionally, a third amino acid, histidine, donates a formimino group (–CH=NH) to produce 5-formimino-THF. Finally, in a reaction catalysed by 10-formyl-THF - synthetase, the carboxylate anion formate (HCO2 ) becomes bound to the N10 position of THF, forming 10-formyl-THF. The chemical structures of these THF derivatives,

8 Chapter 1 including the nitrogen positions where the various one-carbon groups affix, are depicted in Figure 1.2.

Tetrahydrofolate (THF)

5-Methyl-THF 5,10-Methylene-THF 5,10-Methenyl-THF

5-Formimino-THF 5-Formyl-THF 10-Formyl-THF

Figure 1.2 Tetrahydrofolate and its derivatives. Single-carbon moieties are conjugated to the 5′ or 10′ nitrogen of the pterin ring, or both. The R group denotes p- aminobenzoic acid and (poly)glutamate.

Once in possession of these one-carbon moieties, THF derivatives can undergo enzyme- mediated transformation interconverting C1 units from one oxidation state to another or from one nitrogen bonding position to another. This group of folate derivatives constitutes the folate pool. Specifically, the serine or glycine-derived CH2 group bonded to the N5 and N10 positions of 5,10-methylene-THF can be converted to 5-methyl-THF or 5,10-methenyl-THF by their respective reductases. Additionally, histidine-derived moieties and those sourced from formate can be converted to the methenyl form by cyclodeaminase and cyclohydrogenase catalysis respectively. All these above reactions, with the exception of the 5,10-methylene-THF-reductase-catalysed reaction, are reversible in vivo. A diagrammatic representation of the interconversion of folate derivatives is depicted in Figure 1.3.

Introduction 9

5-Methyl-THF Serine Glycine + Serine NAD 5,10-Methylene-THF hydroxymethyl reductase + transferase NADH + H

THF 5,10-Methylene-THF

Histidine Glycine synthase + Glutamate NADP + 5,10-Methenyl-THF + NH4 + Glycine + CO2 + NH4 + reductase + NADPH + H NAD + NADH

5-formimino-THF cyclodeaminase 5-Formimino-THF 5,10-Methenyl-THF dehydrogenase THF -

Formate + ATP ADP + P H2O NH3 i Formyl 10-Formyl-THF 5,10-Methenyl- 5,10-Methenyl- 10 -

THF THF synthetase synthetase cyclohydrolase ATP ADP + Pi

10-Formyl-THF 5-Formyl-THF 10-Formyl- THF isomerase ADP + Pi ATP

Figure 1.3 Interconversion of single-carbon units carried by tetrahydrofolate. Derivative conversion is achieved by a host of enzymes, which are depicted in italics.

As one-carbon units attached to tetrahydrofolate have different metabolic fates depending on their oxidation state, interconversion allows C1 units to enter the pool at different points and yet be accessible to all downstream pathways. Interconversion also enables the various folates to enter a storage pool as 5-formyl-THF, which is the only derivative that doesn’t directly donate its methyl group to biosynthetic reactions. It is the most stable of the natural folates, and while it accounts for less than 15% of folates in mammals (Stover et al., 1993), it acts as a store of formylated folates in the cell (Field et al., 2006). It also acts as an inhibitor of serine hydroxymethyl transferase, thereby regulating C1 units entering the folate pool. Additionally, the 5-formyl-THF store buffers 10-formyl-THF required for maintaining purine synthesis, keeping the

10 Chapter 1 production of DNA immune, to some extent, from perturbations in the folate network (Field et al., 2006).

1.1.6 Metabolic destinations of folate C1 units

Folate can receive and donate single carbon units in many states of oxidation, thus is a particularly versatile cofactor; as alluded to already, one-carbon units covalently attached to folate coenzymes can be transferred to multiple substrates (see Figure 1.4).

The particular one-carbon acceptor depends on the oxidation state of the C1 moiety held by folate. Ergo there are three exits whereby single-carbon units leave the folate pool, each facilitated by a cofactor with a C1 group at a different oxidation level. The derivatives concerned are 10-formyl-THF, 5,10-methylene-THF and 5-methyl-THF.

The 10-formyl-THF derivative is responsible for contributing two formyl groups to the formation of purines, or more specifically, during the synthesis of inosine monophosphate, abbreviated as IMP (Fox and Stover, 2008). Inosine monophosphate is the precursor of adenosine monophosphate (AMP) and guanosine monophosphate (GMP) and IMP’s synthesis is a ten-step process that takes place in the cytoplasm. In the third reaction, catalysed by glycinamide ribotide formyltransferase (GARFT), the formyl group from 10-formyl-THF is transferred to IMP precursor glycinamide ribotide (GAR). The latter is transformed through five more reactions to aminoimidazolecarboxomide ribotide (AICAR) whereupon another formyl group from 10-formyl-THF is added by aminoimidazolecarboxamide ribonucleotide formyltransferase (AICARFT). The carbons donated in these two folate-facilitated reactions become the C8 and C2 positions of the purine ring respectively (Voet and Voet, 1995).

With protein synthesis in mitochondria, peptide formation begins with the addition of the initiator tRNA formylmethionyl-tRNA (fMet-tRNA). This differs from cytosolic translation initiation, which uses methionyl-tRNA (RajBhandary, 1994). The formate added to fMet-tRNA is obtained from 10-formyl-THF and is catalysed by methionyl- tRNA formyltransferase. This is the only known biosynthetic reaction, other than interconversions of amino acids, to occur in the mitochondrion. The organelle possesses

Introduction 11

its own enzymatic machinery to recruit C1 units from substrates such as serine and glycine to produce 10-formyl-THF for fMet-tRNA production (Christensen and Mackenzie, 2008).

The second folate derivative listed above, 5,10-methylene-THF, is required to convert uridylate (uridine monophosphate, UMP) to thymidylate (thymidine monophosphate, TMP). The enzyme that catalyses the reaction is thymidylate synthase (TS). Thymidylate is subsequently incorporated into DNA; the thymidylate synthase- mediated reaction is the rate limiting step during DNA synthesis (Fukushima et al., 2003). Thymidylate synthase effectively “competes” for 5,10-methylene-THF in the folate pool against 5,10-methylene-THF reductase (MTHFR), which reduces the methylene moiety to methyl. Competition arises from the fact that cellular concentrations of folate are lower than that of folate-binding proteins, hence there is very little free folate in the cell (Suh et al., 2001).

Lastly, 5-methyl-THF is required for the methylation of homocysteine to methionine (Loenen, 2006). As well as helping maintain the intracellular levels of both these amino acids, the reaction forms part of the methionine cycle, required for the production of S- adenosylmethionine (SAM), the universal methyl donor in the cell.

1.1.7 Overview of the methionine cycle

The methionine cycle consists of four reactions (see Figure 1.4): conversion of homocysteine to methionine, adenosylation of methionine to S-adenosylmethionine (SAM), demethylation of SAM to S-adenosylhomocysteine (SAH) and hydrolysis of adenosine to regenerate homocysteine (Finkelstein, 2007). Methionine synthase (MS) catalyses the homocysteine-methionine reaction, methionine adenosyltransferase, also known as SAM synthetase, conjugates the adenosine moiety (from ATP) to methionine, and S-adenosylhomocysteine hydrolase restores SAH to homocysteine. The transfer of the folate-derived methyl group from SAM to other substrates is facilitated by a host of methyltransferases (MTases) – fifteen superfamilies in all, with more than 120 classified members (Martin and McMillan, 2002). These MTases provide specific one-carbon additions to almost all molecule types in the cell (Loenen, 2006). In fact, SAM is the

12 Chapter 1 most used enzyme substrate in the cell after ATP (Fontecave et al., 2004). A list of SAM co-substrates includes halometabolites containing bromide and iodine (Kozbial and Mushegian, 2005), the membrane phospholipid phosphatidylcholine (Chiang et al., 1996), neurotransmitters and catecholamines such as noradrenaline (Thomas and Fenech, 2008), ribosomal and transfer ribonucleic acid, the 6’ nitrogen of adenine, and post-translational methylation of proteins. S-adenosylmethionine also provides methyl moieties for the modification of amino groups of lysine or arginine in histones (Ragsdale, 2008) and, importantly with regards to this thesis, DNA methyltransferases (DNMTs) which catalyse the addition of methyl groups to the 5′ carbon of cytosine in cytosine-guanine dinucleotide pairs of genomic DNA. This DNA methylation is involved in the control of gene expression, among other functions, and is discussed further in section 1.5.

Folate pool Purines Histidine

10-Formyl-THF fMet-tRNA Phosphatidylcholine

5,10-Methenyl-THF Thymidylate (dTMP) Catecholamines

Methylcytosine 5,10-Methylene-THF Homocysteine

SAH Proteins 5-Methyl-THF Methionine cycle Histones SAM THF Methionine Ribonucleotides

Figure 1.4 Biosynthetic fates of single-carbon units in the THF pool, incorporating methionine metabolism and the multiple methyl acceptors of S-adenosylmethionine. Interconverting folate derivatives in the folate pool are represented within the shaded box. Dashed arrows indicate multiple enzymatic steps between substrates and products. The group of SAM methyl acceptors is by no means meant to be an exhaustive list, but gives an indication of the major categories to which methyl groups are provided. THF: tetrahydrofolate; fMet-tRNA: formylated methionyl-tRNA; dTMP: deoxythymidine monophosphate; SAM: S-adenosylmethionine; SAH: S- adenosylhomocysteine.

Introduction 13

Although 5-methyl-THF is the principle source of methyl groups for homocysteine remethylation, in certain species choline and its metabolite betaine (trimethylglycine) are also contributors, principally in the liver and kidney. Betaine is the methyl donor in a second homocysteine-methionine reaction catalyzed by betaine homocysteine methyltransferase (BHMT). While dietary choline and betaine contributions to methionine synthesis are of significance in rodents, humans lack any appreciable expression of BHMT outside the liver, and the enzyme choline oxidase, which catalyzes the conversion of choline to betaine is effectively absent in this organ. Consequently, the effect of betaine and choline on overall homocysteine metabolism is slight. Conversely, the enzyme methionine synthase, which utilises 5-methyl-THF for the generation of methionine from homocysteine, is ubiquitiously expressed across all tissues, suggesting that folate-derived moieties are the dominant source of methyl groups for homocysteine remethylation. (Fox and Stover, 2008; Lieber and Packer, 2002).

The commitment of C1 units from the folate pool to the methionine cycle and SAM generation is thus an important step in maintaining methylation homeostasis in the cell. The enzyme responsible for committing these methyl groups is 5,10-methylene-THF reductase, or MTHFR.

1.1.8 Methylenetetrahydrofolate reductase (MTHFR)

As mentioned in section 1.1.5, the reaction catalysed by MTHFR proceeds only in one direction in vivo – 5,10-methylene-THF is reduced to 5-methyl-THF. This means that MTHFR is responsible for “committing” one-carbon moieties to the methionine cycle. (Appling, 1991, Wagner 1995). The MTHFR enzyme exists as a dimer, with catalysis taking place at the N-terminal domain of each subunit. The C-terminal domain contains a binding site for SAM, which acts as an allosteric inhibitor of the enzyme (Jencks and Mathews, 1987). Catalysis requires both NADPH and FAD as cofactors (Thomas and Fenech, 2008).

The gene for human MTHFR resides on chromosome 1 at p36.3 (Goyette et al., 1994). Initial assessment of the MTHFR gene structure revealed a gross coding region of

14 Chapter 1

20.3 kb and a resulting 70 kDa enzyme consisting of 656 amino acids (Goyette et al., 1998). Further investigation has revealed multiple start sites producing transcripts between 7.5 kb and 9.5 kb and alternative splicing in exon 1 producing a 77 kDa isoform of the enzyme. Large transcript heterogeneity is apparent, with multiple polyadenylation sites, varying 3′UTRs between 0.2 and 5.0 kb, and variance in 5′ UTR length, which alters the speed of translation initiation (Gaughan et al., 2000; Tran et al., 2002). The aforementioned allosteric inhibition by SAM and phosphorylation of a threonine residue within the N-domain, which partially mitigates SAM inhibition, regulates the enzyme at the protein level (Yamada et al., 2005). Together, these complex transcriptional and translational regulatory mechanisms underline the importance of MTHFR in drafting one-carbon flux towards the methionine cycle.

Fifteen mutations were identified in the initial description of the MTHFR gene (Goyette et al., 1998). A further nine have been reported since (Sibani et al., 2000). Of these, the C to T substitution at position 677 is the most common and produces a protein with a alanine to valine substitution at position 222. This reduces the binding efficiency of the FAD cofactor and as a result, the enzyme has reduced catalytic activity when compared to wildtype – a decrease of 35% and 70% for heterozygotes and homozygotes of the mutation respectively (Frosst et al., 1995; James et al., 1999; Yamada et al., 2001). Distribution of the altered allele varies considerably between populations around the globe but in Australian caucasians, the frequency of the allele is approximately 0.29 (Wilcken et al., 2003). It is classed accordingly as a polymorphism; in fact, mild MTHFR deficiency is regarded as the “most common inborn error in folate metabolism” with the C677T polymorphism being the primary cause (Fox and Stover, 2008; Pejchal et al., 2006).

Given MTHFR’s pivotal role in committing methyl moieties to homocysteine’s remethylation and subsequent SAM-facilitated methylation reactions, it follows that such a change in enzyme activity could manifest as a redistribution of folate-derived one-carbon units in the folate pool which could consequently affect downstream methylation events. Indeed, presence of the A→V MTHFR isoform results in increased intracellular 5,10-methyleneTHF, decreased 5-methylTHF and produces mild homocysteinaemia (elevated blood homocysteine), particularly under conditions of low

Introduction 15 dietary folate (Jacques et al., 1996). Additionally, the SNP has been implicated in numerous disorders including neural tube defects (van der Put et al., 1995), cardiovascular disease (Klerk et al., 2002), and adverse pregnancy outcomes like preeclampsia (Dekker et al., 1995; Rajkovic et al., 1997). The outcomes of certain cancers may also be modified by the polymorphism, although its influence is not necessarily detrimental (Ma et al., 1997; Slattery et al., 1999).

To some degree, this pleiotropism mirrors the multiple disorders of folate inadequacy, which is unsurprising given their relationship. Section 1.2 discusses the ramifications of both insufficient dietary folate and the MTHFR C677T polymorphism in the context of disease pathogenesis.

1.2 Folate’s role in the pathogenesis of human disease

As alluded to in the opening comments of this chapter, poor dietary intake or errors in folate metabolism can manifest as an array of different pathologies and disease. Mention has already been made of the causative mechanisms by which folate deficiency results in megaloblastic anaemia (section 1.1.1). Indeed, folate deprivation has been classed as one of the most common forms of vitamin deficiency; around 10% of the U.S. population were affected prior to mandatory folic acid fortification (Hercberg and Galan, 1992) and in this country, inadequate folate intake is in the order of 7%. Depletion is consistently higher in females with a prevalence of inadequate intake in those between 15 and 24 years estimated to be greater than 20% (Russell et al., 1999).

Other diseases associated with impaired folate metabolism include neural tube defects (anencephaly, encephalocoele and spina bifida), cardiovascular disease, adverse pregnancy outcomes, neural and neuropsychiatric disorders like Alzheimer disease and depression, and multiple subtypes of cancer. This chapter will focus on folate’s relationship with NTDs and cancer in particular but will also briefly address other pathologies in which folate is implicated.

16 Chapter 1

1.2.1 General effects of folate deficiency and reduced MTHFR efficiency

Poor dietary intake of folate leads to decreased cellular folate. Principal ramifications of this are lowered methionine and elevated homocysteine concentrations, or homocysteinaemia, a principle risk factor in multiple illnesses, which are discussed below in more detail (Durand et al., 2001; Sauls et al., 2007; Ubbink et al., 1993). As adequate 5-methyl-THF is required as a methyl source for methionine synthase to remethylate homocysteine to methionine, an insufficient intracellular folate pool can disrupt this process (see Figure 1.4). Correspondingly, low folate affects S-adeonsyl homocysteine and S-adenosyl methionine, which are elevated and decreased respectively. Increasing folate intake can correct low cellular folate status and reverse elevated homocysteine (Homocysteine Lowering Trialists' Collaboration, 1998).

Integrity of DNA in particular is affected by folate deficiency, with chromosome fragmentation, uracil misincorporation and possible DNA methylation defects all features of inadequate cellular folate levels. For example, hypomethylation of centromeric repeats, leading to chromosome dysfunction during cell division, can arise from folate depletion (Beetstra et al., 2005). Collectively referred to as genomic instability, this topic will be discussed further in sections 1.2.7 and 1.7.3.

Physiological effects of reduced MTHFR catalysis are similar to those of folate deficiency. Individuals who are homozygous for the MTHFR C677T polymorphism, often described as possessing a MTHFR “T/T” genotype, can have lower blood folate and higher homocysteine concentrations in the presence of folate deficiency than heterozygous (C/T) or wildtype (C/C) individuals (Ashfield-Watt et al., 2002; Molloy et al., 1997). Supplementation with folic acid normalizes blood folate status and homocysteine concentrations in T/T homozygotes (Ashfield-Watt et al., 2002).

At a cellular level, various mechanisms are in place to partially ameliorate the effects of folate deficiency. Cell division can be limited to reduce purine and thymidylate requirements, and remaining methyl groups are shuttled towards SAM-facilitated methylation reactions. Enzymes involved in repairing DNA defects are upregulated in response to a decay of structural integrity (Duthie et al., 2010). Folate transporter expression is also suppressed and polyglutamylation increased by raised

Introduction 17 folylpolyglutamate synthetase activity to limit cellular folate efflux (Ifergan and Assaraf, 2008). However prolonged folate deficiency may stymie these processes and result in the breakdown of folate-mediated one carbon metabolism, leading to disease.

1.2.2 Folate and atherosclerosis

Atherosclerosis is a progressive hardening and occlusion of artery walls caused by the accumulation of macrophages, lipoproteins and other atherogenic materials which form plaques. The rupturing of plaques can produce thromboses, often leading to infarction events where blockage of blood flow to critical tissues results in tissue death. Infarctions such as heart attacks and strokes are the biggest cause of death in the developed world (Murray and Lopez, 1997). Atherosclerosis is a multifactorial disease with numerous genetic and environmental contributors. Although not a confirmed cause, elevated plasma homocysteine is one strong and independent risk factor (Durand et al., 2001; Rasouli et al., 2005). Homocysteine is known to be cytotoxic, damaging vascular endothelium and enhancing platelet aggregation, presumably through the production of reactive oxygen species during the oxidation of excess homocysteine to homocystine, and disulphide formation with nonspecific proteins (Austin et al., 2004; Ramakrishnan et al., 2006; Sauls et al., 2007; Weiss et al., 2002).

Because homocysteine is an acceptor of folate-derived methyl-groups, folate status and the MTHFR C677T polymorphism have been a natural focus of investigation apropos of atherosclerotic development. Interestingly, the first description of the “thermolabile” MTHFR (later confirmed as the A→V isoform) also cited an increased incidence of confirmed coronary artery disease accompanying the presence of the enzyme variant (Kang et al., 1991). Similar reports were duly published (Kluijtmans et al., 1996; Morita et al., 1997) and the MTHFR C677T polymorphism was also found to be often associated with hyperhomocysteinaemia (Jacques et al., 1996) and thromboses (Keijzer et al., 2002; Quere et al., 2002). Large meta-analyses collating more than 70 studies describing MTHFR genotype and atherosclerotic outcomes have subsequently reported a higher risk for the polymorphism, however the extent of this risk is disputed and potentially marginal (Lewis et al., 2005a; Wald et al., 2002).

18 Chapter 1

As dietary folate supplementation can reduce elevated serum homocysteine, it has been suggested that raising folate status may, by extension, reduce the incidence of cardiovascular disease (Graham and O'Callaghan, 2002; Refsum et al., 1998). Recent studies assessing this proposition have not supported this hypothesis however – the incidence of atherosclerotic end-point events such as heart attacks has not decreased with supplementation, although homocysteine has been greatly reduced (Albert et al., 2008; Ntaios et al., 2008; Zhang et al., 2008). Increasing folate status in one trial was even deemed to be detrimental (Imasa et al., 2009). These study outcomes stress the multi-factorial nature of atherosclerotic development. As such, the present consensus is that low folate status and the C677T polymorphism may amplify, through homocysteine, other risk factors or predispositions of the disease, although this hypothesis has existed for some time (Refsum and Ueland, 1998).

1.2.3 Adverse pregnancy outcomes

Pre-eclampsia, the occurrence of severe hypertension during pregnancy, can cause maternal endothelium, kidney and liver damage and even cerebral haemorrhage potentially causing death. Hyperhomocysteinaemia is also a risk factor for this condition, presumably because of the amino acid’s oxidative propensity, causing placental endovascular damage (Dekker et al., 1995). Due to folate’s reciprocal relationship with homocysteine, it perhaps not surprising that folate supplementation can lower the re-occurrence of pre-eclampsia (Leeda et al., 1998). The MTHFR C677T polymorphism has not been positively associated with pre-eclampsia but an increased incidence of the T/T genotype has been found in cases of placental abruption, where the placental lining separates from the uterus, fetal growth retardation and still births (Kupferminc et al., 1999). Additionally, a study of spontaneously aborted embryos reported a higher incidence of the C677T polymorphism (Zetterberg et al., 2002).

1.2.4 Neuropsychiatric disorders

Folate inadequacy is implicated in a number of brain-related disorders, although often the relationship is tenuous. These include dementia, incorporating vascular dementia and Alzheimer disease, schizophrenia and depression. As with atherosclerosis, homocysteine is an independent risk factor for dementia or cognitive decline (Selhub et

Introduction 19 al., 2000). Elevated homocysteine has been observed in individuals with vascular dementia (Nilsson et al., 2009) and confirmed Alzheimer disease compared to controls (Clarke et al., 1998; McCaddon et al., 1998) and dementia is more likely to develop when the circulating concentration of homocysteine is high (Seshadri et al., 2002). Cognitive impairment is also observed in the presence of folate deficiency (McMahon et al., 2006).

Although no definitive mechanism is confirmed by which low folate and elevated homocysteine induce dementia, it has been postulated that the vasotoxic effects of the latter could damage vascular endothelial cells in the brain, allow breakdown of the blood-brain-barrier or act as a neurotoxin by generating oxygen radicals, leading to neuronal apoptosis (Mattson and Shea, 2003; Miller, 1999; Tchantchou and Shea, 2008). Low folate, in addition to contributing to raised homocysteine, may exert its effects via the genomic instability processes listed in section 1.2.1 (Kronenberg et al., 2009). Consequently, many studies have attempted to improve dementia and cognitive outcomes by lowering homocysteine through increased folate intake. Generally, however, these have not been successful (Clarke et al., 2003; Lewerin et al., 2005; McMahon et al., 2006).

Publications assessing the MTHFR C677T polymorphism and risk of dementia are conflicting in their findings. A recent meta-analysis of eleven studies concluded that the TT genotype exhibited an increased risk (odds ratio and 95% confidence interval of 1.41 and 1.06-1.88 respectively), although the authors recommended further investigation due to many studies’ small sample size (Liu et al., 2010).

An association between low folate and depression is more robust and was first acknowledged during the 1960s, with a high incidence of folate deficiency in patients with depression and psychosis described in many studies (Bottiglieri, 2005). Folate supplementation can improve the clinical outcome of such patients and increase the efficacy of antidepressants (Coppen and Bailey, 2000; Passeri et al., 1993). Mechanisms are unknown, however. The MTHFR C677T polymorphism has not been implicated in depressive disorders due to lack of investigation.

20 Chapter 1

With schizophrenia, low folate, elevated homocysteine and MTHFR C677T are linked by association but, once again, have no defined causal relationship. A recent meta- analysis of eight studies found a 5 µmol/L increase in homocysteine correlated with a 70% increased risk of developing the disease (Muntjewerff et al., 2006). The same publication assessed ten further studies comparing the C677T polymorphism and disease risk, reporting a 36% increase compared with wildtype. This increased risk has been reported by others (Lewis et al., 2005b).

1.2.5 Neural tube defects

During the fourth week of human embryo gestation, initiation of brain and spinal cord development takes place with the formation of the neural tube (Blom, 2009). The thickened dorsal surface of the ectoderm folds up and around to fuse as a tube, within which neurulation takes place. Partial failure of neural tube closure results in exposure of neural tissue which causes neurological damage (Pitkin, 2007). Closure initiates at multiple points along the midline, and the position of the initiation failure determines the type and severity of the presenting defect (Copp and Greene, 2010). Anterior defects substantially affecting the skull and cerebrum (anencephaly) are usually lethal at birth, whereas some encephalocoeles (protrusions of brain and membranes from the back and front of skull) and spina bifidas (fusion failure of the sacral end of the neural tube which forms the spine) are often survivable but still produce severe neurological symptoms including mental retardation, seizures and limb paralysis.

The incidence of NTDs varies between 0.5 and 2 per 1000 pregnancies (Mitchell, 2005), with the New Zealand rate at the lower end of this range. Correspondingly, around 30 cases of NTDs occur annually in this country. Global estimates of NTDs approximate 240,000 cases per year (Oakley et al., 2004a). Severe NTDs are less common, with anencephaly occurring in around 1 in 200,000 births.

The exact causal mechanisms underlying NTDs are undefined, but the disease is multifactorial with both genetic and non-genetic contributions (Eskes, 1998). Using mouse models, multiple mutations associated with NTDs (around 200) have been identified in genes involved in cytoskeletal proteins and cell cycle, apoptotic and

Introduction 21 signalling pathways – the mutations may disrupt neural tube formation via these processes (Copp and Greene, 2010). Genes involved in folate metabolism, presumably because of its central role in maintaining correct DNA synthesis and methylation, are also implicated. The most well documented instance is the MTHFR C677T polymorphism. The SNP is over-represented in NTD-affected cases (van der Put et al., 1995; Whitehead et al., 1995), with homozygosity for the T allele associated with an almost doubling of NTD risk (Botto and Yang, 2000; Shields et al., 1999). Additionally, the MTHFR polymorphism’s predisposition to hyperhomocysteinaemia is reflected in the elevated homocystiene frequently observed in NTD cases (Mills et al., 1995; Steegers-Theunissen et al., 1995; Steegers-Theunissen et al., 1994).

From a nutritional perspective, replete dietary folate intake and folic acid supplementation, culminating in high cellular folate, and are the principal agents in mitigating NTD formation, with periconceptual folic acid supplements reducing NTD incidence by at least 50% (although the extent of reduction is dependent on the background rate of NTDs). This has lead to major public health initiatives to supplement diets of women of child-bearing age with folic acid (see section 1.2.6). Also, folate status and the MTHFR C677T polymorphism can interact, with low maternal or fetal folate further elevating the risk associated with the T allele. The most acute example is a study of NTD cases and controls whereby a combined T/T genotype and red blood cell folate level less than 400 nmol/L conferred a 13-fold increase in NTD risk (Christensen et al., 1999).

While folic acid supplementation reduces NTD incidence, it does not prevent NTDs entirely, with a significant percentage remaining unresponsive to folate – around 30%. Alternatively, a proportion of pregnancies following non-folate responsive NTD occurrences can instead be rescued by inositol supplementation (Cogram et al., 2004; Groenen et al., 2003). Here, supplementing inositol stimulates protein kinase C activity and upregulates retinoic acid receptor β expression in the fetus which are required for correct closure of the caudal portion of the neural tube. As such, NTDs are not specifically folate-deficiency conditions, but instead it is supposed that folate supplementation can ameliorate effects of pre-existing genetic or environmental disturbances. These contributors may be related to folate (as in MTHFR) or be

22 Chapter 1 perturbations of other growth-related pathways (Spiegelstein et al., 2004). It is probable that inadequate folate impedes correct DNA synthesis and methylation events required for the rapid cellular growth and division that accompanies neural tube closure. Mechanistic evidence, however, is lacking. Interestingly, mice in the absence of a genetic predisposition do not develop NTDs, even when dietary folate is severely depleted (Burren et al., 2008).

Despite this and because human observational data is so compelling, there is considerable impetus to introduce folic acid fortification in this country to reduce NTD incidence. How the call for fortification has arisen in public health circles is described below.

1.2.6 Folic acid, neural tube defects and public health

A seminal study published in 1976 was the first to describe a relationship between folate and neural tube defects. From a cohort of over 900 pregnant women, mothers giving birth to children with NTDs displayed lower serum and red cell folate during their first trimester than controls (Smithells et al., 1976). The authors subsequently proposed that periconceptual vitamin supplementation could ameliorate this phenomenon (Smithells et al., 1981). This was confirmed with additional comprehensive randomised controlled trials where folic acid was administered (Czeizel and Dudas, 1992; MRC Vitamin Study Research Group, 1991). Neural tube defects were reduced by around 50% when consuming 400 µg folic acid daily and 70% when consuming ten times this amount. In reaction to this data, guidelines recommending 400 µg supplementation before and during pregnancy were implemented in America (Centers for Disease Control and Prevention, 1992). However, given that around half of pregnancies in that country are unplanned (Forrest, 1994) and due to neural tube formation often occurring before a pregnancy is known, these recommendations failed to decrease NTD rates to any large extent (Shane, 2003). Consequently, mandatory folic acid fortification in North America was introduced soon after.

Implemented on January 1, 1998, folic acid was added at a level of 1.4 mg / kg of grain (DHHS, 1996) in order to achieve folic acid fortification of all bread, pasta, flour,

Introduction 23 breakfast cereals and rice. This was predicted to increase the daily recommended average intake of folate by approximately 100 µg per person, however actual increases have been reported to be more than double this amount, effectively doubling total folate intake (Lewis et al., 1999; Quinlivan and Gregory, 2003). Fortification in America and Canada has produced a dramatic reduction in the incidence of folate deficiency – it is now classed as a rare event (Joelson et al., 2007). The incidence of neural tube defects has also declined by 20-30% (Erickson, 2002; Mathews et al., 2002; Mills and Signore, 2004).

There have been calls in North America for an increase in fortification levels to decrease NTD incidence further (Brent and Oakley, 2006), but others advise caution, arguing the limit of ameliorating folate-responsive NTDs may have been reached (Shane, 2003). Indeed NTDs still occur in cases where mothers have been supplemented. Advocates of fortification also argue that a reduction in cardiovascular disease incidence as a result of lowered homocysteine would be a concomitant outcome of added folic acid (Oakley et al., 2004b). However, as discussed already, recent studies investigating the lowering of homocysteine by folate supplementation and its effects on atherosclerosis have demonstrated that this is not necessarily correct (Albert et al., 2008; Ntaios et al., 2008; Zhang et al., 2008).

Additionally, there is emerging evidence to suggest that increased folate may accelerate the growth of pre-existing neoplasms (Eichholzer et al., 2006). Consequently, there is some speculation that fortification to reduce NTD incidence may inadvertently expose a population to increased cancer risk (Kim, 2007b). Conversely, adequate folate status may prevent normal tissue from becoming neoplastic in the first place (Kim, 2003). Indeed, folate has a very convoluted relationship with cancer, which will be explained more thoroughly in the next section, and more research in these areas is required to build a more complete picture of the ramifications of national folic acid fortification.

Partly because of these reservations, many countries, including New Zealand, have been hesitant to introduce mandatory folic acid fortification. In this country, reversals of decisions by successive governments and lobbying from various quarters have received

24 Chapter 1 a lot of mainstream media interest and generated considerable public debate. The introduction of folic acid fortification in New Zealand is currently delayed.

Fittingly, as one of the few remaining western countries without mandatory folic acid fortification and with relatively low levels of voluntary fortification, New Zealand is an attractive setting in which to study the wider effects of supplementation in a population not already exposed to high folate levels.

1.2.7 Folate and Cancer

The nature of folate’s influence upon tumorigenesis was realised in the decade immediately following the vitamin’s structural discovery. As folate’s biochemical role in growth and development became clear, folate presented itself as an important requisite for the proliferation of neoplastic lesions. Subsequently, synthetic folate analogues were developed as antiproliferative agents which became the first cancer chemotherapies (Hoffbrand and Weir, 2001). Notably, the antimetabolite and dihydrofolate reductase inhibitor methotrexate (4-amino-10-methylpteroylglutamic acid), which was developed almost 60 years ago, is still used today conjunction with other chemotherapies to treat cancers such as childhood acute lymphoblastic leukaemia (ALL) and non-hodgkin’s lymphoma (Assaraf, 2007; Ifergan and Assaraf, 2008).

Although the disruption of folate metabolism can be beneficial in limiting tumour growth, folate depletion can also deleteriously affect DNA integrity, invoking a propensity towards a malignant cell phenotype. Low levels of cellular 5,10-methylene- THF impede thymidylate synthase’s ability to generate thymidine from uridine, resulting in inadequate amounts of the former and excessive concentrations of the latter. As these two species are chemically similar, uracil becomes a substitute for thymine during DNA synthesis (Olinski et al., 2010). While the cell has mechanisms for excising and replacing misincorporated bases, excessive uracil substitution (up to ten fold) can occur under depleted folate conditions, overwhelming corrective machinery (Blount et al., 1997). This leads to DNA strand branching or breaks, which fragment chromosomes and produce micronuclei (Blount and Ames, 1995; Blount et al., 1997; Duthie et al.,

Introduction 25

2002). In turn, these chromosomal aberrations manifest in the malignant transformation of affected cells (Crott et al., 2007; Melnyk et al., 1999).

Given that folate depletion can invoke such genetic damage, it is unsurprising that numerous observational studies have associated changes in folate status with a modified risk for a wide variety of human cancers. For the most part, a high folate status is protective and a low folate intake increases the risk of contracting the disease; this is true of breast (Ericson et al., 2007; Zhang et al., 1999; Zhang et al., 2003), ovarian (Kelemen et al., 2004; Larsson et al., 2004) and pancreatic (Larsson et al., 2006c) malignancies. In some tumour types the association is weak, for example lung cancer (Cho et al., 2006), or not significant, as with gastric (Larsson et al., 2006a; Vollset et al., 2007), or prostate cancers (Stevens et al., 2006), although often such observational studies are hampered by a lack of statistical power.

The relationship between folate status and cancer risk is not invariably linear, however. As already mentioned, folate is an important component of the biosynthetic pathways responsible for generating DNA and methylating multiple components of the cell and a large folate intake may in fact facilitate accelerated tumour growth (Ulrich and Potter, 2007). Thus, excessive folate intake may negate a lower relative risk afforded by a moderate folate status. Furthermore, some studies have reported an increased risk with high folate status as a result of supplementation, reinforcing folate’s role as a facilitator of cell proliferation (Ebbing et al., 2009; Stolzenberg-Solomon et al., 2006).

This seemingly bimodal relationship between folate and carcinogenesis is exemplified in studies documenting colorectal neoplasia, probably the most thoroughly investigated of all cancers with respect to folate status (Duthie 2010). Collectively, human epidemiological studies suggest that an adequate folate status or folic acid supplementation may prevent colorectal cancer development; a risk reduction in the order of 20–40% is typical (Kim, 2007a, 2008; Sanjoaquin et al., 2005). There is some evidence of low folate levels conferring a decreased risk, however (Van Guelpen et al., 2006). Large prospective cohorts – approximately 16,000 women followed over 16 years, and over 88,000 women assessed over 14 years, for example – add considerable

26 Chapter 1 weight to the supposition that high folate status is protective (Fuchs et al., 2002; Giovannucci et al., 1998). However, recent substantial randomised controlled trials attempting to reduce adenoma recurrence through folic acid supplementation have been conflicting. One study observed a lower incidence of polyps following supplementation when baseline folate status was low, but found no benefit when participants’ blood folate was already adequate (Wu et al., 2009). This result was not observed in other studies – folate supplementation either had no effect (Logan et al., 2008), or was in fact detrimental, increasing the frequency of advanced lesions found in follow-up endoscopy examinations (Cole et al., 2007).

It has been suggested that the timing of folic acid administration may be important in determining whether the vitamin’s effects are favourable or harmful (Eichholzer et al., 2006; Kim, 2007a; Ulrich and Potter, 2007). Whereas folate deficiency may predispose tissues to neoplastic transformation and an adequate folate status is protective against this event, neoplasms that have already developed may have their growth inhibited by a lower folate status and supplementation may instead increase their proliferative rate. This hypothesis is borne out by animal studies. In colorectal cancer mouse models, folate supplementation suppressed tumour formation if administered prior to the development of aberrant foci, the earliest precursor of cancer. After the development of polyps, tumour progression was enhanced when folate status was high (Kim, 2004; Song et al., 2000a; Song et al., 2000b). Similar observations have been made in rat mammary tumours (Baggott et al., 1992; Kotsopoulos et al., 2003; Lindzon et al., 2009).

1.2.7.1 MTHFR C677T as a risk modifier

The C677T polymorphism of the methylenetetrahydrofolate reductase (MTHFR) gene is a potential modifier of risk for several cancers, although once again its effect is not unequivocal. Additionally, the influence of MTHFR may be dependent upon folate status (Bailey and Gregory, 1999). Individuals homozygous for the polymorphism may have a reduced relative risk of contracting cervical and hepatocellular carcinomas and acute lymphocytic leukaemia (Skibola et al., 1999; Wiemels et al., 2001). Conversely the polymorphism is associated with increased risk of breast, esophageal, gastric and

Introduction 27 pancreatic cancers (Larsson et al., 2006b; Ueland et al., 2001). Associations are often tenuous, as with gastric cancer (Vollset et al., 2007).

For the most part, the polymorphism appears protective against colorectal cancer when folate status is replete, with studies reporting up to three-fold reduced risk (Chen et al., 1996; Ma et al., 1997). Curiously, a low folate status has been associated with an increase in neoplasm incidence in some cases (Sharp and Little, 2004) but just the opposite in others (Sharp et al., 2008; Van Guelpen et al., 2006).

1.2.7.2 Folic acid as a public health measure – potentially deleterious ramifications of supplementation

Owing to the aforementioned modulatory effects of folate on tumour development, it is unsurprising that there has been increasing discussion surrounding secondary consequences of folic acid fortification programmes, particularly in North America (Oakley and Mandel, 2004; Shane, 2003; Solomons, 2007). While primarily instigated to ameliorate neural tube defects (see section 1.2.6), the resulting increases in folate status in Canada and the United States as a consequence of fortification have also produced a lowering of blood homocysteine (Pfeiffer et al., 2005) and purportedly a decline in lip and palate defects and cerebrovascular deaths (Yang et al., 2006; Yazdy et al., 2007).

Countering these observations, however, has been the supposed increase in colorectal cancer incidence subsequent to the initiation of fortification. National colorectal cancer rates in both North American countries were assessed between 1986 and 2006 (Mason et al., 2007). Steadily declining rates were observed in both datasets prior to fortification. Coinciding with folic acid introduction, diagnosis rates increased for a number of years before resuming their downward trend. The onset of the spike in Canada trailed the American data uptick by around a year – the time difference in fortification implementation between these countries – reinforcing the possibility that folic acid might be implicated. The analysis calculated a net increase of 5 cases per 100,000 individuals per year, which equates to approximately 15,000 additional cases in the United States per year. This work has been predictated by some to highlight the

28 Chapter 1 unforeseen complexity of folic acid fortification (Kim, 2007b; Solomons, 2007). Furthermore, some have argued that the possible accelerated transformation of polyps into adenomas as a result of fortification may decrease the window of opportunity for a diagnosis to be made, further increasing risk (Scott, 2006).

There have also been concerns regarding the chemical nature of folic acid, compared to natural folates. The fully oxidized form, although not having been exposed to humans before its chemical synthesis, has been traditionally regarded as identical to natural folates once reduction and glutamate addition has taken place (Gregory, 2001). However, as folic acid is converted to tetrahydrofolate by dihydrofolate reductase, a rate-limiting enzyme lowly expressed in the liver and proximal small intestine, high doses of folic acid may saturate this process and the supplement could end up unconverted in the systemic circulation (Abali et al., 2008; Wright et al., 2007). Indeed, folic acid exposures as low as 200 µg have been found to saturate intestinal metabolic capacity (Sweeney et al., 2007). Furthermore, the folate receptor alpha (FOLRα) has an increased affinity for folic acid and is known to be excessively expressed in certain breast and ovarian malignancies associated with poor outcome (Hartmann et al., 2007; Kelemen, 2006). Therefore, growth in these tumours may be exacerbated against a background of folic acid fortification. There is limited clinical data in support of this notion (Charles et al., 2004).

While such assertions should be tempered by the fact that they are indeed observations and therefore do not necessarily infer causality, it does raise important implications for countries planning to initiate folic acid fortification. Additionally, it is clear from the totality of findings presented in this section that the connection between folate and tumorigenesis is a fairly convoluted one, with many mechanisms at play. Indeed, there is another well-described and important mechanism, epigenetically mediated by alterations in DNA methylation, which further confounds the relationship between folate and cancer. Before this is discussed however, it is pertinent to first chronicle what epigenetics and DNA methylation are and how these intersect with folate metabolism.

Introduction 29

1.3 An introduction to epigenetics

DNA methylation is a fundamental component of the phenomenon of epigenetics. Epigenetics refers to the heritable variation in gene expression or cellular phenotype not attributable to differences in primary DNA sequence. While an organism’s cells all hold the same genetic code, it is axiomatic that this information is not expressed uniformly across all cell types. Instead, the modification of additional layers of information residing “on top of” the bare genomic sequence helps contribute to cellular phenotypic variance. The prefix epi literally means “above” or “in addition to” (Tost, 2008). It is important to note that epigenetic modification is a normal and ubiquitous process.

Initially, certain divergent genes become epigenetically suppressed during embryonic differentiation. This “locks in” the cell’s basic expression profile and contributes to the resultant morphological and functional diversity among cell phenotypes. Furthermore, once established, these epigenetic changes can be stable and have the ability to persist through multiple cellular divisions of the differentiated line; in other words, they are heritable (Jablonka and Raz, 2009). In some circumstances, epigenetic marks may even be persistent across generations, passed meiotically from parent to offspring.

The term epigenetics was originally coined in the 1940s to describe the interplay of genes and environmental factors or stimuli that produce a resultant phenotype, long before the mechanisms underscoring such events were known (Holliday, 2006). Today, there is indeed growing evidence that external factors such as diet, for example, have the ability to alter phenotypic outcomes via changes in recently-discovered epigenetic machinery (McGowan et al., 2008). Interestingly, some have suggested that such plasticity under the jurisdiction of environmental cues could constitute an organism’s ability to cognizantly integrate environmental information into its genetic function – in effect, a new kind of Lamarckism (Gorelick, 2004; Handel and Ramagopalan, 2010).

In any case, epigenetic processes allow the cell to enhance and augment the information residing in its DNA and to add nuance to the cell’s control over gene expression. For example, epigenetic mechanisms are responsible for the dosage compensation phenomena of random X chromosome inactivation and gene imprinting. The nature of these processes is discussed below.

30 Chapter 1

1.4 Components of epigenetics

At a molecular level, epigenetics involves two main constituents, these being histone modification and DNA methylation, although it is increasingly evident that non-coding RNAs are implicated in epigenetic maintenance (Yu, 2008). Collectively, these mechanisms facilitate the expression or silencing of targeted genes that ultimately culminate in the desired cell state, referred to as the epigenome (Fazzari and Greally, 2010). This section will briefly introduce histone modification and non-coding RNA, before a more thorough review of DNA methylation in section 1.5.

1.4.1 Histones

The histone family (H2A, H2B, H3 and H4) are highly conserved proteins that, when arranged in an octamer, permit 146 bp lengths of genomic DNA to wind around them to form a nucleosome. Millions of nucleosomes together with other scaffold proteins, collectively referred to as chromatin, allow for the efficient packing of DNA into the nucleus (Sims et al., 2008). The degree to which localised sectors of DNA are packaged also dictates how readily accessible the underlying genetic information is to transcription machinery responsible for gene expression. Generally, condensed chromatin, or heterochromatin, restricts access to DNA sequence and thus represses gene expression, whereas euchromatin is in an open conformation that allows binding of transcription factors and the RNA polymerase complex (Khan and Khan, 2010).

Several covalent modifications of histone N-terminal tails exist which are thought to control the extent of chromatin packing as well as to recruit and regulate attachment of other DNA binding proteins capable of initiating or suppressing transcription. Presently, around thirty acetylation, methylation, phosphorylation and ubiquitination additions are documented, which are facilitated by multiple corresponding acetylases and transferases (Sims et al., 2008). Consequently, deciphering this regulation is quite complex and can be seemingly contradictory. For example, methylation of histone H3-Lys4 leads to a open chromatin structure (Noma et al., 2001; Santos-Rosa et al., 2002) whereas methylation of Lys9 of the same histone invokes chromatin condensation (Nakayama et al., 2001). Such residues can be mono-, di- or tri-methylated, further complicating the understanding of such events (Rice and Allis, 2001). Certainly, the picture describing

Introduction 31 histone modification and control is far from complete. The exact nature of proteins that bind modified histone residues are still being elucidated. Even the mechanism whereby a histone’s modification state is preserved across mitosis is unconfirmed, although division of the octamer in a semi-conservative manner analogous to DNA replication is a possibility (Tagami et al., 2004).

1.4.2 MicroRNAs

The existence of non-coding microRNAs (miRNAs) as epigenetic regulators is a recent discovery (Bergman and lane, 2003). Non-coding RNAs participate in regulation of gene expression at all stages, including transcription modulation, primary transcript processing and mature mRNA stability, and translation repression (Yu, 2008). Notably, miRNAs have also been implicated in modification of chromatin and associated processes like imprinting. For example, H3-Lys9 methylation in both bacteria and mice has been demonstrated to be dependent upon RNA interference (RNAi) mechanisms (Grewal and Rice, 2004; Kanellopoulou et al., 2005). Double stranded RNAs (dsRNAs), transcribed from heterochromatic regions, are processed into small interfering RNAs (siRNAs) and integrated into the RNA-induced transcriptional silencing complex (RITS). This complex directs the histone methyltransferase to further methylate local H3-Lys9 residues, several kilobases around the RNA-targeted site (Yu, 2008).

Additionally, non-coding RNA can assist genomic DNA methylation, a process that also contributes to heterochromatin formation. RNA-directed DNA methylation (RdDM) functions similarly to the aforementioned procedure, but processed dsRNAs are smaller (21-24 nucleotides) and instead recruit DNA methyltransferases to the promoter region homologous to the dsRNA (Kawasaki and Taira, 2004; Morris et al., 2004).

Owing to the fact that only a fraction of non-coding RNAs have been characterised, considerable further investigation is required to determine the extent to which non- coding RNAs control epigenetic events.

32 Chapter 1

1.5 DNA methylation

1.5.1 Introduction

DNA methylation is the most genuine epigenetic mark, with methyl groups covalently attached to DNA being faithfully inherited through cell division together with the “carrier” DNA molecule. However, these moieties are still able to be added or removed independently of genetic duplication (Paulsen et al., 2008).

DNA methylation also contributes to the condensation of chromatin and thus gene suppression. Similarly to the mechanisms by which histone modifications are thought to induce changes in chromatin, the modification of cytosines aids the binding of a family of methyl binding domain proteins (MBDs) that facilitate chromatin modification or act as linkers to recruit DNA methyltransferases to further methylate adjacent cytosines (Van den Veyver, 2002). Alternatively, the presence of a methyl group can interfere with trans factor binding, thus altering the ability of the underlying genetic material to be expressed. Generally, methylation in the promoter or first exon of a particular gene silences that gene. This process is depicted in Figure 1.5.

1.5.2 5′-Methylcytosine and the CpG dinucleotide

A methyl moiety covalently bonded to the 5′ carbon of the pyrimidine ring of cytosine is the predominant form of DNA methylation. In mammals, cytosine methylation generally only occurs when that residue is adjacent to a 3′ guanosine, termed a CpG dinucleotide (Bird, 2002). The “p” represents the phosphodiester bond between nucleotides. Significant methylation of other cytosine-containing di- and trinucleotides does exist however. For example, plants exhibit more complicated CpNpG and CpNpN patterns, where N is any nucleotide (Zilberman et al., 2007). Another example is during fetal development, where non CpG methylation is witnessed (Lister et al., 2009). This section focuses on the CpG dinucleotide.

Introduction 33

Transcription Histone DNA A

Promoter DNMTs accessible methylate cytosines to initiate chromatin state change condensation Chromatin

MBDs recruit histone Histone methyl deacetylases group addition Binding of MBDs

Co-repressor binding Histone No Transcription acetyl group removal C

Promoter Chromatin inaccessible compaction

Figure 1.5 General mechanism of transcriptional silencing involving DNA methylation. Panel A: Active gene promoters are characterised by an open chromatin structure, facilitated by acetylated histones which permit a low nucleosome occupancy (acetyl groups denoted by cyan spheres). Cytosine methylation of the DNA locus is performed by DNA methyltransferases (gold cylinder). Panel B: Addition of methyl groups attracts Trans-acting factors with methyl binding domains (MBDs, denoted as green ovals), a family comprised of MeCP2, MBD1, MBD2 and MBD4. These proteins recruit additional factors to interfere with transcriptional machinery (co-repressors shown as red cylinders in Panel C) and condense chromatin by altering histone tags (histone deacetylases and methylated histone tags represented by yellow cylinders and red cubes, respectively), which also occludes DNA sequence from the polymerase complex. Panel C: As a result, transcription is repressed. Note, the temporal sequence of these events varies and in some instances, histone modification precedes DNA methylation. This is discussed in section 1.5.4. Diagram based on that of Jones and Laird, 1999. Renders composed in Blender 2.49b.

34 Chapter 1

In essence, each CpG dinucleotide pair is a palindrome and the methyl moiety shares this reciprocity with cytosines of both DNA strands being methylated. The methyl groups of complimentary CpGs both protrude into the major groove of the helix (see Figure 1.6, panel A). As DNA is replicated semi-conservatively, the newly synthesised daughter strand at S phase is devoid of cytosine methylation; while adenosine, cytosine, guanine and thymine are present in the cell as free nucleotides, 5′-methylcytosine only exists as a modification of cytosine once the latter has been integrated into the DNA molecule. Therefore, following replication, hemi-methylated CpGs are detected and the remaining dinucleotide is enzymatically methylated (Jeltsch, 2006). This method allows faithful epigenetic transmission from parent to daughter cell. The process is catalysed by DNA methyltransferases (DNMTs).

The methyltransferases are coded by three genes, DNMT1, DNMT3A and DNMT3B, all of which contain several conserved motifs (Van Emburgh and Robertson, 2008). Interestingly, these motifs are present across all studied prokaryotes and eukaryotes (Posfai et al., 1989). Multiple splice variants of each gene’s transcripts exist in humans. Some isoforms only perform key methylation events during embryogenesis, or are sex- specific, while DNMT1, the prime “maintenance” methyltransferase, is responsible for the routine remethylation of the bulk of DNA following replication (Van Emburgh and Robertson, 2008). However, in addition to primarily recognising hemimethylated DNA, DNMT1 can also conduct de novo methylation (Pradhan et al., 1999). Conversely, DNMT3 families are also primarily de novo methyltransferases, but may contribute to maintenance methylation (Liang et al., 2002). DNMT2, originally presumed to act exclusively on DNA, has recently been re-categorised as a RNA methylase (Goll et al., 2006).

Blah

The catalytic mechanism by which DNMTs operate is worth mentioning. Following substrate recognition, one strand of the DNA is flipped out of the helix by the enzyme to expose the acceptor carbon of the unmethylated cytosine. A methyl moiety is then transferred from S-adenosylmethionine to the cytosine (Cheng and Roberts, 2001; Klimasauskas et al., 1994). Although only established in bacteria, this mechanism is

Introduction 35 presumed to occur in mammals due to the very close sequence homology between pro- and eukaryotes (see Figure 1.6, panel B).

A B

Cytosine Methylcytosine Thymine

Figure 1.6 Attributes of 5′-methylcytosine. Panel A: Spatial representation of methyl groups of a CpG dinucleotide pair within the DNA double helix. Methyl moieties protrude into the major groove, which suggests their steric interaction with methyl- binding proteins or transcription factors. Methyl groups are depicted in red. Panel B: Cartoon rendering of the 3-dimensional structure of the Haemophilus parahaemolyticus DNMT-DNA complex (PDB ID: 1MHT). The enzyme (green) “flips” one strand of the DNA (light blue) outwards to expose the unmethylated cytosine residue. The methyl moiety (magenta) is then transferred from S-adenosylmethionine to the 5′ carbon of cytosine. As there is a very high degree of conservation between bacterial and higher eukaryote DNA methyltransferases, this mechanism is also assumed to occur in mammals. Images in panels A and B rendered using PyMol version 0.94. Panel C: Chemical structures of cytosine, methylcytosine and thymine, highlighting the propensity of methylcytosine to deaminate to thymine. The R-group denotes the deoxyribose common to all three species.

36 Chapter 1

Less is known about the process of methylated cytosine removal. Rapid and large scale demethylation is known to occur early during embryonic development (see section 1.6.2), and adult somatic tissues can display a gradual loss of methylation as part of senescence (Issa, 2003; Maegawa et al., 2010) and cancer (discussed in part 1.7.3). This loss of methylation is generally believed to be a passive event, perhaps as a result of a breakdown in DNMT-facilitated methylation over multiple cell divisions. However, active demethylation mechanisms may exist outside of fetal development. The nuclear protein Gadd45a in conjunction with a DNA repair endonuclease XPG, leaves damaged CpGs unmarked after repair, which could be an attractive mechanism to explain age- related hypomethylation (Barreto et al., 2007). There is also accumulating evidence that DNA methylation is dynamic in the brain and involved in memory formation (Levenson et al., 2006; Miller and Sweatt, 2007), and recent work in plants has discovered multiple DNA glycosylases capable of actively performing demethylation via excision ′ of 5 - methylcytosine (Agius et al., 2006; Morales-Ruiz et al., 2006). It is presently unclear if this process extends to humans.

1.5.3 Genomic distribution of 5′-methylcytosine

Under conditions of local strand separation, genomic cytosine is prone to spontaneous deamination, becoming uracil (Fryxell and Moon, 2005). Fortuitously, the cell has the ability to detect, excise and replace uracil in DNA with uracil glycosylase (Lindahl et al., 1997). However, if the cytosine is methylated, the end product of deamination is thymine (see Figure 1.6, panel C), a legitimate base which is invisible to the aforementioned corrective machinery. Consequently, over time, methylated cytosines have transmogrified to thymine residues and CpGs are statistically underrepresented throughout the human genome (Lander et al., 2001).

Correspondingly, around 1% of nucleotides in adult mammalian genomes are 5′- methylcytosine (and, as it follows, approximately 4.5% of cytosines). In a given human somatic cell, this equates to around 30 million CpG dinucleotides (Costello and Brena, 2008). The distribution of cytosine methylation throughout the genome is far from even however, with the bulk of methylation occurring in repetitive elements including transposons and satellite sequences (see Table 1.1). Approximately 75-80% of CpGs are

Introduction 37 methylated (Dahl and Guldberg, 2003). Clusters of unmethylated CpGs exist, however, usually within gene promoter CpG islands.

CpG islands were originally defined as a region with at least 200 bp and with a GC percentage that is greater than 50% and with an observed/expected CpG ratio that is greater than 0.6 (Gardiner-Garden and Frommer, 1987). More stringency has since been added to this definition to exclude much of the CpG content of repetitive elements while retaining islands associated with gene regulation (Takai and Jones, 2002). The revised classification specifies that CpG islands are deemed to possess greater than 60% GC content and be between 0.5 – 2 kb in length. This is often relaxed in circumstances such as algorithmic genome searches (Yamada et al., 2004).

Table 1.1 Distribution of CpG dinucleotides in the human genome. After Paulsen, 2008 (data taken from Rollins, 2006)

Composition of CpG distribution CpG GC content (%) genome (%) (%) Genome 29,848,743 41 100 100

CpG island 1,876,802 62 0.57 5.5 First exon 508,553 56 0.31 1.8

Other exons 1,337,271 48 1.83 4.9

DNA transposons 565,601 29 3.6 2.1

LINE transposons 3,242,225 32 22.3 11.8

SINE transposons 7,479,682 38 16.1 27.2

LTR transposons 1,958,798 37 9.3 7.13

Alpha satellite ~766,000 38 2.07 2.79

Classical satellite ~1,140,000 34 2.1 4.15

Other sequences 8,358,888 42 41.85 32.58

1.5.4 CpG islands help determine gene expression state

The presence of CpG islands around certain gene promoters and first exons, particularly of house-keeping genes, suggests a role in regulation of gene expression. Indeed, the great majority of CpG islands in these regions appear to confer a stable open chromatin conformation and permit ubiquitous expression (Weber et al., 2007). Conversely, many

38 Chapter 1 tissue-specific genes lack CpG islands which may allow them to be shut down more easily upon differentiation. Such genes typically contain a TATA box for a precisely mapped transcription start site and specific transcription factor binding, whereas gene promoters containing CpG islands can have multiple transcription start sites and are frequently devoid of TATA elements (Carninci et al., 2006).

Other factors relating to 5′ CpG islands also contribute to the ultimate expression state of a gene, however. These include the nature of other various regulatory motifs within the island that facilitate transcription factor binding, overall sequence composition of the CpG island and the level of GC content, which can affect the amount of “twist” in the DNA helix and thereby alter the accessibility of associated proteins (Bock et al., 2006). Incidentally, the methylation status of CpG islands further downstream, outside the transcription start site, generally have little bearing on transcription and do not interfere with elongation (Larsen et al., 1993).

Adding complexity, there is considerable crosstalk between CpG island methylation state, histones, trans-acting factors and non-coding RNA. DNA methyltransferases and histone methyltransferases (HMTs) often interact directly to methylate CpGs and alter histone conformation of both genes and satellite repeat sequences. (Lehnertz et al., 2003; Li et al., 2006).

It is important to note too that unmethylated CpG islands are generally, but not always an attribute of positive gene expression. Some strongly unmethylated islands can be devoid of expression (Weber et al., 2007) and conversely, active histone marks have been shown to override methylated CpGs and force transcription (Brinkman et al., 2007). Additionally, it is not clear if there is a specific temporal sequence of histone modification and DNA methylation that culminates in the changed chromatin state (Sims et al., 2008). While the subtleties of CpG island control may not have been completely elucidated, it is clear they play an important part in epigenetic gene regulation.

Introduction 39

1.6 Concomitant roles of DNA methylation

Due to its ability to suppress expression, DNA methylation is utilised by the cell to affect a number of disparate functions that require such repression – the sequential switching off of tissue-specific genes during development, accountancy of various dosage imbalances, or the suppression of parasitic elements. Furthermore, the replication of methylated CpG marks across cell divisions allows DNA methylation to act as a faithful epigenetic inheritance mechanism.

1.6.1 Transposon suppression – DNA methylation as a genetic defence

As shown in Table 1.1, around half of all CpGs reside in repetitive sequences and such sequences constitute a similar proportion of the human genome. The nature of these elements is diverse, including tandem repeats located around centromeres and telomeres and numerous classes of transposable elements. Many of the latter are inactive due to deletions or mutations, but elements such as complete LINE transposons, of which there are almost 100, or LTR retrotransposons (endogenous retroviruses, occupying approximately 10% of the genome) have the potential to become active and invade functional genomic regions (Brouha et al., 2003; Slotkin and Martienssen, 2007). Doing so can cause genomic recombination or chromosome breaks, or disrupt enhancers, promoters, polyadenylation and splicing of adjacent genes (Girard and Freeling, 1999).

Not surprisingly, these elements are suppressed by the cell, being heavily methylated and consequently quarantined by heterochromatin. Flagging of these sequences for methylation is presumed to occur by sequence recognition, but also by virtue of their high copy number – experiments increasing the amount of introduced transgene sequence have been shown to increase its methylation by the host, for example (Garrick et al., 1998).

DNA methylation of these “parasitic” elements occurs across animals, plants and fungi. As such, it has been proposed that DNA methylation originally evolved as a host defence mechanism to protect against foreign DNA integration and that utilisation of DNA methylation for host gene regulation was a later adoption (Yoder et al., 1997).

40 Chapter 1

1.6.2 DNA methylation and development

The most dramatic changes in genomic CpG methylation have been witnessed during embryonic development in mice (it is assumed that the following events occur in humans as well). Almost immediately after conception, the methyl groups inherited from the gametes are scrubbed, to a level of around 30% of a typical somatic cell (Kafri et al., 1992). The paternal genome is actively demethylated within one cell division, whereas the maternal DNA undergoes passive demethylation where multiple rounds of mitosis dilute the methylation marks. By the time the embryo reaches an 8-cell stage, this process is completed and methylation patterns are then re-established in the developing blastocyst. This “general wave” of de novo methylation is completed by about the time of implantation (Sandovici et al., 2008). This process is referred to as “epigenetic reprogramming” or the resetting of epigenetic marks and is thought to return the differentiated parental DNA to a pluripotent state, capable of being re-differentiated as the embryo develops.

Interestingly, many CpG islands, particularly in genes governing early development, are protected from this wave of methylation and instead can be methylated later. As embryonic stem cells (ES cells) differentiate into specific cell lineages, genes dictating ES cell phenotype, such as self-renewal, are suppressed and tissue-specific genes are up-regulated (Shafa et al., 2010). Both methylation and demethylation events are associated with differentiation at these specific loci (Shiota, 2004). While not necessarily initiated by methylation, silenced genes are targeted by de novo DNMTs recruited to bring about chromatin compaction (Bird, 2002). Conversely, demethylation events are accompanied by histone modifications associated with an open chromatin structure (Klug et al., 2010).

Once established in the mature lineage, the resulting methylation pattern is then preserved through subsequent cells divisions; de novo methylation of 5′ CpG islands is regarded as a rare event in normal adult somatic tissues (Jones and Laird, 1999). As such, this maintenance has been thought to contribute to the persistence of the transcriptome across mitosis, termed bookmarking (Sarge and Park-Sarge, 2005).

Introduction 41

1.6.3 X chromosome inactivation

Another intriguing process in mammalian embryo development is X inactivation. One of the two X chromosomes in females becomes transcriptionally inactive (Xi) to yield a single X chromosome gene dose comparable to a male XY embryo (Boumil and Lee, 2001). This process is mediated in part by CpG methylation. Specifically, partial inactivation of the paternal X chromosome is observed from the two-cell to the early blastocyst stage. This is retained in the non-embryo-forming cells, which develop into the placenta, but reactivated at the late blastocyst stage in the inner cell mass which later forms the embryo (Sengupta et al., 2008). One of the two active X chromosomes is then randomly inactivated in these cells so that all future embryonic development is dosage- compensated. The initial paternal X inactivation is carried over from that chromosome’s silencing during spermatogenesis. During oogenesis, the maternal X is marked with an imprint to make it resistant to inactivation in extra-embryonic tissues after fertilisation (Tada et al., 2000).

Embryonic X inactivation occurs when the non-translated RNA Xist, transcribed from the chosen Xi chromosome coats itself over the entire chromosome (Clemson et al., 1996). This induces an ordered progression of histone modification, DNA methylation and heterochromatin formation, transcriptionally silencing the chromosome (Kay et al., 1993; Wutz and Jaenisch, 2000). Once established, heavy DNA methylation maintains gene promoter silencing and a heterochromatic state over subsequent cell divisions. A small, variable fraction of genes escape this suppression, however, which may contribute to expression heterogeneity among females (Carrel and Willard, 2005). DNA methylation may also be a determinant of Xist suppression prior to X chromosome inactivation, with DNMT3a recruited to methylate regions flanking the gene, preventing transcription (Panning, 2008).

1.6.4 Imprinting

The phenomenon of imprinting relates to around 50 known genes in the human genome (Morison et al., 2005). Imprinted genes are characterized by gene methylation (or more specifically, regulatory element patterns) that are specific to each of the parental alleles. Consequently, transcription occurs in a monoallelic, parent-of-origin-specific fashion,

42 Chapter 1 i.e. only from the paternally-inherited allele or the maternally-inherited allele (Ferguson-Smith et al., 1993). Imprints are set up during the formation of parental gametes and escape the resetting of methylation marks in the morula (Reik et al., 2001). However, the generation of imprints is often transient and tissue-specific after this point (Frost and Moore, 2010).

In mammals, imprinting is predominantly associated with genes involved in the regulation of fetal growth and development and postnatal behaviour (Reik and Dean, 2001). Interestingly, many paternally expressed genes promote prenatal growth whereas many maternally expressed genes limit it. It has been postulated that this pattern has arisen evolutionarily to balance paternal and maternal resource allocation and gene fitness – enhanced fetal growth gives a selection advantage to help continue the paternal DNA lineage, but the associated increased nutritional cost may compromise the total reproductive success of the mother, hence the maternal desire to limit fetal growth (Haig and Graham, 1991). It should be noted that not all imprinted genes fit this model, however.

Probably the best-characterised example of imprinting is that of the paternally- expressed insulin-like growth factor 2 gene (IGF2), which codes for a hormone that promotes growth during gestation, and the reciprocally regulated, maternally expressed H19 gene, which transcribes a putative tumour suppressor non-coding RNA (Barlow et al., 1991; Rachmilewitz et al., 1992; Zhang and Tycko, 1992). Both genes are co- localised with numerous other imprinted genes at 11p15.5 and are separated by a differentially-methylated, imprinting control region, DMR1. Lack of DMR methylation on the maternal allele allows for expression of the downstream H19 gene, but also facilitates the binding of methylation-sensitive insulator elements (CTCF) to motifs in the DMR which block 3′ enhancers vital for IGF2 expression. Conversely, paternal DMR methylation suppresses H19 expression and prevents CTCF insulators from occluding IGF2 enhancers (Du et al., 2003; Takai et al., 2001; Webber et al., 1998).

Errors in methylation of this region, such as loss of imprinting or hypermethylation of the maternal H19 DMR, result in under or over expression of IGF2 respectively. Overgrowth conditions like Beckwith-Wiedemann Syndrome can often be attributed to

Introduction 43 this dysregulation, which also have a predisposition towards malignancies such as Wilms tumour (Cui et al., 2001; Weksberg et al., 2003). In fact, as described in section 1.7, numerous perturbations in the wider methylome can have multiple disease outcomes.

1.7 Improper DNA methylation culminates as disease

The most striking examples of disrupted DNA methylation or associated machinery manifesting as disease are the Fragile X, ICF (immunodeficiency, centromere instability and facial anomalies) and Rett syndromes, together with the imprinting disorders Prader-Willi syndrome, Angelman syndrome and the aforementioned Beckwith Wiedemann syndrome. Less defined but nonetheless significant are the non-specific changes in DNA methylation that accompany the development and progression of cancer. Additionally, recent studies have tentatively associated improper DNA methylation with widespread chronic diseases such as atherosclerosis (reviewed in section 1.7.2).

1.7.1 Syndromes mediated by errors of DNA methylation, in brief

Fragile X syndrome is characterised by a spectrum of physical, behavioural and intellectual disabilities including memory problems and social anxiety (for a review, see O’Donnell and Warren, 2002). The syndrome arises from a promulgation of a repeated CGG sequence within the Fragile X Mental Retardation 1 (FMR1) gene on the X chromosome; FMR1 is required for correct neuronal development and plasticity. Excessive repetition of the trinucleotide attracts CpG methylation which results in incorrect suppression of the gene. The severity of symptoms depends on repetitive sequence length and the consequent degree of repression.

Not caused by aberrant methylation per se, Rett syndrome is most often attributed to mutations of the MeCP2 gene, the protein of which is normally responsible for binding to methyl-CpGs and inducing heterochromatin and transcriptional repression (see Figure 1.5). Loss of MeCP2 severely affects neuronal function and individuals with Rett

44 Chapter 1 syndrome present with slowed developmental progress, erratic limb movements and asocial, autistic-like behaviour (Weaving et al., 2005).

ICF syndrome, an autosomal recessive condition, is also a product of damaged methylation machinery. Mutations in DNMT3B result in hypomethylation of centromeres, causing aberrant chromosome branching and breakage. As its name suggests, individuals with the disease are prone to severe infections due to immunoglobulin deficiencies, have wide-set eyes, an enlarged tongue, low set ears and are mentally retarded (Xu et al., 1999; Ehrlich, 2003).

The imprinting disorders Angelman and Prader-Willi syndromes both occur because of the loss of an imprinted region in chromosome 15 (15q11-13), which is usually paternally expressed and maternally repressed. Loss of the paternal copy results in PWS whereas maternal deletion produces AS (Nicholls et al., 1998).

Cases of these diseases are sporadic and rare. Conversely, an aberrant methylation profile is implicated in atherosclerosis and is consistently found in multiple common cancer subtypes.

1.7.2 Aberrant methylation in cardiovascular disease and other human pathologies

There are multiple reported instances of DNA hypomethylation occurring in atherosclerosis, specifically in the immediate atherosclerotic lesion (Hiltunen et al., 2002) as well as in peripheral white blood cells (Castro et al., 2003) and smooth muscle (Hiltunen and Yla-Herttuala, 2003; Yideng et al., 2007). Being a multifactorial disease, the hypothesis that hypomethylation could be a contributor to atherosclerosis was raised about a decade ago, although the aberration could also be a secondary event. As already mentioned, elevated homocysteine is a risk factor for the disease. Homocysteine and S- adenosyl homocysteine are efficient inhibitors of DNA methyltransferases which could manifest as genomic hypomethylation, implying a secondary development. However hypomethylation is often observed early in atherosclerotic development (Lund et al., 2004) and 5-lipoxygenase and 15-lipoxygenase genes, key enzymes implicated in

Introduction 45 atherosclerosis development (Zhao and Funk, 2004), are upregulated under promoter hypomethylation (Liu et al., 2004; Uhl et al., 2002).

Limited research has also found associations between genomic hypomethylation and Alzheimer disease (Tohgi et al., 1999; West et al., 1995) and rheumatoid arthritis (Neidhart et al., 2000; Nile et al., 2008).

1.7.3 Improper DNA methylation and cancer – a well established example

Cancer has been traditionally regarded as a cumulative series of genetic alterations resulting in deregulated cell growth. In the early 1980s, hypomethylation of RAS oncogenes in malignant tissue were reported, indicating epigenetic factors may contribute to tumorigenesis (Feinberg and Vogelstein, 1983). Altered DNA methylation is now regarded as one of the most common molecular changes in human neoplasm development (Baylin et al., 1998; Jones and Baylin, 2002).

Interestingly, aberrant methylation coexists as a non-specific global hypomethylation of the genome and defined regions of hypermethylation. A decrease in global methylation occurs early in tumorigenesis, and its extent mirrors the progression of the disease (Hernandez-Blazquez et al., 2000; Watts et al., 2008). Hypomethylation can up-regulate proto-oncogenes as well as alter heterochromatin or satellite DNA. The latter perturbations can activate latent transposons and induce chromosomal rearrangement, instability and poor chromosomal segregation during mitosis (Bouzinba-Segard et al., 2006; Howard et al., 2008; Wong et al., 2006). Cells with this hypomethylated attribute have been described as having a “mutator phenotype”, as the propensity for chromosomal deletion or recombination can elevate mutation rates in other genes (Chen et al., 1998).

For the most part hypermethylation is confined to discrete promoter CpG islands with neighbouring genes being unaffected, however entire chromosomal bands undergoing methylation-mediated suppression have been reported in colorectal and breast cancer cases (Frigola et al., 2006; Novak et al., 2006). Hypermethylation of gene promoters causes their suppression and most aspects of tumorigenesis are affected by such aberrant methylation. Disruption of genes involved in cell cycle (Ferguson et al., 2000; Greger et

46 Chapter 1 al., 1989; Herman et al., 1994; Liang et al., 2000; Merlo et al., 1995; Sakai et al., 1991; Stirzaker et al., 1997), DNA repair mechanisms (Esteller et al., 2001; Esteller et al., 2000; Hegi et al., 2005; Herman et al., 1998) which can induce chromosomal instability (Ahuja et al., 1997), and apoptosis (Conway et al., 2000; Katzenellenbogen et al., 1999; Teitz et al., 2000) are all affected.

Imprinted genes are frequently influenced too, with loss of imprinting (LOI) of IGF2 occurring in colorectal, breast, liver and bladder cancers (Cui et al., 2002; Ito et al., 2008; Takai et al., 2001; Tang et al., 2006), and H19 LOI in colon and lung cancers, for example (Cui et al., 2002; Kondo et al., 1995).

Exactly how aberrant methylation manifests during tumorigenesis and what drivers are responsible is an area of active research. Numerous environmental effectors, particularly diet, have been implicated in both cancer and wider aberrant methylation events. These are elaborated upon in parts 1.8 and 1.9.

1.8 Environmental factors influence (disease) phenotypes through DNA methylation

Subsequent to the highly regulated changes during development, it has been generally assumed that the methylome is fairly immutable and uniform in adult cells. There is, however, the potential for the epigenome to be modulated by external factors, under certain circumstances. In the case of DNA methylation, there is an accumulation of evidence to suggest that exogenous effectors, including diet, have the ability to alter methylation either of the individual or that individual’s progeny. As a caveat, this usually occurs perinatally, when the methylome is particularly plastic. It has been postulated that such changes may manifest as a disease phenotype in the individual or offspring, potentially not until later in life. Additionally, methylation changes brought about by transitory external stimuli may persist for multiple generations.

Introduction 47

1.8.1 Retrospective studies

There are a limited number of epidemiological studies where associations between nutrition and disease outcomes have been attributed to epigenetics, albeit tenuously. For example, epigenetic profiling of 80 monozygotic twins found that variation in global and locus-specific DNA methylation and histone acetylation accumulated between twins with age, mirroring age-dependent morphological differences, or disease susceptibility discordance. A difference in diet between twins was given as one possible explanation (Fraga et al., 2005). Also, a recent Swedish study reported a link between men starting smoking at an early age and a higher body mass index in their sons, which they tentatively attributed to epigenetics (Pembrey et al., 2006). The same laboratory assessed other cohorts that were exposed to poor nutrition at discrete time periods early in life. Participants’ grandchildren subsequently displayed a decreased risk of heart disease, but only if the grandchild was of the same sex. Conversely, an excess of food in the grandfather produced a four-fold increase in risk of death from diabetes in the grandchild (Kaati et al., 2002; Pembrey et al., 2006). The sex-specific results in particular, suggest an epigenetic influence, although changes in DNA methylation were not investigated.

More robust examples are found in the Hunger Winter Families Studies. In 1944, the occupied Western territories of the Netherlands underwent severe famine, due to a particularly harsh winter and a Nazi-imposed food embargo. Throughout this period however, birth registries and healthcare continued and existing food rations were well documented. This allowed the famine to be clearly defined and exposed individuals to be traced. Various studies have investigated the epigenetic effect of the famine on those who were in utero at that time. Interestingly, such individuals later displayed aberrant IGF2 imprinting, six decades after the event, compared to their siblings who were not exposed to such periconceptual shock (Heijmans et al., 2008). Further methylated loci have since been shown to be differentially methylated in these individuals, often in a sex-specific manner (Tobi et al., 2009). Interestingly, colorectal cancer incidence was decreased for individuals who were adolescents during the famine, despite hypermethylation of discrete loci associated with a neoplastic phenotype (Dirx et al., 2003; Hughes et al., 2009).

48 Chapter 1

1.8.2 Experimental studies

The epidemiological nature of the above studies makes delineating specific contributors to disease difficult, whether they be epigenetic or otherwise. More success has been achieved in experimental animal model systems.

Agouti mice are characterised by a variable coat colour that is epigenetically mediated. Mice with the Avy (agouti, viable yellow) allele together with a wildtype allele (a) display a range of coats: yellow, where the agouti gene is overexpressed, through increasingly mottled shades due to partial agouti overexpression, to brown, which is referred to as psuedoagouti and which has minimal aberrant expression (Wolff et al., 1998). Furthermore, ectopic expression of the agouti allele interferes with metabolic and endocrine systems in multiple tissues resulting in increased morbidity later in life; yellow and mottled mice are prone to obesity, cancer and diabetes, while brown psuedoagoutis are not (Wolff et al., 1999).

The variation in agouti gene expression is dependent upon the methylation of a neighbouring upstream LTR promoter, termed an intracesternal A particle, or IAP (Cooney et al., 2002). The degree of methylation in this element produces the coat colour spectrum (Michaud et al., 1994; Morgan et al., 1999). Intriguingly, experiments that altered maternal nutritional supplementation shifted the epigenetic spectrum of the mother’s litter – methyl donors such as methionine fed to the mother increased methylation of the IAP, suppressed agouti gene expression and pushed the colour range of her offspring towards the brown end of the spectrum. This also improved morbidity in the pups when fully grown (Cooney et al., 2002; Waterland and Jirtle, 2003). Furthermore, such modified agouti phenotypes as a result of supplementation persisted for multiple generations (Cropley et al., 2006).

The agouti model has also been used to probe other environmental exposures, such as the effect of maternal alcohol consumption on developing embryos. In this case, early gestational exposure to ethanol produced hypermethylation of the Avy allele and produced more offspring with a pseudo-agouti-coloured coat (Kaminen-Ahola et al., 2010).

Introduction 49

More nebulous effectors like behaviour have also been shown to influence DNA methylation and produce different phenotypic trajectories. Rats that show high levels of care and attention towards their pups in the first week of life, in the form of licking and grooming, produce lower stress responses in the offspring as adults (Francis et al., 1999; Liu et al., 1997). Cross-fostering of pups to mothers that did not groom as much created higher-stressed pups in adulthood. This suggested an epigenetic rather than genetic mechanism. High rates of maternal licking and grooming altered the expression of multiple genes in the hippocampus of the young, including increased hippocampal glucocorticoid receptor (GR) expression, a protein implicated in short and long term response to stressors (Weaver et al., 2004; Weaver et al., 2005). Differences in the DNA methylation and histone acetylation of the GR promoter were apparent by day eight in the pups between high care and low care mothers and these differences persisted through adulthood. Additionally, infusing trichostatin, a histone deacetylase inhibitor, into the brains of adult offspring of low care mothers increased histone acetylation, decreased cytosine methylation in the GR promoter and increased expression of the gene. Response to stress became comparable to rats who received high levels of licking and grooming as pups (Weaver et al., 2004). Conversely, injecting methionine into the brains of offspring from high-care mothers increased the methylation state of the GR promoter, decreased GR expression and increased their anxiety (Weaver et al., 2005; Weaver et al., 2006). Methionine treatment altered methylation patterns of a further 300 hippocampal genes, suggesting that such a phenomenon was not limited to the GR locus (Weaver et al., 2006).

There is a paucity of comparable human experiments that wholly link environmental events to discrete methylation changes which then manifest as a disease phenotype. Some studies have addressed this relationship in part, however. For example, a recent human study drew similar conclusions to the work of Weaver et al. – the level of maternal nurturing of around 500 infants at 8 months predicted their ability to deal with stress as adults. No epigenetic mechanisms were assessed, however (Maselko et al., 2010). Additionally, numerous studies have linked elements of folate metabolism, DNA methylation and disease and these are discussed below and in section 1.9.

50 Chapter 1

1.9 Does inappropriate folate status or the MTHFR C677T polymorphism lead to disrupted DNA methylation, altered gene expression and thus disease?

As indicated in many of the above studies, diet can play a role in modulating DNA methylation. Furthermore, folate is an essential dietary component and an integral part in the generation of methyl groups for cytosine methylation, as alluded to in section 1.1. Therefore, it might be surmised that interrupted folate metabolism, either through poor dietary intake or polymorphisms such as MTHFR C677T, may result in an impaired ability to correctly methylate genomic cytosine residues, which in turn may be a driver of disease development. This may also be a mechanistic explanation for many of the aetiologies detailed throughout this introduction. Certainly, diseases associated with inadequate folate status or MTHFR inefficiency, such as cancer, neural tube defects or atherosclerosis, are also associated with DNA methylation disruption.

1.9.1 Evidence of folate as an environmental influence

Again, rodent models provide good evidence of a relationship between disease, DNA methylation and folate. With regard to the agouti mouse experiments outlined in the previous section, folic acid (5-15 mg per kg of feed) was a component of the diet given to dams that improved one-carbon metabolism and methionine synthesis, induced IAP hypermethylation and altered coat colour in the offspring (Cooney et al., 2002). Similar results were observed in experiments assessing the mouse AxinFu locus which also contains a retrotransposon. Hypomethylation of this element induces expression of a truncated Axin gene that manifests as a kinked tail. A high methyl diet including folic acid administered periconceptually to dams produced offspring with increased DNA methylation across the AxinFu region and reduced tail kinking by half (Waterland et al., 2006).

Comparable experiments have been performed in other animals. Folic acid supplementation in pregnant female rats fed a nutrition-deficient diet rescued their adult offspring from aberrant methylation of numerous genes, including the glucocorticoid receptor, and late onset diseases that otherwise resulted from maternal malnutrition

Introduction 51

(Burdge et al., 2008; Lillycrop et al., 2005; Lillycrop et al., 2008). Periconceptual restriction of dietary folate in female sheep produced offspring that were fatter, insulin- resistant and had elevated blood pressure as adults. These outcomes were accompanied by appreciable changes in DNA methylation in the liver of the animals (Sinclair et al., 2007).

In relation to specific disorders, limited work has assessed the relationship between folate, DNA methylation and neural tube defects. Conversely, cancer studies probably provide the best evidence to date of folate and MTHFR’s mechanistic involvement in aberrant DNA methylation and disease formation.

1.9.2 Folate, DNA methylation and neural tube defects

The higher incidence of NTDs in females, which is not due to elevated male embryo death, has caused speculation that a chromosomal gender effect is responsible in some cases. The rapidly dividing cells of female embryos during neural tube formation may incur an increased requirement for methyl moieties because of the methylation and inactivation of half of all its X chromosomes (Juriloff and Harris, 2000). This in turn would place increased demands on one-carbon metabolism, including sourcing of folate, when compared to male embryos. Lesser availability of methyl groups for other methylation needs may be a ramification of this, which could lead to disrupted tube closure.

With particular relevance to public health folic acid fortification policies, maternal folic acid supplementation of 400 µg was recently documented to slightly increase CpG methylation of IGF2 by 4.5% and lower birthweight (decrease of 1.7% methylation per standard deviation of birthweight) in the child (Steegers-Theunissen et al., 2009). Whether this methylation phenomenon occurs in a widespread manner across the genome or what other genes, if any, are appreciably affected remains to be investigated.

52 Chapter 1

1.9.3 Folate, the MTHFR polymorphism, methylation and cancer

Section 1.2.7 has described the observational findings and mechanisms that ostensibly associate folate and cancer. Additionally, as detailed in part 1.7.3, aberrant DNA methylation is an early hallmark of tumour progression, contributing to altered regulation of tumour suppressor and oncogenes, as well as inducing chromosomal instability. Owing to the fact that folate plays an important role in providing methyl moieties for DNA methylation, it might be reasonable to assume that irregular folate metabolism may invoke improper methylation of DNA and thus contribute to tumorigenesis. In effect, this would be a second mechanism whereby atypical folate status or metabolism could influence cancer development, in addition to the interruption of dTMP synthesis and resultant uracil incorporation into DNA.

Indeed, there are numerous data that support such a connection. For example, folate depletion of various cancer cell lines perturbs DNA methylation – SV40 immortalised colonocytes and SW620 human colon adenoma lines grown in folate-free medium exhibit global hypomethylation and altered methylation of tumour suppressor genes (Duthie et al., 2000; Wasson et al., 2006). Folate deficiency-induced heptocarcinogenic rat models exist in which global hypomethylation occurs as an early onset event and is progressive over the course of the disease (Pogribny et al., 2004). Increasingly aberrant methylation of p53 and p16 CpG islands also accompanies this tumour type’s progression (Pogribny and James, 2002; Pogribny et al., 1997). In humans, folate deficiency has been reported to cause altered global and gene-specific DNA methylation and increased susceptibility to colorectal, head, and neck cancers (Kim, 2005; Kraunz et al., 2006). Additionally, increased frequency of LINE element hypomethylation has been associated with a low dietary folate intake in human colon tumours (Schernhammer et al., 2010).

However, once again the evidence is not unequivocal. For example, disparate results from other in vitro studies suggest that folate’s effects on DNA methylation may be dependent on the type and temporal stage of the malignancy (Duthie et al., 2008; Stempak et al., 2005). This mirrors the general inconsistencies between folate status and cancer outlined previously.

Introduction 53

Similarly, the MTHFR C677T polymorphism has been proposed to confer cancer risk in a bimodal fashion (Choi and Mason, 2000; Kono and Chen, 2005). On one hand, the enzyme variant’s reduced catalytic capability may reduce the shuttling of methyl groups towards DNA methylation, resulting in hypomethylation and increasing cancer risk. Conversely, lower MTHFR activity may conserve the 5,10-methyleneTHF pool, maintaining correct DNA synthesis and thus reducing cancer risk. Furthermore, risk associated with MTHFR may be under the influence of folate status. These hypotheses attempt to reconcile evidence that, a priori, can often appear contradictory (see Figure 1.7).

Low Dysregulation Low CpG INCREASED of gene folate methylation CANCER Chromosomal expression RISK? loss or rearrangement Low MTHFR activity Low uracil DECREASED Adequate incorporation Chromosomal CANCER folate into DNA stability MTHFR RISK? genotype Normal CpG methylation Correct gene Normal MTHFR expression INCREASED activity Low CANCER folate High uracil RISK? incorporation Accelerated growth into DNA Excessive of pre-existing folate neoplasms?

Figure 1.7 Proposed interrelationship between MTHFR C677T genotype, folate status and DNA methylation, with respect to cancer. Some steps are simplified for brevity – the dysregulation of gene expression, for example, incorporates tumour suppressor gene silencing or oncogene activation events. Based on figure from Thomas and Fenech, 2008.(Thomas and Fenech, 2008)

This is exemplified in a recently published study that examined the effect of the MTHFR polymorphism on DNA methylation in carcinoma cell lines from different tissues (Sohn et al., 2009). Genomic methylation was higher in HCT116 colon adenocarcinoma cells possessing the MTHFR T/T genotype compared to wildtype, but low folate induced almost 10% hypomethylation in the T/T cells. Contrastingly, hypomethylation worsened (over 16% decrease) with increasing folate concentration in

54 Chapter 1

MDA MB-435 breast carcinoma cell lines possessing the MTHFR T/T genotype compared with wildtype. Despite this, no significant difference in methylation of multiple promoter CpG sites was found between genotypes, with the exception of hypomethylation in the promoter of tumour suppressor gene STK11. Expression of this gene was unaffected, however.

1.10 Evidence for perturbations in folate metabolism and MTHFR affecting DNA methylation in humans

Once established, methylation of DNA has been traditionally regarded as fairly immutable. The above examples of perturbed DNA methylation accompanying cancer development are obvious exceptions to this dogma, however. That modulation of folate status can influence DNA methylation in these scenarios is of added significance and raises the question of just how plastic DNA methylation is in response to folate status, not only in a disease context, but also in normal tissue. The possibility that diet could habitually modify epigenetic programming and perhaps induce a propensity towards disease is a powerful notion indeed. This is dependent on substantial evidence implicating folate in the modulation of DNA methylation, however. Correspondingly, a summary of studies assessing how low folate or the MTHFR polymorphism influence DNA methylation in human populations is presented below.

In a small study (n = 8) of post-menopausal women, hypomethylation of lymphocyte- derived DNA was observed when subjects consumed a low folate diet. Folate supplementation reversed this effect (Jacob et al., 1998). Rampersaud and co-workers showed, in a larger group of older women (n=30; 60-85 y), that a moderately folate- depleted diet decreased leukocyte DNA methylation; folate supplementation did not reverse this effect however (Rampersaud et al., 2000). It should be noted that both these studies explored solely the relationship between folate and DNA methylation – decreases in methylation as a function of MTHFR genotype were not investigated. In a study of patients with hyperhomocysteinaemia, a marked hypomethylation of peripheral blood leukocyte DNA was observed compared with controls. This was assumed to be because S-adenosyl-homocysteine is an efficient inhibitor of DNMT1. Folate

Introduction 55 supplementation of these participants restored both global DNA methylation and at the IGF2-H19 locus (Ingrosso et al., 2003).

With regard to the MTHFR C677T polymorphism and of particular relevance to the present thesis, Friso and colleagues, in an observational study of a group of subjects in Italy, reported a 50% lower level of leukocyte DNA methylation in individuals possessing the MTHFR T/T genotype when combined with a low folate status, compared with high folate status T/T individuals. DNA methylation was static and independent of folate status in wild-type individuals (Friso et al., 2002). Similar results in other observational cohorts had been described (Stern et al., 2000), or were subsequently reported following the initiation of the current study (Castro et al., 2004), although participant numbers were low (n = 10 and n = 9 for T/T participants in respective studies). Hypomethylation of DNA in T/T individuals with low folate status was less pronounced in these studies than in the cohort described by Friso and colleagues. Lastly, there have been two small intervention studies (n = 41; n = 43) in which women, divided into MTHFR C/C and T/T subsets, had their leukocyte DNA methylation assessed following restricted and supplemented folate intake (Axume et al., 2007; Shelnutt et al., 2003; Shelnutt et al., 2004). No significant decrease in DNA methylation was observed in either study by genotype following folate depletion. A small increase in DNA methylation after folic acid supplementation occurred in the women with the T/T genotype (n = 10) in one study (Shelnutt et al., 2004). Surprisingly, in the other study DNA methylation fell during the supplemented period in the T/T group (n = 17) (Axume et al., 2007).

1.11 Aims of the study

When the above data is considered together, it is clear that further work is required to assess the effect of folate and the MTHFR C677T genotype on global DNA methylation. Although the studies cited focused on the impact of folic acid on DNA methylation, with some including MTHFR genotype, they were either small in size, purely observational, or only looked at distinct subsets of the population – the elderly or hospital admissions, for example. While Shelnutt et al. had sought to describe the folate/MTHFR/DNA methylation relationship in greater detail, again, the number of

56 Chapter 1 participants possessing the T/T genotype was low (n = 10) and the results were applicable to young women only. Outcomes from this study and the work of Axume and colleagues were conflicting. No studies had assessed a large group of individuals from a healthy general population.

Accordingly, the primary motivation of the current project was to conduct an intervention trial that assessed changes in DNA methylation as a response to modulated folate intake and MTHFR genotype, similar to the University of Florida study (Shelnutt et al., 2004), but with a greater sample size consisting of both men and women. Additionally, the present study would recruit participants from a disease-free general population. Conducting the study outside a metabolic unit setting would enable results to be more clinically realistic, in other words, more reflective of a real world response to modulated folate intake. Differences in global DNA methylation between individuals with all MTHFR C677T genotypes would be measured, and the folate status of participants would be changed by modulating folate intake over the course of the study. Details of the trial are outlined in Chapter 3. The analysis of participants’ total methylation is reported in Chapter 4, which also discusses the merits and limitations of liquid chromatography-coupled mass spectrometry (LC/MS) and high performance liquid chromatography (HPLC), two methods that were considered for the assay.

Secondly, and differing markedly from previous studies, the current work would attempt to investigate changes in DNA methylation of specific genomic elements as well as changes in gene expression potentially resulting from altered methylation of regulatory regions. It is important to note that the aforementioned studies in section 1.10 only assessed global DNA methylation. As already mentioned, changes in total methylation often accompany the progression of diseases such as cancer (Barton et al., 2008; Gronbaek et al., 2007) however a global methylation change phenotype does not necessarily imply an intrinsic causal mechanism for such a disease. This is exemplified by the phenomenon of aging, where gradual decreases in overall DNA methylation are witnessed with age, but do not equate with a specific disease outcome (Issa, 2003; Richardson, 2003).

Introduction 57

A more pertinent observation is that none of the above studies, with the exception of Ingrosso et al., noted any morbidity differences between participants with high and low folate status, where total DNA methylation was, especially in the case of the Italian publication, drastically different between the groups. In short, global DNA methylation analysis provides only a partial picture of any potentially disease-causing epigenetic differences and may not always be relevant. Thus, in the current study, microarrays would be employed to assay for changes in expression of individual genes, with the justification that modified expression could act as a proxy for possible altered promoter methylation. Furthermore, the assessment of methylation in satellite 2 repetitive elements would provide an indication of any changes in chromosomal integrity as a result of the dietary intervention. Owing to the fact that there are multiple pathologies associated with aberrant folate status, it was anticipated that these assays could reveal a more coherent picture of genetic and epigenetic indicators for disease propensity. Microarray analysis is discussed in Chapter 5 and assessment of Satellite 2 methylation is reported in Chapter 6.

Finally, the study would provide pertinent information regarding mandatory folic acid fortification legislation. At the time of planning, a number of countries including New Zealand were considering implementing mandatory folic acid fortification of food staples to help reduce neural tube defects (NTD) rates (Food Standards Australia and New Zealand, 2007; US Department of Health and Human Services, 1996). While it is undisputed that folic acid lowers the incidence of these diseases, the apprehension towards fortification due to potential risks associated with neoplasm development, as discussed in section 1.2.7.2, has delayed the introduction of fortification in this country. As participants would be sourced from a population not already exposed to mandatory folic acid fortification, certain changes in discrete gene expression as a result of increased folate exposure, that may infer an increased propensity toward disease, would be of interest, especially as the majority of studies of folate, the MTHFR C677T polymorphism and DNA methylation interactions have taken place in regions of high dietary folate (including folic acid) intake.

58 Chapter 1

2 Materials & Methods

2.1 Chemicals and materials

All chemicals were purchased from Ajax chemicals, AppliChem, BDH or Scharlau and were analytical grade, unless stated otherwise. Electrophoresis grade agarose (LE) was purchased from Bio-Rad. CyDye monofunctional NHS-reactive fluorescent dyes were obtained from GE Healthcare, formerly Amersham Biosciences. Amicon YM-3 Microcons were from Millipore. PCR primers were obtained from Proligo. Ethidium bromide, polyethylene glycol 500,000 and 2′-deoxycytidine and 5′-methyl-2- deoxycytidine standards were purchased from Sigma.

2.1.1 Enzymes Restriction enzyme HinfI was purchased from New England Biolabs. AmpliTaq Gold™ DNA polymerase and accompanying buffers were purchased from Applied Biosystems (now a subsidiary of Roche). Proteinase K was from Merck. RQ DNase was purchased from Promega. RNase A, RNase T1 and alkaline phosphatase were obtained from Roche. Venom phosphodiesterase I and S1 nuclease were purchased from Sigma. Nuclease P1 was made by US Biological.

2.2 Buffers and solutions Buffers and solutions are listed in their relevant sections in the text.

2.3 Folic acid supplementation trial

A non-randomized intervention study provided the foundation for the current thesis. The trial comprised of two phases, each of approximately 12 weeks in duration. The first consisted of a folic acid depletion period, where participants limited their intake of

60 Chapter 2 dietary folate in order to reduce their folate status to non-fortified levels. For the second phase, folic acid was supplemented to elevate participants’ folate status. Equal numbers of subjects were grouped according to MTHFR C677T genotype to assess any differential gene expression and DNA methylation effects of supplementation among genotypes.

2.3.1 Ethical approval

Ethical approval was sought for genotyping of potential participants and for the intervention study. Both applications were approved by the University of Otago Ethics Committee.

2.3.2 Study preparation

Vitamin capsules were prepared by Vita-Fit Nutritionals Ltd, Christchurch, New

Zealand. Capsules contained 1.6 mg riboflavin, 10 mg vitamin B6 (pyridoxine) and

50 µg vitamin B12 (cyanocobalamin). These vitamins are required in methylation biochemical pathways and were supplied throughout the intervention to ensure that any effects on DNA methylation by substrates other than folic acid were controlled. Capsules for the supplementation phase of the trial contained 400 µg folic acid, in addition to the aforementioned vitamins.

Countdown and New World supermarkets were visited in order to compile a list of products containing added folic acid. Participants were to avoid consuming foods on this list for the duration of the study. Fortified foods consisted mainly of breakfast cereals, but also certain snack and cereal bars, breads and flours, spreads, sports foods and orange juices. A full list, which was provided to study participants, can be found in Appendix A.

Calendars were produced for participants to promote compliance. Individuals were encouraged to mark off days when pills were taken.

Materials & Methods 61

2.3.3 Participant recruitment and screening

Participant recruitment was undertaken during June, 2004. Preliminary genotyping of potential participants for the MTHFR C677T polymorphism was necessary as the study design required equal numbers of participants with each genotype. Due to the low prevalence of individuals homozygous for the polymorphism – approximately 10% in Caucasian populations (Wilcken et al., 2003) – many more individuals were screened than were inducted into the study proper.

Additionally, prospective participants were asked whether they were taking medicines known to interfere with folate metabolism (Methotrexate, Sulfasalazine, Phenytoin), whether they were pregnant, or planning pregnancy, or whether they had epilepsy, coeliac disease, crohn’s disease, kidney disease or inflammatory bowel disease. Individuals matching any of the above criteria were prevented from entering the study. Consent forms outlining these conditions can be found in Appendix B.

Advertisements were placed in Dunedin newspapers The Otago Daily Times and The Star. An answer phone was set up to field advertisement inquiries. Interested parties were encouraged to leave their contact details. All callers were subsequently telephoned back and given further details about the study, including study duration, supplements prescribed and number of blood draws required. For individuals still interested, a time was arranged, during a two-week window, to visit the clinic at the Department of Human Nutrition, The University of Otago, for a buccal swab for MTHFR C677T genotype assessment (buccal swab DNA isolation method described in section 2.5.1, MTHFR C677T genotyping method described in section 2.7). Consent forms were signed prior to performing the swab and date of birth and full contact details were also obtained at this time. A copy of the consent form is listed in Appendix B.

Members of the Department of Biochemistry, University of Otago, were approached to participate in the trial. A University email was also circulated requesting participants. Buccal swabs were taken from consenting individuals for genotyping.

62 Chapter 2

Participants from a previous study conducted in the Department of Human Nutrition, University of Otago, who were known to be homozygous for the MTHFR C677T polymorphism and had stated that they were willing to participate in further studies, were contacted and invited to join the current trial. Also, University of Otago students who had participated as healthy controls in a previous childhood leukaemia study conducted in the Biochemistry Department and who had given their consent to be genotyped for additional polymorphisms in future studies, were screened for the MTHFR C677T polymorphism. Homozygous individuals were contacted and invited to join the present study.

All genotyped individuals from the Biochemistry Department, wider University staff, and the public that responded to the newspaper advertisements, who were homozygous for the MTHFR C677T polymorphism, were sent letters inviting them to join the intervention study. Individuals from those cohorts who could be sex and age matched within two years of the T/T participants were also sent letters inviting them to participate. All other individuals were sent letters thanking them for their cooperation.

2.3.4 Folic acid supplementation trial

Participants of the intervention phase of the study were contacted to arrange a time, within a one week timeframe (June 29 – July 5), to visit the clinic for a blood draw.

Two 4 mL spray-coated K2EDTA Vacutainer tubes (Becton Dickinson) were used for blood collection. Venipuncture was performed by registered nurses. All blood draws were performed between 7:30 and 9:30 am. Participants had been instructed to be fasting 12 hours prior. Following venipuncture, supplements containing 1.6 mg riboflavin, 10 mg pyridoxine and 50 µg cyanocobalamin were distributed to participants, with a request to consume one capsule per day. A second clinic visit was scheduled for 12 weeks hence (during week 28 September – 5 October). Subjects were then provided with breakfast in the clinic kitchen.

Baseline measurements for the following were made from the recovered blood: haematocrit, plasma folate, red cell folate, whole blood folate, total homocysteine and

Materials & Methods 63

vitamin B12. Methods describing these analyses are described in section 2.4. Genomic DNA was also extracted for total methylation analysis, as described in section 2.5.2.

Participants were telephoned two days prior to their second appointment as a reminder. The second blood draw was analogous to the first, except for the following: in addition to the two 4 mL tubes, three 10 mL K2EDTA vacutainers were collected to obtain peripheral blood leukocytes for RNA extraction (described in section 2.6.1). Supplements were also changed to include 400 µg folic acid, in addition to the other vitamins. Existing pills were collected to assess compliance (see section 2.3.5).

A third and final blood draw was scheduled for the week beginning 13 December. Participants were again telephoned prior to their appointments. Two 4 mL and three

10 mL tubes were used to obtain material for folate, homocysteine, B12, DNA methylation and RNA (gene expression) measurements as before. Remaining vitamins were returned and counted. Calendars were returned.

2.3.5 Compliance

Compliance was assessed by counting participants’ remaining capsules. Food questionaires were also completed after the second clinic visit to assess avoidance of folate-fortified foods.

2.4 Biochemical assays

Overnight fasting blood samples were collected at baseline, after the folic acid avoidance phase (12 wk) and post folic acid repletion (24 wk), as stated in section 2.3.4. Plasma was obtained by centrifuging the whole blood at 1650 × g, for 15 minutes at 4°C within 1 hour of collection. Blood samples were stored at –80°C until analyzed.

64 Chapter 2

2.4.1 Folate measurements

Whole blood folate concentrations were determined using a microbiological method on 96-well plates, exactly as described by O’Broin and Kelleher with chloramphenicol resistant Lactobacillus casei as the test microorganism (O'Broin and Kelleher, 1992).

2.4.2 Homocysteine

Total homocysteine was assessed using the IMx (Abbott Laboratories) Fluorescence Polarisation Immunoassay, measuring total L-homocysteine in plasma.

2.4.3 Vitamin B12

Plasma vitamin B12 was measured using the ADVIA Centaur™ vitamin B12 assay, a competitive immunoassay using direct chemiluminescent technology. The assay was performed by Southern Community Laboratories (Plunket House, Dunedin).

2.4.4 Statistics

Sample size was calculated based on a mean difference in genomic DNA methylation status of 30 ng 5′-methylcytosine/µg DNA between T/T and the C/C MTHFR C677T genotypes in an Italian cohort (Friso et al., 2002b). For the current study, a sample size of 27 individuals per genotype group was calculated to be sufficient to detect this difference with 80% power assuming a standard deviation of 33 ng 5′- methylcytosine/µg DNA and a two-sided alpha of 0.05. Differences in biochemical measures between post-depletion and baseline and between post-repletion and baseline were compared using Student’s t-test. Results were considered significant at p < 0.05. Microsoft Excel 2000 was used for basic data analysis.

Materials & Methods 65

2.5 DNA preparations

2.5.1 DNA extraction from buccal cells

Aliquots (1 mL) of STE* buffer with 0.2 mg/mL proteinase K and 0.5% SDS were prepared in 15 mL Falcon tubes and frozen at -20°C. These were removed from the freezer two hours prior to use and allowed to thaw.

Two sterile swabs (cotton buds) were rubbed against the inside of participants’ cheeks, and placed in the 15 mL falcon tubes, with heads immersed in the buffer. The tubes were subsequently incubated at 65°C for two hours. Swabs were partially removed from the tubes and centrifuged at 350 × g for 5 minutes. Swabs were removed and the tubes boiled for 5 minutes. Isopropanol was added (0.6 mL) and the tubes inverted gently. Tubes were then left to incubate at room temperature for 10 minutes. Following centrifugation at 850 × g for 15 minutes, supernatant was removed and the pellet resuspended in 300 µL TE buffer*. DNA concentration was measured using a Nanodrop ND-1000 spectrophotometer system from Nanodrop Technologies.

2.5.2 DNA extraction from peripheral blood

DNA was extracted from whole blood using a method adapted from Bowtell (Bowtell, 1987) and Jeanpierre (Jeanpierre, 1987). Four volumes of 0.15 M ammonium chloride were added to one volume of whole blood and mixed for 5-10 minutes to induce red cell lysis. The sample was then centrifuged at 800 × g for 10 minutes and the supernatant discarded. The pellet was washed with two volumes of 10 mM sodium chloride / 10 mM EDTA by shaking and pipetting to break up cell clumps. The pellet was recovered by centrifugation at 800 × g for 10 minutes and the supernatant removed. The pellet was resuspended into a slurry, followed by the addition of 0.35 volumes of 6 M guanidine hydrochloride and 0.025 volumes of 7.5 M ammonium acetate, and was dissolved completely by pipetting. Proteinase K (10 mg/mL, 0.005 volumes) and 20% sodium sarcosyl (0.025 volumes) were added and the sample was incubated at 60°C for

* STE (sodium, Tris, EDTA) buffer: 100 mM NaCl, 10 mM Tris HCl, pH 8.0, 10 mM EDTA

66 Chapter 2

1 hour. Following incubation, the sample was cooled to room temperature and transferred to a new tube containing 0.25 volumes of pre-chilled chloroform. Samples were vortexed and allowed to stand for 1 minute, then centrifuged at 1000 × g for 3 minutes. The upper layer was collected and added to 1 volume of cold absolute ethanol. The tube was gently inverted to precipitate the DNA, then centrifuged at 1000 × g for 15 minutes and the supernatant was removed. The pellet was washed twice in 70% ethanol and left to dry. The DNA was resuspended in 50 µL TE buffer. DNA concentration was measured using the Nanodrop ND-1000 spectrophotometer system.

2.5.3 DNA preparation for LC/MS and HPLC

Hydrolysis of leukocyte DNA for LC/MS and HPLC was performed using a method based on that of Crain (Crain, 1990). Isolated leukocyte DNA was further purified prior to hydrolysis by removal of ribonucleic acid. S1 nuclease, venom phosphodiesterase I and Alkaline phosphatase were employed to hydrolyze the DNA to single nucleosides.

2.5.3.1 Removal of ribonucleic acid

Twenty micrograms of DNA was digested with 1 unit of RNase A and 2 units of RNase T1 at 37°C for two hours, in a total volume of 200 µL. Subsequent protein removal was achieved by addition of 200 µL 7.5 M ammonium acetate, followed by centrifugation at 10,000 × g for 10 minutes. The supernatant was transferred to a new tube and the DNA precipitated with 1 mL of absolute ethanol and centrifugation at 10,000 × g for 15 minutes. Nucleotides from the hydrolysed ribonucleic acid remained in solution, which was then removed. The pellet was resuspended in 100 µL of 3.3 M ammonium acetate by brief vortexing. DNA was precipitated once again by the addition of 300 µL absolute ethanol. Supernatant was removed and the DNA pellet resuspended in 20 µL TE buffer. Samples were refrigerated until ready for hydrolysis.

* TE buffer: 10 mM Tris-HCl, pH 8.0, 1 mM EDTA

Materials & Methods 67

2.5.3.2 DNA hydrolysis

DNA sample concentrations were determined using a Nanodrop ND-1000 spectrophotometer system. Four micrograms of the RNA-free DNA was added to a PCR tube. MilliQ water was added to a final volume of 100 µL. The tube was heated to 100°C for 4 minutes, then placed on ice. Ten microlitres of 10 × incubation buffer, followed by 20 units of S1 nuclease (4 µL at 5 U/µL, 1:100 dilution of stock) was added. The tube was then incubated for 37°C for 3 hours. Ammonium bicarbonate (11 µL, 1 M) was added to adjust pH for venom phosphodiesterase I digestion, of which 0.002 units were added (1 µL, neat). Tubes were incubated at 37°C for a further 2 hours. Alkaline phosphatase (1.5 units, 3 µL at 0.5 U/µL, 1:60 dilution of stock) was then added and incubated at the same temperature for 1 hour.

YM-3 Amicon Microcon filters (3,000 MW cutoff, Millipore) were prepared immediately after alkaline phosphatase addition. Microcons were rinsed with 0.5 mL MilliQ water at 10,000 × g for 45 minutes to remove any residual cellulose left after manufacture. The filtrate was subsequently discarded. Any residual water retained by the microcon was carefully removed by pipette. The sample was removed from incubation and loaded onto the Microcon. Samples were then centrifuged at 10,000 × g for 80 minutes until membranes were dry. Filtrate was transferred to an autoinjector vial with 150 µL insert (Alltech, Part # 2108720 and 73054, respectively) and frozen at -20°C until use.

2.6 RNA preparation

2.6.1 RNA extraction from peripheral blood leukocytes

RNA was isolated from peripheral blood using a three step process. Leukocytes were first separated from whole blood using a modified dextran sedimentation method (Williams, 1990); dextran has the ability to selectively adsorb to erythrocytes, causing their aggregation and sedimentation (Brooks, 1973). Residual erythrocytes and dextran were removed with a repeated erythrocyte lysis and wash routine. Finally, total RNA

68 Chapter 2 was extracted using a modified TRI Reagent protocol. The procedure is described in detail below.

Peripheral blood was collected by venipuncture in three 10 mL Vacutainers containing EDTA (Beckon Dickinson) from each participant. The three tubes were pooled and transferred to a 50 ml Falcon tube, containing 15 ml of 3.6% dextran 500,000 in PBS. Tubes were mixed thoroughly by gentle rocking. Caps were then removed and any bubbles burst to help minimize erythrocyte carry-over. Tubes were then stood upright and allowed to sediment for 20 to 30 minutes. Following sedimentation the upper layer containing plasma and peripheral blood leukocytes were removed with a disposable plastic Pasteur pipette and transferred into new 50 mL tube. These were spun at 800 × g for 5 minutes at room temperature. Supernatant was removed and discarded.

Residual erythrocytes and dextran were removed by resuspending the leukocyte pellet into a slurry before adding 15 mL of 0.2% (w/v) sodium chloride for 30 seconds, followed by rapid addition of 15 mL of 1.6% sodium chloride. Tubes were centrifuged at 800 × g for 5 minutes and the supernatant removed. One repetition of these lysis steps resulted in a cream leukocyte pellet, essentially free of erythrocytes and dextran. Pellets were resuspended with 1 mL PBS and transferred to three 2.0 mL microcentrifuge tubes whereupon 1 mL of TRI Reagent (Molecular Research Center, Inc. Cincinnati, OH) per tube was added. Thorough pipetting ensured complete dissolution of the pellets. Tubes were left to stand for 5-10 minutes before being frozen at -80°C. All steps up to this point were executed within two hours of venipuncture.

Total RNA was isolated using the TRI Reagent protocol with minor modifications. Tubes were thawed and spun at 12,000 × g for 10 minutes at 4ºC to pellet genomic DNA and particulate material. Supernatants were transferred to new 2.0 mL microcentrifuge tubes containing 0.267 mL chloroform. Tubes were capped tightly and shaken vigorously for 15 seconds before being left to stand for 5 minutes at room temperature. Following centrifugation at 12,000 × g at 4ºC for 15 minutes, the aqueous (upper) layer was transferred to three fresh microcentrifuge tubes, each containing 0.67 mL of propanol, mixed and left to stand for 5-10 minutes. Tubes were then spun at 12,000 × g for 10 minutes at 4ºC, the supernatant removed and the pellet washed with

Materials & Methods 69

1.5 mL 75% ethanol per tube. A final centrifugation at 12,000 × g for 10 minutes at 4ºC was performed, followed by supernatant removal, and the pellet was air dried and resuspended in DEPC water. Tubes were pooled to give a final volume of approximately 25 µL.

Any residual DNA was removed at this point by the addition of 5 µL RQ DNase buffer and by 5 µL RQ DNase (Promega) with incubation at 37ºC for 30 minutes. DNase was removed by a second TRI Reagent procedure: 200 µL DEPC water was added, then 750 µL TRI Reagent. Tubes were shaken to mix and allowed to stand for 5 minutes. Chloroform was added (150 µL) and tubes shaken vigorously for 15 seconds. Centrifugation at 12,000 × g at 4ºC for 15 minutes produced layer separation, with the aqueous layer transferred to a new tube, containing 0.5 mL isopropanol. Tubes were spun at 12,000 × g for 8 minutes at 4ºC, followed by two 75% ethanol washes. Pellets were dried and resuspended in 25 µL DEPC water.

2.6.2 RNA quantification

The quality of RNA extracted from whole blood during method development was initially determined by formaldehyde gel electrophoresis. Subsequent quality assessment, including for all participant RNA, was determined by the Experion automated electrophoresis system, as described in section 2.6.2.2.

2.6.2.1 RNA gels

RNA was visualised using 1% agarose/MOPS/formaldehyde gels*. Ten microlitres of RNA denaturing solution† was added to 5 µL of RNA/DEPC treated water and incubated at 60°C for 10 minutes. Loading dye‡ (2 µL) was added and the samples

*1% agarose/MOPS/formaldehyde gel: 1% agarose (Bio-Rad), 1 × MOPS buffer, 5.4 % deionised formaldehyde

†RNA denaturing solution: 23.8 µL Deionised formaldehyde, 45 µL 10 × MOPS buffer, 225 µL deionised formamide, 6.2 µL DEPC H2O

‡RNA loading dye: 54 µL 6 × RNA loading buffer (40% sucrose, 0.2% bromothymol blue), 6 µL 10 mg/mL ethidium bromide

70 Chapter 2 electrophoresed in a Horizon 58 tank apparatus (Life Technologies) containing 1 × MOPS buffer*. Electrophoresis was performed at 90 V for 90 minutes in a fume hood. The completed gel was photographed under UV transillumination using a GelDoc system from Bio-Rad.

2.6.2.2 Experion automated electrophoresis system

Quality of leukocyte RNA was assessed by an Experion electrophoresis system incorporating a StdSens analysis kit. RNA concentration was measured using a Nanodrop spectrophotometry system and a 10 µL aliquot of each sample was adjusted to 200 µg/mL, of which 2 µL was used in the electrophoresis. Samples and the supplied RNA ladder were denatured by heating for 2 minutes at 70°C, before being placed on ice. The chip was primed and samples loaded, avoiding the generation of bubbles. The chip was vortexed and run on the Experion electrophoresis station. Virtual electrophoretograms were generated by the system software upon completion of the run.

2.7 Assessment of MTHFR C677T genotype status

Buccal cell-derived DNA samples were assessed for the MTHFR C677T genotype. The procedure was derived from the method of Frosst (Frosst et al., 1995) and consisted of PCR amplification of the region of the MTHFR gene containing the polymorphism, followed by polymorphism-dependent cleavage of the 158 bp PCR product by HinfI. Primers were replicated from a publication by Wiemels (Wiemels et al., 2001).

2.7.1 MTHFR C677T Polymerase chain reaction (PCR)

The follow primers were used:

5′-CCTTGAACAGGTGGAGGCC-3′ (forward) 5′-CAAAGAAAAGCTGCGTGATGAT-3′ (reverse)

*10 × MOPS buffer: 41.8 g MOPS (3-[N-morpholino] propane-sulphonic acid), 4.1 g anhydrous sodium acetate, 10 mL 0.5 M EDTA, pH 8.0, water to a total volume of 1 L, pH 7.0.

Materials & Methods 71

A master mix consisting of the following reagents was added to each 2 µL of sample DNA. Master mix component volumes are given for one reaction. All reagents, except for primers, were supplied as a kit from Applied Biosystems.

GeneAmp 10 × PCR Buffer II* 2 µL

MgCl2 (1.5 mM) 1.2 µL dNTPs (2.5 mM) 1 µL Forward primer (10 pmol/µL) 1 µL Reverse primer (10 pmol/µL) 1 µL AmpliTaq Gold polymerase 0.2 µL

H2O 11.6 µL

PCR was carried out on a MJ Research PTC-200 Peltier Thermal Cycler. Cycling conditions were as follows:

Initial denaturation: 95°C, 10 minutes 35 cycles of: 94°C, 30 seconds 60°C, 30 seconds 72°C, 30 seconds Additional extension: 72°C, 7 minutes

PCR product was visualized on a 2% agarose gel.

2.7.2 PCR product digestion

The PCR mixture remaining after removal of the 7 µL aliquot for gel electrophoresis (13 µL) was added to a “master mix” of HinfI enzyme, buffer and water to a total volume of 20 µL. Two microlitres of NEbuffer 2 (New England Biosciences) was used per reaction. The enzyme was added in excess so 2 µL was sufficient for multiple reactions (Note: HinfI does not possess star activity). The digestion was performed at 37˚C for 3 hours in a MJ Research PTC-200 thermocycler.

* GeneAmp 10 × PCR buffer II: 100 mM Tris-HCl, pH 8.3, 500 mM KCl

72 Chapter 2

Digested products were visualized on an agarose gel. The presence of one band of 158 bp indicated two MTHFR 677 C alleles, one band of 130 bp designated two T alleles and two bands indicated heterozygosity.

2.8 Global DNA Methylation Analysis using LC/MS

A combination of high performance liquid chromatography (HPLC) and mass spectrometry (MS) was employed to assess the degree of methylation in genomic DNA in participant leukocytes. The method was based on that of Friso and colleagues (Friso et al., 2002a; Friso et al., 2002b) with minor modifications.

Briefly, purified DNA was hydrolysed to single nucleosides (as described in section 2.4.3) and separated over a reversed-phase column. An electro-spray ion trap mass spectrometer unit was plumbed to the column to establish the mass spectra of peaks eluted and thus identify nucleoside species and their retention times. Additionally, the MS recorded the total ion count of peaks and quantities of 2′-deoxycytosine and 5′-methyl-2-deoxycytosine were determined by peak integration. The fraction of deoxycytosine that existed in methylated form was then calculated.

2.8.1 Equipment

A Phenomenex Luna C8(2), 5 µm, 100 Å, reversed-phase column (250 × 2.0 mm i.d.) with SecurityGuard™ pre column was responsible for nucleoside separation. The HPLC unit consisted of a Finnigan P4000 pump and degasser, AS3000 autoinjector and UV 6000LP diode array detector. This was coupled to a Thermo Finnigan LCQ Deca ion trap mass spectrometer, with electrospray ionisation interface (Thermo Finnigan, San Jose, CA). Both the HPLC unit and MS were controlled by a computer workstation, running Xcalibur version 2.0 software from Thermo Finnigan.

2.8.2 System setup

Prior to running participants’ hydrolyzed DNA, mass spectrometer tuning was performed using direct infusion of 10 µM 2′-deoxycytidine and 5′-methyl-2-

Materials & Methods 73 deoxycytidine. Capillary temperature was set at 350°C to achieve complete fragmentation of the two species into pentose sugar and pyrimidine base components. ESI needle potential was set at 5 kV. Mass/charge ratios (m/z) of 112 and 126, representing the pyrimidine moieties of 2′-deoxycytosine and 5′-methyl-2- deoxycytosine respectively, were tuned for. Separate tune files were saved for each species.

The linear range of the trap and limits of detection were then queried using 2′-deoxycytidine and 5′-methyl-2-deoxycytidine standards from 0.04 pmol to 120 pmol. Autoinjector loading loop volumes from 2 µL to 20 µL were also assessed.

Non-trial digested genomic DNA was run to establish retention times. The mobile phase was 7 mM ammonium acetate pH 6.7 / methanol 5% (v/v), delivered at a flow rate of 0.2 ml/min. The column was periodically washed in 50% acetonitrile (HPLC grade, Mallinckrodt). Scan parameters were set according to peak elution times. The scan routine consisted of two segments, each with one scan event. The first segment was initiated at the beginning of a run and ended following complete elution of the 2′-deoxycytidine peak, typically around 8 minutes. Scan parameters were:

Scan method: MS Scan type: single ion mode (SIM) SIM scan range: 111.50 – 112.50 Polarity: positive Tune method: 2′deoxycytosine

The second segment initiated immediately after the first and ran until acquisition was ceased, following 5′-methyl-2-deoxycytidine elution. Scan parameters were as above except the SIM range was 125.50 – 126.50 and the 5′-methyl-2-deoxycytosine tuning parameters were used. Mass spectrometer acquisition times were varied between 14 and 35 minutes. Total pump runs times were between 20 and 40 minutes.

Nucleoside monitoring by UV diode array was performed at 254 nm, 271 nm (the absorption maxima of 2′-deoxycytidine) and 279 nm (the absorption maxima of 5′-methyl-2-deoxycytidine). Complete hydrolysis of DNA was confirmed by

74 Chapter 2 chromatograms exhibiting only five peaks, correlating to the derivatives of the five bases: cytosine, methylcytosine, guanine, adenine and thymine.

2.8.3 Analysis of participant DNA hydrolysates

Participant DNA hydrolysates acquired from all three trial time points (baseline, post- depletion, and post-supplementation) were then run and analysed. Turnaround time was 35 minutes. Loop volume was 15 µL. As a control for the assay, a standard containing 15 µL of 2 µM 2′-deoxycytidine (30 pmol) and 0.08 µM 5′-methyl-2-deoxycytidine (1.2 pmol) was run after each participant’s samples. Standard curves were run before and after samples to enable total ion counts to be converted to molar amounts so as to avoid discrepancies arising from species flight and trap retention variation. Standard curves ranged from 0 – 37 pmol for 2′-deoxycytidine and 0 – 1.5 pmol for 5′-methyl-2-deoxycytidine. Participant order and sample order were randomised.

Peaks were automatically integrated using the LCQuan batch processing feature of the Xcalibur software. These were verified by inspecting integrated traces using the Quan browser application and adjusted manually if auto-integration had failed. Data was imported as a tab delimited text file into Microsoft Excel 2003. Ion counts were converted to molar amounts and the percentage of global DNA methylation was calculated using the following formula: 5′-methyl-2-deoxycytidine / (5′-methyl- 2-deoxycytidine + 2′-deoxycytidine) × 100.

2.9 Global Methylation Analysis utilising HPLC

Global DNA methylation was also assessed by HPLC alone. DNA hydrolysis was performed as before. The assay was identical to the LC/MS method, except that molar amounts of 2′-deoxycytidine and 5′-methyl-2-deoxycytidine were calculated by integration of chromatographic peaks recorded using a UV diode array. The same Phenomenex Luna C8(2) reversed-phase column was used, however a Agilent 1100 series HPLC unit controlled the separation. The system comprised of a G1379A degasser, G1312A pump, G1329A autoinjector with temperature control, G1316A column temperature control and a G1315B DAD UV diode array. The HPLC was

Materials & Methods 75 controlled by ChemStation for LC software (build 1757, Agilent Technologies). As the same column was used as in the LC/MS method, and parameters such as flow rate were kept constant, chromatographs of hydrolysed DNA between the LC/MS method and the HPLC method were comparable. This allowed initial species peak identification without mass confirmation. Eluted 2′-deoxycytidine and 5′-methyl-2-deoxycytidine were monitored by UV diode array detection at 271 nm and 279 nm respectively. Bandwidth and slit width were both set to 2 nm. To improve separation, column temperature was lowered to 4°C. Standard curves were run as stated in section 2.8.3. Peak areas were converted to molar amounts from the generated standard curves and processed as in section 2.8.3.

2.10 Microarray Analysis

Spotted oligonucleotide microarrays provided the basis for gene expression analysis of the folic acid supplementation trial participants. Gene expression profiles between folate depletion and supplementation phases of the trial were compared, as well as differences in gene expression between participants homozygous for the MTHFR C677T polymorphism and wildtype individuals.

RNA was first reverse-transcribed to cDNA, incorporating aminoallyl-modified nucleotides. These nucleotides were then crosslinked to monoreactive N- hydroxysuccinimide (NHS)-ester fluorescent dyes. The labeled cDNA was then hybridised to the arrays and visualised by laser excitation of the dyes.

2.10.1 Array design

The microarray experiment was set up in accordance with Dobbin and colleagues’ non- reference design for paired samples (Dobbin et al., 2003). This design is typically used for “before treatment” and “after treatment” RNA samples for each participant, which was appropriate for the current experiment.

A total of 30 participants were selected, 15 of which were homozygous for the MTHFR C677T polymorphism with the remainder of subjects having the wildtype version of the

76 Chapter 2 gene. The selection of participant samples is described in section 2.10.2 below. In order to correct for gene-specific dye bias, half the arrays were “flipped”, that is, half of the participant sample pairs were labelled inversely, with the Cy5 dye ascribed to the 3- month sample instead of Cy3, and the Cy3 dye ascribed to the 6-month sample instead of Cy5.

2.10.2 Sample selection for gene expression array studies

Due to monetary and time constraints, not all of the 98 participants had their RNA arrayed. Instead, a subset of C/C and T/T individuals were selected. Each participant had their RNA samples assessed by spectrophotometry (Nanodrop spectrophotometer) and electrophoresis (Experion automated electrophoresis system, described in section 2.6.2.2). Any RNA that had a 260/280 ratio less than 1.85, or which was partially degraded, was excluded from the array experiment. RNA degradation was visualized as an absence of ribosomal RNA banding in an approximate 28S/18S ratio of 1.5:1 or a shift in the RNA smear towards a smaller molecular weight. Heterozygous individuals for the MTHFR C677T were excluded from the array experiment. Participants displaying no modulation of folate data between depletion and supplementation phases of the trial were also excluded. Equal numbers of males were selected for each group, as were females. Age distributions were also matched as closely as possible.

2.10.3 Arrays

Oligonucleotide spotted arrays were used in the experiment. A “20 K” set of oligonucleotides (19,968 elements), produced by MWG Biotech Germany was used. Array printing was performed at the University of Otago Genomics facility by Mr Les McNoe and Dr Dongho Kim.

2.10.4 RNA to cDNA reverse-transcription

Total leukocyte RNA was reverse-transcribed using the Superscript cDNA indirect labeling kit from Invitrogen. (Catalog No. L1014-02). Cyanine Cy3 and Cy5 dyes (Amersham Biosciences) were used in the labeling procedure.

Materials & Methods 77

Ten micrograms of total RNA per sample was first reverse-transcribed into cDNA. Each sample was incubated at 70°C for 5 minutes together with 2 µL of Anchored Oligo

(dT)20 primer (2.5 µg/µL) and DEPC-treated water in a volume of 18 µL in 0.6 mL microcentrifuge tubes. Mixtures were then placed on ice for 3 minutes. The following reagents were then added:

5 × First-strand buffer 6 µL 0.1 M DTT 1.5 µL dNTP mix (including amino-modified nucleotides) 1.5 µL RNaseOUT™ (40 U/µL) 1 µL SuperScript III reverse transriptase (400 U/µL) 2 µL Note: all reagents are provided as part of the kit.

Tubes were then incubated at 46°C for 3 hours. Following cDNA synthesis, RNA was subsequently degraded by addition of 15 µL of 1 N NaOH and incubation at 70°C for 10 minutes. The pH of the solution was neutralized by the addition of 15 µL 1 N HCl.

First strand cDNA was purified using the S.N.A.P. column purification system included with the kit. The cDNA was loaded onto the column with 500 µL of loading buffer and centrifuged at 14,000 × g for 1 minute. Two 700 µL volumes of wash buffer were then passed through the column by centrifuging at the same speed and duration as before with the eluate being discarded. Purified cDNA was then removed from the column by addition of 100 µL DEPC-treated water. Following addition of 10 µL of 3 M sodium acetate and 2 µL of 20 mg/mL glycogen, cDNA was precipitated with 300 µL 100% ethanol and washed with a further 250 µL of 75% ethanol. The pellet was resuspended in 5 µL of 2 × coupling buffer.

2.10.4.1 Labeling of cDNA

Each vial of Cy dye was reconstituted with 80 µL DMSO and separated into 5 µL aliquots. These aliquots were lyophilised using a Savant Speedvac (Savant) and stored in the dark at 4°C until use. Prior to coupling, lyophilised dye aliquots were resuspended in 5 µL DMSO. The rest of the procedure (from this point) was carried out under low light conditions to minimize Cy dye degradation by photobleaching. One

78 Chapter 2

5 µL Cy dye aliquot was added to each cDNA sample. These were then incubated at room temperature in the dark for 1 hour.

Unincorporated dye was removed using S.N.A.P. purification columns. The method for purification was identical to the one stated above, except labelled cDNA was eluted in 50 µL DEPC water into a light-restrictive centrifuge tube. Sample volumes were reduced to 10 µL using a Savant Speedvac (Savant). Samples were used immediately in the hybridisation procedure.

2.10.5 Hybridisation of labeled cDNA

2.10.5.1 Slide preparation

Typically, four slides were prepared per day, with half of the slides having their pre- hybridisation routine initiated 30 minutes later to stagger, and thus make easier, the loading and post hybridisation steps of the procedure.

The identification number of the sample to be hybridized on the slide, and the boundaries of the printed area were marked on the opposite face of the slide with a diamond pencil.

Slides were placed in falcon tubes containing 0.22 µm filtered BSA pre-hybridisation blocking buffer*, preheated for around an hour at 42°C and incubated for 45 minutes at that temperature. This was to ensure reactive groups on the slide surface were blocked, thus preventing non-specific binding of target cDNA to the slide. Following incubation, slides were rinsed five times in ddH2O and spun dry at 500 × g. Subsequent loading of cDNA material onto the slides was performed within half an hour of the slides being dried.

* BSA Pre-hybridisation solution: 0.25 mL 20 × SSC, H2O to 50 mL.

SSC buffer (20 x): 3 M NaCl, 0.3 M sodium citrate

Materials & Methods 79

2.10.5.2 Loading

Slides were placed in hybridisation chambers (Corning) and raised coverslips (Erie Scientific Company) were placed over the printed area. The entire assembly was placed on a heating block, maintained at a temperature of 42°C. SlideHyb buffer (Ambion), pre-warmed to 68°C, was added to the concentrated sample to give a total volume of 40 µL. Samples were left at 68°C until loading, whereupon they were briefly centrifuged to collect the liquid. Samples were loaded at the corners of the coverslip using a 20 µL micropipette where the solution was drawn onto the array by capillary action. A small amount (10 µL) of MilliQ water was placed in the hydration wells of the hybridisation chamber. The chamber was sealed and incubated at 42°C for 16 hours in the dark.

2.10.5.3 Post-hybridisation

After incubation, the slides were removed from their hybridisation chambers and the coverslips were lifted away, taking care not to touch the printed areas. Slides were immediately placed in light restrictive 50 mL falcon tubes (one slide per tube) containing post-hybridisation buffer A*, pre-warmed to 42°C, and rinsed by gentle inversion of the Falcon tubes for eight minutes. Slides were then transferred to Falcon tubes containing post-hybridisation buffer B† at room temperature. The slides were agitated for another six minutes. The arrays were transferred to Falcon tubes containing post-hybridisation buffer C‡ and then post-hybridisation buffer D§, with agitation times of six minutes and three minutes, respectively. Immediately following the washing procedure, the slides were spun dry at 500 × g in falcon tubes with tissue placed at the bottom of the tubes.

Post-hybridisation wash buffers (0.2 µm filtered):

* Buffer A: 2 x SSC and 0.2% SDS

† Buffer B: 1x SSC and 0.2% SDS

‡ Buffer C: 0.5 x SSC

§ Buffer D: 0.1 x SSC

80 Chapter 2

2.10.6 Scanning of arrays

Scanning and image acquisition of completed microarrays was performed on an Axon Instruments GenePix 4000B scanner, interfaced with a computer running Genepix Pro version 6 software. Cy5 and Cy3 channels were scanned simultaneously, at wavelengths of 635 nm and 532 nm, respectively. Preliminary 40 µm per pixel scans were performed to establish correct laser power and photomultiplier tube (PMT) gains for each channel. This was performed by the auto gain feature of the software. Optimal PMT gain was sought by attempting to match total intensities between channels, while maximising intensities without exceeding 0.1% saturated pixels. Once gains were set, a high resolution (5 µm) scan was run, with two lines averaged for an improved signal to noise ratio. The 635 and 532 channels were combined as a multi-image TIFF file and saved.

The Genepix Pro software was then used to determine spot intensities. A grid of the same dimensions as the printed spots was overlaid upon the image, with the software automatically determining spot outlines using the following settings:

Circular features enabled Feature resize: minimum = 85%, maximum = 150% Limit feature movement to 30 µm Block alignment Maximum translation = 10 pixels

Spot outlines incorrectly defined by the software were corrected manually. Spots of very low intensity or quality were flagged and excluded thereafter from the analysis. Spot intensities – the median intensity of each defined feature – were then calculated by the software. A list of Genbank accession numbers associated with spot positions was imported and this information, together with intensity data, was saved as a tab-delimited .gpr file. All files were then converted apropos of normalisation to the tab-delimited .mev filetype using ExpressConverter version 1.9 (TIGR software).

Materials & Methods 81

2.10.7 Data normalisation

Data normalization of array data was performed using MIDAS version 2.19, part of the TIGR 4 software suite from TIGR software. This process is described in detail in Chapter 5 of this thesis.

2.10.8 Expression data analysis

The open source MeV multi experiment viewer version 4.0 developed by TM4 software development team, TM4.org, was used for preliminary data analysis. Numerous other software packages and web-based applications were utilised for analysis, including DAVID, GATHER, NCBI gene and AceView, UCSC genome browser and PubMatrix. These software are discussed in detail in their relevant sections in Chapter 5.

2.11 Satellite 2 DNA methylation assay

The extent of methylation of satellite 2 repeat (Sat2) sequences in DNA obtained from study participants was measured using a methylation-sensitive quantitative PCR method (MS-qPCR). The assay was based on a published method already established in the student’s laboratory (Fukuzawa et al., 2004). The principle for the assay was as follows: digestion of participant Sat2 sequences by a methylation-sensitive restriction enzyme would produce differing amounts of template for qPCR, depending on the degree of methylation of that sequence. The extent of amplification and accompanying quantitation would thus be proportional to the level of methylation in the original sequence.

More specifically, DNA samples from each participant were split, with one aliquot (experimental) subjected to digestion by the methylation-sensitive restriction enzyme BstB1, while the other “mock” aliquot was left undigested. BstB1 recognises the sequence TTCGAA, but only does so if the CpG dinucleotide is unmethylated. The recognition sequence falls within the Sat2 repeat motif. Both aliquots are subjected to qPCR. Primers flanking the sequence are used in the reaction. Amplification only

82 Chapter 2 occurs if the sequence has not been digested. The two aliquots are then quantified and the ratio of experimental to mock calculated, expressed as a percentage methylation.

Satellite 2 primers and restriction sites were derived from the Genbank accession for Sat2 sequence (X72623.1). The primers flanking the cut site were:

5′- CTC ATG AAA TTG AAA TGG ATG G-3′ (forward)

5′- GTC CAT TCG ATG ATT CCA TCT-3′ (reverse)

Assays were kindly run by Jackie Ludgate in the Cancer Genetics Laboratory, University of Otago. Five micrograms of genomic DNA were used for the assay – the high copy number of Sat2 sequences meant only this small amount was necessary. For each sample, two 30 µL digests were set up. These contained 3 µL of appropriate restriction buffer, water and either 25 units of BstB1 for the digest or 0.5 µL of 50% glycerol for the mock reaction. Digestion was accomplished by incubation at 65°C for 3 hours. Each of the 30 µL digests were then divided into three replicates (8.3 µL) to which was added 10 µL of Platinum Quantitative PCR Super-Mix with Rox (Invitrogen), 0.6 µM of each primer and 0.3 µL of STBR green fluorescent dye (at 1:1000 dilution). All pipetting steps following incubation were performed with a liquid handling system (CAS-1200) from Corbett Robotics.

Quantification of the real time PCR of Sat2 sequence was achieved using an ABI Prism 7900HT sequence detection system (Applied Biosystems). ABI software was used to monitor reactions and establish the cycle at which the level of fluorescence reached an arbitrary threshold level, or Ct value. The triplicate Ct values for each of the sample were averaged and standard deviation calculated. Ct values with a standard deviation greater than 0.2 were rejected, whereupon the assay was performed again for those samples.

For each participant sample, the mean Ct value for the digested aliquot was subtracted from the mean Ct of the mock digest, to give a delta Ct value. Due to the exponential nature of PCR, percentage methylation was calculated as 2 ΔCt × 100. For example, if a participant’s Sat2 DNA possessed only 50% methylation, the mock reaction would

Materials & Methods 83 amplify at twice the rate as the digested aliquot, meaning the latter would require an extra cycle of PCR to reach the same Ct threshold. If the mock reaction achieved the Ct threshold in 14 PCR cycles, the experimental aliquot would require 15 cycles. Thus the ΔCt would equal 14-15, or -1. Calculating % methylation would be 2 -1 × 100 = 50%.

Standard curves of known concentrations of DNA were performed as part of every run. As a control for the assay, DNA from the peripheral blood of a normal individual was included in each run. This sample reported a consistent degree of Sat2 methylation across multiple assays on different days. The calculated mean for the control was 88%. Standard error of the assay was 1.9.

84 Chapter 2

3 Folic Acid Supplementation Study

3.1 Trial overview and provisos

A twenty-four week folic acid supplementation intervention upon 93 human participants formed the foundation of this project. Trial participants were sourced from a “free- living” population in Dunedin, New Zealand – a region that had limited exposure to folic acid fortification. Subjects were representative of all adult age groups and sexes. Prospective participants were screened for the MTHFR C677T polymorphism and equal numbers of subjects for each genotype were inducted into the trial proper. The intervention consisted of a 13 week dietary folate avoidance period, followed by an 11 week folate supplementation phase where participants were given a daily supplement of 400 µg folic acid. Folate intake was modulated with the intent to alter peripheral blood folate and related metabolites, and to determine its effect on leukocyte DNA methylation. Any differences in methylation among MTHFR genotypes were also investigated, as well as whether folic acid avoidance or supplementation differentially affected DNA methylation among the MTHFR genotypes. This chapter describes the enactment of the supplementation study and its outcomes related to folate and homocysteine, whereas Chapter 4 discusses the subsequent work related to global DNA methylation analysis. The study overview is represented diagrammatically in Figure 3.1. Detailed methods of the trial are in section 2.3.

3.1.1 Peripheral blood’s suitability as a source and a proxy for whole-body folate status

Peripheral blood erythrocytes and leukocytes were the cell types of choice for procurement of folate measures and DNA and RNA respectively, primarily because of their accessibility, but also because their relatively high turnover made them useful for a

86 Chapter 3 study of limited duration (see section 3.1.3.1). Obtaining alternative material such as hepatic tissue for example, although potentially lucrative from a scientific perspective, was not considered for obvious reasons. Conversely, blood draws were quick to perform and carried good public acceptance.

Fittingly, all the related studies previously mentioned in section 1.10 that investigated the relationship between folate, the MTHFR C677T polymorphism and DNA methylation in humans used peripheral blood as a source for DNA material and folate measurements. Indeed, plasma folate, red cell folate and homocysteine assays have been

Recruitment

MTHFR C677T polymorphism genotyping

Participant induction

Thirteen week dietary folate avoidance period

Blood draw Intervention phase Eleven week folic acid supplementation period

Blood draw

Post-intervention analysis

Blood folate Blood homocysteine Global DNA methylation Global gene expression Loci-specific methylation

Figure 3.1 Intervention study procedural overview. Following recruitment, participants were screened, which included assessment of MTHFR C677T genotype. Participants homozygous for the polymorphism were inducted into the study proper, together with age and sex-matched CT and CC subjects. During the avoidance period, participants were required to avoid eating folic acid-fortified foods, in addition to consuming daily supplements of vitamin B6 and B12, described in more detail in section 3.1.4. These vitamins were continually taken through the folic acid supplementation phase as well.

Folic Acid Supplementation Study 87 used routinely for many years to assess folate status in a diagnostics setting (Hall and Chu, 1990). Using peripheral blood for these assays in the present study was therefore in keeping with the published literature and this consistency would enable better comparison of results.

3.1.2 Compensation for MTHFR allele distribution

As individuals possessing the MTHFR 677 T/T genotype represent only a small fraction (8-10%) of Caucasian populations (Wilcken et al., 2003) and because equal numbers of participants of each genotype were required for the study, a large number of individuals needed to be screened in order to identify a sufficient number of T/T participants. Consequently, considerable effort was placed on obtaining as many people as possible for initial recruitment in order to achieve sufficient study power. As a consequence, only around a third of prospective participants would be eligible for the study proper. It should be noted that wildtype and heterozygous individuals were matched to T/T participants only by sex and age within two years (as described in section 2.3.3). Where more than one prospective individual met these criteria, selection was randomised to avoid any introduction of selection bias.

3.1.3 Determinates of study duration

The duration of the study was initially envisaged as being six months. A delay in the hearing of the ethics application truncated the trial, beginning in July, to an approximately five month timeframe. This was because it was deemed unsuitable to conduct the study over the Christmas break. In addition to the holiday period making participants difficult to retain or contact, potentially high alcohol consumption during this time was identified as an important confounder. Alcohol is known to decrease folate status, by reducing folate receptor expression and impairing kidney conservation of circulating folate, and increase plasma homocysteine (Gibson et al., 2008; Halsted et al., 2002).

88 Chapter 3

3.1.3.1 Peripheral blood cell turnover

Another determinant of study length was the time anticipated for the dietary folate regimen to manifest as a stable intracellular folate level. Peripheral blood was the material used to assay folate and thus the major folate-carrying component was the erythrocyte. Given that erythrocyte folate uptake occurs only during cell development and a RBC’s time in circulation is around 100 days (Bray, 1999), a sufficient length of time was required for adequate turnover of the erythrocyte population before implementing a new dietary folate regimen, in order to measure the full effect of tissue folate status. For example, RBCs synthesised during the folate avoidance phase of the study, presumably having a relatively low intracellular folate concentration, would require a depletion period long enough to allow sufficient accumulation of these cells and achieve a near-homogenous population, prior to end-of-regime red-cell folate assays being conducted. Subsequently, these cells would require an adequate time period to be evacuated from the circulation and replaced with “high folate” erythrocytes before subsequent assays could be performed at the conclusion of the supplementation phase. Although mean durations for depletion and supplementation periods were around 90 and 80 days respectively, and complete erythrocyte turnover was known to be approximately 100 days, the study was judged to be satisfactorily long enough to assess the effects of the regimen on cellular folate.

To add weight to this consideration, Rampersaud and colleagues had observed effects on folate and homocysteine over a period as short as seven weeks, reinforcing the notion that an approximate 24-week time frame would be sufficient to alter the variables assayed for (Rampersaud et al., 2000).

DNA and RNA for the study were sourced from peripheral blood leukocytes. Once in the bloodstream, these cells are much shorter lived than erythrocytes. For example, neutrophils, which form the bulk of leukocytes (from 45 – 75%), are only present in circulation for a matter of days (Pillay, 2010). The ramification of this rapid cell turnover is that DNA and RNA derived from leukocytes at the conclusion of each phase of the study would be from a new cell population generated only under the dietary conditions of that period.

Folic Acid Supplementation Study 89

3.1.4 B vitamin supplementation and diet considerations

Vitamins B2 (riboflavin), B6 (pyridoxine) and B12 (cobalamin) are required cofactors for multiple enzymes in the folate metabolism pathway (see Figure 3.2). Any dietary deficiencies of these cofactors could possibly interfere with folate inter-conversion flux and mask the desired effects of folate depletion and supplementation. Consequently, these vitamins were administered to participants for the duration of the study to ensure that any modulation in peripheral blood folate, homocysteine or DNA methylation was attributable to folic acid modulation and not deficiencies in these vitamins. It should be noted that the dietary intake of these vitamins is generally considered replete in the New Zealand population; inadequacy is estimated to be around 3% for riboflavin and 0.4% for cobalamin. Intake of vitamin B6 is considered more than adequate (Russell et al., 1999). Thus, administering these vitamins to study participants was purely precautionary.

Homocysteine B6 B2 SHM MTHFR B12 THF 5, 10-MethyleneTHF 5-MethylTHF MS

THF Methionine

Figure 3.2 Sites of action of cofactors involved in folate metabolism. B2: riboflavin, B6: pyridoxine, B12: cobalamin, SHM: serine hydroxymethyltransferase, MTHFR: methylenetetrahydrofolate reductase, MS: methionine synthase.

Additionally, participants would be required to avoid folic acid fortified foods during the supplementation phase, as well as during folate avoidance. The rationale for this was to keep subjects’ diets between avoidance and repletion periods as similar as possible. Allowing them to consume folate-fortified food during the supplementation phase may inadvertently induce different dietary habits between avoidance and supplementation, which may potentially manifest as an altered B vitamin intake, for example. With this requirement in place, any changes in plasma and red cell folate between avoidance and supplementation periods could be attributed to the folic acid supplementation regimen with greater confidence.

90 Chapter 3

It must be acknowledged, however, that although given assurances from participants, diet could not be monitored closely given that subjects were not retained in a metabolic laboratory environment. Accordingly, no data of actual folate intake would be available.

New Zealand Ministry of Health statistics were used as estimates for folate and other B vitamin intakes during planning of the study. The implementation of the study’s “washout period” was in response to estimates that up to 70 µg of daily folate intake may be contributed by foods fortified with folic acid (Russell et al., 1999). Accordingly, the washout was regarded as a “folic acid avoidance” phase as opposed to a “depletion” of overall dietary folate intake.

3.1.5 Blood processing

The quality of RNA, inversely determined by the level of RNA degradation present in a sample once it is extracted, can worsen over time due to intrinsic RNase activity, until inhibited chemically or by freezing. As different RNA species are disproportionately affected by this phenomenon, the measured transcriptome can effectively change as a function of sample storage time (Auer et al., 2003). Hence a concerted effort was made to isolate and preserve leukocyte RNA as quickly as practicable following venipuncture to ensure the microarrayed material was an accurate reflection of the RNA present in the cell.

Additionally, various RNA transcripts are known to vary as a function of time of day (Boorman et al., 2005). To minimise RNA variability from circadian genes, blood draws were scheduled for the same time of day to within an hour for each participant.

Folic Acid Supplementation Study 91

3.2 Results

3.2.1 MTHFR polymorphism genotyping

All buccal cell-derived DNA from prospective participants and DNA from peripheral blood were genotyped for the MTHFR C677T polymorphism. Sequences corresponding to the MTHFR gene were amplified by PCR and subsequently digested with HinfI as described in section 2.7. The resulting fragments were run on agarose gels. A typical electrophoretogram is depicted in Figure 3.3.

T/T C/T C/T C/C

158 bp → 130 bp →

Figure 3.3 Electrophoretogram of digested PCR amplicons. PCR amplification of the polymorphic region of the MTHFR gene produces a 158 bp product. Subsequent digestion with HinfI is successful only if the C677T base substitution is present, whereupon the PCR product is cleaved into two fragments, 130 and 28 bp in length. Amplicons from wildtype (C/C) individuals do not possess the thymine at position 677 and hence are not digested by the restriction enzyme. This results in a single band of 158 bp. Heterozygous samples produce both digestible and indigestible PCR products, and so produce all three bands. Individuals possessing both polymorphic alleles produce only the 130 and 28 bp fragments – the 158 bp product is absent. Because 28 bp bands are very faint, MTHFR C677T genotype is deduced by analysis of 158 bp and 130 bp bands alone. Blank lanes are the result of failed PCR, usually because of the impure nature of buccal cell preparations. Further centrifugation of these samples to sediment particulate material typically improved sample purity and allowed for successful amplification when PCR was repeated.

92 Chapter 3

3.2.2 MTHFR C677T allelic distribution

A total of 221 buccal samples were genotyped. Of these, 4 failed to PCR after repeated attempts. Of the remainder, 100 were wildtype (C/C), 99 heterozygous (C/T) and 18 homozygous for the MTHFR polymorphism. The C allele frequency was 0.689 and the T allele frequency was 0.311.

To date, the most comprehensive data on MTHFR allele frequencies (Wilcken et al., 2003) ascribes Caucasian Australians, the subpopulation closest to this study in terms of ethnicity, allele frequencies of 0.714 and 0.286 for C and T alleles respectively, from a sample population of 288. Hence individuals homozygous for the MTHFR polymorphism in the Australian cohort, for example, represented around 8.18% of the total population, which compared favourably with the current study’s 8.29%.

Genotyping of existing laboratory samples yielded similar results. A total of 237 samples were successfully genotyped, with n = 110, 106 and 21 for C/C, C/T and T/T genotypes respectively. Allele frequencies were 0.688 and 0.312 for C and T alleles respectively. A comprehensive breakdown of participant numbers by MTHFR genotype is given in the section below.

3.2.3 Participant flow and follow up

In response to the newspaper advertisements, 174 members of the public were genotyped, in addition to 83 University of Otago staff. Together these individuals (217 following genotyping) comprised the main pool from which subjects for the folic acid intervention phase of the study would be drawn – 18 individuals homozygous for the MTHFR C677T polymorphism and 33 each of heterozygote and wildtype individuals. Six individuals from the childhood leukaemia study control group participated in the present study. From the 237 samples screened, only the 21 individuals homozygous for the MTHFR polymorphism were identified and subsequently invited to participate. Six T/T homozygotes identified in this screening were affiliated with the University of Otago Biochemistry Department and all but one took part in the intervention trial. Of the remainder, just 7 were able to be contacted and only one was able to participate in the supplementation trial, which she subsequently failed to complete.

Folic Acid Supplementation Study 93

Together with 8 participants from a Department of Human Nutrition study known to be homozygous for MTHFR C677T, a total of 32 T/T individuals entered the folic acid supplementation trial. Given that there were many more individuals that were wildtype and heterozygous for the MTHFR polymorphism, sex and age matching the latter groups to polymorphic homozygotes was straightforward. A summary of participant numbers is shown in Figure 3.4.

General public University staff CHLS samples HUNT study (known T/T) n = 174 n = 83 n = 237 n = 8

Genotyped for MTHFR C677T polymorphism CHLS T/T samples n = 221 n = 21

C/C C/T T/T T/T T/T n = 100 n = 99 n = 18 n = 6 n = 8

C/C C/T T/T n = 33 n = 33 n = 32 Folate Avoidance Period

C/C C/T T/T n = 32 n = 33 n = 31 Folate Supplementation Period C/C C/T T/T n = 31 n = 31 n = 31

Figure 3.4 Participant flow and follow up. Shaded area denotes folic acid supplementation trial. CHLS: Childhood Leukaemia Study control samples. HUNT: Department of Human Nutrition, University of Otago.

3.2.4 Study participant statistics & compliance

At the conclusion of the intervention trial, 93 participants remained - a dropout rate of 5.1%. Age and sex data of participants who completed the trial are presented in Table

94 Chapter 3

3.1. Variation of age and sex between genotype groups was deemed satisfactory – analysis of variance determined that no groups were significantly different (p = 0.97).

Table 3.1 Age and sex distributions of participants that completed the study.

All By MTHFR C677T genotype C/C C/T T/T Sex (n) Female 57 (61.3%) 19 20 18 Male 36 (38.7%) 12 11 13 Total 93 31 31 31

Age (years, mean ± SD) Female 44.4 ± 14.2 43.6 ± 14.6 43.8 ± 14.1 45.8 ± 14.5 Male 38.9 ± 13.3 39.8 ± 13.4 38.5 ± 13.1 38.5 ± 14.3 Total 42.2 ± 14.0 42.1 ± 14.1 41.9 ± 13.8 42.7 ± 14.7

Every participant was provided with a container of 100 capsules for each phase of the intervention study. Participants were required to consume one capsule per day and the duration of avoidance and supplementation phases were each well under 100 days. Thus compliance was assessed by counting pills remaining at the end of each period. Due to variations in the scheduling of participants for venipuncture, not all subjects’ intervention periods were of equal length, and consequently, not all participants were capable of consuming the same number of vitamin capsules. The mean duration of the avoidance period was 91 days with 77 days for the supplementation period. The average number of pills taken was 80.6 (88.6%, SD = 10.5) for the avoidance period and 69.6 (90.4%, SD = 10.0) for the supplementation period. Capsule consumption by genotype is listed in Table 3.2.

Of the 93 participants, 83 (89%) took at least 80% of their study capsules. No participant was excluded from subsequent folate, homocysteine or global DNA methylation analysis due to poor compliance. However, restriction criteria were placed on samples apropos of the microarray analysis. These criteria included a participant’s

Folic Acid Supplementation Study 95 level of compliance in addition to various other factors, as described in greater detail in Chapter 5.

Table 3.2 Vitamin capsule consumption by MTHFR C677T genotype.

Avoidance period (¯x = 91 days) Supplementation period (¯x = 77 days) Capsules consumed (mean ± SD) Overall 80.6 ± 10.5 (88.6%) 69.6 ± 10.0 (90.4%) C/C 77.4 ± 12.1 (85.1%) 68.0 ± 8.4 (88.3%) C/T 83.1 ± 8.9 (91.3%) 68.8 ± 10.0 (89.4%) T/T 82.4 ± 9.5 (90.5%) 71.8 ± 8.4 (93.2%)

3.2.5 Blood processing

As previously mentioned, a concerted effort was made to collect and process all blood samples as quickly as possible. This included the time taken from venipuncture to sample freezing, as well as taking blood from all participants within the shortest possible timeframe. Regarding blood collection, 97% of participants were processed within one week of each other, with 75% within three days. All blood samples destined for RNA extraction were processed to the point of leukocyte suspension in TRI Reagent and freezing at -80 °C within two hours of venipuncture. All participants were scheduled for blood draws at the same time of day across all three appointments.

3.2.6 Plasma folate

The avoidance phase of the trial produced a modest overall decrease in plasma folate. The mean plasma folate concentration fell by 7 ± 8 nmol/L (mean ± SD, p < 0.001). Following supplementation, plasma folate rose by 26 ± 12 nmol/L (p < 0.001). This trend was reflected in all MTHFR C677T genotype groups. Notably, plasma folate values for T/T participants were lower than other genotypes at all time points throughout the study, although this difference was not statistically significant.

Following the avoidance period, a significant proportion of individuals displayed a plasma folate status under 12 nmol/L (n = 38, 40.9%). For T/T participants, almost half

96 Chapter 3 had a folate level below 12 nmol/L (n = 15, 48.4%). It should be noted that Friso and colleagues, in their observational study of folate status and global DNA methylation, witnessed hypomethylation occurring in patients with plasma folate concentrations lower than approximately 12 nmol/L, a figure they defined as low folate status (Friso et al., 2002b). Following supplementation, no participant had a folate concentration of less than 12 nmol/L. The coefficient of variation (standard deviation / mean) for the plasma folate assay was 2.6%. Modulation of plasma folate over the course of the study is depicted graphically in Figure 3.5, with summary data provided in Table 3.3.

Figure 3.5 Plasma folate concentration according to MTHFR C677T genotype at baseline, following folic acid avoidance (week 13) and following folic acid supplementation (week 24). Data expressed as mean ± 95% confidence interval.

3.2.7 Red cell folate (erythrocyte folate)

Red blood cell (erythrocyte) folate status was measured in addition to plasma folate. Erythrocyte folate followed a similar trend to plasma folate measurements; concentration decreases across all genotypes were observed following folate avoidance (p < 0.001, except C/C where p < 0.05) and all groups increased markedly following folic acid supplementation (p < 0.001). As with plasma folate measurements, individuals possessing the T/T MTHFR genotype displayed lower red cell folate levels

Folic Acid Supplementation Study 97 than the other participants. The coefficient of variation for the red cell folate assay was 4.2%. Results are depicted in Figure 3.6 and Table 3.3.

3.2.8 Homocysteine

Total plasma homocysteine was also measured, owing to its close antithetical relationship with folate status. Mean homocysteine decreased significantly following folic acid supplementation (p < 0.001). Individuals homozygous for the MTHFR polymorphism displayed the largest decrease – a difference of 1.4 µmol/L between post avoidance and post supplementation (p < 0.001). The folic acid supplementation period was successful at lowering the T/T group’s homocysteine concentration to that of heterozygous and wildtype individuals. Results are represented in Figure 3.7 and Table 3.3.

3.2.9 Additional micronutrient data

As stated in section 3.1.4, riboflavin, pyridoxine and cyanocobalamin were supplemented throughout the study to ensure correct function of folate-related metabolic

98 Chapter 3

pathways. Total Vitamin B12 (cobalamin) was assayed as a proxy for compliance Plasma samples from participants were analyzed by Southern Community Laboratories.

Figure 3.7 Homocysteine concentration according to MTHFR C677T genotype at baseline, following folic acid avoidance (week 13) and following folic acid supplementation (week 24). Data expressed as mean ± 95% confidence interval.

Mean plasma vitamin B12 concentration at the beginning of the study was 278 ± 105 pmol/L (mean ± SD), around twice what is considered a low B12 status. Concentration increased slightly throughout the study in all participant groups due to the inclusion of

B12 in the daily supplements. These increases were statistically significant with the exception of the change from post-avoidance to post-supplementation time points in

C/C and C/T participants. No participants had a B12 concentration below 150 pmol/L

(generally considered as the threshold for low B12 status) at post-avoidance and post- supplementation time points, which suggested that this vitamin was not limiting flux through the pathway outlined in Figure 3.2. Vitamin B12 concentrations are outlined in Table 3.3. Coefficient of variation for this assay was 5.6%.

Folic Acid Supplementation Study 99

Table 3.3 Red blood cell folate, plasma folate, cobalamin and homocysteine concentration by MTHFR C677T genotype at baseline, following folate avoidance and repletion1

MTHFR Genotype Baseline Avoidance Difference2 Repletion Difference3 Plasma folate (nmol/L) Overall 22 ± 12 15 ± 8 -7 ± 8** 41 ± 14 26 ± 12** CC 21 ± 11 15 ± 6 -6 ± 8** 42 ± 13 28 ± 12** CT 25 ± 13 17 ± 10 -8 ± 6** 42 ± 16 26 ± 13** TT 19 ± 11 14 ± 7 -5 ± 8* 39 ± 12 25 ± 10** Red cell folate (nmol/L) Overall 815 ± 309 704 ± 259 -111 ± 154** 1069 ± 244 366 ± 181** CC 818 ± 302 739 ± 263 -78 ± 190* 1145 ± 229 406 ± 175** CT 882 ± 327 734 ± 247 -148 ± 138** 1060 ± 242 326 ± 152** TT 747 ± 292 639 ± 264 -108 ± 122** 1003 ± 245 364 ± 207** Plasma homocysteine (μmol/L) Overall 8.4 ± 2.4 8.3 ± 2.3 -0.1 ± 2.0 7.5 ± 2.0 -0.9 ± 1.9** CC 8.2 ± 2.0 8.2 ± 2.0 0.0 ± 1.4 7.6 ± 2.1 -0.6 ± 1.3* CT 8.3 ± 2.7 7.9 ± 2.7 -0.4 ± 1.7 7.3 ± 2.2 -0.6 ± 2.1* TT 8.9 ± 2.5 8.9 ± 2.2 0.0 ± 2.7 7.5 ± 1.6 -1.4 ± 2.3** Plasma cobalamin (pmol/L) Overall 278 ± 105 345 ± 127 67 ± 81** 355 ± 119 10 ± 71* CC 267 ± 95 363 ± 125 96 ± 83** 360 ± 121 -3 ± 64 CT 275 ± 121 308 ± 121 32 ± 74* 329 ± 127 21 ± 56 TT 292 ± 101 363 ± 131 70 ± 76** 375 ± 107 12 ± 88* 1 Mean ± standard deviation 2 Difference between post-avoidance and baseline (* p < 0.05; ** p < 0.001) 3 Difference between post-supplementation and post-avoidance (* p < 0.05; ** p < 0.001)

3.2.10 Plasma homocysteine as a function of folate, MTHFR C677T genotype and B12 status

Interactions between homocysteine and the other measured variables were also investigated. There was no significant correlation between total homocysteine and vitamin B12 at any point in the study, indicating that the dietary B12 supplement was not responsible for modulating homocysteine (p = 0.44). However, there was a significant negative correlation between folate status and homocysteine at baseline (r = -0.37, p < 0.001). Additionally, the MTHFR polymorphism and folate status interacted to

100 Chapter 3 modulate homocysteine (p = 0.004). As depicted in Figure 3.8, individuals homozygous for the MTHFR polymorphism displayed differential reductions in homocysteine between avoidance and supplementation periods depending on their pre- supplementation folate status. This confirmed that homocysteine remethylation may be more sensitive to low folate status when the MTHFR enzyme is catalytically compromised. Vitamin B12 was not responsible for this effect; there was no significant difference in B12 levels between T/T individuals with plasma folate above and below 12 nmol/L (p = 0.25). This was not surprising given that B12 had been administered for the three months prior to the “pre-supplementation” blood draw.

Figure 3.8 Comparison of homocysteine concentration between pre- and post-folic acid supplementation periods according to MTHFR C677T genotype and stratified by pre- supplementation folate status. Data expressed as mean ± 95% confidence interval. Wildtype and heterozygous participants have been grouped to aid clarity. PF: plasma folate.

Homocysteine also declined to a significantly greater degree in T/T individuals than C/C and C/T participants when pre-supplementation folate status was less than 12 nmol/L, reinforcing the higher folate requirement of the former group. The difference in homocysteine concentration between T/T subjects and other genotypes with a folate status prior to supplementation above 12 nmol/L was not significant (p = 0.098) but was significant when folate status was below 12 nmol/L (p = 0.016).

Folic Acid Supplementation Study 101

3.2.11 Sex effect in measured outcomes Although the study was not explicitly configured to explore sex effects with any great power, all measured variables were analysed nonetheless. No statistically significant differences between males and females were observed for plasma folate, red cell folate, homocysteine or B12, either overall or by MTHFR C677T genotype.

3.3 Discussion of immediate study outcomes

This chapter has described the nature of the cohort recruited for the study as well as the changes in micronutrient concentrations as a result of participant folic acid avoidance and supplementation. The comprehensive recruitment phase of the project and the high number of individuals screened ensured that there were adequate numbers of participants for all MTHFR C677T genotypes and that distributions were commensurate for age and sex among the genotypes. Indeed, the volunteer turnout was among the highest observed by the Department of Human Nutrition for the type of study. Participant retention was also satisfactorily high.

Regarding the biochemical data obtained from the trial, the most salient observations were that supplementation increased red cell and plasma folate by 366 ± 181 and 26 ± 12 nmol/L (p < 0.001) respectively, and reduced homocysteine by 0.9 ± 1.9 µmol/L (p < 0.001). Folate or homocysteine concentrations did not differ by MTHFR genotype. Plasma folate was modulated as expected, decreasing following participant avoidance of folate fortified foods and increasing following supplementation. While other studies have noted a greater decrease from baseline when implementing a dietary restriction of folic acid, it should be noted that participants in the present study were not sourced from an environment with systemic folic acid fortification, so their baseline levels were lower than trials based in countries with fortification (Guinotte et al., 2003; Shelnutt et al., 2003; Shelnutt et al., 2004). Baseline levels were similar to studies from countries without fortification (Ashfield-Watt et al., 2002). Folic acid supplementation in the current study produced a large increase in folate status, a trend reflected in numerous other studies.

102 Chapter 3

Similar changes in red cell folate occurred between avoidance and supplementation periods, although to a lesser degree, possibly owing to the 15 week lifespan of erythrocytes described above. As changes in erythrocyte folate might be a better indicator of intracellular folate fluctuations in leukocytes, the source of RNA for gene expression analysis and the source of DNA for the total methylation analysis, the differences in red cell folate between avoidance and supplementation periods were encouraging. While concentrations following avoidance were not typically in the severely depleted range – generally accepted as being less than 370 nmol/L (Malinow et al., 1999; Robinson et al., 1998) – DNA methylation-induced uracil misincorporation and micronuclei formation has been observed when red cell folate is below 700 nmol/L (Fenech, 2001), a level of which a significant proportion of participants were below at the 13 week point of the current study.

It is interesting to note that individuals possessing the T/T polymorphism had a lower mean folate status at all time points. Although this difference was not significant, it may have been a reflection of the less efficient MTHFR isoform and indicative of the higher dietary folate requirement necessary to overcome this enzyme’s inefficiency.

With regard to homocysteine, the T/T group also displayed elevated concentrations of the amino acid relative to other genotypes throughout the washout period prior to supplementation. Addition of folic acid to the dietary regimen corrected mean homocysteine concentration to that of the other genotypes. However, increased levels during the avoidance period were contributed entirely by T/T participants whose folate status was low. This was in accordance with other studies (Ashfield-Watt et al., 2002; Friso et al., 2002b; Girelli et al., 1998; Jacques et al., 1996).

Friso and colleagues considered homocysteine a preliminary proxy for DNA methylation as the amino acid is an intermediate step between folate and CpG addition and acts as an inhibitor of DNA methyltransferases. Moreover, they witnessed hypomethylation occurring in their MTHFR T/T subjects when folate status was below 12 nmol/L and when homocysteine was elevated, although not defined (Friso et al., 2002b). In the present study, there were a number of participants whose plasma folate status was less than 12 nmol/L (n = 38, 40.9%) and for individuals with the T/T

Folic Acid Supplementation Study 103 genotype, almost half had a folate concentration below this level (n = 15, 48.4%). Thus the data appeared, a priori, to suggest a milieu conducive for the occurrence of genomic hypomethylation. Tempering this prediction, however, was that fact that Friso et al. noted a significant inverse correlation between total homocysteine and vitamin B12 concentrations, something not observed in the current study. Therefore lack of methionine synthase activity due to low B12 status may have contributed to genomic hypomethylation observed in the Italian cohort. In any case, it was imperative to assess global genomic methylation of the current trial participants. This is documented in the next chapter.

104 Chapter 3

4 Global DNA Methylation Analysis

The completion of the folic acid supplementation trial and its associated assays of plasma and whole blood folate confirmed that folate status was indeed modulated by the dietary regimen. Therefore, assays to assess changes in the level of global DNA methylation as a result of the intervention were undertaken. For this to take place, however, a reliable and sensitive assay for total methylation of leukocyte DNA had to be established. This chapter describes the development of such an assay, and reports the results of the study participants’ global DNA methylation.

4.1 Brief overview of existing global DNA methylation assays

Numerous methods exist for assessing the fraction of methylated cytosine residues across the entire genome. Immunochemical and chloroacetaldehyde analyses operate by detecting 5′-methylcytosine with antibodies or chemically annealing fluorescent adducts respectively, which permit subsequent fluorescence quantification (Oakeley et al., 1997; Oakeley et al., 1999). The SssI acceptance assay and the cytosine extension assay both utilise the incorporation of tritium-labelled species to assess genomic methylation. The former involves incubation of DNA with a bacterial methyltransferase which, in the presence of radio-labelled SAM as a methyl donor, adds tritiated methyl moieties to unmethylated cytosines in CpGs. The degree of incorporation, as measured by scintillation, is a direct reflection of the number of unmethylated CpGs (Balaghi and Wagner, 1993). The latter method uses methylation-sensitive restriction endonucleases to digest genomic DNA, leaving 5′ guanine overhangs which form a template for the annealing of radio-labelled cytosine residues. Again, the degree of incorporation is a function of the proportion of unmethylated CpGs in the genome (Pogribny et al., 1999).

While these methods are relatively fast and inexpensive, and in the case of the cytosine extension assay in particular, able to quantify nanogram quantities of DNA, they are

106 Chapter 4 prone to large fluctuations in variation and accuracy (Dahl and Guldberg, 2003; Quinlivan and Gregory, 2008). More pertinently, they are limited by their semi- quantitative nature – they can only assess differences or changes in methylation between two or more comparable samples and cannot provide an absolute measure of the number of cytosine residues that are methylated.

Alternatively, complete hydrolysis of genomic DNA material, followed by high resolution separation and quantification of its constituents, allows for complete base composition profiling of a DNA sample. Typically this is achieved with enzymatic digestion and high performance liquid chromatography with UV detection. The technique is well described and many minor variations of the procedure have been reported since the method was originally published in 1980 (Galushko et al., 1988; Gehrke et al., 1984; Kuo et al., 1980; Lim and Peters, 1989; Magana et al., 2008; Palmgren et al., 1990; Ramsahoye, 2002; Simpson and Brown, 1986). The technique has also been scaled down for capillary electrophoresis units (Fraga et al., 2002). The impediments to this technique are the relatively high cost of specialty equipment, non- parallel nature of running samples, long run times and relatively high amounts of DNA required. Coupling of the chromatography column to a mass spectrometer (LC/MS) with an electrospray ionisation (ESI) source has been used to improve resolution with small amounts of sample (Friso et al., 2002a; Kok et al., 2007; Quinlivan and Gregory, 2008; Song et al., 2005). Mass spectrometry has the added benefit of identifying the mass of each particular species as it is eluted from the column and thus complete separation of bases by chromatography is not critical. For the current study, a LC/MS assay was employed to assess total DNA methylation of trial participants for the above reasons, plus the fact that studies that served as catalysts for the current trial also assessed genomic methylation differences by this technique (Friso et al., 2002a; Friso et al., 2002b). A diagrammatic representation of the procedure is shown in Figure 4.1.

Global DNA Methylation Analysis 107

DNA digestion HPLC separation

2′-Deoxycytidine

5′-Methyl-2- deoxycytidine Strand separation

MS detection

ESI fragmentation Nuclease S1 & ionisation

5′-Methyl-2- deoxycytidine

2′-Deoxycytidine Venom phosphodiesterase 5′-Methyl Cytosine cytosine

Peak identity & Alkaline integration phosphatase Ion count

Free nucleoside 112 products 126

m/z

Figure 4.1 Procedural overview of the global 5′-methylcytosine assay. Double stranded genomic DNA is separated by heat denaturation and digested with a combination of nucleases (first panel). Nuclease S1 possesses both exo- and endolytic activity and cleaves 3′ phosphodiester bonds, producing mono- or short oligonucleotides. Venom phosphodiesterase (phosphodiesterase I) further digests oligonucleotides at the same position. Alkaline phosphatase catalyses the cleavage of 5′ phosphates to yield a pool of single deoxynucleosides. The removal of phosphate groups permits improved chromatographic separation of nucleoside material when using a reversed phase HPLC column. Separation is depicted in the second panel. As nucleosides elute from the column they enter the mass spectrometer (panel three). The electrospray ionisation source of the MS unit facilitates the ionisation and desolvation of eluted species, but also fragments nucleosides into base and deoxyribose daughter ions. Mass to charge ratios pertaining to the applicable base components of the nucleoside fragments are thus selected for and filtered accordingly. Ion intensity is also used to quantify species from which the methylated proportion of cytosine can be calculated.

108 Chapter 4

4.2 Development of LC/MS procedure

Prior to participant samples being analysed, all aspects of the assay had to be verified and optimised. This included ensuring hydrolysis of DNA was complete, chromatographic separation was effective and that eluted nucleosides could be detected and quantified successfully. In order for the first requisite to be fulfilled, the identity of digestion products had to be resolved, requiring the effective operation of the MS. Hence, the MS component of the assay was investigated first.

4.2.1 Principles of mass spectrometry

The process underlying quadrupole and ion trap mass spectrometry involves the motion of ions in an oscillating electric field (Arnott, 2001). Conceptually, if two electrodes are oppositely charged, ions between them will accelerate away from one electrode towards the other. Switching electrode polarity will cause the ions to decelerate and reverse their direction. Thus, if the frequency and amplitude of the electric field is correctly calibrated, ions can be retained by oscillating between the poles. In practice, quadrupole spectrometers confine ions in two dimensions between four (or more in the case of ion trap instruments) rod-shaped electrodes producing vertical and horizontal fields, while allowing the particles to migrate in the third dimension, travelling from an ion source to a detector. Importantly, as the mass of ion species determines the degree of acceleration between poles, field voltages can be configured to destabilise and eject ions of certain weights whilst retaining others, making the quadrupole an effective mass filter. Additionally, ion trap instruments can arrest ions in a third dimension, allowing accumulation of desired particles, which can then be manipulated in a variety of further mass analyses.

4.2.2 Instrument optimisation

In the current assay, only ion products of cytidine and methylcytidine were sought for detection, hence the instrument required “tuning” of octopole and trap field parameters to optimally select these species. This process was determined electronically by the instrument and was largely invisible to the user. Standards of each species (10 µM 2′-deoxycytidine and 5′-methyl-2-deoxycytidine) were directly infused and the base

Global DNA Methylation Analysis 109 peak (most abundant ion) was automatically tuned for by the system to maximise signal at the detector. Tune parameters were saved and used for all subsequent assays.

Similarly, optimisation of the ionisation source was required. The most common procedure for converting on-line species suspended in liquid, provided by HPLC, into gaseous ionic radicals required by MS is achieved by an electrospray ionisation source (Arnott, 2001; de Hoffmann and Stroobant, 2007). In this process, liquid is ejected from a needle across which a large voltage is applied (± 3-6 kV). Analytes become charged, separate by coulomic diffusion and are emitted as a cone of droplets (a Taylor cone). A sheath gas (low pressure nitrogen) directs a portion of the droplets to enter the MS through a heated capillary under vacuum. This process desolvates the analytes and liberates them into the gas phase.

While electrospray ionisation is regarded as a soft ionisation technique, partial fragmentation of deoxynucleosides had been observed previously (Friso et al., 2002a). In that work, ESI conditions were subsequently modified to produce maximum fragmentation and measure the nitrogenous base fragment ion. The same approach was taken in the present study: a needle voltage of 5 kV and a capillary temperature of 350 °C were used to achieve fragmentation at flow rates of up to 200 µL/min. The extent of fragmentation is depicted in Figure 4.2.

4.2.3 Initial assessment of standard separation

Column separation characteristics and tune parameters were assessed for sample mixes of 2′-deoxycytidine and 4% 5′-methyl-2-deoxycytidine which would be more representative of hydrolysed participant samples. Column run conditions stipulated by Friso and colleagues – 7 mM ammonium acetate and 5% methanol (v/v) – provided adequate separation of these species. An exemplar is shown in Figure 4.2.

110 Chapter 4

Figure 4.2 Preliminary trace of ion abundance versus time for 25:1 molar mix of cytidine and methylcytidine standards, as separated by HPLC at 0.2 mL/min. Cytidine was eluted from the column at approximately 4 minutes, followed by methylcytidine at approximately 8 minutes as indicated by retention time (RT). MS total ion count peak integration is denoted as “manually assigned area” (MA). Insets: centroid plots of ion abundance at the monitored m/z ranges for the eluted species and their fragmented components. Measured mass widths were 2 units. Cytidine appeared to be fragmented entirely – ions invariably possessed a m/z of 112, corresponding to the protonated cytosine base, while the unfragmented deoxyribonucleoside (protonated mass of 228) was not present. While, a priori, fragmentation appeared incomplete for methylcytidine due to a small proportion of ions detected with a m/z corresponding to the unfragmented molecule (m/z = 241), much of this was attributable to baseline noise (note the lower intensities plotted for methylcytidine species when compared to cytidine). Thus fragmentation was deemed to be acceptable for both nucleosides. MS ion collection began at 3 minutes and utilized tuning parameters optimized for cytidine. Tune parameters were switched at 6 minutes to optimally select for methylcytidine.

4.2.4 Genomic DNA hydrolysis, separation and species identification

Methods for digesting DNA for separation by HPLC have been well described in the literature (see section 4.1). The protocol employed in the current study is reported in section 2.5.3. Digests were initially assessed by HPLC. These were unfiltered and tended to incur high column back-pressures upon running (> 3500 PSI), requiring frequent guard column replacements. Microcon filters were subsequently used to filter samples prior to loading which stabilised column pressure.

Global DNA Methylation Analysis 111

Figure 4.3 Qualitative MS assessment of hydrolysed DNA following HPLC separation. Top chromatogram depicts separation of nucleoside species as detected by absorbance at 271 nm. Total ion counts are shown below which correspond to eluted material depicted in the top chromatogram as grey boxes. The trap was configured to sequentially retain ions (single ion mode) in seven scan events that encompassed predicted masses of whole nucleosides and their fragments. Note that the scale of each total ion count varies for each elution window – the MS was tuned for cytosine (m/z of 112) and so exhibits the greatest intensity for this species (A). Relative noise floor is higher for less abundant or poorly filtered species which accounts for the presence of neighbouring nondescript peaks. In all cases, the m/z ratio of the dominant peak at each time point corresponds to the predicted mass of a protonated nitrogenous base: 112, 126, 152, 127 and 136 for cytosine, methylcytosine, guanine, thymine and adenine, respectively. The absence of ions at m/z values equating to unfragmented cytidine and methylcytidine (228 and 242 respectively) suggests near- complete fragmentation in the ESI has occurred, although the noise floor tends to decrease as m/z values move away from the mass of the species tuned for.

112 Chapter 4

Initial nucleoside separation under isocratic conditions and a flow rate of 0.2 mL/min was performed and is depicted in the chromatogram of Figure 4.3. Adequate separation of species was achieved, although some overlap of the second and third peaks was apparent. Later runs (see section 4.4) would employ column cooling to further aid separation. Extended running of aqueous solvent through the column altered retention times, thus the column was washed periodically in 50% acetonitrile to restore initial separation characteristics.

Figure 4.4 Species identification of hydrolysed DNA by MS base peak. Peaks depict the ion derivatives of deoxynucleosides eluted from the HPLC column. Only the most abundant ion at any given point in time (the base peak) was plotted by the software. The mass of the most abundant ion in each peak is denoted by the BP prefix. For example, the peak at 4.29 minutes consisted entirely of ions with a mass to charge (m/z) ratio of 112. As the MS is tuned to select for cytosine, the abundance of the other bases is considerably lower. Additionally, the trap was configured to sequentially retain ions with specific m/z ratios pertaining to the six bases (with an isolation width of two). Consequently, each scan event is undertaken one sixth of the time, giving rise to the blocky appearance of base peaks. Peak masses correspond to the following protonated bases, [m+H]+, in order of elution: cytosine (112), methylcytosine (126), guanine (152), thymine (127) and adenine (136). No uracil peak was identified, indicating absence of appreciable RNA contamination. The first and second peaks have been integrated to assess the proportion of recovered cytosine that was methylated, which here equates to 4.49%. Note this is not an accurate reflection of the proportion of methylated genomic cytosine as raw ion abundance does not account for differences in mass filtering of the two species or ion suppression artifacts. RT: retention time. MA: manually integrated peak (area).

Global DNA Methylation Analysis 113

Figure 4.5 Representative run of hydrolysed genomic DNA. Top chromatogram depicts eluted nucleosides as separated by HPLC and detected by diode array at 271 nm. Bottom trace displays resulting MS ion count for cytosine and methylcytosine. Associated centroid graphs confirm the mass identity of each of the eluted species.

114 Chapter 4

Only five major peaks were detected which indicated minimal ribonucleotide contamination and comprehensive DNA hydrolysis. The masses of each of these peaks were assessed by MS (see Figure 4.3 and Figure 4.4). An example assay of hydrolysed genomic DNA following establishment of the procedure is shown in Figure 4.5.

4.2.5 Limits of detection, inter-assay variation and standard curves

The limit of detection for both cytosine and methylcytosine was below an on-column load of 0.08 pmol. Standard curves were obtained from this level to 200 pmol, however these were only linear to approximately 40 pmol (see Figure 4.6). Above this point, the MS unit displayed signal attenuation, presumably due to trap or detector saturation. As cytosine abundance was considerably higher than that of methylcytosine and the latter was unaffected by this phenomenon, suppression of counts of the former at high loads considerably altered cytosine/methylcytosine ratios. Therefore, load volumes of subsequent samples were adjusted to insure that cytosine peak abundance was below 250 million counts.

A corollary of this limited dynamic range, a trait not uncommon of ion trap instruments (Kazakevich and LoBrutto, 2007), was that methylcytosine counts were relatively low when cytosine loads were kept under the aforementioned abundance threshold. Methylcytosine peak integration was less accurate as a result and together with inconststencies in autosampler loading (section 4.2.7), produced considerable variance of calculated methylcytosine percentage among replicates of said standards (see Figure 4.8).

4.2.6 Internal standard considerations

To combat these problems, Friso et al. spiked experimental samples with stable isotopic standards of both 2′-deoxycytidine and 5′-methyl-2-deoxycytidine (Friso et al., 2002a). Ratios of genome-derived species to their internal standard isotopomer were calculated within each run and multiplied against an external standard curve of isotopic standards to produce an adjusted nucleobase amount for each sample.

Global DNA Methylation Analysis 115

An anticipated caveat to this approach would be that the accuracy of such a procedure would immediately become reliant on the proficiency of pipetting the isotopomers,

Figure 4.6 Standard curves of cytosine and 4% methylcytosine. The first graph highlights the attenuation of ion counts with excessive sample due to ion trap saturation (space-charge effect). Methylcytosine also demonstrates improved operating line retention – counts are higher when compared to the equivalent amount of cytosine. The bottom graph shows an exploded section of the first standard curve representing its linear portion. For optimal calculation of the methylated proportion of cytosine, load volumes of sample were adjusted to deliver cytosine ion counts towards the top of the linear range so as to enable reasonable resolution of methylcytosine peaks. Mean correlation coefficient of the linear part of the curve (2-20 pmol): cytosine = 0.987, methylcytosine = 0.966. The assay was performed in triplicate.

116 Chapter 4 whilst additionally being dependent on the proficiency of the instrument to measure the standards in the curve.

15 Despite this reservation, inquiries were sought for the purchase of [ N3] deoxycytidine 15 and [ N3] methyldeoxycytidine. Unfortunately, availability was limited to expensive bespoke chemicals that would invoke long project delays while waiting for their synthesis. Owing to the unknown worth of these species, together with events outlined in the section below, isotopomers were not used.

4.2.7 Equipment issues

Impeding the development of the assay were multiple equipment failures and associated time spent attempting to remedy these issues. Problems included autoinjector leakage and erratic sample volume loading, arcing of the ESI needle causing the circuitry controlling ionisation to trip, replacement of a faulty MS motherboard and various conflicts with the controller software that caused various components of the system to not be recognised or fail to initialise. While these will not be discussed in detail here, it should be noted that these problems played a large part in abandoning the LC/MS assay in favour of another procedure. Quantifying species by integrating chromatographic peaks instead of by ion abundance offered a straightforward surrogate. However, failure of the diode array detector coupled to the MS prior to running participant samples meant that these could be assessed only by ion count and not by UV detection. In order to measure absorbance, switching analysis to a completely new HPLC apparatus was necessary. This process is documented in section 4.4.

4.3 Results of LC/MS assay of total DNA methylation

Despite the inaccuracies of the LC/MS system, participant samples were run regardless and cytosine and methylcytosine were assessed by ion abundance. Molar amounts were interpreted from averaged standard curves run before and after experimental samples. Although of limited worth, the results of these assays are reported below in Table 4.1.

Global DNA Methylation Analysis 117

No statistically significant differences in total methylation were observed between study phases or by genotype. The assay tended to underestimate methylcytosine percentage – overall means for baseline, avoidance and supplementation periods were 3.42, 3.35 and 3.36 respectively. Unsurprisingly, standard deviations of the mean were very high, including that of differences between study phases, which further indicated excessive error inherent in the assay as opposed to well correlated intra-participant data with large inter-participant variance.

Table 4.1 LC/MS assay of global DNA methylcytosine by MTHFR C677T genotype at baseline, following folate avoidance and following repletion (mean ± SD)

MTHFR n Baseline Avoidance Difference* Repletion Difference* genotype (Wk 0) (Wk 13) (Wk 13 – Wk 0) (Wk 24) (Wk 24 - Wk 13) Methylcytosine (%) Overall 93 3.42 ± 0.59 3.35 ± 0.53 0.08 ± 0.44 3.36 ± 0.48 -0.01 ± 0.44 CC 31 3.32 ± 0.50 3.30 ± 0.44 0.02 ± 0.39 3.32 ± 0.38 -0.02 ± 0.41 CT 31 3.59 ± 0.74 3.36 ± 0.60 0.23 ± 0.46 3.39 ± 0.62 -0.03 ± 0.53 TT 31 3.39 ± 0.51 3.39 ± 0.56 0.00 ± 0.45 3.38 ± 0.47 0.01 ± 0.37 *Standard deviations were calculated from the total of individual differences between phases. No differences were statistically significant.

4.4 High performance liquid chromatography

A HPLC-only technique was alternatively used to quantify DNA methylation. An Agilent 1100 series automated HPLC system was available for use in the Department of Human Nutrition, University of Otago. The instrument included a G1315B diode array detector with a rated sample limit of detection of less than one picogram. Having already established base identities and column retention times, utilisation of the Agilent HPLC system for the total DNA methylation assay was relatively straightforward, given that the same column was used.

Some limitations existed for an HPLC-only assay: as species were identified by UV absorbance alone, eluted peaks had to be free of overlap. To ameliorate the imbricated retention times of methylcytidine and deoxyguanosine (as alluded to in Figure 4.3), the column was cooled to 4°C. This prolonged retention times and allowed complete separation of the two species. Additionally, it was anticipated that increased column

118 Chapter 4 loads would be required, given that MS instruments possess a greater sensitivity than diode arrays. In reality, the nature of the ESI interface dictated that only a small portion of the eluent from the column entered the MS, with the majority going to waste. Consequently, little more hydrolysed DNA was required for the HPLC-based assay than the LC/MS procedure. A 20 µL loop was used, compared to a 15 µL loop for LC/MS, which conferred adequate detection of methylcytosine – for example, typical signal to noise ratios were above 20 for 1.6 pmol (200 pg) of on-column methylcytosine controls (20 µL load at 0.08 µM). Standard curves possessed good linearity across the assayed range, as shown in Figure 4.7. Replicates also had improved variance compared to the LC/MS method (see Figure 4.8).

4.5 HPLC results

As described in section 2.9, amounts of 2′-deoxycytidine and 5′-methyl-2-deoxycytidine were measured by HPLC for the calculation of 5′-methylcytosine as a percentage of total cytosine, in all study participants at all trial time points. Results are listed in

Figure 4.7 Standard curves of cytidine and 5′-methylcytidine as performed by HPLC. Cytidine data are expressed in blue, methylcytidine in red.

Global DNA Methylation Analysis 119

Table 4.2. Importantly, no statistically significant changes in the proportion of methylated genomic cytosine were observed, either between dietary phases of the study, or by MTHFR genotype. Methylated cytosine percentages were very stable, despite significant changes in participant plasma folate and homocysteine – the overall mean of methylated cytosine remained within 0.01% of baseline throughout the study.

The stability of genomic methylcytosine content throughout the study was judged to be a genuine observation and not because of type two error due to inadequate assay sensitivity, a trait of the LC/MS experiment (Figure 4.8). Interpreting individual differences between dietary phases as absolute values revealed that the overall mean change was 0.17 percentage points between baseline and avoidance, and 0.12 points between avoidance and supplementation periods. The highest individual changes witnessed in T/T participants, for example, were from 4.35% to 3.97% (baseline to avoidance, a difference of 0.38) and 3.85% to 4.18% (avoidance to supplementation, a

% methylcytosine % % methylcytosine % Mean = 3.88 Mean = 3.85 SD = 0.35 SD = 0.031

LC/MS Control replicates HPLC Control replicates

Figure 4.8 Comparison of replicate variation between LC/MS and HPLC assays. Data points represent consecutive replicates of a standard containing 40 µM 2′-deoxycytidine and 1.6 µM 5′-methyl-2-deoxycytidine (25:1 mix). Runs were interleaved between individual participant samples (baseline, avoidance and supplementation), hence time between standard runs was approximately 2 hrs. Cytidine (cytosine for MS) and methylcytidine (methylcytosine) raw data from each run were converted to their respective molar amounts from standard curves. These adjusted data were then used to calculate the percentage of methylated cytosine. Yellow, orange and red horizontal lines represent the mean, one standard deviation and two standard deviations from the mean respectively. The graph at left highlights the large run-to-run variance apparent with the LC/MS assay – within this sampling, coefficient of variation is 0.0902. Variance was deemed unacceptable for an assay responsible for detecting potentially subtle changes in global DNA methylation. Conversely, the HPLC-only technique exhibited a coefficient of variation approximately one order of magnitude lower than the LC/MS procedure (0.0081).

120 Chapter 4

Table 4.2 Global DNA methylcytosine by MTHFR C677T genotype at baseline, following folate avoidance and following repletion (mean ± SD), according to HPLC

MTHFR n Baseline Avoidance Difference Repletion Difference genotype (Wk 0) (Wk 13) (Wk 13 – Wk 0) (Wk 24) (Wk 24 - Wk 13) Methylcytosine (%) Overall 93 4.03 ± 0.25 4.03 ± 0.24 0.00 ± 0.23 4.02 ± 0.21 -0.03 ± 0.21 CC 31 4.04 ± 0.14 4.05 ± 0.16 0.01 ± 0.12 4.05 ± 0.18 0.01 ± 0.11 CT 31 4.00 ± 0.30 4.01 ± 0.35 0.01 ± 0.34 4.00 ± 0.25 -0.05 ± 0.32 TT 31 4.05 ± 0.27 4.03 ± 0.22 -0.03 ± 0.20 4.01 ± 0.22 0.02 ± 0.13

difference of 0.33). This was despite methylcytosine content ranging from 3.19% to 5.18% across all MTHFR genotypes and trial phases.

When methylation of DNA was assessed as a function of folate status, no association was found (see Figure 4.9). DNA methylation was independent of folate status for all MTHFR genotypes and at all study time points. Of the 14 participants in the T/T group that had a plasma folate≤ 12 nmol /L, there was an insignificant trend towards lower DNA methylation between baseline and post-depletion (-0.1%, p = 0.08).

Figure 4.9 DNA methylation stratified by post-avoidance plasma folate concentration (folate status), and separated according to MTHFR C677T genotype. The proportions of cytosine that are methylated for each participant at baseline, following dietary folate avoidance and post- folic acid supplementation are represented as triplets of stacked diamonds, squares and circles, respectively. Total folate concentrations post-avoidance for each participant are denoted by black crosses. Heterozygous individuals have been omitted for brevity. For all MTHFR genotypes, the extent of DNA methylation is independent of blood folate status at all study time points.

Global DNA Methylation Analysis 121

4.6 Discussion

This section addresses both the technical merits of the two assays and the assay outcome. How the results compare to the work of Friso et al. is briefly canvassed; a more thorough critique regarding other related literature is presented the final chapter of this thesis.

4.6.1 Results of assay and comparison to data from Friso et al.

For the current thesis, it was hypothesised that avoidance of folic acid fortified foods by participants would lower blood folate concentrations and correspondingly induce DNA hypomethylation of peripheral blood leukocytes. Subsequent folic acid supplementation would increase blood folate and increase DNA methylation. While changes in peripheral blood folate concentrations were observed, modulation did not extend to global DNA methylation, as documented in this chapter. Instead, mean total

C/C T/T

Figure 4.10 Comparison of global DNA methylation data between Italian and New Zealand populations according to MTHFR genotype and plasma folate status. Data obtained by Friso and colleagues, 2002, is depicted in blue whereas the current study data is labelled in red. Friso et al stratified their data by folate status (split by tertiles) and the current study data is thus separated accordingly – data from the avoidance phase of the trial is used. Data points from the Dunedin cohort are expressed as mean ± standard deviation. Error bar type was not defined for the data from the Friso et al. publication (Friso et al., 2002b).

122 Chapter 4

5′-methylcytosine remained static throughout the duration of the intervention, including in participants homozygous for the MTHFR C677T polymorphism with low plasma folate (< 12 nmol/L).

The impetus for undertaking the present study was due in large part to the results from Friso and colleagues, who described a large decrease in global DNA methylation (approximately 50%) in subjects who had both a low folate status (< 12 nmol/L) and were homozygous for the MTHFR polymorphism. Data from that study and the present work are compared in Figure 4.10. As the two datasets are seemingly at odds for T/T individuals with low folate status, some possibilities are offered to explain the differences observed. Firstly, it should be noted that in the present study, healthy individuals were recruited, whereas, Friso et al. used subjects from a hospital in Northern Italy, where two thirds of the cohort had confirmed coronary atherosclerosis. Elevated homocysteine is a hallmark of this disease and homocysteine measurements were 14.7 and 21.3 μmol/L for C/C and T/T participants respectively. These were approximately twice what was observed in the current study, despite T/T individuals displaying similar depleted plasma folate levels. Homocysteine and S-adenosylhomocysteine are efficient inhibitors of DNA methyltransferases. Interestingly, Friso et al. made no mention of any differences in morbidity between methyl replete and depleted individuals.

Additionally, in the present study, participants were supplemented with other vitamins involved in homocysteine remethylation. As no intervention was undertaken in the Italian study, deficiencies in other parts of the pathway, for example inhibition of methionine synthase activity due to vitamin B12 deficiency, may have contributed to the observed hypomethylation.

4.6.2 Comments on the merits of the LC/MS and HPLC assays

The originally-anticipated method to deduce global DNA methylation proved to be problematic. While the limits of detection of the mass spectrometry apparatus were very high, the dynamic range of the system was inadequate for the accurate quantification of two species occurring in disparate abundances. Increasing sample loads resulted in

Global DNA Methylation Analysis 123 marked ion suppression artifacts for cytosine species, while restricting loads reduced methylcytosine peak resolution and tested the limits of the associated autoinjector, introducing unnecessary error. In a sense, the LC system and the MS were mismatched, with the former not able to supply the sample resolution required by the spectrometer.

Other groups have also attempted to replicate the LC/MS method of Friso et al. In New Zealand, projects at the Christchurch School of Medicine (Magzhal, Brown & Winterbourne, personal communication, June 2005) and The Liggins Institute of the University of Auckland (van Zijl, personal communication, June 17, 2005) have also attempted to reproduce the MS-based methylation assay, but without success. In the second instance, the Liggins Institute experimenters reverted to a HPLC-only procedure, and while requiring a greater amount of hydrolyzed DNA to achieve a satisfactory sensitivity, offered increased system stability and greater success.

The implementation of the LC/MS procedure has been reported with some success in other instances, however. Shelnutt and colleagues used a LC/MS assay analogous to the present method which utilised an external standard curve (Shelnutt et al., 2004). They were able to achieve an intra-assay coefficient of variation of 4% (compared with 9% for the current study). Song and colleagues used deoxyguanosine present in DNA digests as a pseudo internal standard to achieve impressive standard CVs – reportedly as low as 0.97% (Song et al., 2005). Methods similar to Friso and colleagues using isotopic standards have been documented (Kok et al., 2007; Quinlivan and Gregory, 2008), with CVs typically 2.5%.

Despite the reported success of the method, it is clear the nature of the LC/MS procedure is complex, perhaps unnecessarily so; in order to function successfully, compensations need to be made for partial fragmentation of nucleoside species, ion count drift and poor dynamic range which all affect mCyt/total Cyt ratios.

By comparison, an HPLC-only method suffered from none of the above issues. It was only reliant on complete separation of nucleoside species. Both protocols still required complete removal of residual RNA. Additionally, the HPLC-based method was already well-described in the literature: multiple papers reporting HPLC’s use for determining

124 Chapter 4 methylcytosine content have been published, the earliest dating back to the early 1980s (Gama-Sosa et al., 1983). Furthermore, comprehensive studies reporting the separation and quantitation of nucleosides using HPLC columns of similar chemistry to the present study have been published as recently as 2008 (Magana et al., 2008).

When describing the LC/MS technique, Friso et al. discounted HPLC-only methods solely because they required large amounts of genomic DNA (5-50 µg) and that run times were relatively long (Friso et al., 2002a). They identified no inherent flaw in existing HPLC methodology. As sufficient quantities of DNA material had been collected in this study and were available for HPLC, and good chromatographic separation was achieved with an approximate increase in run time of only ten minutes, LC/MS offered no real advantages over the HPLC technique used in this study. Additionally, advances in detection meant that the newer HPLC system rivalled the LC/MS setup available in terms of sensitivity at the column size and flow rates used for the study.

5 Microarray Analysis

5.1 Introduction

From the outset of this project, it was anticipated that an investigation into the gene expression of peripheral blood leukocyte DNA would be undertaken, in addition to the global DNA methylation analysis described in the previous chapter. The rationale for this was as follows: the hypothesised effects of restricted dietary folate intake upon the study participants would be a reduction in leukocyte total CpG methylation resulting from a reduced methyl supply. Additionally, restoration of the methyl source would allow newly synthesised leukocytes to attain complete methylation of receptive CpG dinucleotides. As one function of DNA methylation is to act as a component of gene repression, it was supposed that altered DNA methylation, through the study’s dietary regimen, may modify discrete gene regulatory regions and thus modify expression of those genes. As evidence of dietary inputs affecting gene expression through direct alteration of DNA methylation is scant, demonstrating a relationship in this study would be pertinent.

However, the predictions regarding modulation of genomic methylation by dietary intervention were not substantiated by the global methylation analysis. Instead these assays produced no observable change in total methylation between repletion and depletion periods, or between MTHFR genotypes. A priori, it would be reasonable to assume that discrete methylation would have remained unaffected also. However, reworking of discrete elements’ methylation may still have occurred, whilst being too subtle to have been observed with a global methylation assay. Thus work was undertaken to assess any changes in gene expression that might infer a possible reworking of promoter methylation. Chapter 5 details this work.

126 Chapter 5

An early objective was to establish a suitable method for assessing changes in distinct CpG elements in the DNA obtained from study participants. Methods to analyse discrete gene methylation patterns are numerous and wide ranging in their application (Dahl and Guldberg, 2003; Zilberman and Henikoff, 2007), from the querying of an individual island (Clark et al., 2006) to massively parallel analyses of CpGs across a genome (Bentley, 2006). Methodologies at the larger end of this spectrum that were considered for use in the current project are discussed below.

5.1.1 Large scale discrete DNA methylation analysis methods

A host of methods have been developed to measure states or changes in state of multiple DNA methylation elements across the genome. Recently, truly genome-wide methylation profiling techniques have become available, with technologies such as the Solexa fluorescent nucleotide-based system from Illumina and the 454/Roche high- throughput pyrosequencing system producing near-complete methylation mapping with single-CpG resolution in a rapid timeframe (Pomraning et al., 2009). These systems were unavailable to the student at the time the study was undertaken. Other techniques that were already published were more limited in scope. They were all based on one of three core methodologies to distinguish methylated DNA – digestion of DNA with methyl-sensitive nucleases, affinity purification of methylated DNA and bisulphite conversion – and then upscaled to assess multiple CpGs in parallel.

Of the methyl-sensitive digestion techniques, Restriction Landmark Genomic Scanning, or RLGS is the most established (Costello et al., 2009; Rush and Plass, 2002). In this method, genomic DNA is digested with a methylation-sensitive restriction enzyme to produce fragments that are then radiolabeled and separated by two dimensional gel electrophoresis. Gels from case and control samples are compared and changes in methylation are depicted as the presence or absence of spots. Spots of interest are then excised for sequencing to determine their identity (Costello et al., 2000). Similar techniques, Amplification of Intermethylated Sites, or AIMS (Frigola et al., 2002) and Differential Methylation Hybridisation, or DMH (Huang et al., 1999; Yan et al., 2001), instead assess changes in methylation by separating fragments by high resolution electrophoresis, and competitive hybridisation of amplified fragments, respectively.

Microarray Analysis 127

Methylation amplification DNA chip (MAD) is a similar technology to DMH, in that it also employs enrichment of methylated fragments by methylation-sensitive enzymatic cleavage, and subsequent amplification using PCR and array hybridisation (Hatada et al., 2002). Different endonucleases are used to better limit amplified fragments to CpG islands only. A third array-based technique, Methylation-specific oligonucleotide microarray for DNA methylation analysis, or MSO (Gitan et al., 2002), is closely related to DMH and MAD. Instead of restriction digests, DNA is first treated with sodium bisulphite, followed by PCR amplification. This process deaminates unmethylated cytosines to uracils, which are then replicated as thymines during PCR. This results in dissimilar primary sequence between samples which differ in their methylation profile. After labeling, the targets are then hybridised to a microarray of custom-designed probes, where hybridisation is dependent on the “methylation” state of the PCR product – the presence or absence of the altered bases.

The fundamental limitation of these restriction enzyme-based methods is that they are constrained to certain digestion motifs and therefore have a limited ability to exclusively select regulatory elements. Consequently, these methods assess only a small fraction of the methylome. The MSO bisulphite method, while working around this problem, is laborious, requires custom probes and consequently extensive probe design, and thus also produces arrays containing relatively few elements. For example, published studies using MSO methodology constructed arrays of just a few hundred probes (Adorjan et al., 2002). DMH arrays fared better with publications describing arrays with several thousand elements (Yan et al., 2001; Yan et al., 2002). RLGS is similarly able to query only a few thousand elements.

More recently, methods for methylation enrichment by affinity have been developed which avoid the problem of CpG analysis being limited to elements within certain restriction motifs. The MeDIP (or occasionally mDIP) procedure uses a monoclonal antibody raised against 5-methylcytosine to pull out methylated DNA that has been sonicated into fragments of 300 to 600 bp (Weber et al., 2005). This enrichment procedure has been used in microarrays, where recently-developed tiling arrays (Weber et al., 2005), or arrays containing upwards of 13,000 promoters have been produced (Keshet et al., 2006). The MeDIP procedure held the most promise of the above

128 Chapter 5

techniques, and interestingly, it forms the basis of the Solexa genome-wide methylation analysis technology (Down et al., 2008). At the time of planning the current study, however, institutional experience with the MeDIP method was limited, and it was not chosen for the present work.

The MeDIP procedure aside, microarray technology was an attractive platform to use for the present study because of its potential to assess multiple elements simultaneously. Instead of attempting to assay methylation directly, another approach was considered which was simply to observe changes in gene expression using microarray technology. Any differential gene expression could then be used to infer possible alterations in DNA methylation which could be investigated further using methylation-sensitive sequencing methods.

5.1.2 DNA microarrays – gene expression as a proxy for DNA methylation

The differential expression between depletion and repletion periods, for example, of a gene whose expression was under methylation control could indicate that the gene’s regulatory region may have undergone some change in methylation state, resulting in the modified expression. Microarray methods for analysing gene expression are well- established and capable of assessing expression in a highly parallel fashion. Moreover, the facilities and specialised equipment required to conduct microarray experiments were available in the laboratory. With suitable filtering of results to select for relevant genes, this approach was considered ideal for the current study. Also, observation of changes in expression of genes involved with folate metabolism, not necessarily under methylation control, was of additional interest and could be assayed with this method.

Of course, as the methylation status of DNA elements would not be directly assayed in the microarray process, this method could only be treated as a triage tool to select for potential altered-methylation targets. Additionally, the differential expression of most genes would not, in all likelihood, be attributable to reworked promoter methylation and so further analysis would be required to select for only genes that represented good candidates for methylation-sensitive expression.

Microarray Analysis 129

5.2 Experimental design

The overall rationale of the array experiments was to elucidate differences in gene expression amongst the participants, which might then infer potential differences in participants’ methylation profiles, or propensity towards disease conditions associated with low folate and the MTHFR C677T genotype. This aim would thus encompass multiple parts. Firstly, differentially expressed (DE) genes immediately involved in folate metabolism and related cellular processes, which would be expected to have altered expression as a result of the study’s dietary regime, would be queried. Secondly, the analysis would be widened to assess all statistically significant DE genes and discover elements not directly associated with folate metabolism but which may be indirectly affected through altered methylation-controlled regulation. Owing to the potentially large number of DE genes in the entire set, genes would require filtering to select for elements relevant to folate-related pathways. Within each of the respective categories, three distinct approaches would be undertaken. Firstly, differences in gene expression between depletion and repletion phases common to the whole group would be investigated. Secondly, differential gene expression as a function of MTHFR C677T genotype would be examined under conditions of low and replete dietary folate. Finally, differential expression as a function of folate status following the folic acid avoidance phase would be assessed. All of these considerations could be encompassed in the microarray design, elaborated in the sections below.

5.2.1 Reversed-dye microarray platform

Printed oligonucleotide arrays were used in this experiment. These arrays allow for two samples to be labeled with different fluorescent adjuncts and competitively hybridised on a single slide. The degree of fluorescence of each species reflects the amount of hybridised material and comparison of the signals at any given probe is used to establish differences or changes in expression for that gene.

Typically two-colour array studies use a reference design, where all experimental samples are identically labeled and hybridised against common reference, labeled with a second dye. This eliminates the effect of dye bias (a phenomenon whereby a cDNA

130 Chapter 5

displays a greater binding affinity to one dye over another, thus producing a dissimilar signal intensity), as all sample cDNAs are labeled with the same chromophore.

This experimental design adds cost however, as only one sample may be hybridised per slide and as such, depletion/repletion sample pairs from each participant would require two arrays. With same-participant samples spread over multiple slides, variation between arrays would dictate that duplicates of all slides be performed. This would result in four hybridisations per participant which would be prohibitive in terms of cost and RNA amounts required for both reference and experimental samples. An alternative scheme is discussed in the section below.

5.2.2 Paired sample non-reference design

The array design was instead based on a well documented method described as a non- reference design for paired samples (Dobbin et al., 2003). Paired samples typically consist of “before” and “after” treatments for individual participants, a definition which was applicable to the present study; the folate avoidance period and the folic acid supplementation period of the trial were the equivalent of the “before” and “after” phases, respectively. The design is geared to identify genes affected by the treatment, or in this case, between avoidance and supplementation, while at the same time minimising person-to-person population variation. This design has been used successfully in numerous array studies (Kim et al., 2006; Niculescu et al., 2005; You et al., 2008).

The design was as follows: paired samples were differentially labeled and co-hybridised on one array per pair. To avoid gene-specific dye biasing, half the arrays were run forward and half were reversed, that is, for labeling of avoidance and supplementation cDNA, the Cy3 and Cy5 dyes were swapped. Arrays that had been “reversed” had their data channels swapped back or restored following their intensity data acquisition. Selection of reversal candidates was randomised with the exception that equal numbers of C/C and T/T MTHFR genotypes fell in each group. The procedure is depicted in Figure 5.1.

Microarray Analysis 131

5.2.3 Additional considerations

Due to the cost of arrays, the decision was made to compare expression profiles of participants possessing only C/C and T/T genotypes, as it was assumed these groups would display the largest variance.

C/C T/T n = 15 n = 15

“Forward” “Reverse” Labeling C/C T/T C/C T/T n = 8 n = 8 n = 7 n = 7

Post-avoidance Post-avoidance “Reversed” slides inversely Post-supplementation Post-supplementation labeled

Hybridisation & scanning

Raw data channels swapped to restore polarity

Processing

Figure 5.1 Array procedure overview. Participants of each MTHFR genotype (C/C and T/T) are randomly assigned to “forward” and “reverse” groups, in which samples are reciprocally labeled, as denoted by green and red circles. Following hybridisation and scanning, Cy3 and Cy5 intensity values are swapped in “reversed” samples to restore data polarity.

5.3 RNA quality assessment

All participant RNA samples obtained during the intervention trial were examined to insure they were of sufficient quality for use in the array experiments. Traditionally, RNA quality has been assessed by running samples on acrylamide gels. The RNA is visualised under UV exposure as a smear with ribosomal RNA coalescing in distinct bands. Recently, automated electrophoresis systems have been developed which

132 Chapter 5

minimise exposure to harmful reagents and have sped up preparation and run times when compared to traditional gels. Their major benefit is the small amount of RNA material required to perform the electrophoresis (~0.4 µg). Gels were used during initial testing of the RNA isolation protocol and the automated system was used for all experimental samples. Examples of spectrophotometer and Experion output are detailed in Figure 5.2.

A B Participant 6 (avoidance)

260/280 = 1.92 260/230 = 2.13

Figure 5.2 Assessment of RNA quality - output of Nanodrop spectrophotometer and Bio-Rad Experion systems. Panel A: Typical RNA chromatogram of participant total RNA produced by the Nanodrop spectrophotometer. Primarily, this assay was conducted to determine the amount of sample obtained after purification. 260/280 nm ratios of arrayed samples ranged from 1.90 - 1.94. This was deemed acceptable as further purification took place during the labeling process. 260/230 nm ratios were all above two, indicating no phenol contamination (Bowtell and Sambrook, 2003). Panel B: Representative output from the Experion electrophoresis system. From each participant sample assayed, chromatograms were produced depicting the relative abundance of RNA species. This was then interpreted as an electronic “virtual” electrophoretogram by the Experion software. Ten participant samples and an artificial RNA ladder were run simultaneously. The sample used in panel A of this figure is shown in lane 3. All lanes indicate good quality RNA with the exception of lane 5, which is an example of a degraded sample. Six samples (three each of C/C and T/T) of the 62 C/C and T/T samples assessed were deemed to be degraded and were excluded from the array experiment.

5.4 Selection of participant samples for microarray analysis

As stated in section 3.2.4, all participants were included for folate, homocysteine or global DNA methylation analysis. However, a subset of 30 participants was selected for the microarray analysis. Individuals heterozygous for the MTHFR C677T

Microarray Analysis 133

polymorphism were excluded from the analysis; wildtype and homozygotes were compared alone in an attempt to maximise differences in gene expression as a function of the polymorphism.

Some individuals were excluded on the basis of suboptimal RNA quality, as mentioned in the previous section. Pill compliance for remaining participant inclusion was set at greater than 80%. Additionally some participants had failed to display a significant decrease in their plasma folate during the folic acid avoidance period; dietary questionnaires suggested these participants persistently consumed foods high in folic acid. These individuals were also excluded. In all, around half of the study participants not heterozygous for MTHFR C677T were included in the array analysis. A breakdown of participant details is given in Table 5.1.

Table 5.1 Arrayed participant statistics

Array analysis participants All homozygous participants C/C T/T C/C T/T Sex distribution (m/f) 6/9 6/9 12/19 13/18 Average age (yr) 45 42 42 43 Mean depleted plasma folate (nmol/L) 10.7 8.4 14.5 13.8 Mean change in plasma folate (nmol/L) 31.5 24.9 27.8 25.1 Mean compliance (%)* 91.8 93.7 86.5 91.5

* Compliance is defined as the number of capsules consumed per maximum allowance across both phases of the study, expressed as a percentage. Mean compliance is the average compliance of each subset of participants.

5.5 Data filtering, normalisation and collation

Raw data obtained from DNA microarray experiments are invariably subjected to a pre- processing or filtering and normalisation procedure prior to the analysis proper. This is to insure that poor quality spots are excluded (filtering) and systematic bias in the hybridisation procedure is compensated for (normalisation) prior to downstream analysis (Quackenbush, 2002).

134 Chapter 5

Following hybridisation, slides were scanned as described in section 2.10.6 and the resulting images were assessed by GenePix Pro software to identify spots and calculate raw spot intensity. Following automated spot detection by the software, arrays were manually inspected for malformed or misaligned spots improperly detected by the identification algorithm. Malformed spots were flagged for exclusion from subsequent analysis and misaligned spots were corrected.

The MIDAS software was also used for log2 transformations of the data to make “up” and “down” regulated gene ratios symmetrical, in addition to low threshold filtering and normalisation. A low intensity filter was applied to the data, where spots less than twice the intensity of each array’s median background intensity value were excluded. Two types of normalisation were implemented. The first, total intensity normalisation, was required to correct for unequal quantities of starting RNA, or any differences in labeling or detection efficiencies between the fluorescent dyes used, so that the sum of intensities for each channel is equalled (Quackenbush, 2002). This allowed for more accurate comparisons to be made between slides. Secondly, microarrays have a tendency to display intensity-dependent effects, a skewing of an element’s ratio as intensity decreases. Locally weighted linear regression, or lowess (Yang et al., 2002), was used to remove these effects, and was applied block by block. Following the removal of low- quality genes by this filtering process, 17792 elements remained in the dataset and were used for subsequent analysis.

5.5.1 Accession number streamlining using DAVID and SOURCE

The global annotation file supplied with the array oligos contained a significant proportion of incomplete data; for example, multiple accession number types, accession numbers pertaining to contigs and genes of unknown name or function. Data ID tags were brought up to date and streamlined to include Unigene names and symbols with the help of the DAVID gene ID conversion tool (Huang da et al., 2008) and the SOURCE web application (Diehn et al., 2003) and the data were then exported to the MeV multi experiment viewer application. Where gene subsets were required for analysis, relevant filtering and ontology tags were applied using Microsoft Excel 2007 prior to exporting.

Microarray Analysis 135

5.6 Expression analysis of genes implicated in folate metabolism and DNA methylation

As already alluded to, the first part of microarray data analysis sought to establish which genes associated with core areas of interest, principally folate metabolism, were differentially expressed (DE) between avoidance and repletion phases, and between MTHFR genotypes. As well as giving an indication of how these genes were modulated throughout the study in response to dietary folate intervention, cataloguing DE genes could potentially point to promoter elements that may be under methylation control. A wider analysis of the complete gene set to detect DE genes regardless of their function (not necessarily involved in folate transport and metabolism and biosynthetic methylation pathways) constituted the second half of the work and is described in section 5.7.

The schema for the first section was as follows. Lists of genes relevant to folate metabolism and related fields were compiled from various sources. Significant genes from the array experiments, as determined by SAM analysis (Significance Analysis of Microarray), were then queried against these lists to identify significantly differentially expressed genes involved in these aforementioned areas. This procedure is described in greater detail below.

5.6.1 Generation of folate- and DNA methylation-centric gene lists

The first step in this process was to produce lists of genes involved in folate transport and metabolism and associated areas such as choline and homocysteine pathways, DNA methylation and wider one-carbon metabolism. Next, these lists would be queried against the array data to establish any modulated expression of these genes. Pertinent genes were gleaned from multiple sources: relevant literature, the curated NCBI Entrez Gene database, and by using the PubMatrix literature mining tool (http://pubmatrix.grc.nia.nih.gov/). The latter is a web-based application for assessing the relevance of genes in user-designated categories (Becker et al., 2003). Gene lists and terms of interest, for example diseases or gene ontologies, are queried against the PubMed literature database of the National Centre for Biotechnology Information

136 Chapter 5

(NCBI). The software returns a frequency matrix of co-occurrence between all comparison pairs. The number of papers associated with a particular gene provides an indicator of the degree of relevance of that gene to the papers’ common topic. Genes relevant to those topics of interest can then be investigated further.

In the context of this project, PubMatrix was used to compile a list of genes of potential interest or significance by submitting Unigene entries together with the same keywords as used in the Entrez Gene database searches – terms relevant to folate and folate metabolism, one-carbon metabolism and DNA methylation. These were: AdoMet, AdoHcy, B group vitamin, choline, CpG, cobalamin, DHFR (dihydrofolate reductase), DNA methylation, epigenetics, folate, folate transport, folic acid, homocysteine, methylcytosine, MS (methionine synthase), MTHFR (methylenetetrahydrofolate reductase), neural tube defects, one carbon metabolism, purine biosynthesis, pyroxidine, RFC (reduced folate carrier), riboflavin, single carbon metabolism, SAH (S- adenosylhomocysteine), SAM (S-adenosylmethionine), SHM (serine hydroxymethyltransferase) and vitamin B. Genes to be included in the final list had to exceed a threshold of two papers.

Entrez Gene queries apportioned 261 genes to “DNA methylation” and sub categories, 60 to “one-carbon metabolism” and 76 to “folate metabolism” and associated categories, with many genes being present in more than one category.

The final list of folate- and DNA methylation-centric genes was then assembled from all sources. Overlap between sources was present; all duplicates were removed. Three genes – ATP13A2, MTFMT and NUDT8 – were not present on the array and were also excluded. A total of 302 genes were included in the list. It should be noted that many of these genes related only tangentially to core areas of interest but were included to ensure downstream analysis remained comprehensive. The list of folate- and DNA methylation-related genes is given in Table 5.2.

Microarray Analysis 137

Table 5.2 Folate- and DNA methylation-centric gene list for microarray data inquiry

Symbol Gene Name NCBI Gene ID complex, transactivator adiponectin, C1Q and collagen COMT catechol-O-methyltransferase 1312 ADIPOQ 9370 domain containing CRABP2 cellular retinoic acid binding protein 2 1382 ADORA2A adenosine A2a receptor 135 CSNK2A1 casein kinase 2, alpha 1 polypeptide 1457 AFF2 AF4/FMR2 family, member 2 2334 CCCTC-binding factor (zinc finger CTCF 10664 AHCY adenosylhomocysteinase 191 protein) v-akt murine thymoma viral CCCTC-binding factor (zinc finger AKT1 207 CTCFL 140690 oncogene homolog 1 protein)-like activated leukocyte cell adhesion cubilin (intrinsic factor-cobalamin ALCAM 214 CUBN 8029 molecule receptor) aldehyde dehydrogenase 1 family, cytochrome P450, family 19, ALDH1L1 10840 CYP19A1 1588 member L1 subfamily A, polypeptide 1 aldehyde dehydrogenase 1 family, cytochrome P450, family 1, subfamily ALDH1L2 160428 CYP1A1 1543 member L2 A, polypeptide 1 anaplastic lymphoma receptor cytochrome P450, family 1, subfamily ALK 238 CYP1A2 1544 tyrosine kinase A, polypeptide 2 ALOX15 arachidonate 15-lipoxygenase 246 cytochrome P450, family 24, CYP24A1 1591 ALPI alkaline phosphatase, intestinal 248 subfamily A, polypeptide 1 alkaline phosphatase, cytochrome P450, family 27, ALPL 249 CYP27B1 1594 liver/bone/kidney subfamily B, polypeptide 1 ALPP alkaline phosphatase, placental 250 cytochrome P450, family 2, subfamily CYP2D6 1565 ALPPL2 alkaline phosphatase, placental-like 2 251 D, polypeptide 6 ALX4 ALX homeobox 4 60529 DAPK1 death-associated protein kinase 1 1612 AMT aminomethyltransferase 275 DAPK2 death-associated protein kinase 2 23604 APC adenomatous polyposis coli 324 DHFR dihydrofolate reductase 1719 APOE apolipoprotein E 348 dihydrofolate reductase pseudogene DHFRP1 573971 ankyrin repeat, SAM and basic leucine 1 ASZ1 136991 zipper domain containing 1 DNA methyltransferase 1 associated DMAP1 55929 activating transcription factor 7 protein 1 ATF7IP 55729 interacting protein DMGDH dimethylglycine dehydrogenase 29958 5-aminoimidazole-4-carboxamide DnaJ (Hsp40) homolog, subfamily C, ATIC 471 DNAJC15 29103 ribonucleotide formyltransferase member 15 alpha thalassemia/mental retardation DNA (cytosine-5-)-methyltransferase ATRX 546 DNMT1 1786 syndrome X-linked 1 AURKB aurora kinase B 9212 DNA (cytosine-5-)-methyltransferase DNMT3A 1788 AVP arginine vasopressin 551 3 alpha bromodomain adjacent to zinc finger DNA (cytosine-5-)-methyltransferase BAZ2A 11176 DNMT3B 1789 domain, 2A 3 beta breast cancer anti-estrogen DNA (cytosine-5-)-methyltransferase BCAR1 9564 DNMT3L 29947 resistance 1 3-like BCL2L2 BCL2-like 2 599 DYNLRB1 dynein, light chain, roadblock-type 1 83658 BCL6 B-cell CLL/lymphoma 6 604 euchromatic histone-lysine N- EHMT1 79813 BDNF brain-derived neurotrophic factor 627 methyltransferase 1 BECN1 beclin 1, autophagy related 8678 euchromatic histone-lysine N- EHMT2 10919 betaine--homocysteine S- methyltransferase 2 BHMT 635 methyltransferase EP300 E1A binding protein p300 2033 BIRC5 baculoviral IAP repeat containing 5 332 EPCAM epithelial cell adhesion molecule 4072 BMP6 bone morphogenetic protein 6 654 EPHX2 epoxide hydrolase 2, cytoplasmic 2053 BRCA1 breast cancer 1, early onset 672 ESR1 estrogen receptor 1 2099 capping protein (actin filament), FABP4 fatty acid binding protein 4, adipocyte 2167 CAPG 822 gelsolin-like Fas (TNF receptor superfamily, FAS 355 CASP1 caspase 1, apoptosis-related cysteine 834 member 6) CBS cystathionine-beta-synthase 875 FGFR2 fibroblast growth factor receptor 2 2263 CBX3 chromobox homolog 3 11335 FGFR3 fibroblast growth factor receptor 3 2261 CBX4 chromobox homolog 4 8535 FHIT fragile histidine triad gene 2272 CBX5 chromobox homolog 5 23468 FLI1 Friend leukemia virus integration 1 2313 CCL2 chemokine (C-C motif) ligand 2 6347 FMR1 fragile X mental retardation 1 2332 CDH13 cadherin 13, H-cadherin (heart) 1012 FOLH1 folate hydrolase 1 2346 CDK10 cyclin-dependent kinase 10 8558 FOLH1B folate hydrolase 1B 219595 cyclin-dependent kinase inhibitor 1A FOLR1 folate receptor 1 (adult) 2348 CDKN1A 1026 (p21) FOLR1P1 folate receptor 1 (adult) pseudogene 1 390221 cyclin-dependent kinase inhibitor 1B FOLR2 folate receptor 2 (fetal) 2350 CDKN1B 1027 (p27) FOLR3 folate receptor 3 (gamma) 2352 cyclin-dependent kinase inhibitor 1C folate receptor 3 (gamma) CDKN1C 1028 FOLR3P1 100288543 (p57) pseudogene 1 cyclin-dependent kinase inhibitor 2A FBJ murine osteosarcoma viral CDKN2A 1029 FOS 2353 (p16) oncogene homolog cyclin-dependent kinase inhibitor 2B FPGS folylpolyglutamate synthase 2356 CDKN2B 1030 (p15) fragile site, folic acid type, rare, FRA10AC1 118924 CDO1 cysteine dioxygenase, type I 1036 fra(10)(q23.3) candidate 1 CDX1 caudal type homeobox 1 1044 fragile site, folic acid type, rare, FRA12A 2448 CDX2 caudal type homeobox 2 1045 fra(12)(q13.1) CENPC1 centromere protein C 1 1060 FST follistatin 10468 cystic fibrosis transmembrane formiminotransferase CFTR 1080 FTCD 10841 conductance regulator cyclodeaminase chromodomain helicase DNA binding FTH1 ferritin, heavy polypeptide 1 2495 CHD5 26038 protein 5 fucosyltransferase 2 (secretor status FUT2 2524 CIITA class II, major histocompatibility 4261 included)

138 Chapter 5

FXN frataxin 2395 MAPK14 mitogen-activated protein kinase 14 1432 G0S2 G0/G1switch 2 50486 methionine adenosyltransferase I, MAT1A 4143 gamma-aminobutyric acid (GABA) A alpha GABRA1 2554 receptor, alpha 1 MBD2 methyl-CpG binding domain protein 2 8932 phosphoribosylglycinamide MBD3 methyl-CpG binding domain protein 3 53615 GART 2618 formyltransferase MBD4 methyl-CpG binding domain protein 4 8930 GATA zinc finger domain containing MBD5 methyl-CpG binding domain protein 5 55777 GATAD2A 54815 2A methyl CpG binding protein 2 (Rett MECP2 4204 GCH1 GTP cyclohydrolase 1 2643 syndrome) GDF10 growth differentiation factor 10 2662 O-6-methylguanine-DNA MGMT 4255 GGH gamma-glutamyl hydrolase 8836 methyltransferase glucosamine (UDP-N-acetyl)-2- MME membrane metallo-endopeptidase 4311 GNE 10020 epimerase MT3 metallothionein 3 4504 GNMT glycine N-methyltransferase 27232 MTAP methylthioadenosine phosphorylase 4507 GSN gelsolin 2934 methylenetetrahydrofolate MTHFD1 4522 GSTP1 glutathione S-transferase pi 1 2950 dehydrogenase (NADP+ dependent) 1 H19, imprinted maternally expressed methylenetetrahydrofolate H19 283120 transcript (non-protein coding) MTHFD1L dehydrogenase (NADP+ dependent) 1- 25902 HELLS helicase, lymphoid-specific 3070 like HemK methyltransferase family methylenetetrahydrofolate HEMK1 51409 MTHFD2 10797 member 1 dehydrogenase (NADP+ dependent) 2 HHIP hedgehog interacting protein 64399 methylenetetrahydrofolate HHIPL1 HHIP-like 1 84439 MTHFD2L dehydrogenase (NADP+ dependent) 2- 441024 HHIPL2 HHIP-like 2 79802 like hypoxia inducible factor 1, alpha methylenetetrahydrofolate reductase HIF1A 3091 MTHFR 4524 subunit (NAD(P)H) HINFP histone H4 transcription factor 25988 5,10-methenyltetrahydrofolate MTHFS 10588 HIST2H4A histone cluster 2, H4a 8370 synthetase major histocompatibility complex, 5-methyltetrahydrofolate- HLA-DRB1 3123 MTR 4548 class II, DR beta 1 homocysteine methyltransferase major histocompatibility complex, 5-methyltetrahydrofolate- HLA-DRB5 3127 class II, DR beta 5 MTRR homocysteine methyltransferase 4552 HLTF helicase-like transcription factor 6596 reductase HNF1A HNF1 homeobox A 6927 MTSS1 metastasis suppressor 1 9788 HNF1B HNF1 homeobox B 6928 mucin 2, oligomeric mucus/gel- MUC2 4583 heterogeneous nuclear forming HNRNPA2B1 3181 ribonucleoprotein A2/B1 v-myc myelocytomatosis viral MYC 4609 heterogeneous nuclear oncogene homolog HNRNPD 3184 ribonucleoprotein D MYST2 MYST histone acetyltransferase 2 11143 HOXA1 homeobox A1 3198 NANOG Nanog homeobox 79923 HOXA4 homeobox A4 3201 N-acetyltransferase 1 (arylamine N- NAT1 9 hydroxysteroid (11-beta) acetyltransferase) HSD11B2 3291 dehydrogenase 2 neuroblastoma breakpoint family, NBPF3 84224 HSPB8 heat shock 22kDa protein 8 26353 member 3 HTT huntingtin 3064 NDRG1 N-myc downstream regulated 1 10397 hydatidiform mole associated and nuclear factor of activated T-cells, HYMAI 57061 NFATC1 4772 imprinted (non-protein coding) cytoplasmic, calcineurin-dependent 1 IFNG interferon, gamma 3458 NKX2-1 NK2 homeobox 1 7080 insulin-like growth factor 2 NKX2-2 NK2 homeobox 2 4821 IGF2 3481 (somatomedin A) NOS2 nitric oxide synthase 2, inducible 4843 insulin-like growth factor binding NOS3 nitric oxide synthase 3 (endothelial) 4846 IGFBP7 3490 protein 7 NPPA natriuretic peptide A 4878 IL10 interleukin 10 3586 nuclear receptor subfamily 5, group NR5A1 2516 IL13 interleukin 13 3596 A, member 1 IL1A interleukin 1, alpha 3552 NRF1 nuclear respiratory factor 1 4899 IL1B interleukin 1, beta 3553 opioid binding protein/cell adhesion OPCML 4978 IL27 interleukin 27 246778 molecule-like IL4 interleukin 4 3565 OPRM1 opioid receptor, mu 1 4988 IL7R interleukin 7 receptor 3575 OSMR oncostatin M receptor 9180 IL8 interleukin 8 3576 oxoglutarate (alpha-ketoglutarate) OXGR1 27199 IRF8 interferon regulatory factor 8 3394 receptor 1 ITGA4 integrin, alpha 4 3676 PAX3 paired box 3 5077 ITGAL integrin, alpha L 3683 platelet-derived growth factor beta PDGFB 5155 JUNB jun B proto-oncogene 3726 polypeptide potassium voltage-gated channel, PDLIM4 PDZ and LIM domain 4 8572 KCNA3 3738 shaker-related subfamily, member 3 phosphatidylethanolamine N- PEMT 10400 KCNQ1 overlapping transcript 1 (non- methyltransferase KCNQ1OT1 10984 protein coding) PICK1 protein interacting with PRKCA 1 9463 KDM1B lysine (K)-specific demethylase 1B 221656 phosphoinositide-3-kinase, catalytic, PIK3CG 5294 KLK10 kallikrein-related peptidase 10 5655 gamma polypeptide LCMT2 leucine carboxyl methyltransferase 2 9836 PITX2 paired-like homeodomain 2 5308 LCN2 lipocalin 2 3934 prostate transmembrane protein, PMEPA1 56937 LEP leptin 3952 androgen induced 1 luteinizing PODXL podocalyxin-like 5420 LHCGR hormone/choriogonadotropin 3973 POMC proopiomelanocortin 5443 receptor POU5F1 POU class 5 homeobox 1 5460 LPL lipoprotein lipase 4023 peroxisome proliferator-activated PPARA 5465 melanoma antigen family A, 1 (directs receptor alpha MAGEA1 4100 expression of antigen MZ2-E) peroxisome proliferator-activated PPARG 5468 MAOB monoamine oxidase B 4129 receptor gamma MAPK1 mitogen-activated protein kinase 1 5594 PPP1R13B protein phosphatase 1, regulatory 23368

Microarray Analysis 139

(inhibitor) subunit 13B structural maintenance of preferentially expressed antigen in SMCHD1 chromosomes flexible hinge domain 23347 PRAME 23532 melanoma containing 1 PRB2 proline-rich protein BstNI subfamily 2 653247 synuclein, alpha (non A4 component SNCA 6622 PR domain containing 2, with ZNF of amyloid precursor) PRDM2 7799 domain SOCS3 suppressor of cytokine signaling 3 9021 PRDM5 PR domain containing 5 11107 SOX10 SRY (sex determining region Y)-box 10 6663 PRMT5 protein arginine methyltransferase 5 10419 SP1 Sp1 transcription factor 6667 PRMT7 protein arginine methyltransferase 7 54496 sepiapterin reductase (7,8- PROM1 prominin 1 8842 SPR dihydrobiopterinNADP+ 6697 PRSS50 protease, serine, 50 29122 oxidoreductase) protein tyrosine phosphatase, non- signal transducer and activator of PTPN6 5777 STAT6 6778 receptor type 6 transcription 6, interleukin-4 induced PTS 6-pyruvoyltetrahydropterin synthase 5805 STX11 syntaxin 11 8676 PWWP2A PWWP domain containing 2A 114825 sulfotransferase family, cytosolic, 1A, SULT1A1 6817 PWWP2B PWWP domain containing 2B 170394 phenol-preferring, member 1 PYCARD PYD and CARD domain containing 29108 transcobalamin I (vitamin B12 binding TCN1 6947 QDPR quinoid dihydropteridine reductase 5860 protein, R binder family) RAB6C RAB6C, member RAS oncogene family 84084 TCN2 transcobalamin II 6948 RAG2 recombination activating gene 2 5897 TDRD1 tudor domain containing 1 56165 RARB retinoic acid receptor, beta 5915 TDRD9 tudor domain containing 9 122402 retinoic acid receptor responder TERT telomerase reverse transcriptase 7015 RARRES1 5918 (tazarotene induced) 1 TET1 tet oncogene 1 80312 Ras protein-specific guanine TET2 tet oncogene family member 2 54790 RASGRF1 5923 nucleotide-releasing factor 1 TET3 tet oncogene family member 3 200424 Ras association (RalGDS/AF-6) TFAP2A transcription factor AP-2 alpha 7020 RASSF1 11186 domain family member 1 TFAP2B transcription factor AP-2 beta 7021 RELN reelin 5649 TFAP2C transcription factor AP-2 gamma 7022 regulatory factor X, 1 (influences HLA TFPI2 tissue factor pathway inhibitor 2 7980 RFX1 5989 class II expression) THBS1 thrombospondin 1 7057 RTBDN retbindin 83546 TIMP1 TIMP metallopeptidase inhibitor 1 7076 RUNX3 runt-related transcription factor 3 864 TMEM8B transmembrane protein 8B 51754 SEPT9 septin 9 10801 TNF tumor necrosis factor 7124 serpin peptidase inhibitor, clade B TP53 tumor protein p53 7157 SERPINB5 5268 (ovalbumin), member 5 TP53BP2 tumor protein p53 binding protein, 2 7159 SET SET nuclear oncogene 6418 TP73 tumor protein p73 7161 SFN stratifin 2810 TRDMT1 tRNA aspartic acid methyltransferase 1 1787 serine hydroxymethyltransferase 1 TYMS thymidylate synthetase 7298 SHMT1 6470 (soluble) UBE3A ubiquitin protein ligase E3A 7337 serine hydroxymethyltransferase 2 UDP glucuronosyltransferase 1 family, SHMT2 6472 UGT1A1 54658 (mitochondrial) polypeptide A1 solute carrier family 19 (folate ubiquitin-like with PHD and ring SLC19A1 6573 UHRF1 29128 transporter), member 1 finger domains 1 solute carrier family 19 (thiamine USF1 upstream transcription factor 1 7391 SLC19A2 10560 transporter), member 2 upstream transcription factor 2, c-fos USF2 7392 SLC19A3 solute carrier family 19, member 3 80704 interacting solute carrier family 22 (organic anion vitamin D (1,25- dihydroxyvitamin D3) SLC22A8 9376 VDR 7421 transporter), member 8 receptor SLC25A32 solute carrier family 25, member 32 81034 VHL von Hippel-Lindau tumor suppressor 7428 solute carrier family 46 (folate Williams Beuren syndrome SLC46A1 113235 WBSCR22 114049 transporter), member 1 chromosome region 22 solute carrier family 5 (iodide WW domain containing SLC5A8 160728 WWOX 51741 transporter), member 8 oxidoreductase solute carrier family 8 (sodium/calcium YY1 YY1 transcription factor 7528 SLC8A2 6543 exchanger), member 2 zinc finger and BTB domain ZBTB33 10009 solute carrier organic anion containing 33 SLCO1B3 28234 transporter family, member 1B3 ZEB1 zinc finger E-box binding homeobox 1 6935

5.6.2 Analysis of folate-centric genes using t-tests, SAM and clustering

To gauge the degree of differential expression in the microarray data, a variety of methods were considered. Historically, significance inference initially consisted of ranking differential expression gene by gene and then assigning an arbitrary cut-off, for example a two-fold decrease or increase. More recently a “significance of differential expression” statistic, such as Student’s t-test, generated from replicate arrays has been used to better cope with biological variation and reduce type I and II error (Nadon and Shoemaker, 2002). To further combat inclusion of false positives, false-discovery

140 Chapter 5

correction algorithms are often applied to array t-tests, an example of which is the Bonferroni correction (Long et al., 2001). However, the stringency of these tests is often severe and excludes legitimate DE genes, decreasing the power of the test (Chuaqui et al., 2002). Therefore, it is preferable to produce a larger set of “significant” genes at the risk of allowing a few false-positives to be included in the set. A method for assigning differential expression cut-offs that is routinely used is the Significance Analysis of Microarray, or SAM analysis (Tusher et al., 2001). Cut-offs can be set according to the degree of false positives predicted in the generated gene set. As stringency is increased by the experimenter, the gene set decreases as well as the corresponding false discovery rate (FDR), allowing the system to be tuned to a desired level of accuracy versus gene set size.

Comprehensive SAM analysis using TM4’s MEV software (TMEV) was performed. Additional t-tests were calculated to confirm gene significance and to generate p-values. Raw data was first filtered to include only genes with 70% present calls or more (genes with missing spot intensities on more than 30% of the arrays were excluded from analysis). Alpha for t-tests was set at 0.05 for most enquiries.

SAM analysis was performed upon the genes identified as being involved in folate metabolism or DNA methylation across all 30 microarray samples to establish differential expression between folate avoidance and repletion periods. Additionally, array data from C/C and T/T MTHFR genotypes were run both together and independently to gauge differences in differential gene expression between the two genotype groups. Genes significantly differentially expressed in one set but not the other, or differentially expressed in opposite directions in each set met the definition of differentiation by MTHFR genotype. Same-direction DE genes present in both groups were discounted. Between-genotype analyses were also run a second time with a relaxed FDR to identify genes falling just outside the FDR cut off, to ensure significant genes were grouped correctly. Finally, SAM analyses were run that compared participants above and below the median plasma folate value following folic acid avoidance. Median false discovery rates used were up to 20%. Significant genes were clustered hierarchically for the generation of heat maps, optimising gene order and sample order where applicable.

Microarray Analysis 141

5.6.3 Statistical analysis outcomes

Results from the SAM analysis of the folate- and DNA methylation-centric dataset are shown in Table 5.3. Immediately apparent was the small number of genes identified as significantly modulated. Additionally, genes that were deemed differentially expressed showed low mean fold expression changes between avoidance and repletion phases, as a function of folate status following folic acid avoidance or between MTHFR genotypes. It should be noted that t-tests were run both with and without Bonferroni correction; however the inclusion of the correction algorithm yielded no DE genes. The genes identified by SAM and listed in Table 5.3 were all identified using t-tests when error correction was not enabled. Notably, the majority of genes whose products form the core machinery of folate metabolism and DNA methylation varied insignificantly in their expression throughout the intervention.

Together, these initial data suggested that the dietary regimen had little impact on the gene expression of folate-related machinery. Additionally, very few uniform changes in expression were witnessed for genes relating to methionine metabolism or known to be under methylation control. Despite this, the characteristics of the genes identified by SAM as being significantly differentially expressed are described below.

5.6.4 Biological relevance of identified DE genes

A total of 13 genes of the folate and DNA methylation-centric dataset were determined to be differentially expressed by SAM, when generous false discovery rates were implemented. Those immediately involved in the uptake and metabolism of folate were folyl-polyglutatmate carboxypeptidase, also termed folate hydrolase (FOLH1) and the reduced folate carrier (RFC, encoded by the SLC19A1 gene). The former is an extracellular membrane-bound enzyme and is responsible for deconjugating folate’s glutamate tail to facilitate cellular uptake of the vitamin. The latter transports the resultant folate monoglutamate across the cell membrane (see section 1.1.4).

142 Chapter 5

Table 5.3 Statistically significant differentially-expressed genes as determined by Significance Analysis of Microarray

Whole group analysis (n = 30) Gene symbol Gene name p-value** Fold change* Greater expression during avoidance

IL13 Interleukin 13 – immune response 0.004 1.078 CDKN1B Cyclin-dependent kinase inhibitor 1 B 0.010 1.061 IL7R Interleukin 7 receptor – immune response 0.031 1.105 MBD2 Methyl-CpG-binding domain protein 2 0.031 1.067 Greater expression during repletion SERPINB5 Serine peptidase inhibitor, clade B, member 5 0.017 1.047 LCMT2 Leucine carboxyl methyltransferase 2 0.035 1.065 THBS1 Thrombospondin 1 0.031 1.020 By MTHFR genotype (T/T = 15, C/C = 15) Gene symbol Gene name p-value Fold difference C/C up-regulated during avoidance, T/T down-regulated during avoidance DNMT3b DNA methyltransferase 3 beta < 0.001 1.272 Solute carrier family 19 member 1 < 0.001 1.175 SLC19A1 (folate transporter) C/C up-regulated during avoidance, T/T no significant change - - - - C/C no significant change, T/T down-regulated during avoidance LCMT2 Leucine carboxyl methyltransferase 2 0.029 (T/T) 1.042 C/C no significant change, T/T up-regulated during avoidance ALDH1L1 Aldehyde dehydrogenase 1 family, member L1 0.003 (T/T) 1.209 ITGAL Integrin, alpha L (antigen CD11A, p180) < 0.001 (T/T) 1.195 C/C down-regulated during avoidance, T/T up-regulated during avoidance MTHFD1L Formyl-THF synthetase (monofunctional) 0.010 1.081 C/C down-regulated during avoidance, T/T no significant change - - - - By plasma folate status following folic acid avoidance (<10 nmol/L = 15, >10 nmol/L = 15) Gene symbol Gene name p-value Fold change Up-regulated during avoidance in low folate status individuals FOLH1 Folyl-polyglutamate carboxypeptidase 0.009 1.103

*Fold change is expressed linearly. For the calculation of fold change, antilogs of mean log ratios have been used. **Although differential expression significance was assigned by SAM, p-values, as calculated by Student’s t-test, have been added to further clarify the relevance of expression change. The SAM false discovery rate was set at 20%.

Microarray Analysis 143

Folate hydrolase showed increased expression in individuals with a low folate status (< 10 nmol/L, the median level of plasma folate in assayed participants) following the folic acid avoidance phase, with no significant change in expression over the duration of the study in participants whose folate status remained above 10 nmol/L. Low folate status participants were predominantly homozygous for the MTHFR polymorphism (11 out of 15, 73%), despite the gene not being found as significantly differentially expressed in analyses comparing C/C and T/T individuals. While the transcriptional up- regulation was slight, this appeared biologically plausible due to FOLH1’s role in cellular folate uptake and that individuals with a low folate status would require increased folate uptake. Indeed, such expression changes due to dietary folate deficiency have been observed by others, although not in peripheral blood tissue (Said et al., 2000).

Expression of the reduced folate carrier (encoded by SLC19A1) is known to parallel that of folate hydrolase in response to low folate status (Said et al., 2000; Suh et al., 2001). Correspondingly, wildtype individuals demonstrated increased transcription following the folic acid avoidance phase of the trial. Given that the majority of participants possessing the T/T MTHFR genotype fell into the low folate status bracket, it would be logical to surmise that RFC expression would be increased in T/T individuals during the avoidance period. In fact the opposite was true, although again, the changes in expression were minimal (a mean fold difference of 1.07 for T/T subjects between avoidance and repletion).

Regulation of SLC19A1 is known to be complex: three TATA-less promoters with binding sites for multiple transcription factors control regulation of transcription and variable splicing and alternative non-coding exons have been reported (Matherly and Goldman, 2003; Whetstine et al., 2002). Exact feedback mechanisms are unknown, but input from particular intracellular THF derivatives may explain the observed direction of expression change. Differences in relative amounts of intracellular folate derivatives have been reported between MTHFR C677T genotypes; T/T individuals display diminished 5-methyl-THF and an antipodal excess of other forms (Bagley and Selhub, 1998). Thus it might be construed that intracellular levels of THF forms other than

144 Chapter 5

5-methyl-THF, that would be present in greater proportions in T/Ts, may be responsible for suppressing RFC expression relative to wildtype.

Of the enzymes involved in tetrahydrofolate interconversion, 10-formyl-THF dehydrogenase, encoded by the ALDH1L1 gene and responsible for converting 10-formyl-THF to THF, and the monofunctional, mitochondrial form of formyltetrahydrofolate synthetase (MTHFD1L), which performs the reciprocal reaction of THF to 10-formyl-THF, were identified as being upregulated in T/T participants under folic acid avoidance conditions. Up-regulation of ALDH1L1 may again have been a reflection of these individuals’ excess formylated folates being shuttled towards attempted methyl-THF conversion to ameliorate a deficiency of the latter form. The increased expression of MTHFD1L may have been to mitigate this equilibrium shift, preserving a pool of 10-formyl-THF for fMet and purine synthesis.

Further downstream in the one-carbon network, the de novo DNA methyltransferase enzyme DNMT3b was identified as having a marginally increased expression in T/T participants following folic acid supplementation. Additionally, the putative DNA demethylase, methyl DNA-binding domain protein 2 (MBD2) exhibited an increase in expression across all participants following dietary folic acid avoidance. These results tentatively indicated that methylation remodelling of DNA may have occurred during the intervention trial. To this end, any potential correlation between global DNA methylation (described in the previous chapter) and the expression of these two genes was investigated. For MBD2, global methylation was compared between pre- supplementation and post-supplementation periods across all participants. No significant statistical difference was observed (p = 0.37). For DNMT3b, global methylation was compared between pre-supplementation and post-supplementation periods for T/T participants only. Again no significant difference was found (p = 0.44). Additionally, no correlation was found when comparing individuals’ magnitude of change in gene expression against the magnitude of total DNA methylation change for either DNMT3b or MBD2 (R2 < 0.051).

Unsurprisingly, many of the identified genes were known to be variably expressed by leukocytes. The integrin CD11a (encoded by ITGAL), interleukin IL13 and interleukin

Microarray Analysis 145

receptor IL7R were identified by SAM analysis. Although potentially interesting as methylation candidates given that they contained promoter CpG islands, the genes for these proteins were treated with caution due to their role in the human immune response and that they were expressed by cell types whose proportions in the blood are known to fluctuate considerably (Fan and Hegde, 2005). Unfortunately, no account was made of relative proportions of cell types or any variation in cell counts between avoidance and supplementation periods. Additionally, given that the first blood draws took place during the winter flu season, while the second draws were during summer, differences in expression due to changes in the proportions of WBC subtypes could not be ruled out. The platelet-derived THBS1, which coded for the thrombospondin 1 glycoprotein, responsible for platelet aggregation, was similarly excluded. Regardless, differential expression of these genes was again minimal.

The rationalization of other genes identified as differentially expressed by SAM was less forthright. Leucine carboxyl methyltransferase 2 (LCMT2), as its name implies, was identified by PubMatrix because it may methylate the carboxyl group of leucine residues to form alpha-leucine esters, not because its expression is regulated by methylation. However it is also thought to act as a G(1)/S and G(2)/M phase checkpoint regulator. Interestingly, although LCMT2’s differential expression was identified in the whole group analysis, this difference was mainly attributable to T/T individuals. Cyclin- dependent kinase inhibitor 1B (CDKN1B) also acts as a cell cycle regulator – its inhibitory activity was weakly upregulated during the dietary folic acid avoidance part of the trial. Whether these genes were modulated to restrict growth as a result of limited folate would be difficult to establish.

Lastly, the serpin peptidase inhibitor, clade B member 5 (SERPINB5) or maspin gene, is known to have its expression dictated in a cell type specific manner by DNA methylation of its promoter (Futscher et al., 2002), which warranted its inclusion in the truncated dataset. However, it is expressed only in epithelial tissue, particularly bronchial, thus the gene’s appearance in the current experiment’s significance table was surprising. Closer investigation revealed raw spot intensities were low and seven of the thirty arrays failed to register a present call. Given that the change in expression was only 1.047, its identification by SAM was most probably spurious.

146 Chapter 5

5.6.5 Summary of folate-related differential gene expression

While many of the genes identified offered a conceivable biological explanation for their inclusion, it should be emphasised that the degree of mean expression change in these genes was very small. Additionally, identification was only achieved with relaxed algorithm parameters – inclusion of error correction mechanisms meant no DE genes could be identified. Consequently, the veracity of these results was treated with a degree of caution. As the data subset contained just a few hundred genes, however, it was anticipated that extending SAM analysis to the entire dataset would identity a greater number of genes with an increased expression divergence.

5.7 Assessment of complete array dataset

The second half of the microarray analysis assessed the total array gene set. The aim of this work was to identify differentially expressed genes, not necessarily involved in folate metabolism, which may have had their expression altered by way of differential methylation of CpG elements in their regulatory regions. It was originally conceived that variations in the dietary intake of folate during the intervention study may have affected such promoter methylation. Given that aberrant DNA methylation is also associated with certain pathologies (see section 1.7), modulated expression of genes by way of altered promoter methylation could potentially highlight possible disease gene candidates.

It was initially envisaged that genes identified as differentially expressed by SAM would be analysed by software capable of ascribing ontologies to the genes, as well as mapping them to biological pathways. It was hoped that these techniques would enrich for genes with a propensity towards disease development and that such a gene shortlist could then be queried for the presence of regulatory CpG elements. Indeed, whole significant gene lists compiled using SAM were run through the ontology software application Expression Analysis Systematic Explorer (EASE), as well as DAVID, BIORAG pathway miner and KEGG pathway tools (Huang da et al., 2009; Kanehisa et al., 2006; Pandey et al., 2004). This was met with limited initial success; ontology filtering did not appreciably focus gene candidates while pathway applications did not

Microarray Analysis 147

assign any more than one gene to neighbouring steps in any pathways. Consequently, an alternative process of filtering was sought.

5.7.1 Utilisation of genome-wide profiling data

It was reasoned that a more direct approach to detecting genes with potentially altered promoter methylation could be to survey changes in expression of genes already known to possess such promoter methylation. While most CpG islands in gene regulatory regions are unmethylated, a subset of promoter islands are partially methylated or heavily methylated. As such, these methylated loci could undergo hypomethylation under low methyl group availability, a scenario potentially brought about under conditions of low dietary folate intake.

This approach required that a database of genes known to possess methylated CpG islands be available, in order to be compared against DE genes identified by SAM in the current study. Fortuitously, a comprehensive census of CpG island methylation in leukocytes was published around the time of analysis of the present data. Shen and colleagues described the profiling of over 5,000 CpG island-dense promoter regions of genes in peripheral blood (Shen et al., 2007). While the majority of CpG islands were found to be unmethylated, a distinct subset of genes possessed methylated promoter islands – around 400. Approximately 260, or 4% of genes with CpG islands, were densely methylated.

By querying the present study’s DE gene list to the methylated gene list of Shen et al., the complete DE gene set could be directly and effectively filtered to contain only elements capable of having their expression modulated by changes in promoter DNA methylation. Thus, it was decided that this filtered gene set would form the basis for subsequent analysis. Genes upregulated following the folic acid avoidance phase were the primary targets of the current analysis, as decreased promoter methylation resulting from the dietary regimen would supposedly manifest as increased gene expression. The list of genes described as possessing methylated CpG islands in peripheral blood by Shen et al. is listed below in table 5.4

148 Chapter 5

Table 5.4 Loci possessing putative methylated promoter CpG islands, as reported by Shen and colleagues, used to probe differentially expressed genes identified in current dataset

CpG island-dense loci* N1114 Gene DMRTA1 DMRT-like family A1 Name Symbol DMRTB1 DMRT-like family B with proline-rich C-terminal, 1 ATP-binding cassette, sub-family C (CFTR/MRP), EFNB1 Ephrin-B1 ABCC4 member 4 EIF2C4 Eukaryotic translation initiation factor 2C, 4 ABCD1 ATP-binding cassette, sub-family D (ALD), member 1 EML2 Echinoderm microtubule associated protein like 2 ABLIM2 Actin binding LIM protein family, member 2 EPB41L2 Erythrocyte membrane protein band 4.1-like 2 ADAM metallopeptidase with thrombospondin type 1 ESX1L ESX homeobox 1 ADAMTS17 V-ets erythroblastosis virus E26 oncogene homolog 1 motif, 17 ETS1 ADARB1 Adenosine deaminase, RNA-specific, B1 (avian) AGXT2L1 Alanine-glyoxylate aminotransferase 2-like 1 FADS1 Fatty acid desaturase 1 Aldo-keto reductase family 7, member A2 (aflatoxin FAM51A1 Gem (nuclear organelle) associated protein 8 AKR7A2 aldehyde reductase) FANK1 Fibronectin type III and ankyrin repeat domains 1 FERM, RhoGEF (ARHGEF) and pleckstrin domain protein ANKRD26P1 Ankyrin repeat domain 26 pseudogene 1 FARP1 ANKRD30A Ankyrin repeat domain 30A 1 (chondrocyte-derived) ANXA11 Annexin A11 FARP2 FERM, RhoGEF and pleckstrin domain protein 2 ARFRP1 ADP-ribosylation factor related protein 1 FBXL2 F-box and leucine-rich repeat protein 2 ARHGAP10 Rho GTPase activating protein 10 FGF13 Fibroblast growth factor 13 ARHGEF18 Rho/Rac guanine nucleotide exchange factor (GEF) 18 FGF22 Fibroblast growth factor 22 ATPase, aminophospholipid transporter, class I, type FLJ20449 Chromosome 13 open reading frame 38 ATP8A2 8A, member 2 FLJ22662 Phospholipase B domain containing 1 BAIAP2L1 BAI1-associated protein 2-like 1 FLJ25471 Peroxidasin homolog (Drosophila)-like BHLHB4 Basic helix-loop-helix family, member e23 FLJ31164 T-SNARE domain containing 1 BHMT2 Betaine--homocysteine S-methyltransferase 2 FLJ35773 Major facilitator superfamily domain containing 6-like BIN1 Bridging integrator 1 FLJ36445 chromosome 19 open reading frame 46 BIRC4 X-linked inhibitor of apoptosis FLJ39553 Chromosome 8 open reading frame 47 BLP2 TM2 domain containing 3 FLJ40201 Chromosome 1 open reading frame 177 BOK BCL2-related ovarian killer FLJ40773 Phospholipase D family, member 5 Nucleus accumbens associated 1, BEN and BTB (POZ) FLJ44968 YjeF N-terminal domain containing 3 BTBD14B domain containing FN3K Fructosamine 3 kinase C10orf93 Chromosome 10 open reading frame 93 FOXJ2 Forkhead box J2 C12orf12 Chromosome 12 open reading frame 12 FOXP4 Forkhead box P4 C13orf23 Chromosome 13 open reading frame 23 FTH1 Ferritin, heavy polypeptide 1 Pancreatic progenitor cell differentiation and FTMT Ferritin mitochondrial C20orf149 proliferation factor homolog (zebrafish) GABRQ Gamma-aminobutyric acid (GABA) receptor, theta UDP-N-acetyl-alpha-D-galactosamine:polypeptide N- C9orf3 Chromosome 9 open reading frame 3 GALNT11 CABP7 Calcium binding protein 7 acetylgalactosaminyltransferase 11 UDP-N-acetyl-alpha-D-galactosamine:polypeptide N- CALB2 Calbindin 2 GALNTL4 CALM3 Calmodulin 3 (phosphorylase kinase, delta) acetylgalactosaminyltransferase-like 4 CALML3 Calmodulin-like 3 GAS1 Growth arrest-specific 1 CAMK1D Calcium/calmodulin-dependent protein kinase ID GLS2 glutaminase 2 CBX3 Chromobox homolog 3 GLT8D1 Glycosyltransferase 8 domain containing 1 CDC34 Cell division cycle 34 homolog (S. cerevisiae) GMIP GEM interacting protein ArfGAP with coiled-coil, ankyrin repeat and PH domains N-acetylglucosamine-1-phosphate transferase, gamma CENTB2 GNPTG 2 subunit CMYA5 Cardiomyopathy associated 5 GPC3 Glypican 3 COMMD3 COMM domain containing 3 GPC4 Glypican 4 COPB2 Coatomer protein complex, subunit beta 2 (beta prime) GPR G protein-coupled receptor 176 COP9 constitutive photomorphogenic homolog subunit GPRC5C G protein-coupled receptor, family C, group 5, member C COPS6 6 (Arabidopsis) GPS1 G protein pathway suppressor 1 Cytochrome c oxidase subunit VIIa polypeptide 1 GRK5 G protein-coupled receptor kinase 5 COX7A1 (muscle) GRTP1 Growth hormone regulated TBC protein 1 CPXM2 Carboxypeptidase X (M14 family), member 2 GSDMDC1 Gasdermin D Cysteine rich transmembrane BMP regulator 1 HCCS Holocytochrome c synthase CRIM1 (chordin-like) HNRPAB Heterogeneous nuclear ribonucleoprotein A/B CSNK1G2 Casein kinase 1, gamma 2 HNRPC Heterogeneous nuclear ribonucleoprotein C (C1/C2) CSTB Cystatin B (stefin B) HOOK3 Hook homolog 3 (Drosophila) CTL2 Solute carrier family 44, member 2 HPS6 Hermansky-Pudlak syndrome 6 CXorf20 BEN domain containing 2 HRB ArfGAP with FG repeats 1 CXorf22 Chromosome X open reading frame 22 HS6ST1 Heparan sulfate 6-O-sulfotransferase 1 CXorf53 BRCA1/BRCA2-containing complex, subunit 3 HSCARG NmrA-like family domain containing 1 CYB561 Cytochrome b-561 HYPE FIC domain containing CYP46A1 Cytochrome P450, family 46, subfamily A, polypeptide 1 IGHMBP2 Immunoglobulin mu binding protein 2 DACH2 Dachshund homolog 2 (Drosophila) IL17R Interleukin 17 receptor A DALRD3 DALR anticodon binding domain containing 3 INSL6 Insulin-like 6 Potassium voltage-gated channel, Isk-related family, DDX4 DEAD (Asp-Glu-Ala-Asp) box polypeptide 4 KCNE4 DDX43- member 4 DEAD (Asp-Glu-Ala-Asp) box polypeptide 43 Potassium inwardly-rectifying channel, subfamily J, DPPA5 KCNJ14 DKFZP434B member 14 Tectonin beta-propeller repeat containing 1 Potassium inwardly-rectifying channel, subfamily J, 0335 KCNJ9 DKFZP434P member 9 Glutamine rich 2 Potassium voltage-gated channel, KQT-like subfamily, 0316 KCNQ1 DKFZp761A member 1 OTU domain containing 5 052 KIAA0998 Tubulin tyrosine ligase-like family, member 5 DKFZp761 Major facilitator superfamily domain containing 4 KIAA1344 Thioredoxin domain containing 16 KIAA1463 DIP2 disco-interacting protein 2 homolog B

Microarray Analysis 149

(Drosophila) (phosphoinositide binding specific) member 3 KRT18 Keratin 18 PODXL Podocalyxin-like L3MBTL L(3)mbt-like 1 (Drosophila) PPIF Peptidylprolyl isomerase F LATS2 LATS, large tumor suppressor, homolog 2 (Drosophila) PPM2C Pyruvate dehyrogenase phosphatase catalytic subunit 1 LIMK1 LIM domain kinase 1 Proteasome (prosome, macropain) 26S subunit, PSMC1 Lin10 Chromosome 16 open reading frame 70 ATPase, 1 LLGL1 Lethal giant larvae homolog 1 (Drosophila) PTCHD1 Patched domain containing 1 LMO6 Prickle homolog 3 (Drosophila) PTHR1 Parathyroid hormone 1 receptor LOC124491 Transmembrane protein 170A PTPN9 Protein tyrosine phosphatase, non-receptor type 9 Low density lipoprotein receptor class A domain PYGL Phosphorylase, glycogen, liver LOC143458 containing 3 RAB28 RAB28, member RAS oncogene family LOC158572 Hypothetical LOC158572 RAB34 RAB34, member RAS oncogene family LOC340602 Chromosome X open reading frame 67 RAET1L Retinoic acid early transcript 1L LOC554226 ankyrin repeat domain 30B-like RELB V-rel reticuloendotheliosis viral oncogene homolog B LOC92162 Transmembrane protein 88 RGN Regucalcin (senescence marker protein-30) LRRC36 Leucine rich repeat containing 36 RHPN2 Rhophilin, Rho GTPase binding protein 2 LTBP2 Latent transforming growth factor beta binding protein 2 RNF113B Ring finger protein 113B LY6K Lymphocyte antigen 6 complex, locus K RNF126 Ring finger protein 126 MAP3K10 Mitogen-activated protein kinase kinase kinase 10 RNF8 Ring finger protein 8 MAPK11 Mitogen-activated protein kinase 11 ROD1 ROD1 regulator of differentiation 1 (S. pombe) MAPK12 Mitogen-activated protein kinase 12 RP11- Kelch-like 34 (Drosophila) MATR3 Matrin 3 450P7.3 ME1 Malic enzyme 1, NADP(+)-dependent, cytosolic RP13- Transmembrane protein 164 MEST Mesoderm specific transcript homolog (mouse) 360B22.2 MGC15677 AlkB, alkylation repair homolog 6 (E. coli) RPL13A Ribosomal protein L13a MGC16635 Kelch domain containing 7B SAT Spermidine/spermine N1-acetyltransferase 1 MGC27016 RNA binding motif protein 46 SCNN1G Sodium channel, nonvoltage-gated 1, gamma MGC35140 Leucine rich repeat containing 43 SCRN1 Secernin 1 MGC39900 Thymosin beta 15B Ctr9, Paf1/RNA polymerase II complex component, SH2BP1 MGC40179 Chromosome 3 open reading frame 24 homolog (S. cerevisiae) MGC52010 Ribosomal protein S19 binding protein 1 SIVA SIVA1, apoptosis-inducing factor MGC52057 LY6/PLAUR domain containing 6 SLC39A14 Solute carrier family 39 (zinc transporter), member 14 MGC61550 Chromosome 15 open reading frame 38 SLC6A16 Solute carrier family 6, member 16 MGC75360 Zinc finger protein 774 SWI/SNF related, matrix associated, actin dependent SMARCA1 MICAL MICAL-like 1 regulator of chromatin, subfamily a, member 1 MIR193b MicroRNA 95 SMO Smoothened homolog (Drosophila) MIR373 MicroRNA 373 SOX3 SRY (sex determining region Y)-box 3 MIR95 MicroRNA 95 SPAG4 Sperm associated antigen 4 MIST1 Basic helix-loop-helix family, member a15 Sparc/osteonectin, cwcv and kazal-like domains SPOCK Mov10l1, Moloney leukemia virus 10-like 1, homolog proteoglycan (testican) 1 MOV10L1 (mouse) SPOP Speckle-type POZ protein MPST Mercaptopyruvate sulfurtransferase ST7 Suppression of tumorigenicity 7 MTM1 Myotubularin 1 STX8 Syntaxin 8 MYADML yeloid-associated differentiation marker-like SUMO2 SMT3 suppressor of mif two 3 homolog 2 (S. cerevisiae) V-myb myeloblastosis viral oncogene homolog (avian)- SURF5 mediator complex subunit 22 MYBL2 like 2 TESSP2 Protease, serine, 42 MYO18B Myosin XVIIIB TEX14 Testis expressed 14 NAG6 Coiled-coil domain containing 136 TIGD5 Tigger transposable element derived 5 NDN Necdin homolog (mouse) TMCC3 Transmembrane and coiled-coil domain family 3 NEBL Nebulette TMEM54 Transmembrane protein 54 Nuclear factor of activated T-cells, cytoplasmic, TMLHE Trimethyllysine hydroxylase, epsilon NFATC2IP calcineurin-dependent 2 interacting protein TNFRSF25 tumor necrosis factor receptor superfamily, member 25 NINJ1 Ninjurin 1 TNFSF7 CD70 molecule NIPSNAP3B Nipsnap homolog 3B (C. elegans) TSSK1 Testis-specific serine kinase 1B NKAP NFKB activating protein TTLL1 Tubulin tyrosine ligase-like family, member 1 NOG Noggin TUBGCP3 Tubulin, gamma complex associated protein 3 NOVA2 Neuro-oncological ventral antigen 2 TUSC3 Tumor suppressor candidate 3 NPAS3 Neuronal PAS domain protein 3 TYRO3 TYRO3 protein tyrosine kinase NPR2 natriuretic peptide receptor B/guanylate cyclase B Ubiquitin-conjugating enzyme E2I (UBC9 homolog, UBE2I NRK Nik related kinase yeast) NT5C 5', 3'-nucleotidase, cytosolic UCN Urocortin OPLAH 5-oxoprolinase (ATP-hydrolysing) UMPS Uridine monophosphate synthetase OSBPL2 Oxysterol binding protein-like 2 WDR45 WD repeat domain 45 PALM Paralemmin WDR5 WD repeat domain 5 PC Pyruvate carboxylase WNK3 WNK lysine deficient protein kinase 3 Pterin-4 alpha-carbinolamine XKRX XK, Kell blood group complex subunit-related, X-linked PCBD2 dehydratase/dimerization cofactor of hepatocyte ZBED4 Zinc finger, BED-type containing 4 nuclear factor 1 alpha (TCF1) 2 ZDHHC14 Zinc finger, DHHC-type containing 14 PDXK Pyridoxal (pyridoxine, vitamin B6) kinase ZEC Zinc finger protein 628 PFN2 Profilin 2 ZIC3 Zic family member 3 (odd-paired homolog, Drosophila) PGRMC1 Progesterone receptor membrane component 1 ZNF297 Zinc finger and BTB domain containing 22 PITPNA Phosphatidylinositol transfer protein, alpha ZNF545 Zinc finger protein 82 homolog (mouse) PIWIL3 Piwi-like 3 (Drosophila) ZNF788 zinc finger family member 788 PLAC2 Placenta-specific 2 (non-protein coding) ZNF84 Zinc finger protein 84 PLEKHA3 Pleckstrin homology domain containing, family A

CpG island-sparse loci** ARPM2 Actin-related protein T2 Gene ASGR2 Asialoglycoprotein receptor 2 Name Symbol AVPR2 Arginine vasopressin receptor 2 ABCA2 ATP-binding cassette, sub-family A (ABC1), member 2 C14orf173 Inverted formin, FH2 and WH2 domain containing

150 Chapter 5

C19orf21 Chromosome 19 open reading frame 21 KIAA0892 MAU2 chromatid cohesion factor homolog (C. elegans) C20orf166 Chromosome 20 open reading frame 166 KLK6 Kallikrein-related peptidase 6 C21orf121 Chromosome 21 open reading frame 121 KRT8 Keratin 8 C21orf125 Chromosome 21 open reading frame 125 LOC115098 Coiled-coil domain containing 124 C9orf132 Family with sequence similarity 102, member A LOC197336 WD repeat domain 90 Spermatogenesis and oogenesis specific basic helix- LOC389791 Hypothetical LOC389791 C9orf157 loop-helix 1 LOC440955 Transmembrane protein 89 CACNG1 Calcium channel, voltage-dependent, gamma subunit 1 LOC492304 Insulin-like growth factor 2 (somatomedin A) CCND3 Cyclin D3 MAGEA10 Melanoma antigen family A, 10 CEBPE CCAAT/enhancer binding protein (C/EBP), epsilon MAPK13 Mitogen-activated protein kinase 13 CLSPN Claspin MBP Myelin basic protein COL8A2 Collagen, type VIII, alpha 2 MGC13102 Transmembrane protein 79 CRAMP1L Crm, cramped-like (Drosophila) MGC17337 Chromosome 9 open reading frame 30 CRYAA Crystallin, alpha A MGC20983 Coiled-coil domain containing 151 CTAG2 Cancer/testis antigen 2 MIR202 MicroRNA 202 CTBP2 C-terminal binding protein 2 MIR208 MicroRNA 208 CYP2W1 Cytochrome P450, family 2, subfamily W, polypeptide 1 MIR26b MicroRNA 26b DCST2 DC-STAMP domain containing 2 MIR381 MicroRNA 381 DHRS2 Dehydrogenase/reductase (SDR family) member 2 MIR382 MicroRNA 382 DIDO1 Death inducer-obliterator 1 MIR99b MicroRNA 99b DKFZP434H1 MMP26 Matrix metallopeptidase 26 Chromosome 15 open reading frame 39 32 MPFL Mesothelin-like EBI3 Epstein-Barr virus induced 3 MSLN Mesothelin EEIG1 Family with sequence similarity 102, member A MYH6 Myosin, heavy chain 6, cardiac muscle, alpha ENO3 Enolase 3 (beta, muscle) NFKBIL2 Tonsoku-like, DNA repair protein EP400NL EP400 N-terminal like NME2 Non-metastatic cells 1, protein (NM23A) Coagulation factor VII (serum prothrombin conversion NTF3 Neurotrophin 3 F7 accelerator) OPTN Optineurin FAM9B Family with sequence similarity 9, member B OR2Z1 Olfactory receptor, family 2, subfamily Z, member 1 FAM9C Family with sequence similarity 9, member C P2RY11 Purinergic receptor P2Y, G-protein coupled, 11 FBN3 Fibrillin 3 PAPD1 Mitochondrial poly(A) polymerase FBS1 Fibrosin PASD1 PAS domain containing 1 FBXO17 F-box protein 17 PLA2G2E Phospholipase A2, group IIE Ficolin (collagen/fibrinogen domain containing) 3 PLEC1 Plectin FCN3 (Hakata antigen) Pleckstrin homology domain containing, family A PLEKHA9 FKBP10 FK506 binding protein 10, 65 kDa member 8 pseudogene 1 FKHL18 Forkhead box S1 PLXNB1 Plexin B1 FLJ14213 Proline rich 5 like PPAN Purinergic receptor P2Y, G-protein coupled, 11 FLJ31528 Chromosome 17 open reading frame 56 PRSSL1 Protease, serine-like 1 FLJ33814 Coiled-coil domain containing 117 PRTN3 Proteinase 3 FLJ38984 Chromosome 1 open reading frame 216 QPRT Quinolinate phosphoribosyltransferase FLJ46072 Family with sequence similarity 83, member H RASGEF1C RasGEF domain family, member 1C FN3KRP Fructosamine 3 kinase related protein RP1L1 Retinitis pigmentosa 1-like 1 FTCD Formiminotransferase cyclodeaminase RXRA Retinoid X receptor, alpha FURIN Furin (paired basic amino acid cleaving enzyme) SEPT6 Septin 6 FXYD1 FXYD domain containing ion transport regulator 1 SHANK1 SH3 and multiple ankyrin repeat domains 1 GAL3ST2 Galactose-3-O-sulfotransferase 2 SIRT4 Sirtuin 4 GFAP Glial fibrillary acidic protein SLC39A4 Solute carrier family 39 (zinc transporter), member 4 Glycosylphosphatidylinositol anchored molecule like SMOC1 SPARC related modular calcium binding 1 GML protein STAU Staufen, RNA binding protein, homolog 1 (Drosophila) GP9 Glycoprotein IX (platelet) TBC1D16 TBC1 domain family, member 16 GPR172B G protein-coupled receptor 172B Transglutaminase 1 (K polypeptide epidermal type I, TGM1 GR6 Chromosome 3 open reading frame 27 protein-glutamine-gamma-glutamyltransferase) GRIN1 Glutamate receptor, ionotropic, N-methyl D-aspartate 1 TMPRSS6 Transmembrane protease, serine 6 GRK1 G protein-coupled receptor kinase 1 TRAF7 TNF receptor-associated factor 7 GRK7 G protein-coupled receptor kinase 7 Transient receptor potential cation channel, subfamily TRPV1 HLA Major histocompatibility complex, class I, F V, member 1 HRAS V-Ha-ras Harvey rat sarcoma viral oncogene homolog TSPYL5 TSPY-like 5 HRH2 Histamine receptor H2 UNQ2541 Lipocalin 15 HSPC142 Chromosome 19 open reading frame 62 URP2 Fermitin family member 3 IFRG28 Receptor (chemosensory) transporter protein 4 VIL1 Villin 1 ISG20L1 Apoptosis enhancing nuclease VSIG2 V-set and immunoglobulin domain containing 2 JAKMIP3 Janus kinase and microtubule interacting protein 3 XAGE3 X antigen family, member 3 KCNK7 Potassium channel, subfamily K, member 7 YIF1B Yip1 interacting factor homolog B (S. cerevisiae) Potassium intermediate/small conductance calcium- ZNF541 Zinc finger protein 541 KCNN1 activated channel, subfamily N, member 1 ZNF646 Zinc finger protein 646 KIAA0664 KIAA0664

Genes were identified by Shen and colleagues as possessing methylated promoter CpG islands. *Loci were classed as CpG island-dense if at least 500 bp of sequence contained a GC content greater than 55% and a CpG ratio above 0.65. **CpG island-sparse entries were defined as possessing at least 200 bp of GC content over 50% and a 0.60 CpG ratio or higher. The CpG ratio was defined as: (number of CpGs × the number of bases analysed) ÷ (number of C × number of G).

Microarray Analysis 151

5.7.2 SAM analysis and t-tests of complete expression dataset

As with the folate-centric list analysis described in section 5.6, the 20 K set was queried using t-tests and SAM. The proportion of genes scored as showing significant change across all participants by t-tests approximated expectations: with an alpha of 0.01, 84 genes were considered upregulated and 89 genes down-regulated during folate avoidance relative to supplementation across all assayed participants (a total of 173 genes, or approximately 1%). Differential expression between C/C and T/T genotypes identified almost 500 significant genes at p = 0.01. However, when standard Bonferroni or modified, less stringent, Bonferroni corrections were applied, no significant genes could be identified, either between genotype, between avoidance and supplementation, or by folate status at either 0.01 or 0.05 levels of significance.

Results of SAM also conferred minimal differential expression; high false discovery rates (up to 20%) were required to identify enough DE genes to allow successful matches to the methylated CpG island gene list of Shen et al. At this FDR level, five genes were identified as possessing methylated CpG islands across all participants, from T/T participants only and from individuals with a plasma folate status less than 10 nmol/L. A breakdown of this process is depicted in Table 5.5 and a brief description of the functions of resultant genes are listed below.

5.7.3 Characteristics of identified genes

Attributes of the five genes identified at high false discovery rates were investigated using various online tools, including NCBI databases such as Gene and AceView (ncbi.nlm.nih.gov), the UCSC Genome Browser (genome.ucsc.edu), BioGPS gene portal (biogps.gnf.org), and literature searches. All genes were confirmed to possess CpG islands; single dense CpG islands spanning genes’ transcription start sites and′ 5 UTRs were identified. CpG island characteristics are reported in Table 5.6. All genes with the exception of the zinc transporter SLC39A14 are known to be highly expressed in peripheral blood, despite the methylated nature of their promoter regions.

152 Chapter 5

Table 5.5 Significant DE genes containing methylated CpG islands as identified by SAM

False Differentially DE genes upregulated DE genes containing Log mean ± SD** discovery expressed following folate methylated promoter (linear mean, p-value) rate* genes avoidance CpG islands All participants

0.13 ± 0.21 HNRPC (1.10, p = 0.002) < 20% FDR 96 91 0.21 ± 0.44 FLJ22662 (1.16, p = 0.024)

< 10% FDR 14 10 - - < 5% FDR - - < 2% FDR 9 9 - - T/T participants only 0.12 ± 0.22 < 20% FDR 724 173 BIN1 (1.09, p = 0.052) < 10% FDR 621 142 BIN1 (As above) < 5% FDR 241 66 - - < 2% FDR 61 0 - - Low plasma folate status (< 10 nmol/L)

0.18 ± 0.42 ETS1 (1.06, p = 0.006) < 20% FDR 176 115 0.03 ± 0.12 SLC39A14 (1.02, p = 0.005)

< 10% FDR 44 20 - - < 5% FDR 19 2 - - < 2% FDR 2 2 - -

* False discovery rate is defined as the calculated median number of falsely significant genes divided by the number of genes deemed significant. Therefore, as no FDRs were exact integers, delta values were chosen that approximated the above rates as closely as possible. ** For all-participant analyses, mean changes in expression between avoidance and supplementation periods for identified genes are expressed as log values ± standard deviation. Antilogs of means, expressed as fold change, are depicted in brackets, accompanied by p-values calculated by Student’s t-test. For significant genes identified only in T/T individuals, p-values are calculated from those individuals, and do not indicate the level of significance between T/T and C/C participants for that gene’s expression. P-values are equivalently calculated for participants whose plasma folate level was below 10 nmol/L following folic acid avoidance. No methylation candidates were flagged when data was queried by MTHFR genotype.

Microarray Analysis 153

Table 5.6 CpG island attributes* of genes identified by SAM

Genomic size (bp) GC content (%) CpG count CpG % BIN1 1646 66.5 146 17.7 ETS1 899 54.6 79 17.6 HNRPC 688 56.7 50 14.5 FLJ22662 845 56.9 65 15.4 SLC39A14 725 71.7 86 23.7

*CpG island features retrieved from UCSC Genome Browser

Heterogeneous nuclear ribonucleoprotein C, or HNRPC, and Protein C-ets-1, or ETS1, form part of the core transcriptional machinery of PBMCs: HNRPC is one of a family of ubiquitous nuclear proteins involved in mRNA binding, pre-processing (splicing) and transport and ETS1 is a transcription factor that is relatively highly expressed in T- lymphocytes, participating in their differentiation (Bhat et al., 1990). Both HNRPC and ETS1 possess complex promoters. The former is capable of producing 30 different mRNA transcripts across tissues, and has multiple splice variants and various truncations of the mRNA at both 5′ and 3′ ends, which affect translation. Similarly, ETS1 has five probable alternative promoters, produces 11 different mRNAs, has 9 validated polyadenylation sites and encodes 9 protein isoforms.

The small up-regulation of ETS1 during the avoidance phase of the trial in low-folate status participants, together with the previously identified ITGAL in T/T individuals, who represented the majority of subjects with low folate status, and the increased expression of other immunity-related genes suggested a possible change in leukocyte sub-type proportions resulting from a mounted immune response in some study participants during this period. With the complex promoter arrangement of the ETS1 locus and its multiple known interactions with other molecules, it was unlikely the small observed change in expression (1.06 fold) was due to changes in CpG island methylation.

The FLJ22662 locus produces transcripts of unknown function, although it contains a putative phospholipase B domain. Once more, transcription appears complex with

154 Chapter 5

8 mRNA variants expressed. Regulation has been suggested to rely partially upon overlapping antisense non-coding genes. It is not known to be implicated in any disease.

The SLC39A14 gene encodes a group of zinc transporters. Transcription produces 13 different mRNAs, depending on tissue type. There are four probable promoters. Given that the gene possesses this complex regulation independent of methylation, is minimally expressed in peripheral blood, that mean expression changes were very small (1.02 fold change), that the gene was only flagged as significant when false discovery rates were set above 20%, and that it is not expressly implicated in any disease, this locus was considered unworthy of further investigation.

Potentially the most interesting of all genes identified, ubiquitous expression of BIN1 encodes multiple tissue-specific isoforms of Bridging integrator 1, nuclear and cytoplasmic adaptor proteins that purportedly plays a role in numerous cellular functions including endocytosis and cellular differentiation and cell cycle regulation. In blood, it is moderately expressed in T-cells. Again, the locus is complex, with 7 possible alternate promoters leading to the production of 24 mRNA variants. Aberrant methylation of a part of the BIN1 CpG island in breast and prostate has been associated with up-regulation and cancer in those tissues (Kuznetsova et al., 2007). Interestingly, the authors noted no methylation of the island in normal lymphocytes, which conflicted with what was reported by Shen and colleagues. Additionally, this implied that hypomethylation in the island as a result of folate depletion could not have occurred if already unmethylated and thus hypomethylation was therefore not responsible for the increase in expression observed in T/T participants under conditions of dietary folic acid avoidance. Conversely, decreased expression potentially resulting from increased methylation during supplementation did not suggest that BIN1 was a marker of increased cancer risk in these participants.

5.7.4 Interpretation of expression data

As with the folate-centric data, identified loci in section 5.7.2 all exhibited marginal changes in expression between folic acid avoidance and folic acid supplementation periods. This was reflective of all genes identified by the wider SAM analysis. For

Microarray Analysis 155

whole-group analyses, for example, high false discovery rates (20%) yielded 96 significant genes. However, these genes were absent when the data was filtered by a linear mean fold change cut-off of 1.3. Maximum individual intensity ratios (antilogs) were no more than 2.6. No genes were identified when SAM false discovery rates were below 5%.

Most importantly with regard to the five genes identified after filtering against the methylated CpG island dataset , multiple oligo probes were recognized as being present on the arrays for many of these elements. Discouragingly, upon investigation duplicate probes for BIN1, ETS1 and HNRPC all failed to exhibit any significant differential expression.

Collectively, these observations indicated that uniform changes in expression across participants throughout the course of the study were negligible. Thus, the proposition that modulated methylation of CpG islands might cause differential gene expression as a result of the study’s dietary regimen was an unlikely scenario in the study cohort. As global DNA methylation analysis also indicated that the dietary regimen had no impact on DNA methylation, it was decided that further investigation into the promoter methylation status of the genes identified in Table 5.5 would be of little benefit. Accordingly, follow-up investigations upon shortlisted genes using RT-PCR were not pursued.

5.8 Discussion

The microarray analysis described in this chapter sought to ascertain changes in peripheral blood leukocyte gene expression as a result of changes in participants’ dietary intake of folic acid. Specifically, it attempted to identify modulated expression of genes involved in folate-mediated one carbon metabolism, as well as identify differentially-expressed genes, not related to wider one carbon metabolism but with the potential to have their expression changes dictated by CpG island methylation changes as a result of the dietary regimen.

156 Chapter 5

The analysis identified a small group of genes involved in folate metabolism and DNA methylation pathways. However, changes in expression across the cohort were only slight. Similarly, a few methylation candidates outside of these processes were identified - five shortlisted genes were found to possess methylated CpG islands in their 5′ regulatory regions - but differential expression was very small and subsequent inspection suggested that the possibility of actual expression change invoked by altered promoter methylation was remote.

5.8.1 Further comments on identified genes

Of the genes related to folate metabolism, DNA methylation and one-carbon metabolism, just 12 were deemed to be differentially expressed following initial SAM analysis. It might be surmised that genes in these categories, particularly the ones related to folate metabolism, would be over-represented among all differentially expressed genes, given the nature of the study, and the presumed propensity to influence folate metabolism genes. In fact, mean expression intensities for almost all of the immediate machinery responsible for folate metabolism displayed insignificant change between avoidance and supplementation periods. Of the folate and DNA methylation related genes identified in the SAM analysis, only three – the reduced folate carrier, folate hydrolase and formyl-THF hydrolase – were directly involved in the transport or metabolism of folate, and even the case of SLC19A1, the direction of expression change was counter-intuitive, although its magnitude was very small.

In retrospect then, it was perhaps unsurprising to find that many of the genes involved in folate-mediated one carbon metabolism have been reported to be constitutively expressed in low abundance. For example, the MTHFR gene is known to produce low levels of mRNA transcript in most tissues (Gaughan et al., 2000). Additionally, regulation and effective catalytic capacity of MTHFR is governed by control mechanisms other than the transcriptional level (see introduction). Indeed, this is true of other enzymes involved in folate-mediated one carbon metabolism, where allosteric binding of folate acts both as a reservoir and govern catalytic activity – as intracellular free folate concentrations decrease, enzymes are unencumbered and thus mitigate folate derivative depletion (Nijhout et al., 2004).

Microarray Analysis 157

As already stated, mean expression differences between the folic acid avoidance phase and folic acid supplementation were small for all genes identified. The largest observed difference was in DNMT3b – a fold change of 1.27 – and over half had fold changes less than 1.1. As reported in section 5.7.4, many of those genes identified had duplicate probes that displayed no significant difference. Accordingly, they could only be regarded as inapposite and to be treated with caution. Thus, the microarray assay was unable to identify any genes that were differentially expressed as a result of potential methylation changes that affected those genes’ regulatory elements.

5.8.2 Assay limitations – inter-subject variability

Even though allowances were made during the intervention trial to minimise potential variation between avoidance and supplementation phlebotomies, such as scheduling blood draws for the same time of day and standardising RNA isolation, the heterogeneity of participants in terms of age, sex and the limited control over factors such as diet meant that inter-subject variation was apparent in the array data. This was exemplified by the large coefficient of variance apparent in genes defined as uniformly differentially expressed (BIN1 equalled 183%, through to SLC39A14 at 400%). This contributed to the difficulty of isolating uniformly differentially methylated genes across the cohort. A larger number of arrays would have increased study power and may have improved the ability of the assay to detect changes in expression.

Additionally, the use of peripheral blood as an RNA source, although convenient, may have contributed greater mRNA variance compared to a more uniform tissue – blood is a heterogeneous mix of many different cell types known to vary in proportion by time and by individual (Fan and Hegde, 2005). For example, monocytes can range from 1 to 5% and lymphocytes can vary anywhere from 20 to 50% of all leukocytes. Many of the genes identified as differentially expressed in this experiment were known to be highly expressed in leukocytes, which would dictate that comparatively high mRNA transcripts would be present in collected participant samples. Accordingly, changes in proportions of leukocyte types would result in appreciable changes in the said gene transcripts.

158 Chapter 5

In contrast to the observed variation between participants, there was relatively low variability within subjects between avoidance and supplementation periods. Eighty percent of genes (10th to 90th percentile) within individual arrays displayed expressional fold change of less than 1.2. Indeed, this phenomenon has been observed by others. In a comprehensive assessment of microarray gene expression variation in leukocytes among 18 healthy volunteers, Eady and colleagues found a remarkable consistency in expression of the majority of genes across five weekly samplings (Eady et al., 2005). Conversely, they reported marked expression differences between individuals, including as a function of age sex and leukocyte subset proportions.

In the current assay, the daily 400 µg folic acid supplement that was the only controlled differential input between study phases was acknowledged to be subtle. Together with the aforementioned variance between volunteers, it might be argued that the cohort used to produce material for the array analysis was insufficiently capable of generating identifiable DE genes. It should be made clear, however, that the primary aim of the experiment was to discover expression attributes of a healthy population subjected to modulated folic acid intake – whether this generated significant differential expression at all was a secondary outcome.

5.8.3 Genes involved in methylation remodelling – implications for the wider methylome?

While their expression changes were only 1.07 fold and 1.27 fold respectively, the appearance of MBD2 and DNMT3b in the DE lists introduced the possibility that a wider reworking of genomic methylation elements may have occurred as a result of modulated folate intake, outside of gene promoter CpG islands. While no correlation between global changes in DNA methylation and changes in expression of these genes was observed, it was possible that certain components of the methylome outside of gene promoter regions were altered, such as repetitive elements. Therefore it was pertinent to assess these aspects of the genome further. Accordingly, an investigation was undertaken to query the degree of methylation of these elements – specifically, the methylation status of repetitive Satellite 2 methylation. This is duly reported in the next chapter.

6 Satellite 2 Methylation Analysis

6.1 Introduction

As alluded to in Chapter 1, CpGs occurring as islands in the promoter regions of genes account for a small fraction of the total CpG content of the human genome (see Table 1.1). The bulk of these dinucleotides instead reside in high copy number elements such as transposons and so-called “satellite” DNA regions – long sequences of short repeating motifs occurring in or near the centromeres of chromosomes. The CpGs in these satellite regions are usually highly methylated in normal somatic tissues, but can become hypomethylated in some disease states. The fact that satellite 2 (Sat2) sequences, of all repetitive DNA elements, appear to be particularly susceptible to aberrant methylation, including in conjunction with the MTHFR C677T polymorphism, as will be described below, made Sat2 an attractive candidate to analyse in the study cohort as a proxy for potential changes in wider DNA methylation as a result of the dietary intervention. Chapter 6 briefly elaborates on these aspects of satellite 2 DNA and reports the results of an assay which investigated the extent of Sat2 methylation as a function of the trial’s modulation of dietary folate intake.

6.1.1 Characteristics of satellite sequences

Satellite DNA sequences are high copy number tandem repeats that contribute to around 4% of the human genome (Paulsen et al., 2008; Rollins et al., 2006). These elements are subdivided into groups according their particular sequence motifs; classical satellites together with alpha, beta and gamma satellite and Sn5 sub-families have been classified (Lee et al., 1997). Satellite 2 repeats form part of the classical satellite group, together with Sat1 and Sat3 sequences. These sequences were first identified about a quarter of a century ago (Prosser et al., 1986).

160 Chapter 6

Sat2 elements consist of poorly conserved tandem repeats of the sequence ATTCC – more explicitly (ATTCC ATTCG)2, followed by one or two ATG trinucleotide motifs (Deininger et al., 1981; Wilson et al., 2007). These sequences are predominantly found in the variable heterochromatic sections of chromosome 1 and chromosome 16 (Schwarzacher-Robinson et al., 1988; Tagarro et al., 1994). To a lesser extent, Sat2 elements also occur in chromosomes 2 and 10 (Lee et al., 1997; Tagarro et al., 1994). These regions are typically defined as being juxta- or pericentromeric. Accordingly, Sat2 sequences are thought to be involved in chromosomal organisation, particularly during replication.

6.1.2 Aberrant Sat2 methylation and disease

Importantly, satellite repeats are highly methylated in normal tissues (Wilson et al., 2007). However, there are numerous instances of hypomethylation of these elements occurring in diseased states. Such hypomethylation is known to bring about chromosomal instability, a term encompassing poor recombination fidelity and chromosomal breaks, rearrangements or translocations (Wilson et al., 2007). Additionally, Sat2 demethylation tends to mirror observed global hypomethylation events (Weisenberger et al., 2005).

There are numerous reports of Sat2 hypomethylation occurring in cancer. In a study of primary breast adenocarcinomas, for example, almost 50% of samples exhibited Sat2 hypomethylation (Narayan et al., 1998). In ovarian cancer, Qu and colleagues identified a significant correlation between the extent of hypomethylation of Sat2 repeats and the degree of malignancy (Qu et al., 1999). Comparable findings have been reported by others - a recent study assessing colorectal carcinogenesis reported a stepwise hypomethylation of Sat2 CpGs accompanying disease progression (Kwon et al., 2010). An assessment of 115 ovarian cancers and 26 non-neoplastic samples came to the same conclusion (Widschwendter et al., 2004). Additionally, Sat2 demethylation proved to be a good indicator of prognosis, with more severe hypomethylation correlated with poor outcome.

Satellite 2 Methylation Analysis 161

Interestingly, the hypomethylation of Sat2 sequences in many cases appears to be preferential to the reduction of methylation in other repetitive elements. Lee and colleagues reported that hypomethylation of Sat2 occurred earlier in hepatocellular carcinomas than that of LINE-1 or ALU repetitive elements. The same group observed progressive Sat2 methylation reduction but no differential methylation of LINE-1 elements as premalignant bile duct epithelia transitioned to extrahepatic cholangiocarcinoma (Kim et al., 2009). These experiments suggest that Sat2 hypomethylation events may be a particularly sensitive and early indicator of changes in wider CpG methylation.

In a study of glioblastoma tissue, Cadieux and colleagues reported that the witnessed genome-wide hypomethylation present in these tumours consisted of demethylation of multiple repeat sequences, including Sat2 elements. Interestingly, hypomethylation of these sequences was exacerbated in samples from participants possessing one or two copies of the polymorphic MTHFR C677T allele, which is perhaps unsurprising given the enzyme’s role in the DNA methylation pathway. The authors accordingly suggested a model for glioblastoma progression, whereby excessive cell growth in a milieu of a low methyl supply, brought about by MTHFR inefficiency, could promote Sat2 hypomethylation and thus further genomic instability.

The other area where aberrant methylation in Sat2 sequences invokes disease is the immunodeficiency, centromere instability, and facial anomalies (ICF) syndrome, as briefly touched on in section 1.7.1. ICF is a manifestation of irregular DNMT3b activity, usually borne out of mutations or deletions of that gene, and results in severe hypomethylation of Sat2 and Sat3 sequences and a more modest reduction in the methylation content of Satα sequences (Ehrlich et al., 2006; Ehrlich et al., 2008).

Serendipitously, DNMT3b was one of two genes identified by microarray analysis as down-regulated in MTHFR T/T participants during the folic acid avoidance period of the current study. Although its change in expression was marginal, DNMT3b’s association with Sat2 methylation reinforced the latter as an attractive genomic element to investigate. A methylation-sensitive quantitative PCR assay was employed to this end.

162 Chapter 6

6.2 Satellite 2 methylation assay

Few assays have been described that quantitatively measure the methylation content of repetitive DNA sequences. Traditionally Southern blot analyses have been employed, however more recently Combined Bisulphite Restriction Analysis (COBRA) and bisulphite-converted DNA pyrosequencing methods and the MethyLight real-time PCR- based assay have been used (Eads et al., 2000; Weisenberger et al., 2005). The latter assay employs methylation-sensitive primers to initiate PCR amplification, together with a TaqMan probe that binds to sequence between the primers. During PCR extension, Taq polymerase liberates a fluorophore attached to the probe which is detected by the real-time PCR instrument. The degree of fluorescence after a given number of cycles is proportional to the amount of methylation in the DNA sequence complementary to the primers.

A similar method to MethyLight, termed methylation-sensitive quantitative PCR (MSQ- PCR) and first developed by Fukuzawa et al., incorporates an endonuclease step prior to DNA amplification (Fukuzawa et al., 2004). The method has subsequently been reported by others as an assay for DNA methylation quantification (Bruce et al., 2008; Oakes et al., 2006; Oakes et al., 2009). By using a methylation-sensitive enzyme with a recognition site corresponding to the sequence of interest, the degree of CpG methylation in the sequence in effect determines the amount of starting template for the PCR, thereby enabling relative quantitation by fluorescence.

As this method was already established in the student’s laboratory, it was chosen to assay DNA samples from the current study’s participants. In a slightly modified version of the above procedure, a precursory methylation-sensitive BstB1 digestion was employed, which digested Sat2 sequence only if it was unmethylated. Accordingly, the higher the proportion of Sat2 CpGs that were methylated, the more sequence was left intact to act as a template for QPCR. Primers complementary to the Sat2 motif were used for amplification. Additionally, the fluorescent DNA interchelator SYBR green was used as a fluorescent agent, instead of the TaqMan probe. Details of the procedure are listed in section 2.11.

Satellite 2 Methylation Analysis 163

6.3 Results

The results of the Sat2 analysis are depicted in Figure 6.1. Following the folic acid avoidance phase of the trial, participants homozygous for the MTHFR C677T polymorphism displayed a significantly reduced percentage of Sat2 methylation compared to wildtype or heterozygous individuals (p = 0.050 and p = 0.033 for C/C versus T/T and C/T versus T/T, respectively). The mean difference between T/T individuals and other participants was only five percentage points, however. There was no difference between MTHFR genotypes following folic acid supplementation. Between post avoidance and post supplementation, mean Sat2 methylation insignificantly increased in T/T participants from 86% to 88%.

Counter-intuitively, mean Sat2 methylation decreased in the other two genotypes over the course of the study – from 91% to 88% for C/C participants and 91% to 89% for C/T participants. No putative cause can be offered for these decreases, although it should be noted that these changes were also statistically insignificant. Interestingly, males exhibited lower Sat2 methylation than females following the repletion period – means were 90% and 86% for females and males respectively (p = 0.014). No

Figure 6.1 Satellite 2 methylation according to MTHFR C677T genotype following folic acid avoidance and folic acid supplementation. Data is expressed as mean ± 95% confidence interval.

164 Chapter 6 significant difference between sexes was observed following folic acid avoidance. No significant correlation was found between Sat2 methylation and age at either post- avoidance or post-supplementation.

A potential relationship between T/T individual Sat2 methylation and DNMT3b expression was investigated, given that T/T participants exhibited lower Sat2 methylation during the avoidance phase of the study and the expression of DNMT3b was also lower relative to other participants during this time. No correlation was found however – DNMT3b expression was completely independent of Sat2 levels (R2 = 0.001). Even though there was a sex effect observed in post avoidance Sat2 data, this was not apparent for DNMT3b expression (p=0.65). It should be noted of course, that the study was never designed to be powered sufficiently to detect sex effects.

Satellite 2 methylation was compared to global DNA methylation results as reported in Chapter 4 (see Figure 6.2). No statistically significant correlation was found between the two datasets overall, during either avoidance or repletion periods. A weak correlation was identified in participants whose plasma folate was below 12 nmol/L during the avoidance phase however, with the level of Sat2 methylation appearing proportional to global methylation in these individuals (see Figure 6.2, panel C). Further dividing this subset by MTHFR genotype revealed stronger correlations in wildtype and heterozygous subjects (panels E and F). The correlation between Sat2 and global methylation was highly significant in this latter group. Interestingly, T/T individuals in this subset displayed a more stable Sat 2 methylation throughout the range of assayed global DNA methylation (see panel G).

Satellite 2 Methylation Analysis 165

A Avoidance B Repletion

C [Plasma folate] < 12 nmol/L D [Plasma folate] ≥ 12 nmol/L

E C/C F C/T G T/T

Figure 6.2 Global DNA methylation versus satellite 2 methylation, subdivided by plasma folate status and MTHFR genotype following folic acid avoidance. The uppermost graphs depict individual Sat2 values plotted against their corresponding total DNA methylation results. Panels A and B correspond to data obtained following the trial’s folic acid avoidance phase and the folic acid supplementation phase respectively. Those participants who reported a low plasma folate status (< 12 nmol/L) during the former period also displayed a weak positive correlation between the two assayed methylation variables (panel C). This was not found in subjects whose folate status was above this level (panel D). Closer inspection of low folate data revealed C/C and C/T individuals alone contributed to this relationship (E and F) – T/T subjects displayed low Sat2 methylation regardless of total DNA methylation content. Some Sat2 values were greater than 100% but these were within the expected limits of the assay. 166 Chapter 6

6.4 Discussion regarding Sat2 results

This chapter has reported the effects of folic acid avoidance and supplementation upon satellite 2 CpG methylation in leukocytes of the study’s participants. Although differences in mean Sat2 methylation were small, individuals homozygous for the T allele displayed a significantly lower methylation level under conditions of folic acid avoidance than either C/C or C/T participants. This was despite no differences in global DNA methylation being observed between these groups when assayed using high performance liquid chromatography.

Interestingly, the lack of correlation between repetitive element methylation and total genomic methylation is in contrast to what has been observed by others. Weisenberger and colleagues regarded their MethyLight quantitative PCR-based assay, similar to the one used above, as a reliable surrogate marker for global methylation, when taking into account both Sat2 and Alu elements (Weisenberger et al., 2005). In the present study, parallel changes in Sat2 methylation and global DNA methylation were instead not typically observed. Only wildtype and heterozygous individuals whose plasma folate status was below 12 nmol/L exhibited this phenomenon.

The current results instead offered that the measurement of discrete components of genomic methylation – in this case the assessment of Sat2 repetitive elements – may be a more sensitive indicator of subtle changes occurring across the genome.

It is worthwhile comparing the Sat2 data with the outcomes of the previous chapter. One rationale for conducting the microarray analysis was to investigate whether changes in gene expression could provide clues for potential methylation control of certain genes. This work reported negligible changes in gene expression in the study participants. Thus, if any genes were in fact capable of having their expression modulated by differential methylation of their control elements, then this phenomenon was not observed in this study; negligible change in methylation-sensitive expression inferred that methylation of CpG islands was effectively static by proxy, or at least changes in CpG methylation contributed little to changes in gene expression. Conversely, CpGs in Sat2 sequences, which comprise a similar percentage of all genomic CpGs to promoter CpGs (approximately 5%) displayed some variation in

Satellite 2 Methylation Analysis 167 methylation. As such, it could be supposed that various CpG elements across the genome may be differentially or preferentially methylated as the input of dietary methyl donors is altered.

This could provide a possible explanation for the loss of association between global DNA methylation and Sat2 methylation in T/T participants with plasma folate under 12 nmol/L (Figure 6.2, panel G). The supposed decrease in availability of methyl groups for DNA methylation arising from the MTHFR / low folate status interaction may have manifested as the occurrence of lower Sat2 methylation, even though high methylation levels were maintained globally. This hypothesis fails to take into account the lack of association between Sat2 methylation and global methylation at higher folate concentrations however, and the correlation observed in panels E and F of Figure 6.2 may simply be artefactual due to low participant numbers in these groups.

In any case, the analysis of Sat2 elements provided a more complete picture of the changes occurring in methylation across the genome than could have been provided by global DNA methylation analysis alone.

That said, it should be reinforced that these changes as a result of the dietary intervention were relatively small. Overall mean change between avoidance and supplementation periods was from 90% to 88%. Five percentage points separated wildtype and T/T means following avoidance. Individually, 80% of all participants displayed changes in Sat2 methylation of less than 17%.

168 Chapter 6

7 Discussion

7.1 Synopsis of the project

This study sought to establish whether differences in dietary folate intake or impaired folate metabolism attributed to the MTHFR C677T polymorphism altered leukocyte DNA methylation in a healthy New Zealand adult population, not already exposed to folic acid fortification. It assessed hypothesised changes in blood folate and homocysteine concentrations as a result of varied dietary folate intake. The work also sought to gauge potential changes in gene expression as a result of possibly altered CpG island methylation, as well as modification of global DNA methylation and satellite 2 repetitive elements, which together might have inferred increased disease propensity.

The work was undertaken in response to often conflicting data in human observational and intervention studies surrounding the relationship between folate status, the MTHFR C677T polymorphism and DNA methylation (as reviewed in section 1.10). Additionally, the project aimed to provide information concerning changes in folic acid- induced methylation apropos of the potential introduction of national folic acid fortification in New Zealand.

7.1.1 Intervention trial and micronutrient data

Ninety three individuals completed the 24 week intervention trial, consisting of a 13 week folic acid avoidance, or “washout” period, followed by an 11 week folic acid supplementation phase, where daily supplements of 400 µg folic acid were taken. Equal numbers of participants possessing each MTHFR genotype were included.

Between baseline and folic acid avoidance samplings, plasma folate and red cell folate concentrations decreased significantly across all MTHFR genotypes, as anticipated

170

(mean decreases of 7 nmol/L and 111 nmol/L for plasma and red cell folate, respectively). Given that New Zealand does not currently mandatorily fortify foods with folic acid, observed decreases were not as pronounced as studies describing folate avoidance in folate-fortified countries. However, following the avoidance period, a significant proportion of participants recorded plasma and erythrocyte folate levels at which DNA hypomethylation had been observed previously (below 12 and 700 nmol/L). Participants homozygous for the MTHFR polymorphism exhibited lower folate concentrations throughout the intervention.

It was hypothesized that following the avoidance of folic acid fortified foods, supplementation with folic acid would increase blood folate concentrations, lower plasma homocysteine concentrations and increase DNA methylation. Indeed, folic acid supplementation successfully and significantly increased plasma and erythrocyte folate concentrations across all participants. Plasma total homocysteine (tHcy) also declined in all participants following supplementation. Homocysteine concentrations decreased to a greater extent in the MTHFR T/T genotype group than in heterozygous or wildtype individuals, especially in people whose pre-supplementation folate status was less than 12 nmol/L. This differential reduction in tHcy between MTHFR genotypes suggested the potential for a greater cellular effect on the folate dependent methylation cycle in the group homozygous for the polymorphism.

7.1.2 Genomic DNA methylation analysis

Despite these initial results, however, lowered folate status through dietary avoidance had no appreciable effect on peripheral blood leukocyte global DNA methylation, which remained well conserved. Similarly, overall genomic 5′-methylcytosine levels were static during folic acid supplementation. No significant differences between MTHFR genotypes were identified at either avoidance of supplementation periods. Individuals possessing a low folate status also exhibited no difference in genomic methylcytosine content compared to other participants.

171

7.1.3 Microarray analysis

Analysis of gene expression exhibited low fold expression change between avoidance and supplementation samples. Additionally, relatively high inter-participant gene-by- gene variability was apparent, indicating little uniform effect of the dietary regime. All differentially expressed genes identified as significant had low fold expression differences – both between dietary periods and between MTHFR genotypes. The small number of genes identified consisted predominantly of those known to be highly and variably expressed in leukocytes, or immediately involved in folate uptake, which indicated a small transcriptional response to the dietary intervention.

Of genes known to possess methylated CpG islands, five were identified in the current analysis as being differentially expressed either by MTHFR genotype, by plasma folate status at post-avoidance, or between avoidance and supplementation periods. The extent of this differential expression was slight – maximum fold expression change was 1.16. Furthermore, statistical analyses to identify these genes required high false discovery rates. The most promising gene, Bridging integrator 1 (BIN1), which was upregulated during the avoidance period in T/T individuals, had been previously identified as possessing aberrant promoter methylation in breast and prostate malignancies. However, given that transcript variation was very small between avoidance and supplementation (1.09 linear fold change) and that another probe for the gene on the array reported no differential expression, this gene was not regarded as a strong marker of carcinoma development in the study cohort.

Collectively, the microarray analysis data inferred that uniform expression changes as a result of the modulated folic acid intake were negligible in the trial participants.

7.1.4 Satellite 2 methylation analysis

Closer inspection of specific genomic loci – satellite 2 repeats – revealed some nuanced changes in CpG methylation. Individuals homozygous for the MTHFR polymorphism exhibited a mean Sat2 methylation status 5.5% lower than heterozygous or wildtype subjects following the avoidance period. This differential methylation was not observed following supplementation. Although not significant, there was a positive linear

172 correlation between global DNA methylation and Sat2 methylation in individuals whose folate status was less than 12 nmol/L (p = 0.09). This was driven by non-T/T participants.

7.1.5 Summary

These results implied that firstly, medium term folate avoidance within a milieu of the average New Zealand diet did not result in severe blood folate deficit or increase plasma homocysteine. Secondly, medium term folic acid supplementation was able to raise folate status and also lowered homocysteine but did not appreciably affect genomic DNA methylation or alter gene expression due to folate-coupled reworking of promoter methylation. Methylation of pericentromeric elements was also minimally affected by this dietary intervention. Therefore, it was concluded that the hypothesised mechanism of disrupted folate metabolism leading to improper DNA methylation events and increased disease propensity was an unlikely event in this cohort of healthy New Zealanders.

7.2 Limitations of the study

It is clear, however, that the observed effects of folate avoidance on related metabolites, principally plasma homocysteine, were not as pronounced in this study as they may have been in a dietary-controlled environment. It must be acknowledged that conducting the study outside a metabolic unit setting introduced potentially confounding elements. There were also other aspects of the trial, such as the choice of tissue used to source DNA, that dictated how widely the study’s findings could be interpreted. These are discussed below.

7.2.1 Potential confounders of the immediate trial

Steps were taken to limit many confounding variables; conducting venipuncture at the same time of day to minimise circadian variation in RNA, for example. However, many confounders could not be controlled. Because the study was designed to assess people

173 in a “free-living” general population and not within the confines of a metabolic unit setting, the trial did sacrifice complete control over dietary intake.

Some confounding factors implicated in folate, homocysteine or wider one-carbon metabolism that may have affected study outcomes were not controlled for. Specifically, hypertension or a history of cardiovascular disease were not explicitly defined as exclusion criteria for the study. This was probably an oversight upon reflection, although during the study no participants reported plasma homocysteine levels greater than 15 µmol/L, the threshold for moderate hyperhomocysteinemia. Less critically, measures of choline and betaine metabolism were not obtained. While the contributions of these factors to homocysteine remethylation are assumed to be slight in peripheral blood (as discussed in section 1.1.7) measuring these would have provided a more complete picture of one-carbon metabolism and the relative effects of folate supplementation.

All participants were supplemented with riboflavin, B6 and B12 to ensure the optimal function of one carbon metabolism. Consequently, many participants presented with a decrease in total plasma homocysteine concentration from baseline to the avoidance period. This implied that providing the above supplements appeared to increase the efficiency of homocysteine to methionine conversion, even before folic acid was introduced. Ironically, this may have reduced the potential of folic acid supplementation to invoke large changes in homocysteine between avoidance and supplementation phases (Graham and O'Callaghan, 2002).

No differences between sexes were found in all measured outcomes, with the exception of Sat2 during depletion. It is possible that the study was not sufficiently powered to determine the difference between males and females for the measurable outcomes of the trial. Furthermore, it has been suggested that utilizing a more strictly defined subset of the population, for example women aged 29-40, may have provided more uniform data and thus provided more power, particularly for the microarray analysis, which failed to identify any noteworthy differentially-expressed genes. While a creditable suggestion, this would have severely limited the size of the study, given that individuals possessing the MTHFR T/T genotype made up less than 10% of the screened participants – T/T

174 female participants falling in that age bracket numbered just six. If during the planning of the trial it was decided that the MTHFR C677T genotype would not have been included, using a subset of the population may have been a feasible alternative study design.

Also, it has been suggested that greater numbers of women than men participating in the study may have attributed to the differences seen between the present study and the results published by Friso and colleagues. While it is plausible that the greater proportion of women, with their lower folate requirement (212 µg/day versus 278 µg/day for males) may have responded less dramatically than men to the folate avoidance period, this is not borne out by data collected by the Ministry of Health. Instead, a higher proportion of New Zealand women than men have an inadequate dietary intake, regardless of age (Russell et al., 1999). Ergo, the greater number of females in the study should have theoretically pushed peripheral blood folate concentrations closer to what was reported by Friso and colleagues. In actuality, plasma folate levels in the present study’s female participants showed no difference to men post folic acid avoidance, including those with plasma folate concentrations less than 12 nmol/L. Arguably, variations in participants’ diets or socioeconomic differences between the two cohorts would have more bearing on outcomes than sex.

Excessive alcohol consumption has been cited by peers, who reviewed a manuscript arising from this work, as potentially interfering with intestinal folate absorbance and metabolism (Mason and Choi, 2005). Although participants were asked to avoid alcohol 24 hours prior to providing a blood sample, they were not asked to exclude alcohol throughout the study. Thus, excessive alcohol intake may have interfered with folate status. However, the purpose of this study was to examine the effects of folic acid supplementation in free-living participants rather than in a controlled setting. Additionally, time restrictions were imposed on the intervention in order to ensure the trial was completed before Christmas, traditionally a time of increased alcohol consumption.

Finally, the study did not provide an estimate of background folate intake, but this was unlikely to have varied throughout the trial and very unlikely that folate intake varied by

175

MTHFR genotype. The median intake of natural folate for participants greater than 15 years of age in the 1997 New Zealand Nutrition Survey was 242 µg/d (Russell et al., 1999).

7.2.2 Assay limitations

The outcomes of the microarray analysis raised questions as to the suitability of gene expression to act as a proxy for methylation detection. The ability to identify changes in CpG island methylation using this method was always going to be difficult: the majority of CpG islands are unmethylated, with only a small fraction being methylated (approximately 4%) and thus possessing the ability to become hypomethylated in response to depleted dietary folate. Additionally, the homeostatic capability to buffer variations of dietary folate intake could conceivably maintain a steady state of methyl group provision for DNA methylation (Nijhout et al., 2004). Furthermore, many CpG islands are located in genes that have a tissue-restricted expression pattern. Thus the extent of methylation of many CpG islands does not necessarily correlate with the associated gene’s expression. For example, the tissue-specifically expressed human α-globin and α2 collagen genes have CpG islands that remain unmethylated in all tested tissues, regardless of expression (Bird, 2002). Combined with the heterogeneous nature of the cohort, assessing gene expression was perhaps not the most appropriate method of discovering CpG island methylation changes. Despite this concession, however, the array data provided interesting information about the immediate expression response of folate uptake machinery.

7.2.3 Tissue suitability

Peripheral blood leukocytes were used as a source of DNA for global DNA methylation and Sat2 analyses and RNA from these cells were used to construct gene expression profiles. Peripheral blood was used owing to its ease of extraction from participants and that phlebotomy was accepted by most individuals as being minimally invasive. Furthermore, most comparative studies (reviewed in section 1.10) assayed samples derived from blood and given the high turnover of leukocytes, using peripheral blood allowed changes in assayed variables to occur in a short frame of time.

176

It should be made clear, however, that the results derived from peripheral blood are not necessarily representative of other tissue types. Indeed, exploration of methylation between different cell types has shown global methylation to be far from uniform. This is exemplified in a recent paper by Sohn and colleagues. The group monitored genomic DNA methylation and uracil incorporation in HCT116 colon and MDAMB-435 breast adenocarcinoma cell lines, both of which possessed the MTHFR C677T polymorphism, under high and low folate supply (Sohn et al., 2009). In the colon cell line, DNA methylation was increased under high folate conditions and decreased under depleted conditions in the polymorphic line, but conversely in the polymorphic breast cell line, DNA methylation was decreased under folate replete conditions and elevated when folate supply was low, when compared to wildtype. Uracil misincorporation also trended in opposite directions between the cell types, however this was nonsignificant. As another example, it has been postulated that neuronal tissue may utilise CpG methylation in the role of memory formation, reinforcing that methylation changes can occur independently of dietary effectors (Korzus, 2010).

Therefore, with regard to the current study, limited conclusions would be able to be made about the effects of folate depletion and supplementation on other body systems.

7.3 Novel aspects of the study and comparisons with associated literature

Many of the results generated by the current project were in agreement with those of previous studies which analyzed effects of folate supplementation in human subjects, as outlined in part 1.10. Additionally, initial micronutrient data were in keeping with the wider consensus surrounding folate biochemistry – for example, plasma and red blood cell folate concentrations acted in accordance with changes in dietary folate intake. However, other outcomes did not always confirm previous observations of related studies, such as the extent of variation between baseline and post-depletion plasma homocysteine concentrations. Most importantly, folate depletion and repletion as well as MTHFR C677T genotype had no significant effect on global DNA methylation in this work, in contrast to some prior observations made in other cohorts. These apparent

177 discordances and explanations for why they might have occurred are discussed in greater detail below.

7.3.1 Studies conducted under a background of folic acid

The present intervention took place in a country not already exposed to mandatory folic acid fortification, counter to previous longitudinal studies undertaken in North America, where fortification was introduced in 1998 (Axume et al., 2007b; Guinotte et al., 2003; Rampersaud et al., 2000; Shelnutt et al., 2004). This could account for the discrepancies seen in micronutrient data between baseline and avoidance/depletion periods. The present study recorded relatively low blood folate concentrations at baseline. Conversely, the United States-based studies, conducted in metabolic unit settings, witnessed large reductions in blood folate concentrations between baseline and depletion periods. For example, Rampersaud et al. found their participants’ serum folate decreased from 45.7 to 13.7 nmol/L after consuming a controlled low-folate diet (188 µg per day) for 51 days. Their baseline data was set against a high level of background fortification, so the controlled diet environment was able to minimise the effects of such fortification. It should be noted that post depletion folate levels were similar between the above studies and the current trial.

Incidentally, New Zealand is not completely devoid of fortified foods, however, hence the reason why the avoidance period was still implemented in this study. It has been estimated that voluntary folic acid fortification by food manufacturers provides an additional 69 µg per day of folic acid to the New Zealand adult population (Food Standards Australia and New Zealand, 2007).

7.3.2 Metabolic unit studies versus “free-living’ cohort

Principally, the current project sought to examine effects of dietary folate avoidance and supplementation in a general population setting. This was in contrast to many related reports that conducted trials in a controlled dietary environment (Abratte et al., 2008; Axume et al., 2007a, b; Guinotte et al., 2003; Shelnutt et al., 2003; Shelnutt et al., 2004). Thus, by proxy, the present study did not control all potential dietary effectors of

178

DNA methylation, or at least ones that were regulated in other trials. For example, Jacob and colleagues designed their diet to limit choline, an exogenous source of methyl groups that may have added to the effect of folate restriction (Jacob et al., 1998). Folate supplementation was able to reverse hypomethylation in that study. As the present study only modulated folic acid, this may have limited the potential to affect DNA methylation. However, in the present work, the relatively low baseline folate status combined with the folic acid avoidance phase – essentially a “washout” period to remove any residual folic acid consumption – arguably produced a cohort that more appropriately reflected true dietary intake of natural folates alone, when compared to artificially controlled-diet studies.

7.3.3 Cohort differences

Similarly, there were differences in participant characteristics between the current study and similar intervention studies that may have contributed to the different outcomes regarding global DNA methylation. Arguably, the present cohort could be described as the most complete and thus a fairer representation of a general population; it consisted of 93 participants of both sexes and across a wide age range (19-70 years), representative of all three MTHFR C677T genotypes. Other comparative studies examined one gender only, and had a narrow age range. For example, Axume et al. and Guinotte et al., and also Shelnutt et al. investigated young women only (age ranges 18- 45 and 20-30 years respectively; a specified mean age of 25 in Axume et al.) whereas Rampersaud et al. and Jacob et al. assessed older women only: 60-85 years and 49-63 years, respectively (Axume et al., 2007b; Guinotte et al., 2003; Jacob et al., 1998; Rampersaud et al., 2000; Shelnutt et al., 2004). Participant numbers were also low for the latter two studies (n = 33 and n = 8 for Rampersaud et al. and Jacob et al., respectively).

7.3.4 DNA methylation assay methods

The present work canvassed genomic CpG methylation in greater detail than other studies that investigated folate status, MTHFR and DNA methylation in humans. Typically those studies only sought to describe changes in global methylation by way of

179 semi-quantitative [3H] methyl-acceptance or cytosine extension assays (see section 4.1 for an explanation of these methods and their limitations). This work used a more comprehensive method of measuring global DNA methylation that had a low limit of detection and good reproducibility. The inclusion of satellite 2 methylation analysis as well as gene expression and in silico analysis of selected promoter CpG islands allowed for a more complete picture to be constructed of changes in DNA methylation as a result of the dietary regime. Even though global DNA methylation analysis failed to find any significant changes as a function of the dietary regime, Sat2 analysis indicated some subtle DNA methylation modification, in accordance with wider DNA hypomethylation observed by others.

7.3.5 Comparison to cross sectional studies

As this thesis was in large part directly related and instigated in response to the work of Friso et al, it is applicable to further compare differences between these two studies. It should be noted that in all respects, apart from the MTHFR T/T subset during depletion, results were in general agreement. In wildtype individuals and polymorphic individuals with adequate plasma folate status, DNA methylation was conserved in both studies. It was only in subjects possessing the T/T genotype with poor folate status that Friso et al. witnessed genomic hypomethylation (Friso et al., 2002).

Partial mention has already been made in section 4.6.1 of the differences between cohorts that may have explained the discordant DNA methylation: the Italian study’s participants were older (mean age 60 years), predominantly male (around 80%), sourced from a hospital and the majority of them had confirmed coronary atherosclerosis, which implied contributing factors in addition to folate status.

The Italian cohort’s total homocysteine measurements were also around twice what was observed in the New Zealand study. So while plasma folate concentrations were comparable – 44% of Italian T/T individuals and 42% of T/T participants in the current study following folic acid avoidance were under 10 nmol/L – the differences in total homocysteine suggested folate status was not solely responsible for modulating the

180 latter. Therefore, this inferred that perhaps folate status may not necessarily be a reliable predictor of DNA methylation content.

Also, homocysteine and S-adenosylhomocysteine are efficient inhibitors of DNA methyltransferases. Excessive levels of these metabolites are known to induce DNA hypomethylation via methyltransferase inhibition (Yi et al., 2000). Friso et al. reported a mean total homocysteine concentration above the threshold for diagnosis of hyperhomocysteinemia (> 12 µmol/L). Therefore, it is conceivable that the methyltransferase inhibition mechanism was a driver for DNA hypomethylation in that study’s participants. Consequently, a lack of S-adenosyl methionine, due to inadequate MTHFR processing and low folate, may not have been a principal cause, despite being suggested as such by the authors.

The age of the Italian cohort coupled with the negative association found between total homocysteine and vitamin B12 by Friso and colleagues indicated that B12 deficiency also may have been a contributing factor to the hyperhomocysteinaemia and thus DNA hypomethylation. Older people tend to be at greater risk of vitamin B12 deficiency

(Institute of Medicine, 2000) and given that vitamin B12 is required for the functioning of methionine synthase which converts homocysteine to methionine for downstream

DNA methylation (see section 3.1.4), a vitamin B12 deficiency could impact on such methylation. Incidentally, no data was given about differences in age between the MTHFR genotypes, if any, in the study by Friso and colleagues, which could, at least in part, account for such observed differences in methylation. In the current study the requirement to assess biochemical changes solely attributable to the folic acid intervention meant all participants received supplements containing vitamin B12 to control for this cofactor. Consequently, there was no association between tHcy and vitamin B12 in the current cohort.

Other, external sources of methylation were not controlled for in either study. For example, dietary choline, although not a major contributor in methylation events in peripheral blood cells (the tissue from which all measurements in this study were taken), can incorporate one-carbon moieties into the DNA methylation pathway via betaine homocysteine methyltransferase (BHMT), transforming homocysteine to

181 methionine (Fox and Stover, 2008; Institute of Medicine, 2000). Differences in choline intake may have contributed to the discordant findings of the two studies. Unfortunately, choline and betaine were not measured in the current study’s participants. As much of the Italian cohort was deficient in folate and vitamin B12, it might be surmised that these individuals possessed a diet that was chronically deficient in multiple nutrients. No information was given regarding diet history, however.

The most astounding finding of the study by the Italian group was the degree of hypomethylation present in individuals homozygous for the MTHFR C677T polymorphism – decreases in the order of 50% were reported when plasma folate concentration was below 10 nmol/L. As has been stated already in section 1.7, severe disorders such as immunodeficiency, centromere instability and facial anomalies syndrome (ICF) can manifest where decreases in DNA methylation are just 15%. Even small reductions in the DNA methylation of single gene promoters have been implicated in severe disease phenotypes (Kondo et al., 2000). What is peculiar in the Italian study is that there were no reported associations between those individuals with severe peripheral blood leukocyte hypomethylation and any increased morbidity, compared to other study participants.

One might surmise that a large decrease in genomic methylation, a critical requirement for chromosomal integrity and correct gene expression, would have a severe detrimental effect on correct cell function. A more plausible notion would be that the cell would instead seek to conserve DNA methylation. Indeed, modelling studies report that one- carbon redistribution and the multiple levels of control and feedback allow for a remarkably consistent supply of methyl groups to their final molecular destinations, even under fluctuation of dietary methyl sources (Nijhout et al., 2008). This makes the report by Friso and colleagues describing such markedly reduced genomic methylation in human participants all the more remarkable.

Two observational studies have reported similar findings to that of Friso et al. where lower DNA methylation was measured in homozygotes for the MTHFR polymorphism under conditions of low folate status compared to wildtype individuals (Castro et al., 2004; Stern et al., 2000). Neither of these studies managed to replicate the 50%

182 reduction in DNA methylation as observed by the Italian group, however. Castro et al. and Stern et al. used semi-quantitative and highly variable cytosine extension and methyl acceptance assays, respectively (see part 4.1). Both studies possessed low numbers of participants homozygous for the MTHFR polymorphism (n = 9 and n = 10, respectively). Interestingly, a cited reference search of the paper by Friso and colleagues, performed January 2011, delivered 322 hits; however, none of these papers, with the exception of Castro et al., had replicated the Italian experiment and witnessed any accompanying demethylation of genomic DNA.

When all the above information is taken into account, it is apparent that the finding of Friso et al. regarding genomic hypomethylation was not solely the result of low dietary folate intake, but a reflection of wider health deprivation in those participants. It is clear that the Italian cohort was not representative of a general healthy population, which was what the present study sought to examine. Thus caution should be taken when directly comparing the outcomes of the current work with those of the Italian cohort.

7.3.6 Literature describing no relationship between folate status and DNA methylation

So far, this section has focussed on studies that have reported positive changes in DNA methylation as a response to dietary folic acid intervention and the MTHFR polymorphism. It is important to note that other studies have been published that describe no such modulation in DNA methylation. For example, in a study investigating potential disease markers in 61 healthy volunteers, 1.2 mg folic acid supplementation significantly improved folate status and lowered homocysteine but DNA methylation remained unchanged (Basten et al., 2006). The experimenters reported that other markers for cancer susceptibility, namely uracil misincorporation, could be regarded as more sensitive indicators for disease propensity than global DNA methylation.

Perhaps not surprisingly, supplementing individuals already replete (high blood folate status) had no effect on DNA methylation in another study of healthy Australian participants (Fenech et al., 1998). Also Narayanan and colleagues, investigating measures of DNA stability, reported that genome wide DNA methylation was

183 comparable between all MTHFR genotypes in lymphocytes from healthy individuals (Narayanan et al., 2004).

Most applicable to the current thesis, a study of 28 young women subjected to a seven week folate restricted diet (135 µg per day, as dietary folate equivalents), followed by a seven week folic acid treatment, found no change in global DNA methylation throughout the study (Axume et al., 2007a). Interestingly this work was performed by the same group and under the same dietary conditions as a study reporting marginal changes in DNA methylation as a result of folate intervention and the MTHFR genotype (Axume et al., 2007b).

7.3.7 Summary

From the above analysis of the literature, it is apparent that extreme folate depletion, especially within a milieu of wider nutritional deficit, or disease states as described in detail in the introduction of this thesis, can result in genomic hypomethylation. It is also clear that the MTHFR C677T polymorphism can influence DNA methylation under these conditions. Arguably, the study of Friso and colleagues falls under this banner, although participant aetiologies were not explicitly canvassed. On the other hand, studies describing disease-free populations appear equivocal in their results with regard to the effect of folic acid supplementation and the MTHFR C677T genotype on DNA methylation. Marginal and incongruous differences in global DNA methylation exist between the studies of Axume et al., Guinotte et al. and Shelnutt et al., for example (Axume et al., 2007a, b; Guinotte et al., 2003; Shelnutt et al., 2003; Shelnutt et al., 2004).

That the present work found no indication of changes in genomic DNA methylation as a result of modulated folic acid intake or the MTHFR C677T genotype did not therefore represent a failed experiment. The motive for undertaking the current work was always to assess a healthy cohort representative of a New Zealand population. In such healthy participants, it was entirely valid that DNA methylation aberrations were not common in response to modulated folic acid intake. What this study did suggest was that low

184 folate status coupled with the MTHFR C677T genotype may not necessarily be reliable or predictive indicators of changes in DNA methylation in disease-free individuals.

7.4 Brief comment on the Folic acid fortification debate

Mention has already been made in Chapter 1 of the impetus to introduce mandatory folic acid fortification into New Zealand. Successive governments here have debated the merits of fortification, which has recently garnered a lot of media attention. In June 2007, following consultation with local and Australian authorities, the Labour government legislated for folic acid to be added to bread, with then Food Safety Minister Annette King proclaiming the decision as “a triumph for humanity and common sense”. Fortification was to commence from September 2009, however a change of government and spirited debate among health professionals, bakers and lobbyists, saw this decision rescinded, with a revised “wait and see” option taken instead (NZPA, 2009). This time, concerns surrounding associations between supplementation, masking of vitamin B12 deficiency and cancer development were among the reasons given for delaying fortification.

With the interest in concomitant effects of folic acid supplementation regimes, results from the current study are of some value. This study found that moderate folic acid supplementation, in the short term at least, had no influence on gene expression through altered genomic methylation in a cohort representative of a healthy New Zealand population. Additionally, methylation of repetitive elements was not appreciably affected. These results suggested that folic acid fortification at current or proposed levels would be unlikely to have an effect on DNA methylation in the medium term.

7.5 Future research directions

Of course, whether long term supplementation – a timeframe of many years – could modify disease propensity, would need to be an ongoing area of research in this country if mandatory fortification was to be introduced. Follow-up studies that probed epigenetic modifications as a result of long term folic acid fortification would be required. Suitably, the use of recently developed next-generation sequencing platforms,

185 as described in section 5.1.1, could generate a wealth of data for such research. That these instruments can generate near-complete mapping of DNA methylation at single- CpG resolution is a boon for the epigenetic researcher. Already, recent work with less sophisticated techniques has begun to describe the extent of inter-individual variance in the methylome (Schneider et al., 2010; Yang et al., 2010; Yuen et al., 2009). However, new high throughput methods could provide a detailed picture of differences in DNA methylation and, potentially, separate natural variation from subtle epigenetic changes resulting from altered folate-mediated one-carbon metabolism and any associated increase in disease risk.

More broadly, epigenetic research is destined to be an exciting field of research over the next decade. Presently, DNA methylation is still somewhat enigmatic; a dichotomy exists between the remarkable preservation of the epigenome through vertical transmission and the increasingly described transient malleability of CpG methylation. The latter may be only sometimes true in situations such as modulated dietary folate intake, as such a result was not witnessed in the current study, but there is increasing evidence that subtle and persistent remodelling of genomic DNA methylation in somatic tissues may be a widespread phenomenon.

For example, Métivier and colleagues have recently reported dynamic and cyclical remodelling of CpG methylation in promoters of a group of genes that coincides with their expression (Kangaspeska et al., 2008; Metivier et al., 2008). Perhaps more revelatory is that epigenetic remodelling may be involved in the formation of memories (Dulac, 2010; Miller et al., 2010). This immediately posits questions surrounding the potential for environmental influences, particularly nutrients like folate that are involved in one-carbon metabolism, to be involved in influencing memory retention.

Finally, further study needs to be undertaken to establish if any differential rates of hypomethylation in response to methyl group deficit exist. The current study attempted to assess short term fluctuations in DNA methylation as a function of varied folate intake. While overall DNA methylation remained unchanged, repetitive element methylation, namely Satellite 2 methylation, displayed some fluctuation throughout the study. It may therefore be possible that changes in Sat2 either precede a global DNA

186 methylation shift, or that DNA hypomethylation across genome may be localised in response to a compromised methyl supply. A propensity for repetitive DNA to preferentially undergo hypomethylation over gene promoter methylation, for example, could be a well-designed response to one-carbon deficient and make DNA methylation in CpG islands resilient to changes from perturbations in folate metabolism.

7.6 Conclusions

With regard to the proposed introduction of folic acid fortification, the present study found that global DNA methylation remained constant throughout the intervention period and Satellite 2 methylation deviated minimally among MTHFR C677T genotypes and between folic acid avoidance and supplementation. Thus, it might be surmised that longer term exposure to lower dose folic acid, as would be the case if mandatory folic acid fortification was introduced in this country, would similarly have little to no effect.

There are some caveats to the results generated in this study, however, and so the study’s findings should be interpreted with some degree of caution. Firstly, these results were obtained from peripheral blood leukocytes, and so might not necessarily be reflective of DNA methylation events in other tissues. Additionally, the analysis of global DNA was a low resolution assessment of total CpG methylation and the microarray and Sat2 analyses assessed only a fraction of the entire methylome. Therefore these results do not provide an absolute assurance that detrimental changes in methylation did not occur.

These scenarios are, however, unlikely. When the results of this study are taken on balance, it is apparent that aberrant DNA methylation or gene expression were not prevalent outcomes of altered folic acid intake. In any case, the MTHFR C677T genotype or low folate status were not necessarily reliable indicators of changes in DNA methylation. In fact, DNA methylation – global DNA methylation in particular – was exceptionally well conserved in this group of healthy New Zealanders.

187

188

Appendices

Appendix A

List of foods containing added folic acid. This list was given to study participants at beginning of the intervention trial. The list was current for the duration of the trial. Amounts of added folic acid (µg / 100 g) are also listed below where known.

Breakfast cereals Basics (Countdown)

Cocoa Poppas 333 Mini Wheats 333 Power Grain 167 Nutri-Grain 167 Rice Poppas 333 Rice Bubbles 167 Special K 333 Hubbards Sultana Bran 222 Vortex 167 Sustain 222 Home Sweet Home 167 Wheat Biscuits 333 Bugs 'n' Mud 333 K* Pows 167 Bountiful Breakfast 167

Cocoa Cornflakes 333 Pams Gisbourne Gold cornflakes 333 Corn Flakes 333 Rice Pops 167 Nutra Bites 333

Kelloggs Sanitarium All-Bran 222 Light 'n' Tasty 222 Bran Flakes 333 Maximise 333 Chex 167 Skippy 333 Coco Chex 167 Up & Go 40 Coco Pops 167 Weet Bix 333 Coco Slams 167 Weet Bix Oat Bran 250 Corn Flakes 333 Crunchy Nut Corn Flakes 167 Signature Range Frosties 167 Honey Nut cornflakes 200 Froot Loops 167 Cornflakes 333 Honey Rice Bubbles 167 Bran and Sultanas 222 Just Right 222 Power Stars 333

Breads Countdown Elfin In-store bakery breads Yes Soft white bread premix Yes

North's Tip Top Extra 200 Holsom's range 200 Mighty White 200

190

Meal substitutes Complan 75 Joe Weider Lean Nutrimeal 220 Fast Burner 300 NutraFibre Diet 156 UltraSlim 50 Healtheries Slim 360 Vitaplan 55

Snack bars Kelloggs Sanitarium K-time bars Yes Light ‘n’ Tasty cereal bars 167 LCM bars Yes Skippy cereal bars 167 Sustain bars Yes Uncle Toby’s Fruesli bars Yes

Spreads Sanitarium Marmite 2000

Supplements (µg/capsule) Apotex Kordel's Apo-folic 800 Hair and skin nutrition 100 Mega-time 300 Blackmores Men's multi 300 B complex 20 Senior-time 75 B12 Yes Super B executive 180 B plus C 50 Women's multi 300 Executive stress 150 Folic acid 300 Natural Nutrition Multi vitamin and mineral 20 Achievement plus Yes Multi for women 300 Executive anti-stress 100 Golden years seniors 100 Biodesign Mega vitamin 300 One a day multivitamin 100 Women's mega vitamin 200

Centrum Nature's Own From A to Zinc 200 B complex forte 75 Bug busters 10 Douglas dietary Chelated iron-plus 65 TMG folic 200 Hair therapeutic formula 100 Maxi B-100 100 Healtheries Mega B-150 150 Boost multi 300 Mega health for women 300 Boost women 300 Multi vitamin and mineral 300 Essentials for men / women 300 Super B complex 300 Executive B 150 Men's multi 300 Senior multi 300 Women's multi 300

191

Nutra Life Thompson's Active men's multi 300 Energise for men 150 Active women's multi 300 Extra B + C 300 Cardio-life 150 Fatigue relief 100 Executive B 100 Femmefort 300 Women's iron complete 200 Immunofort 50 Men's multi 300 Radiance Multifort 300 Multi powder 200 Organic iron complete 100 Vitamin B complex 400 Stress formula 100 Women's iron complex 200 Ultra-B 100 Women's multi 300 Red Seal Everyday multi 100 Unichem Executive stress B 100 Men's multi 100 Men's multi 300 Women's multi 300 One a day multi 100 Women's multi 300

192

Appendix B

Study information sheet for participants, questionnaire and consent forms.

INFORMATION SHEET FOR PARTICIPANTS

Thank you for showing an interest in this project. We encourage you to read this information sheet carefully and to ask questions if clarification is needed. If you decide to participate we thank you. If you decide not to take part we thank you for considering our request.

Participation involves both screening and intervention.

Screening

Vitamin requirements may differ among people depending on a person’s genes. The enzyme methylenetetrahydrofolate reductase (or MTHFR for short) comes in three types. We would like to determine which type you have. This is a simple test in which we will take some cells from the inside your cheek by swabbing. Depending on your genotype, we may invite you to take part in a multivitamin study. All people who are screened will be advised of their MTHFR genotype.

Multivitamin intervention

Depending on your MTHFR genotype we may invite you to take part in a 6 month multivitamin

intervention study. We will provide the supplement that will contain vitamin B12 (50 µg),

vitamin B6 (10 mg), and riboflavin (1.6 mg) for the first 3 months after which the supplement will also contain 400µg folic acid. During the entire study period (6 months) we would ask you not to eat foods containing added folic acid. These are mainly fortified breakfast cereals and we will provide you with a list of foods to avoid. Three blood tests will be required, at the beginning, at 3 months, and at 6 months. From these samples we will provide you with test

results for blood folate and vitamin B12.

Who May Participate?

We are seeking 90 individuals (aged 18+) to participate in the multivitamin intervention. Women planning a pregnancy are not eligible to participate. Some diseases and medication use would also preclude participation (see Personal Information Sheet).

What Does Participation Involve?

You will be asked to consume a multivitamin supplement every day for six months and provide three blood samples (baseline, 3 months, and 6 months). Blood samples will be collected by a registered nurse in the Department of Human Nutrition clinic.

193

Time Required

Three sessions of 15 minutes each.

Possible Risks

You may experience slight pain during the drawing of blood and in rare cases bleeding around the puncture but this generally disappears in about one day.

What Personnel Data or Information will be collected on me?

We will be collecting contact details (address, telephone number, email) and personnel information regarding age, sex, and medication related to this study.

What use will be made of this personal information, who will have access to it, and how long will it be kept?

The purpose of collecting this information is so that we are able to describe the overall characteristics of the study group. The handwritten form will be kept in a locked cabinet and your contact details transferred onto a computer that is password protected for access by James Elworthy only. You will be given a unique study code (ID). Staff involved in laboratory tests and data analysis will only have access to your ID and information relevant to group descriptives (age, sex etc.); they will not be able to identify you by name. Results of this project will be published as group data only. At the end of the study personal information will be destroyed immediately except that, as required by the University's research policy, any raw data on which the results of the study depend will be retained in secure storage for five years, after which it will be destroyed.

If you have any questions about the study please contact:

James Elworthy 479 7868 or Tim Green 479 5068

Study reviewed and approved by the Ethics Committee of the University of Otago

194

CONSENT FORM FOR PARTICIPANTS (screening)

I have read the Information Sheet and understand the study. All my questions have been answered to my satisfaction. I am free to request further information at any stage.

I consent to:

having cheek cells swabbed for the purpose of folate genotype analysis

I know that:

My participation is entirely voluntary;

I will be advised of my MTHFR genotype together with recommendations regarding folate intake;

I agree to be contacted if I am eligible to take part in the multivitamin intervention;

......

(Signature of participant) (Date)

This project has been reviewed and approved by the Ethics Committee of the University of Otago

195

CONSENT FORM FOR PARTICIPANTS (intervention)

I have read the Information Sheet and understand the study. All my questions have been answered to my satisfaction. I am free to request further information at any stage.

I consent to: taking part in the multivitamin supplementation study for 6 months as described providing two overnight fasting blood samples

I know that:

My participation is entirely voluntary and that I am free to withdraw at any time;

I should withdraw from the study if I feel that I may have become pregnant no matter how remote the possibility and begin taking folic acid supplements immediately;

I may experience slight pain during the drawing of blood and in rare cases bruising around the puncture but this generally disappears in about one day;

I will receive blood test results for folate and vitamin B12;

The results of the project may be published but my anonymity will be preserved;

I will receive $5 for each visit to the clinic to cover parking costs

......

(Signature of participant) (Date)

This project has been reviewed and approved by the Ethics Committee of the University of Otago

196

PERSONAL INFORMATION

Title: Mr / Mrs / Ms / Dr Surname ………………………………………..

First name(s) ………………………………… Date of birth …………………………

Address …………………………………………………………………………………

………………………………………………………………………………

Telephone: Home …………… Work ……………….. Mobile …………………………

Email ……………………………………………………………………………………

Do you regularly eat fortified breakfast cereals (see list)? Yes / No

If yes, how often? (please circle) Daily / Up to 3 times a week / less than once a week

Are you taking vitamin supplements? Yes / No

If yes, please list them:

Would you be willing to stop eating foods and supplements containing folic acid for 6 months? Yes / No

Have you been diagnosed with any of the following or had gastric surgery? Yes / No

Epilepsy, caeliac disease, crohn’s disease, kidney disease, inflammatory bowel disease

Are you taking any of these drugs? Yes / No

Methotrexate, Sulfasalazine, Phenytoin, Bezalip retard, Fibalip

197

198

References

Abali, E.E., Skacel, N.E., Celikkaya, H., and Hsieh, Y.C. (2008). Regulation of human dihydrofolate reductase activity and expression. Vitam Horm 79, 267-292.

Abratte, C.M., Wang, W., Li, R., Moriarty, D.J., and Caudill, M.A. (2008). Folate intake and the MTHFR C677T genotype influence choline status in young Mexican American women. J Nutr Biochem 19, 158-165.

Adorjan, P., Distler, J., Lipscher, E., Model, F., Muller, J., Pelet, C., Braun, A., Florl, A.R., Gutig, D., Grabs, G., et al. (2002). Tumour class prediction and discovery by microarray-based DNA methylation analysis. Nucleic Acids Res 30, e21.

Ahuja, N., Mohan, A.L., Li, Q., Stolker, J.M., Herman, J.G., Hamilton, S.R., Baylin, S.B., and Issa, J.P. (1997). Association between CpG island methylation and microsatellite instability in colorectal cancer. Cancer Res 57, 3370-3374.

Albert, C.M., Cook, N.R., Gaziano, J.M., Zaharris, E., MacFadyen, J., Danielson, E., Buring, J.E., and Manson, J.E. (2008). Effect of folic acid and B vitamins on risk of cardiovascular events and total mortality among women at high risk for cardiovascular disease: a randomized trial. JAMA 299, 2027-2036.

Angier, R.B., Boothe, J.H., Hutchings, B.L., Mowat, J.H., Semb, J., Stokstad, E.L., Subbarow, Y., Waller, C.W., Cosulich, D.B., Fahrenbach, M.J., et al. (1945). Synthesis of a Compound Identical with the L. Casei Factor Isolated from Liver. Science 102, 227-228.

Antony, A.C. (1996). Folate receptors. Annu Rev Nutr 16, 501-521.

Arnott, D. (2001). Basics of Triple-Stage Quadrupole/Ion-Trap Mass Spectrometry: Precursor, Product and Neutral-Loss Scanning. Electrospray Ionisation and Nanospray Ionisation. In Proteome Research: Mass Spectrometry, P. James, ed. (New York, Springer-Verlag), pp. 11-31.

200

Ashfield-Watt, P.A., Pullin, C.H., Whiting, J.M., Clark, Z.E., Moat, S.J., Newcombe, R.G., Burr, M.L., Lewis, M.J., Powers, H.J., and McDowell, I.F. (2002). Methylenetetrahydrofolate reductase 677C-->T genotype modulates homocysteine responses to a folate-rich diet or a low-dose folic acid supplement: a randomized controlled trial. Am J Clin Nutr 76, 180-186.

Assaraf, Y.G. (2006). The role of multidrug resistance efflux transporters in antifolate resistance and folate homeostasis. Drug Resist Updat 9, 227-246.

Assaraf, Y.G. (2007). Molecular basis of antifolate resistance. Cancer Metastasis Rev 26, 153-181.

Auer, H., Lyianarachchi, S., Newsom, D., Klisovic, M.I., Marcucci, G., and Kornacker, K. (2003). Chipping away at the chip bias: RNA degradation in microarray analysis. Nat Genet 35, 292-293.

Austin, R.C., Lentz, S.R., and Werstuck, G.H. (2004). Role of hyperhomocysteinemia in endothelial dysfunction and atherothrombotic disease. Cell Death Differ 11 Suppl 1, S56-64.

Axume, J., Smith, S.S., Pogribny, I.P., Moriarty, D.J., and Caudill, M.A. (2007a). Global leukocyte DNA methylation is similar in African American and Caucasian women under conditions of controlled folate intake. Epigenetics 2, 66-68.

Axume, J., Smith, S.S., Pogribny, I.P., Moriarty, D.J., and Caudill, M.A. (2007b). The MTHFR 677TT genotype and folate intake interact to lower global leukocyte DNA methylation in young Mexican American women. Nutr Res 27, 1365-1317.

Baggott, J.E., Vaughn, W.H., Juliana, M.M., Eto, I., Krumdieck, C.L., and Grubbs, C.J. (1992). Effects of folate deficiency and supplementation on methylnitrosourea-induced rat mammary tumors. J Natl Cancer Inst 84, 1740-1744.

Bagley, P.J., and Selhub, J. (1998). A common mutation in the methylenetetrahydrofolate reductase gene is associated with an accumulation of formylated tetrahydrofolates in red blood cells. Proc Natl Acad Sci U S A 95, 13217- 13220.

201

Bailey, L.B., and Gregory, J.F., 3rd (1999). Polymorphisms of methylenetetrahydrofolate reductase and other enzymes: metabolic significance, risks and impact on folate requirement. J Nutr 129, 919-922.

Bailey, S.W., and Ayling, J.E. (2009). The extremely slow and variable activity of dihydrofolate reductase in human liver and its implications for high folic acid intake. Proc Natl Acad Sci U S A 106, 15424-15429.

Balaghi, M., and Wagner, C. (1993). DNA methylation in folate deficiency: use of CpG methylase. Biochem Biophys Res Commun 193, 1184-1190.

Barlow, D.P., Stoger, R., Herrmann, B.G., Saito, K., and Schweifer, N. (1991). The mouse insulin-like growth factor type-2 receptor is imprinted and closely linked to the Tme locus. Nature 349, 84-87.

Barton, C.A., Hacker, N.F., Clark, S.J., and O'Brien, P.M. (2008). DNA methylation changes in ovarian cancer: implications for early diagnosis, prognosis and treatment. Gynecol Oncol 109, 129-139.

Basten, G.P., Duthie, S.J., Pirie, L., Vaughan, N., Hill, M.H., and Powers, H.J. (2006). Sensitivity of markers of DNA stability and DNA repair activity to folate supplementation in healthy volunteers. Br J Cancer 94, 1942-1947.

Baylin, S.B., Herman, J.G., Graff, J.R., Vertino, P.M., and Issa, J.P. (1998). Alterations in DNA methylation: a fundamental aspect of neoplasia. Adv Cancer Res 72, 141-196.

Becker, K.G., Hosack, D.A., Dennis, G., Jr., Lempicki, R.A., Bright, T.J., Cheadle, C., and Engel, J. (2003). PubMatrix: a tool for multiplex literature mining. BMC Bioinformatics 4, 61.

Beetstra, S., Thomas, P., Salisbury, C., Turner, J., and Fenech, M. (2005). Folic acid deficiency increases chromosomal instability, chromosome 21 aneuploidy and sensitivity to radiation-induced micronuclei. Mutat Res 578, 317-326.

Bentley, D.R. (2006). Whole-genome re-sequencing. Curr Opin Genet Dev 16, 545-552.

202

Bhat, N.K., Thompson, C.B., Lindsten, T., June, C.H., Fujiwara, S., Koizumi, S., Fisher, R.J., and Papas, T.S. (1990). Reciprocal expression of human ETS1 and ETS2 genes during T-cell activation: regulatory role for the protooncogene ETS1. Proc Natl Acad Sci U S A 87, 3723-3727.

Bird, A. (2002). DNA methylation patterns and epigenetic memory. Genes Dev 16, 6- 21.

Birn, H. (2006). The kidney in vitamin B12 and folate homeostasis: characterization of receptors for tubular uptake of vitamins and carrier proteins. Am J Physiol Renal Physiol 291, F22-36.

Blom, H.J. (2009). Folic acid, methylation and neural tube closure in humans. Birth Defects Res A Clin Mol Teratol 85, 295-302.

Blount, B.C., and Ames, B.N. (1995). DNA damage in folate deficiency. Baillieres Clin Haematol 8, 461-478.

Blount, B.C., Mack, M.M., Wehr, C.M., MacGregor, J.T., Hiatt, R.A., Wang, G., Wickramasinghe, S.N., Everson, R.B., and Ames, B.N. (1997). Folate deficiency causes uracil misincorporation into human DNA and chromosome breakage: implications for cancer and neuronal damage. Proc Natl Acad Sci U S A 94, 3290-3295.

Bock, C., Paulsen, M., Tierling, S., Mikeska, T., Lengauer, T., and Walter, J. (2006). CpG island methylation in human lymphocytes is highly correlated with DNA sequence, repeats, and predicted DNA structure. PLoS Genet 2, e26.

Boorman, G.A., Blackshear, P.E., Parker, J.S., Lobenhofer, E.K., Malarkey, D.E., Vallant, M.K., Gerken, D.K., and Irwin, R.D. (2005). Hepatic gene expression changes throughout the day in the Fischer rat: implications for toxicogenomic experiments. Toxicol Sci 86, 185-193.

Bottiglieri, T. (2005). Homocysteine and folate metabolism in depression. Prog Neuropsychopharmacol Biol Psychiatry 29, 1103-1112.

Botto, L.D., and Yang, Q. (2000). 5,10-Methylenetetrahydrofolate reductase gene variants and congenital anomalies: a HuGE review. Am J Epidemiol 151, 862-877.

203

Boumil, R.M., and Lee, J.T. (2001). Forty years of decoding the silence in X- chromosome inactivation. Hum Mol Genet 10, 2225-2232.

Bouzinba-Segard, H., Guais, A., and Francastel, C. (2006). Accumulation of small murine minor satellite transcripts leads to impaired centromeric architecture and function. Proc Natl Acad Sci U S A 103, 8709-8714.

Bowtell, D., and Sambrook, J., eds. (2003). DNA Microarrays (Cold Spring Harbor, New York, Cold Spring Harbor Laboratory Press).

Bowtell, D.D. (1987). Rapid isolation of eukaryotic DNA. Anal Biochem 162, 463-465.

Bray, J.J. (1999). Lecture notes on human physiology, 4th ed. edn (Malden, Mass, Blackwell Science).

Brent, R.L., and Oakley, G.P., Jr. (2006). Triumph and/or tragedy: the present Food and Drug Administration program of enriching grains with folic acid. Pediatrics 117, 930- 932.

Brinkman, A.B., Pennings, S.W., Braliou, G.G., Rietveld, L.E., and Stunnenberg, H.G. (2007). DNA methylation immediately adjacent to active histone marking does not silence transcription. Nucleic Acids Res 35, 801-811.

Brooks, D.E. (1973). The Effect of Neutral Polymers on the Electrokinetic Potential of Cells and Other Charged Particles. J Colliod Iinterf Sci 43, 700-713.

Brouha, B., Schustak, J., Badge, R.M., Lutz-Prigge, S., Farley, A.H., Moran, J.V., and Kazazian, H.H., Jr. (2003). Hot L1s account for the bulk of retrotransposition in the human population. Proc Natl Acad Sci U S A 100, 5280-5285.

Bruce, S., Hannula-Jouppi, K., Lindgren, C.M., Lipsanen-Nyman, M., and Kere, J. (2008). Restriction site-specific methylation studies of imprinted genes with quantitative real-time PCR. Clin Chem 54, 491-499.

Burdge, G.C., Lillycrop, K.A., Jackson, A.A., Gluckman, P.D., and Hanson, M.A. (2008). The nature of the growth pattern and of the metabolic response to fasting in the rat are dependent upon the dietary protein and folic acid intakes of their pregnant dams and post-weaning fat consumption. Br J Nutr 99, 540-549.

204

Burren, K.A., Savery, D., Massa, V., Kok, R.M., Scott, J.M., Blom, H.J., Copp, A.J., and Greene, N.D. (2008). Gene-environment interactions in the causation of neural tube defects: folate deficiency increases susceptibility conferred by loss of Pax3 function. Hum Mol Genet 17, 3675-3685.

Carmel, R. (2005). Folic Acid. In Modern Nutrition in Health and Disease, M.E. Shils, M. Shike, A.C. Ross, B. Caballero, and R.J. Cousins, eds. (Philadelphia, Lippincott Williams & Wilkins), pp. 470-480.

Carninci, P., Sandelin, A., Lenhard, B., Katayama, S., Shimokawa, K., Ponjavic, J., Semple, C.A., Taylor, M.S., Engstrom, P.G., Frith, M.C., et al. (2006). Genome-wide analysis of mammalian promoter architecture and evolution. Nat Genet 38, 626-635.

Carrel, L., and Willard, H.F. (2005). X-inactivation profile reveals extensive variability in X-linked gene expression in females. Nature 434, 400-404.

Castro, R., Rivera, I., Ravasco, P., Camilo, M.E., Jakobs, C., Blom, H.J., and de Almeida, I.T. (2004). 5,10-methylenetetrahydrofolate reductase (MTHFR) 677C-->T and 1298A-->C mutations are associated with DNA hypomethylation. J Med Genet 41, 454-458.

Castro, R., Rivera, I., Struys, E.A., Jansen, E.E., Ravasco, P., Camilo, M.E., Blom, H.J., Jakobs, C., and Tavares de Almeida, I. (2003). Increased homocysteine and S- adenosylhomocysteine concentrations and DNA hypomethylation in vascular disease. Clin Chem 49, 1292-1296.

Caudill, M.A. (2010). Folate bioavailability: implications for establishing dietary recommendations and optimizing status. Am J Clin Nutr.

Chandler, C.J., Wang, T.T., and Halsted, C.H. (1986). Pteroylpolyglutamate hydrolase from human jejunal brush borders. Purification and characterization. J Biol Chem 261, 928-933.

Charles, D., Ness, A.R., Campbell, D., Davey Smith, G., and Hall, M.H. (2004). Taking folate in pregnancy and risk of maternal breast cancer. BMJ 329, 1375-1376.

205

Chen, J., Giovannucci, E., Kelsey, K., Rimm, E.B., Stampfer, M.J., Colditz, G.A., Spiegelman, D., Willett, W.C., and Hunter, D.J. (1996). A methylenetetrahydrofolate reductase polymorphism and the risk of colorectal cancer. Cancer Res 56, 4862-4864.

Chen, R.Z., Pettersson, U., Beard, C., Jackson-Grusby, L., and Jaenisch, R. (1998). DNA hypomethylation leads to elevated mutation rates. Nature 395, 89-93.

Cheng, X., and Roberts, R.J. (2001). AdoMet-dependent methylation, DNA methyltransferases and base flipping. Nucleic Acids Res 29, 3784-3795.

Chiang, P.K., Gordon, R.K., Tal, J., Zeng, G.C., Doctor, B.P., Pardhasaradhi, K., and McCann, P.P. (1996). S-Adenosylmethionine and methylation. FASEB J 10, 471-480.

Cho, E., Hunter, D.J., Spiegelman, D., Albanes, D., Beeson, W.L., van den Brandt, P.A., Colditz, G.A., Feskanich, D., Folsom, A.R., Fraser, G.E., et al. (2006). Intakes of vitamins A, C and E and folate and multivitamins and lung cancer: a pooled analysis of 8 prospective studies. Int J Cancer 118, 970-978.

Choi, S.W., and Mason, J.B. (2000). Folate and carcinogenesis: an integrated scheme. J Nutr 130, 129-132.

Christensen, B., Arbour, L., Tran, P., Leclerc, D., Sabbaghian, N., Platt, R., Gilfix, B.M., Rosenblatt, D.S., Gravel, R.A., Forbes, P., et al. (1999). Genetic polymorphisms in methylenetetrahydrofolate reductase and methionine synthase, folate levels in red blood cells, and risk of neural tube defects. Am J Med Genet 84, 151-157.

Christensen, K.E., and Mackenzie, R.E. (2008). Mitochondrial methylenetetrahydrofolate dehydrogenase, methenyltetrahydrofolate cyclohydrolase, and formyltetrahydrofolate synthetases. Vitam Horm 79, 393-410.

Chuaqui, R.F., Bonner, R.F., Best, C.J., Gillespie, J.W., Flaig, M.J., Hewitt, S.M., Phillips, J.L., Krizman, D.B., Tangrea, M.A., Ahram, M., et al. (2002). Post-analysis follow-up and validation of microarray experiments. Nat Genet 32 Suppl, 509-514.

Clark, S.J., Statham, A., Stirzaker, C., Molloy, P.L., and Frommer, M. (2006). DNA methylation: bisulphite modification and analysis. Nat Protoc 1, 2353-2364.

206

Clarke, R., Harrison, G., and Richards, S. (2003). Effect of vitamins and aspirin on markers of platelet activation, oxidative stress and homocysteine in people at high risk of dementia. J Intern Med 254, 67-75.

Clarke, R., Smith, A.D., Jobst, K.A., Refsum, H., Sutton, L., and Ueland, P.M. (1998). Folate, vitamin B12, and serum total homocysteine levels in confirmed Alzheimer disease. Arch Neurol 55, 1449-1455.

Clemson, C.M., McNeil, J.A., Willard, H.F., and Lawrence, J.B. (1996). XIST RNA paints the inactive X chromosome at interphase: evidence for a novel RNA involved in nuclear/chromosome structure. J Cell Biol 132, 259-275.

Centers for Disease Control and Prevention. (1992). Recommendations for the use of folic acid to reduce the number of cases of spina bifida and other neural tube defects. MMWR Recomm Rep 41, 1-7.

Cogram, P., Hynes, A., Dunlevy, L.P., Greene, N.D., and Copp, A.J. (2004). Specific isoforms of protein kinase C are essential for prevention of folate-resistant neural tube defects by inositol. Hum Mol Genet 13, 7-14.

Cole, B.F., Baron, J.A., Sandler, R.S., Haile, R.W., Ahnen, D.J., Bresalier, R.S., McKeown-Eyssen, G., Summers, R.W., Rothstein, R.I., Burke, C.A., et al. (2007). Folic acid for the prevention of colorectal adenomas: a randomized clinical trial. JAMA 297, 2351-2359.

Collins, S.J. (2002). The role of retinoids and retinoic acid receptors in normal hematopoiesis. Leukemia 16, 1896-1905.

Conway, K.E., McConnell, B.B., Bowring, C.E., Donald, C.D., Warren, S.T., and Vertino, P.M. (2000). TMS1, a novel proapoptotic caspase recruitment domain protein, is a target of methylation-induced gene silencing in human breast cancers. Cancer Res 60, 6236-6242.

Cooney, C.A., Dave, A.A., and Wolff, G.L. (2002). Maternal methyl supplements in mice affect epigenetic variation and DNA methylation of offspring. J Nutr 132, 2393S- 2400S.

207

Copp, A.J., and Greene, N.D. (2010). Genetics and development of neural tube defects. J Pathol 220, 217-230.

Coppen, A., and Bailey, J. (2000). Enhancement of the antidepressant action of fluoxetine by folic acid: a randomised, placebo controlled trial. J Affect Disord 60, 121- 130.

Costello, J.F., and Brena, R.M. (2008). Cancer Epigenetics. In Epigenetics, J. Tost, ed. (Norfolk, Caister Academic Press), pp. 303-330.

Costello, J.F., Fruhwald, M.C., Smiraglia, D.J., Rush, L.J., Robertson, G.P., Gao, X., Wright, F.A., Feramisco, J.D., Peltomaki, P., Lang, J.C., et al. (2000). Aberrant CpG- island methylation has non-random and tumour-type-specific patterns. Nat Genet 24, 132-138.

Costello, J.F., Hong, C., Plass, C., and Smiraglia, D.J. (2009). Restriction landmark genomic scanning: analysis of CpG islands in genomes by 2D gel electrophoresis. Methods Mol Biol 507, 131-148.

Crain, P.F. (1990). Preparation and enzymatic hydrolysis of DNA and RNA for mass spectrometry. Methods Enzymol 193, 782-790.

Cropley, J.E., Suter, C.M., Beckman, K.B., and Martin, D.I. (2006). Germ-line epigenetic modification of the murine A vy allele by nutritional supplementation. Proc Natl Acad Sci U S A 103, 17308-17312.

Crott, J.W., Liu, Z., Choi, S.W., and Mason, J.B. (2007). Folate depletion in human lymphocytes up-regulates p53 expression despite marked induction of strand breaks in exons 5-8 of the gene. Mutat Res 626, 171-179.

Crott, J.W., Mashiyama, S.T., Ames, B.N., and Fenech, M.F. (2001). Methylenetetrahydrofolate reductase C677T polymorphism does not alter folic acid deficiency-induced uracil incorporation into primary human lymphocyte DNA in vitro. Carcinogenesis 22, 1019-1025.

208

Cui, H., Niemitz, E.L., Ravenel, J.D., Onyango, P., Brandenburg, S.A., Lobanenkov, V.V., and Feinberg, A.P. (2001). Loss of imprinting of insulin-like growth factor-II in Wilms' tumor commonly involves altered methylation but not mutations of CTCF or its binding site. Cancer Res 61, 4947-4950.

Cui, H., Onyango, P., Brandenburg, S., Wu, Y., Hsieh, C.L., and Feinberg, A.P. (2002). Loss of imprinting in colorectal cancer linked to hypomethylation of H19 and IGF2. Cancer Res 62, 6442-6446.

Czeizel, A.E., and Dudas, I. (1992). Prevention of the first occurrence of neural-tube defects by periconceptional vitamin supplementation. N Engl J Med 327, 1832-1835.

Dahl, C., and Guldberg, P. (2003). DNA methylation analysis techniques. Biogerontology 4, 233-250. de Hoffmann, E., and Stroobant, V. (2007). Mass spectrometry: principles and applications (Chichester, John Wiley & Sons). de Vogel, S., Wouters, K.A., Gottschalk, R.W., van Schooten, F.J., de Goeij, A.F., de Bruine, A.P., Goldbohm, R.A., van den Brandt, P.A., Weijenberg, M.P., and van Engeland, M. (2009). Genetic variants of methyl metabolizing enzymes and epigenetic regulators: associations with promoter CpG island hypermethylation in colorectal cancer. Cancer Epidemiol Biomarkers Prev 18, 3086-3096.

Deininger, P.L., Jolly, D.J., Rubin, C.M., Friedmann, T., and Schmid, C.W. (1981). Base sequence studies of 300 nucleotide renatured repeated human DNA clones. J Mol Biol 151, 17-33.

Dekker, G.A., de Vries, J.I., Doelitzsch, P.M., Huijgens, P.C., von Blomberg, B.M., Jakobs, C., and van Geijn, H.P. (1995). Underlying disorders associated with severe early-onset preeclampsia. Am J Obstet Gynecol 173, 1042-1048.

Diehn, M., Sherlock, G., Binkley, G., Jin, H., Matese, J.C., Hernandez-Boussard, T., Rees, C.A., Cherry, J.M., Botstein, D., Brown, P.O., et al. (2003). SOURCE: a unified genomic resource of functional annotations, ontologies, and gene expression data. Nucleic Acids Res 31, 219-223.

209

Dirx, M.J., van den Brandt, P.A., Goldbohm, R.A., and Lumey, L.H. (2003). Energy restriction early in life and colon carcinoma risk: results of The Netherlands Cohort Study after 7.3 years of follow-up. Cancer 97, 46-55.

Dobbin, K., Shih, J.H., and Simon, R. (2003). Statistical design of reverse dye microarrays. Bioinformatics 19, 803-810.

Down, T.A., Rakyan, V.K., Turner, D.J., Flicek, P., Li, H., Kulesha, E., Graf, S., Johnson, N., Herrero, J., Tomazou, E.M., et al. (2008). A Bayesian deconvolution strategy for immunoprecipitation-based DNA methylome analysis. Nat Biotechnol 26, 779-785.

Du, M., Beatty, L.G., Zhou, W., Lew, J., Schoenherr, C., Weksberg, R., and Sadowski, P.D. (2003). Insulator and silencer sequences in the imprinted region of human chromosome 11p15.5. Hum Mol Genet 12, 1927-1939.

Dulac, C. (2010). Brain function and chromatin plasticity. Nature 465, 728-735.

Durand, P., Prost, M., Loreau, N., Lussier-Cacan, S., and Blache, D. (2001). Impaired homocysteine metabolism and atherothrombotic disease. Lab Invest 81, 645-672.

Duthie, S.J., Grant, G., Pirie, L.P., Watson, A.J., and Margison, G.P. (2010). Folate deficiency alters hepatic and colon MGMT and OGG-1 DNA repair protein expression in rats but has no effect on genome-wide DNA methylation. Cancer Prev Res (Phila Pa) 3, 92-100.

Duthie, S.J., Mavrommatis, Y., Rucklidge, G., Reid, M., Duncan, G., Moyer, M.P., Pirie, L.P., and Bestwick, C.S. (2008). The response of human colonocytes to folate deficiency in vitro: functional and proteomic analyses. J Proteome Res 7, 3254-3266.

Duthie, S.J., Narayanan, S., Blum, S., Pirie, L., and Brand, G.M. (2000). Folate deficiency in vitro induces uracil misincorporation and DNA hypomethylation and inhibits DNA excision repair in immortalized normal human colon epithelial cells. Nutr Cancer 37, 245-251.

Duthie, S.J., Narayanan, S., Brand, G.M., Pirie, L., and Grant, G. (2002). Impact of folate deficiency on DNA stability. J Nutr 132, 2444S-2449S.

210

Eads, C.A., Danenberg, K.D., Kawakami, K., Saltz, L.B., Blake, C., Shibata, D., Danenberg, P.V., and Laird, P.W. (2000). MethyLight: a high-throughput assay to measure DNA methylation. Nucleic Acids Res 28, E32.

Eady, J.J., Wortley, G.M., Wormstone, Y.M., Hughes, J.C., Astley, S.B., Foxall, R.J., Doleman, J.F., and Elliott, R.M. (2005). Variation in gene expression profiles of peripheral blood mononuclear cells from healthy volunteers. Physiol Genomics 22, 402- 411.

Ebbing, M., Bonaa, K.H., Nygard, O., Arnesen, E., Ueland, P.M., Nordrehaug, J.E., Rasmussen, K., Njolstad, I., Refsum, H., Nilsen, D.W., Tverdal, A., Meyer, K., Vollset, S. E. (2009). Cancer incidence and mortality after treatment with folic acid and vitamin B12. JAMA 302, 2119-2126.

Ehrlich, M. (2003). The ICF syndrome, a DNA methyltransferase 3B deficiency and immunodeficiency disease. Clinical Immunology 109, 17–28.

Ehrlich, M., Jackson, K., and Weemaes, C. (2006). Immunodeficiency, centromeric region instability, facial anomalies syndrome (ICF). Orphanet J Rare Dis 1, 2.

Ehrlich, M., Sanchez, C., Shao, C., Nishiyama, R., Kehrl, J., Kuick, R., Kubota, T., and Hanash, S.M. (2008). ICF, an immunodeficiency syndrome: DNA methyltransferase 3B involvement, chromosome anomalies, and gene dysregulation. Autoimmunity 41, 253- 271.

Eichholzer, M., Tonz, O., and Zimmermann, R. (2006). Folic acid: a public-health challenge. Lancet 367, 1352-1361.

Einstein, F., Thompson, R.F., Bhagat, T.D., Fazzari, M.J., Verma, A., Barzilai, N., and Greally, J.M. (2010). Cytosine methylation dysregulation in neonates following intrauterine growth restriction. PLoS One 5, e8887.

Erickson, J.D. (2002). Folic acid and prevention of spina bifida and anencephaly. 10 years after the U.S. Public Health Service recommendation. MMWR Recomm Rep 51, 1-3.

211

Ericson, U., Sonestedt, E., Gullberg, B., Olsson, H., and Wirfalt, E. (2007). High folate intake is associated with lower breast cancer incidence in postmenopausal women in the Malmo Diet and Cancer cohort. Am J Clin Nutr 86, 434-443.

Eskes, T.K. (1998). Neural tube defects, vitamins and homocysteine. Eur J Pediatr 157 Suppl 2, S139-141.

Esteller, M., Risques, R.A., Toyota, M., Capella, G., Moreno, V., Peinado, M.A., Baylin, S.B., and Herman, J.G. (2001). Promoter hypermethylation of the DNA repair gene O(6)-methylguanine-DNA methyltransferase is associated with the presence of G:C to A:T transition mutations in p53 in human colorectal tumorigenesis. Cancer Res 61, 4689-4692.

Esteller, M., Silva, J.M., Dominguez, G., Bonilla, F., Matias-Guiu, X., Lerma, E., Bussaglia, E., Prat, J., Harkes, I.C., Repasky, E.A., et al. (2000). Promoter hypermethylation and BRCA1 inactivation in sporadic breast and ovarian tumors. J Natl Cancer Inst 92, 564-569.

Fan, H., and Hegde, P.S. (2005). The transcriptome in blood: challenges and solutions for robust expression profiling. Curr Mol Med 5, 3-10.

Fazzari, M.J., and Greally, J.M. (2010). Introduction to epigenomics and epigenome- wide analysis. Methods Mol Biol 620, 243-265.

Feinberg, A.P., and Vogelstein, B. (1983). Hypomethylation of ras oncogenes in primary human cancers. Biochem Biophys Res Commun 111, 47-54.

Fenech, M. (2001). The role of folic acid and Vitamin B12 in genomic stability of human cells. Mutat Res 475, 57-67.

Fenech, M., Aitken, C., and Rinaldi, J. (1998). Folate, vitamin B12, homocysteine status and DNA damage in young Australian adults. Carcinogenesis 19, 1163-1171.

Ferguson-Smith, A.C., Sasaki, H., Cattanach, B.M., and Surani, M.A. (1993). Parental- origin-specific epigenetic modification of the mouse H19 gene. Nature 362, 751-755.

212

Ferguson, A.T., Evron, E., Umbricht, C.B., Pandita, T.K., Chan, T.A., Hermeking, H., Marks, J.R., Lambers, A.R., Futreal, P.A., Stampfer, M.R., et al. (2000). High frequency of hypermethylation at the 14-3-3 sigma locus leads to gene silencing in breast cancer. Proc Natl Acad Sci U S A 97, 6049-6054.

Field, M.S., Szebenyi, D.M., and Stover, P.J. (2006). Regulation of de novo purine biosynthesis by methenyltetrahydrofolate synthetase in neuroblastoma. J Biol Chem 281, 4215-4221.

Finkelstein, J.D. (2007). Metabolic regulatory properties of S-adenosylmethionine and S-adenosylhomocysteine. Clin Chem Lab Med 45, 1694-1699.

Finkelstein, J.D., and Martin, J.J. (1986). Methionine metabolism in mammals. Adaptation to methionine excess. J Biol Chem 261, 1582-1587.

Fontecave, M., Atta, M., and Mulliez, E. (2004). S-adenosylmethionine: nothing goes to waste. Trends Biochem Sci 29, 243-249.

Food Standards Australia and New Zealand (2007). Final Assessment Report, Proposal P295, Consideration of Mandatory Fortification with Folic Acid (Canberra). Available at http://www.foodstandards.gov.au/standardsdevelopment/proposals/proposalp295 considerationofmandatoryfortificationwithfolicacid/index.cfm.

Forrest, J.D. (1994). Epidemiology of unintended pregnancy and contraceptive use. Am J Obstet Gynecol 170, 1485-1489.

Fox, J.T., and Stover, P.J. (2008). Folate-mediated one-carbon metabolism. Vitam Horm 79, 1-44.

Fraga, M.F., Ballestar, E., Paz, M.F., Ropero, S., Setien, F., Ballestar, M.L., Heine- Suner, D., Cigudosa, J.C., Urioste, M., Benitez, J., Boix-Chornet, M., Sanchez- Aguilera, A., Ling, C., Carlsson, E., Poulsen, P., Vaag, A., Stephan, Z., Spector, T. D., Wu, Y. Z., Plass, C., Esteller, M., (2005). Epigenetic differences arise during the lifetime of monozygotic twins. Proc Natl Acad Sci U S A 102, 10604-10609.

213

Fraga, M.F., Uriol, E., Borja Diego, L., Berdasco, M., Esteller, M., Canal, M.J., and Rodriguez, R. (2002). High-performance capillary electrophoretic method for the quantification of 5-methyl 2'-deoxycytidine in genomic DNA: application to plant, animal and human cancer tissues. Electrophoresis 23, 1677-1681.

Francis, D., Diorio, J., Liu, D., and Meaney, M.J. (1999). Nongenomic transmission across generations of maternal behavior and stress responses in the rat. Science 286, 1155-1158.

French, H.J., Attenborough, R., Hardy, K., Shannon, M.F., and Williams, R.B. (2009). Interindividual variation in epigenomic phenomena in humans. Mamm Genome 20, 604-611.

Frigola, J., Ribas, M., Risques, R.A., and Peinado, M.A. (2002). Methylome profiling of cancer cells by amplification of inter-methylated sites (AIMS). Nucleic Acids Res 30, e28.

Frigola, J., Song, J., Stirzaker, C., Hinshelwood, R.A., Peinado, M.A., and Clark, S.J. (2006). Epigenetic remodeling in colorectal cancer results in coordinate gene suppression across an entire chromosome band. Nat Genet 38, 540-549.

Friso, S., Choi, S.W., Dolnikowski, G.G., and Selhub, J. (2002a). A method to assess genomic DNA methylation using high-performance liquid chromatography/electrospray ionization mass spectrometry. Anal Chem 74, 4526-4531.

Friso, S., Choi, S.W., Girelli, D., Mason, J.B., Dolnikowski, G.G., Bagley, P.J., Olivieri, O., Jacques, P.F., Rosenberg, I.H., Corrocher, R., Selhub, J. (2002b). A common mutation in the 5,10-methylenetetrahydrofolate reductase gene affects genomic DNA methylation through an interaction with folate status. Proc Natl Acad Sci U S A 99, 5606-5611.

Frosst, P., Blom, H.J., Milos, R., Goyette, P., Sheppard, C.A., Matthews, R.G., Boers, G.J.H., denHeijer, M., Kluijtmans, L.A., van den Heuvel, L.P., Rosen, R. (1995). A candidate genetic risk factor for vascular disease: a common mutation in methylenetetrahydrofolate reductase. Nature Genetics 10, 111-113.

214

Frost, J.M., and Moore, G.E. (2010). The importance of imprinting in the human placenta. PLoS Genet 6, e1001015.

Fryxell, K.J., and Moon, W.J. (2005). CpG mutation rates in the human genome are highly dependent on local GC content. Mol Biol Evol 22, 650-658.

Fuchs, C.S., Willett, W.C., Colditz, G.A., Hunter, D.J., Stampfer, M.J., Speizer, F.E., and Giovannucci, E.L. (2002). The influence of folate and multivitamin use on the familial risk of colon cancer in women. Cancer Epidemiol Biomarkers Prev 11, 227- 234.

Fukushima, M., Morita, M., Ikeda, K., and Nagayama, S. (2003). Population study of expression of thymidylate synthase and dihydropyrimidine dehydrogenase in patients with solid tumors. Int J Mol Med 12, 839-844.

Fukuzawa, R., Breslow, N.E., Morison, I.M., Dwyer, P., Kusafuka, T., Kobayashi, Y., Becroft, D.M., Beckwith, J.B., Perlman, E.J., and Reeve, A.E. (2004). Epigenetic differences between Wilms' tumours in white and east-Asian children. Lancet 363, 446- 451.

Futscher, B.W., Oshiro, M.M., Wozniak, R.J., Holtan, N., Hanigan, C.L., Duan, H., and Domann, F.E. (2002). Role for DNA methylation in the control of cell type specific maspin expression. Nat Genet 31, 175-179.

Galushko, S.V., Shishkina, I.P., and Pilipenko, A.T. (1988). Reversed-phase high- performance liquid chromatography of pyrimidine bases and nucleosides. Application of solvophobic theory. J Chromatogr 445, 59-70.

Gama-Sosa, M.A., Slagel, V.A., Trewyn, R.W., Oxenhandler, R., Kuo, K.C., Gehrke, C.W., and Ehrlich, M. (1983). The 5-methylcytosine content of DNA from human tumors. Nucleic Acids Res 11, 6883-6894.

Gardiner-Garden, M., and Frommer, M. (1987). CpG islands in vertebrate genomes. J Mol Biol 196, 261-282.

Garrick, D., Fiering, S., Martin, D.I., and Whitelaw, E. (1998). Repeat-induced gene silencing in mammals. Nat Genet 18, 56-59.

215

Gaughan, D.J., Barbaux, S., Kluijtmans, L.A., and Whitehead, A.S. (2000). The human and mouse methylenetetrahydrofolate reductase (MTHFR) genes: genomic organization, mRNA structure and linkage to the CLCN6 gene. Gene 257, 279-289.

Gehrke, C.W., McCune, R.A., Gama-Sosa, M.A., Ehrlich, M., and Kuo, K.C. (1984). Quantitative reversed-phase high-performance liquid chromatography of major and modified nucleosides in DNA. J Chromatogr 301, 199-219.

Gibson, A., Woodside, J.V., Young, I.S., Sharpe, P.C., Mercer, C., Patterson, C.C., McKinley, M.C., Kluijtmans, L.A., Whitehead, A.S., and Evans, A. (2008). Alcohol increases homocysteine and reduces B vitamin concentration in healthy male volunteers--a randomized, crossover intervention study. QJM 101, 881-887.

Giovannucci, E., Stampfer, M.J., Colditz, G.A., Hunter, D.J., Fuchs, C., Rosner, B.A., Speizer, F.E., and Willett, W.C. (1998). Multivitamin use, folate, and colon cancer in women in the Nurses' Health Study. Ann Intern Med 129, 517-524.

Girard, L., and Freeling, M. (1999). Regulatory changes as a consequence of transposon insertion. Dev Genet 25, 291-296.

Girelli, D., Friso, S., Trabetti, E., Olivieri, O., Russo, C., Pessotto, R., Faccini, G., Pignatti, P.F., Mazzucco, A., and Corrocher, R. (1998). Methylenetetrahydrofolate reductase C677T mutation, plasma homocysteine, and folate in subjects from northern Italy with or without angiographically documented severe coronary atherosclerotic disease: evidence for an important genetic-environmental interaction. Blood 91, 4158- 4163.

Gitan, R.S., Shi, H., Chen, C.M., Yan, P.S., and Huang, T.H. (2002). Methylation- specific oligonucleotide microarray: a new potential for high-throughput methylation analysis. Genome Res 12, 158-164.

Goll, M.G., Kirpekar, F., Maggert, K.A., Yoder, J.A., Hsieh, C.L., Zhang, X., Golic, K.G., Jacobsen, S.E., and Bestor, T.H. (2006). Methylation of tRNAAsp by the DNA methyltransferase homolog Dnmt2. Science 311, 395-398.

216

Gomes, M.V., and Waterland, R.A. (2008). Individual epigenetic variation: when, why, and so what? Nestle Nutr Workshop Ser Pediatr Program 62, 141-150; discussion 151- 145.

Gorelick, R. (2004). Neo-Lamarckian medicine. Med Hypotheses 62, 299-303.

Goyette, P., Pai, A., Milos, R., Frosst, P., Tran, P., Chen, Z., Chan, M., and Rozen, R. (1998). Gene structure of human and mouse methylenetetrahydrofolate reductase (MTHFR). Mamm Genome 9, 652-656.

Goyette, P., Sumner, J.S., Milos, R., Duncan, A.M., Rosenblatt, D.S., Matthews, R.G., and Rozen, R. (1994). Human methylenetetrahydrofolate reductase: isolation of cDNA mapping and mutation identification. Nat Genet 7, 551.

Graham, I.M., and O'Callaghan, P. (2002). Vitamins, homocysteine and cardiovascular risk. Cardiovasc Drugs Ther 16, 383-389.

Greger, V., Passarge, E., Hopping, W., Messmer, E., and Horsthemke, B. (1989). Epigenetic changes may contribute to the formation and spontaneous regression of retinoblastoma. Hum Genet 83, 155-158.

Gregory, J.F., 3rd (1997). Bioavailability of folate. Eur J Clin Nutr 51 Suppl 1, S54-59.

Gregory, J.F., 3rd (2001). Case study: folate bioavailability. J Nutr 131, 1376S-1382S.

Grewal, S.I., and Rice, J.C. (2004). Regulation of heterochromatin by histone methylation and small RNAs. Curr Opin Cell Biol 16, 230-238.

Groenen, P.M., Peer, P.G., Wevers, R.A., Swinkels, D.W., Franke, B., Mariman, E.C., and Steegers-Theunissen, R.P. (2003). Maternal myo-inositol, glucose, and zinc status is associated with the risk of offspring with spina bifida. Am J Obstet Gynecol 189, 1713-1719.

Gronbaek, K., Hother, C., and Jones, P.A. (2007). Epigenetic changes in cancer. APMIS 115, 1039-1059.

217

Guinotte, C.L., Burns, M.G., Axume, J.A., Hata, H., Urrutia, T.F., Alamilla, A., McCabe, D., Singgih, A., Cogger, E.A., and Caudill, M.A. (2003). Methylenetetrahydrofolate reductase 677C-->T variant modulates folate status response to controlled folate intakes in young women. J Nutr 133, 1272-1280.

Habuchi, H., Miyake, G., Nogami, K., Kuroiwa, A., Matsuda, Y., Kusche-Gullberg, M., Habuchi, O., Tanaka, M., and Kimata, K. (2003). Biosynthesis of heparan sulphate with diverse structures and functions: two alternatively spliced forms of human heparan sulphate 6-O-sulphotransferase-2 having different expression patterns and properties. Biochem J 371, 131-142.

Haig, D., and Graham, C. (1991). Genomic imprinting and the strange case of the insulin-like growth factor II receptor. Cell 64, 1045-1046.

Hall, C.A., and Chu, R.C. (1990). Serum homocysteine in routine evaluation of potential vitamin B12 and folate deficiency. Eur J Haematol 45, 143-149.

Halsted, C.H. (1980). Intestinal absorption and malabsorption of folates. Annu Rev Med 31, 79-87.

Halsted, C.H., Villanueva, J.A., Devlin, A.M., and Chandler, C.J. (2002). Metabolic interactions of alcohol and folate. J Nutr 132, 2367S-2372S.

Handel, A.E., and Ramagopalan, S.V. (2010). Is Lamarckian evolution relevant to medicine? BMC Med Genet 11, 73.

Hartmann, L.C., Keeney, G.L., Lingle, W.L., Christianson, T.J., Varghese, B., Hillman, D., Oberg, A.L., and Low, P.S. (2007). Folate receptor overexpression is associated with poor outcome in breast cancer. Int J Cancer 121, 938-942.

Hatada, I., Kato, A., Morita, S., Obata, Y., Nagaoka, K., Sakurada, A., Sato, M., Horii, A., Tsujimoto, A., and Matsubara, K. (2002). A microarray-based method for detecting methylated loci. J Hum Genet 47, 448-451.

Havlis, J., and Trbusek, M. (2002). 5-Methylcytosine as a marker for the monitoring of DNA methylation. Journal of Chromatography B 781, 373-392.

218

Hegi, M.E., Diserens, A.C., Gorlia, T., Hamou, M.F., de Tribolet, N., Weller, M., Kros, J.M., Hainfellner, J.A., Mason, W., Mariani, L., et al. (2005). MGMT gene silencing and benefit from temozolomide in glioblastoma. N Engl J Med 352, 997-1003.

Heijmans, B.T., Tobi, E.W., Stein, A.D., Putter, H., Blauw, G.J., Susser, E.S., Slagboom, P.E., and Lumey, L.H. (2008). Persistent epigenetic differences associated with prenatal exposure to famine in humans. Proc Natl Acad Sci U S A 105, 17046- 17049.

Henderson, G.B., Tsuji, J.M., and Kumar, H.P. (1988). Mediated uptake of folate by a high-affinity binding protein in sublines of L1210 cells adapted to nanomolar concentrations of folate. J Membr Biol 101, 247-258.

Hercberg, S., and Galan, P. (1992). Nutritional anaemias. Baillieres Clin Haematol 5, 143-168.

Herman, J.G., Latif, F., Weng, Y., Lerman, M.I., Zbar, B., Liu, S., Samid, D., Duan, D.S., Gnarra, J.R., Linehan, W.M., et al. (1994). Silencing of the VHL tumor- suppressor gene by DNA methylation in renal carcinoma. Proc Natl Acad Sci U S A 91, 9700-9704.

Herman, J.G., Umar, A., Polyak, K., Graff, J.R., Ahuja, N., Issa, J.P., Markowitz, S., Willson, J.K., Hamilton, S.R., Kinzler, K.W., et al. (1998). Incidence and functional consequences of hMLH1 promoter hypermethylation in colorectal carcinoma. Proc Natl Acad Sci U S A 95, 6870-6875.

Hernandez-Blazquez, F.J., Habib, M., Dumollard, J.M., Barthelemy, C., Benchaib, M., de Capoa, A., and Niveleau, A. (2000). Evaluation of global DNA hypomethylation in human colon cancer tissues by immunohistochemistry and image analysis. Gut 47, 689- 693.

Hiltunen, M.O., Turunen, M.P., Hakkinen, T.P., Rutanen, J., Hedman, M., Makinen, K., Turunen, A.M., Aalto-Setala, K., and Yla-Herttuala, S. (2002). DNA hypomethylation and methyltransferase expression in atherosclerotic lesions. Vasc Med 7, 5-11.

Hiltunen, M.O., and Yla-Herttuala, S. (2003). DNA methylation, smooth muscle cells, and atherogenesis. Arterioscler Thromb Vasc Biol 23, 1750-1753.

219

Hoffbrand, A.V., and Weir, D.G. (2001). The history of folic acid. Br J Haematol 113, 579-589.

Hofmann, M., Harris, M., Juriloff, D., and Boehm, T. (1998). Spontaneous mutations in SELH/Bc mice due to insertions of early transposons: molecular characterization of null alleles at the nude and albino loci. Genomics 52, 107-109.

Holliday, R. (2006). Epigenetics: a historical overview. Epigenetics 1, 76-80.

Homocysteine Lowering Trialists' Collaboration (1998). Lowering blood homocysteine with folic acid based supplements: meta-analysis of randomised trials. BMJ 316, 894- 898.

Howard, G., Eiges, R., Gaudet, F., Jaenisch, R., and Eden, A. (2008). Activation and transposition of endogenous retroviral elements in hypomethylation induced tumors in mice. Oncogene 27, 404-408.

Huang da, W., Sherman, B.T., and Lempicki, R.A. (2009). Systematic and integrative analysis of large gene lists using DAVID bioinformatics resources. Nat Protoc 4, 44-57.

Huang da, W., Sherman, B.T., Stephens, R., Baseler, M.W., Lane, H.C., and Lempicki, R.A. (2008). DAVID gene ID conversion tool. Bioinformation 2, 428-430.

Huang, T.H., Perry, M.R., and Laux, D.E. (1999). Methylation profiling of CpG islands in human breast cancer cells. Hum Mol Genet 8, 459-470.

Hughes, L.A., van den Brandt, P.A., de Bruine, A.P., Wouters, K.A., Hulsmans, S., Spiertz, A., Goldbohm, R.A., de Goeij, A.F., Herman, J.G., Weijenberg, M.P., et al. (2009). Early life exposure to famine and colorectal cancer risk: a role for epigenetic mechanisms. PLoS One 4, e7951.

Hung, J., Yang, T.L., Urrutia, T.F., Li, R., Perry, C.A., Hata, H., Cogger, E.A., Moriarty, D.J., and Caudill, M.A. (2006). Additional food folate derived exclusively from natural sources improves folate status in young women with the MTHFR 677 CC or TT genotype. J Nutr Biochem 17, 728-734.

Ifergan, I., and Assaraf, Y.G. (2008). Molecular mechanisms of adaptation to folate deficiency. Vitam Horm 79, 99-143.

220

Imasa, M.S., Gomez, N.T., and Nevado, J.B., Jr. (2009). Folic acid-based intervention in non-ST elevation acute coronary syndromes. Asian Cardiovasc Thorac Ann 17, 13- 21.

Ingrosso, D., Cimmino, A., Perna, A.F., Masella, L., De Santo, N.G., De Bonis, M.L., Vacca, M., D'Esposito, M., D'Urso, M., Galletti, P., et al. (2003). Folate treatment and unbalanced methylation and changes of allelic expression induced by hyperhomocysteinaemia in patients with uraemia. Lancet 361, 1693-1699.

Institute of Medicine, The Standing Committee on the Scientific Evaluation of Dietary Reference Intakes and its Panel on Folate, other B vitamins, and Choline, and the Food and Nutrition Board. (2000). Dietary Reference Intakes for Thiamin, Riboflavin, Niacin, Vitamin B6, Folate, Vitamin B12, Pantothenic Acid, Biotin, and Choline (Washington, D.C., National Academies Press).

Issa, J.P. (2003). Age-related epigenetic changes and the immune system. Clin Immunol 109, 103-108.

Ito, Y., Koessler, T., Ibrahim, A.E., Rai, S., Vowler, S.L., Abu-Amero, S., Silva, A.L., Maia, A.T., Huddleston, J.E., Uribe-Lewis, S., et al. (2008). Somatically acquired hypomethylation of IGF2 in breast and colorectal cancer. Hum Mol Genet 17, 2633- 2643.

Jablonka, E., and Raz, G. (2009). Transgenerational epigenetic inheritance: prevalence, mechanisms, and implications for the study of heredity and evolution. Q Rev Biol 84, 131-176.

Jacob, R.A., Gretz, D.M., Taylor, P.C., James, S.J., Pogribny, I.P., Miller, B.J., Henning, S.M., and Swendseid, M.E. (1998). Moderate folate depletion increases plasma homocysteine and decreases lymphocyte DNA methylation in postmenopausal women. J Nutr 128, 1204-1212.

Jacques, P.F., Bostom, A.G., Williams, R.R., Ellison, R.C., Eckfeldt, J.H., Rosenberg, I.H., Selhub, J., and Rozen, R. (1996). Relation between folate status, a common mutation in methylenetetrahydrofolate reductase, and plasma homocysteine concentrations. Circulation 93, 7-9.

221

James, S.J., Pogribna, M., Pogribny, I.P., Melnyk, S., Hine, R.J., Gibson, J.B., Yi, P., Tafoya, D.L., Swenson, D.H., Wilson, V.L., et al. (1999). Abnormal folate metabolism and mutation in the methylenetetrahydrofolate reductase gene may be maternal risk factors for Down syndrome. Am J Clin Nutr 70, 495-501.

Jansen, G., Westerhof, G.R., Jarmuszewski, M.J., Kathmann, I., Rijksen, G., and Schornagel, J.H. (1990). Methotrexate transport in variant human CCRF-CEM leukemia cells with elevated levels of the reduced folate carrier. Selective effect on carrier- mediated transport of physiological concentrations of reduced folates. J Biol Chem 265, 18272-18277.

Jeanpierre, M. (1987). A rapid method for the purification of DNA from blood. Nucleic Acids Res 15, 9611.

Jeltsch, A. (2006). On the enzymatic properties of Dnmt1: specificity, processivity, mechanism of linear diffusion and allosteric regulation of the enzyme. Epigenetics 1, 63-66.

Jencks, D.A., and Mathews, R.G. (1987). Allosteric inhibition of methylenetetrahydrofolate reductase by adenosylmethionine. Effects of adenosylmethionine and NADPH on the equilibrium between active and inactive forms of the enzyme and on the kinetics of approach to equilibrium. J Biol Chem 262, 2485- 2493.

Joelson, D.W., Fiebig, E.W., and Wu, A.H. (2007). Diminished need for folate measurements among indigent populations in the post folic acid supplementation era. Arch Pathol Lab Med 131, 477-480.

Jones, P.A., and Baylin, S.B. (2002). The fundamental role of epigenetic events in cancer. Nat Rev Genet 3, 415-428.

Jones, P.A., and Laird, P.W. (1999). Cancer epigenetics comes of age. Nat Genet 21, 163-167.

Juriloff, D.M., and Harris, M.J. (2000). Mouse models for neural tube closure defects. Hum Mol Genet 9, 993-1000.

222

Kaati, G., Bygren, L.O., and Edvinsson, S. (2002). Cardiovascular and diabetes mortality determined by nutrition during parents' and grandparents' slow growth period. Eur J Hum Genet 10, 682-688.

Kafri, T., Ariel, M., Brandeis, M., Shemer, R., Urven, L., McCarrey, J., Cedar, H., and Razin, A. (1992). Developmental pattern of gene-specific DNA methylation in the mouse embryo and germ line. Genes Dev 6, 705-714.

Kamen, B.A., and Capdevila, A. (1986). Receptor-mediated folate accumulation is regulated by the cellular folate content. Proc Natl Acad Sci U S A 83, 5983-5987.

Kaminen-Ahola, N., Ahola, A., Maga, M., Mallitt, K.A., Fahey, P., Cox, T.C., Whitelaw, E., and Chong, S. (2010). Maternal ethanol consumption alters the epigenotype and the phenotype of offspring in a mouse model. PLoS Genet 6, e1000811.

Kamiya, H., Kawakami, K., Miyanaga, T., Omura, K., Oda, M., Murakami, S., and Watanabe, Y. (1998). A methylenetetrahydrofolate reductase polymorphism is associated with expression of p16 in human lung cancer. Oncol Rep 5, 911-914.

Kanehisa, M., Goto, S., Hattori, M., Aoki-Kinoshita, K.F., Itoh, M., Kawashima, S., Katayama, T., Araki, M., and Hirakawa, M. (2006). From genomics to chemical genomics: new developments in KEGG. Nucleic Acids Res 34, D354-357.

Kanellopoulou, C., Muljo, S.A., Kung, A.L., Ganesan, S., Drapkin, R., Jenuwein, T., Livingston, D.M., and Rajewsky, K. (2005). Dicer-deficient mouse embryonic stem cells are defective in differentiation and centromeric silencing. Genes Dev 19, 489-501.

Kang, S.S., Wong, P.W., Susmano, A., Sora, J., Norusis, M., and Ruggie, N. (1991). Thermolabile methylenetetrahydrofolate reductase: an inherited risk factor for coronary artery disease. Am J Hum Genet 48, 536-545.

Kangaspeska, S., Stride, B., Metivier, R., Polycarpou-Schwarz, M., Ibberson, D., Carmouche, R.P., Benes, V., Gannon, F., and Reid, G. (2008). Transient cyclical methylation of promoter DNA. Nature 452, 112-115.

223

Katzenellenbogen, R.A., Baylin, S.B., and Herman, J.G. (1999). Hypermethylation of the DAP-kinase CpG island is a common alteration in B-cell malignancies. Blood 93, 4347-4353.

Kawasaki, H., and Taira, K. (2004). Induction of DNA methylation and gene silencing by short interfering RNAs in human cells. Nature 431, 211-217.

Kay, G.F., Penny, G.D., Patel, D., Ashworth, A., Brockdorff, N., and Rastan, S. (1993). Expression of Xist during mouse development suggests a role in the initiation of X chromosome inactivation. Cell 72, 171-182.

Kazakevich, Y., and LoBrutto, R. (2007). HPLC for pharmaceutical scientists (Wiley- Interscience).

Keijzer, M.B., den Heijer, M., Blom, H.J., Bos, G.M., Willems, H.P., Gerrits, W.B., and Rosendaal, F.R. (2002). Interaction between hyperhomocysteinemia, mutated methylenetetrahydrofolatereductase (MTHFR) and inherited thrombophilic factors in recurrent venous thrombosis. Thromb Haemost 88, 723-728.

Kelemen, L.E. (2006). The role of folate receptor alpha in cancer development, progression and treatment: cause, consequence or innocent bystander? Int J Cancer 119, 243-250.

Kelemen, L.E., Sellers, T.A., Vierkant, R.A., Harnack, L., and Cerhan, J.R. (2004). Association of folate and alcohol with risk of ovarian cancer in a prospective study of postmenopausal women. Cancer Causes Control 15, 1085-1093.

Kerkel, K., Spadola, A., Yuan, E., Kosek, J., Jiang, L., Hod, E., Li, K., Murty, V.V., Schupf, N., Vilain, E., et al. (2008). Genomic surveys by methylation-sensitive SNP analysis identify sequence-dependent allele-specific DNA methylation. Nat Genet 40, 904-908.

Keshet, I., Schlesinger, Y., Farkash, S., Rand, E., Hecht, M., Segal, E., Pikarski, E., Young, R.A., Niveleau, A., Cedar, H., et al. (2006). Evidence for an instructive mechanism of de novo methylation in cancer cells. Nat Genet 38, 149-153.

224

Khan, S.N., and Khan, A.U. (2010). Role of histone acetylation in cell physiology and diseases: An update. Clin Chim Acta.

Kim, B.H., Cho, N.Y., Shin, S.H., Kwon, H.J., Jang, J.J., and Kang, G.H. (2009). CpG island hypermethylation and repetitive DNA hypomethylation in premalignant lesion of extrahepatic cholangiocarcinoma. Virchows Arch 455, 343-351.

Kim, H.R., Hwang, K.A., Kim, K.C., and Kang, I. (2007). Down-regulation of IL- 7Ralpha expression in human T cells via DNA methylation. J Immunol 178, 5473-5479.

Kim, S.J., Lee, S.Y., Lee, C., Kim, I., An, H.J., Kim, J.Y., Baek, K.H., Kim, E.J., Kim, J.M., Lee, J.B., et al. (2006). Differential expression profiling of genes in a complete hydatidiform mole using cDNA microarray analysis. Gynecol Oncol 103, 654-660.

Kim, Y.I. (2003). Role of folate in colon cancer development and progression. J Nutr 133, 3731S-3739S.

Kim, Y.I. (2004). Folate, colorectal carcinogenesis, and DNA methylation: lessons from animal studies. Environ Mol Mutagen 44, 10-25.

Kim, Y.I. (2005). Nutritional epigenetics: impact of folate deficiency on DNA methylation and colon cancer susceptibility. J Nutr 135, 2703-2709.

Kim, Y.I. (2007a). Folate and colorectal cancer: an evidence-based critical review. Mol Nutr Food Res 51, 267-292.

Kim, Y.I. (2007b). Folic acid fortification and supplementation--good for some but not so good for others. Nutr Rev 65, 504-511.

Kim, Y.I. (2008). Folic acid supplementation and cancer risk: point. Cancer Epidemiol Biomarkers Prev 17, 2220-2225.

Kimura, M., Umegaki, K., Higuchi, M., Thomas, P., and Fenech, M. (2004). Methylenetetrahydrofolate reductase C677T polymorphism, folic acid and riboflavin are important determinants of genome stability in cultured human lymphocytes. J Nutr 134, 48-56.

225

Klerk, M., Verhoef, P., Clarke, R., Blom, H.J., Kok, F.J., and Schouten, E.G. (2002). MTHFR 677C-->T polymorphism and risk of coronary heart disease: a meta-analysis. JAMA 288, 2023-2031.

Klimasauskas, S., Kumar, S., Roberts, R.J., and Cheng, X. (1994). HhaI methyltransferase flips its target base out of the DNA helix. Cell 76, 357-369.

Klug, M., Heinz, S., Gebhard, C., Schwarzfischer, L., Krause, S.W., Andreesen, R., and Rehli, M. (2010). Active DNA demethylation in human postmitotic cells correlates with activating histone modifications, but not transcription levels. Genome Biol 11, R63.

Kluijtmans, L.A., van den Heuvel, L.P., Boers, G.H., Frosst, P., Stevens, E.M., van Oost, B.A., den Heijer, M., Trijbels, F.J., Rozen, R., and Blom, H.J. (1996). Molecular genetic analysis in mild hyperhomocysteinemia: a common mutation in the methylenetetrahydrofolate reductase gene is a genetic risk factor for cardiovascular disease. Am J Hum Genet 58, 35-41.

Kok, R.M., Smith, D.E., Barto, R., Spijkerman, A.M., Teerlink, T., Gellekink, H.J., Jakobs, C., and Smulders, Y.M. (2007). Global DNA methylation measured by liquid chromatography-tandem mass spectrometry: analytical technique, reference values and determinants in healthy subjects. Clin Chem Lab Med 45, 903-911.

Kondo, M., Suzuki, H., Ueda, R., Osada, H., Takagi, K., and Takahashi, T. (1995). Frequent loss of imprinting of the H19 gene is often associated with its overexpression in human lung cancers. Oncogene 10, 1193-1198.

Kondo, T., Bobek, M.P., Kuick, R., Lamb, B., Zhu, X., Narayan, A., Bourc'his, D., Viegas-Pequignot, E., Ehrlich, M., and Hanash, S.M. (2000). Whole-genome methylation scan in ICF syndrome: hypomethylation of non-satellite DNA repeats D4Z4 and NBL2. Human molecular genetics 9, 597-604.

Kono, S., and Chen, K. (2005). Genetic polymorphisms of methylenetetrahydrofolate reductase and colorectal cancer and adenoma. Cancer Sci 96, 535-542.

Korzus, E. (2010). Manipulating the brain with epigenetics. Nat Neurosci 13, 405-406.

226

Kotsopoulos, J., Sohn, K.J., Martin, R., Choi, M., Renlund, R., McKerlie, C., Hwang, S.W., Medline, A., and Kim, Y.I. (2003). Dietary folate deficiency suppresses N- methyl-N-nitrosourea-induced mammary tumorigenesis in rats. Carcinogenesis 24, 937- 944.

Kozbial, P.Z., and Mushegian, A.R. (2005). Natural history of S-adenosylmethionine- binding proteins. BMC Struct Biol 5, 19.

Kraunz, K.S., Hsiung, D., McClean, M.D., Liu, M., Osanyingbemi, J., Nelson, H.H., and Kelsey, K.T. (2006). Dietary folate is associated with p16(INK4A) methylation in head and neck squamous cell carcinoma. Int J Cancer 119, 1553-1557.

Kronenberg, G., Colla, M., and Endres, M. (2009). Folic acid, neurodegenerative and neuropsychiatric disease. Curr Mol Med 9, 315-323.

Kuo, K.C., McCune, R.A., Gehrke, C.W., Midgett, R., and Ehrlich, M. (1980). Quantitative reversed-phase high performance liquid chromatographic determination of major and modified deoxyribonucleosides in DNA. Nucleic Acids Res 8, 4763-4776.

Kupferminc, M.J., Eldor, A., Steinman, N., Many, A., Bar-Am, A., Jaffa, A., Fait, G., and Lessing, J.B. (1999). Increased frequency of genetic thrombophilia in women with complications of pregnancy. N Engl J Med 340, 9-13.

Kuznetsova, E.B., Kekeeva, T.V., Larin, S.S., Zemlyakova, V.V., Khomyakova, A.V., Babenko, O.V., Nemtsova, M.V., Zaletayev, D.V., and Strelnikov, V.V. (2007). Methylation of the BIN1 gene promoter CpG island associated with breast and prostate cancer. J Carcinog 6, 9.

Kwon, H.J., Kim, J.H., Bae, J.M., Cho, N.Y., Kim, T.Y., and Kang, G.H. (2010). DNA methylation changes in ex-adenoma carcinoma of the large intestine. Virchows Arch 457, 433-441.

Lander, E.S., Linton, L.M., Birren, B., Nusbaum, C., Zody, M.C., Baldwin, J., Devon, K., Dewar, K., Doyle, M., FitzHugh, W., et al. (2001). Initial sequencing and analysis of the human genome. Nature 409, 860-921.

227

Larsen, F., Solheim, J., and Prydz, H. (1993). A methylated CpG island 3' in the apolipoprotein-E gene does not repress its transcription. Hum Mol Genet 2, 775-780.

Larsson, S.C., Giovannucci, E., and Wolk, A. (2004). Dietary folate intake and incidence of ovarian cancer: the Swedish Mammography Cohort. J Natl Cancer Inst 96, 396-402.

Larsson, S.C., Giovannucci, E., and Wolk, A. (2006a). Folate intake and stomach cancer incidence in a prospective cohort of Swedish women. Cancer Epidemiol Biomarkers Prev 15, 1409-1412.

Larsson, S.C., Giovannucci, E., and Wolk, A. (2006b). Folate intake, MTHFR polymorphisms, and risk of esophageal, gastric, and pancreatic cancer: a meta-analysis. Gastroenterology 131, 1271-1283.

Larsson, S.C., Hakansson, N., Giovannucci, E., and Wolk, A. (2006c). Folate intake and pancreatic cancer incidence: a prospective study of Swedish women and men. J Natl Cancer Inst 98, 407-413.

Lawrence, M. (2007). Synthetic folic acid vs. food folates. Public Health Nutr 10, 533.

Lee, C., Wevrick, R., Fisher, R.B., Ferguson-Smith, M.A., and Lin, C.C. (1997). Human centromeric DNAs. Hum Genet 100, 291-304.

Lee, D.H., Jacobs, D.R., Jr., and Porta, M. (2009). Hypothesis: a unifying mechanism for nutrition and chemicals as lifelong modulators of DNA hypomethylation. Environ Health Perspect 117, 1799-1802.

Leeda, M., Riyazi, N., de Vries, J.I., Jakobs, C., van Geijn, H.P., and Dekker, G.A. (1998). Effects of folic acid and vitamin B6 supplementation on women with hyperhomocysteinemia and a history of preeclampsia or fetal growth restriction. Am J Obstet Gynecol 179, 135-139.

Lehnertz, B., Ueda, Y., Derijck, A.A., Braunschweig, U., Perez-Burgos, L., Kubicek, S., Chen, T., Li, E., Jenuwein, T., and Peters, A.H. (2003). Suv39h-mediated histone H3 lysine 9 methylation directs DNA methylation to major satellite repeats at pericentric heterochromatin. Curr Biol 13, 1192-1200.

228

Lewerin, C., Matousek, M., Steen, G., Johansson, B., Steen, B., and Nilsson-Ehle, H. (2005). Significant correlations of plasma homocysteine and serum methylmalonic acid with movement and cognitive performance in elderly subjects but no improvement from short-term vitamin therapy: a placebo-controlled randomized study. Am J Clin Nutr 81, 1155-1162.

Lewis, C.J., Crane, N.T., Wilson, D.B., and Yetley, E.A. (1999). Estimated folate intakes: data updated to reflect food fortification, increased bioavailability, and dietary supplement use. Am J Clin Nutr 70, 198-207.

Lewis, S.J., Ebrahim, S., and Davey Smith, G. (2005a). Meta-analysis of MTHFR 677C->T polymorphism and coronary heart disease: does totality of evidence support causal role for homocysteine and preventive potential of folate? BMJ 331, 1053.

Lewis, S.J., Zammit, S., Gunnell, D., and Smith, G.D. (2005b). A meta-analysis of the MTHFR C677T polymorphism and schizophrenia risk. Am J Med Genet B Neuropsychiatr Genet 135B, 2-4.

Li, H., Rauch, T., Chen, Z.X., Szabo, P.E., Riggs, A.D., and Pfeifer, G.P. (2006). The histone methyltransferase SETDB1 and the DNA methyltransferase DNMT3A interact directly and localize to promoters silenced in cancer cells. J Biol Chem 281, 19489- 19500.

Liang, G., Chan, M.F., Tomigahara, Y., Tsai, Y.C., Gonzales, F.A., Li, E., Laird, P.W., and Jones, P.A. (2002). Cooperativity between DNA methyltransferases in the maintenance methylation of repetitive elements. Mol Cell Biol 22, 480-491.

Liang, G., Robertson, K.D., Talmadge, C., Sumegi, J., and Jones, P.A. (2000). The gene for a novel transmembrane protein containing epidermal growth factor and follistatin domains is frequently hypermethylated in human tumor cells. Cancer Res 60, 4907- 4912.

Lieber, C. S., and Packer, L. (2002). S-adenosylmethionine: molecular, biological, and clinical aspects—an introduction. The American journal of clinical nutrition 76, 1148S.

Lillycrop, K.A., Phillips, E.S., Jackson, A.A., Hanson, M.A., and Burdge, G.C. (2005). Dietary protein restriction of pregnant rats induces and folic acid supplementation

229 prevents epigenetic modification of hepatic gene expression in the offspring. J Nutr 135, 1382-1386.

Lillycrop, K.A., Phillips, E.S., Torrens, C., Hanson, M.A., Jackson, A.A., and Burdge, G.C. (2008). Feeding pregnant rats a protein-restricted diet persistently alters the methylation of specific cytosines in the hepatic PPAR alpha promoter of the offspring. Br J Nutr 100, 278-282.

Lim, C.K., and Peters, T.J. (1989). Isocratic reversed-phase high-performance liquid chromatography of ribonucleotides, deoxynucleotides, cyclic nucleotides and deoxycyclic nucleotides. J Chromatogr 461, 259-266.

Lindahl, T., Karran, P., and Wood, R.D. (1997). DNA excision repair pathways. Curr Opin Genet Dev 7, 158-169.

Lindzon, G.M., Medline, A., Sohn, K.J., Depeint, F., Croxford, R., and Kim, Y.I. (2009). Effect of folic acid supplementation on the progression of colorectal aberrant crypt foci. Carcinogenesis 30, 1536-1543.

Lister, R., Pelizzola, M., Dowen, R.H., Hawkins, R.D., Hon, G., Tonti-Filippini, J., Nery, J.R., Lee, L., Ye, Z., Ngo, Q.M., et al. (2009). Human DNA methylomes at base resolution show widespread epigenomic differences. Nature 462, 315-322.

Liu, C., Xu, D., Sjoberg, J., Forsell, P., Bjorkholm, M., and Claesson, H.E. (2004). Transcriptional regulation of 15-lipoxygenase expression by promoter methylation. Exp Cell Res 297, 61-67.

Liu, D., Diorio, J., Tannenbaum, B., Caldji, C., Francis, D., Freedman, A., Sharma, S., Pearson, D., Plotsky, P.M., and Meaney, M.J. (1997). Maternal care, hippocampal glucocorticoid receptors, and hypothalamic-pituitary-adrenal responses to stress. Science 277, 1659-1662.

Liu, H., Yang, M., Li, G.M., Qiu, Y., Zheng, J., Du, X., Wang, J.L., and Liu, R.W. (2010). The MTHFR C677T polymorphism contributes to an increased risk for vascular dementia: a meta-analysis. J Neurol Sci 294, 74-80.

230

Loenen, W.A. (2006). S-adenosylmethionine: jack of all trades and master of everything? Biochem Soc Trans 34, 330-333.

Logan, R.F., Grainge, M.J., Shepherd, V.C., Armitage, N.C., and Muir, K.R. (2008). Aspirin and folic acid for the prevention of recurrent colorectal adenomas. Gastroenterology 134, 29-38.

Long, A.D., Mangalam, H.J., Chan, B.Y., Tolleri, L., Hatfield, G.W., and Baldi, P. (2001). Improved statistical inference from DNA microarray data using analysis of variance and a Bayesian statistical framework. Analysis of global gene expression in Escherichia coli K12. J Biol Chem 276, 19937-19944.

Lucock, M., Yates, Z., Glanville, T., Leeming, R., Simpson, N., and Daskalakis, I. (2003). A critical role for B-vitamin nutrition in human developmental and evolutionary biology. Nutrition Research 23, 1463-1475.

Lund, G., Andersson, L., Lauria, M., Lindholm, M., Fraga, M.F., Villar-Garea, A., Ballestar, E., Esteller, M., and Zaina, S. (2004). DNA methylation polymorphisms precede any histological sign of atherosclerosis in mice lacking apolipoprotein E. J Biol Chem 279, 29147-29154.

Ma, J., Stampfer, M.J., Giovannucci, E., Artigas, C., Hunter, D.J., Fuchs, C., Willett, W.C., Selhub, J., Hennekens, C.H., and Rozen, R. (1997). Methylenetetrahydrofolate reductase polymorphism, dietary interactions, and risk of colorectal cancer. Cancer Res 57, 1098-1102.

Maegawa, S., Hinkal, G., Kim, H.S., Shen, L., Zhang, L., Zhang, J., Zhang, N., Liang, S., Donehower, L.A., and Issa, J.P. (2010). Widespread and tissue specific age-related DNA methylation changes in mice. Genome Res 20, 332-340.

Magana, A.A., Wrobel, K., Caudillo, Y.A., Zaina, S., and Lund, G. (2008). High- performance liquid chromatography determination of 5-methyl-2'-deoxycytidine, 2'- deoxycytidine, and other deoxynucleosides and nucleosides in DNA digests. Anal Biochem 374, 378-385.

231

Malinow, M.R., Bostom, A.G., and Krauss, R.M. (1999). Homocyst(e)ine, diet, and cardiovascular diseases: a statement for healthcare professionals from the Nutrition Committee, American Heart Association. Circulation 99, 178-182.

Mann, J., and Truswell, A.S. (2007). Essentials of human nutrition, 3rd edn (Oxford, Oxford University Press).

Martin, J.L., and McMillan, F.M. (2002). SAM (dependent) I AM: the S- adenosylmethionine-dependent methyltransferase fold. Curr Opin Struct Biol 12, 783- 793.

Martinov, M.V., Vitvitsky, V.M., Banerjee, R., and Ataullakhanov, F.I. (2009). The logic of the hepatic methionine metabolic cycle. Biochim Biophys Acta 1804, 89-96.

Maselko, J., Kubzansky, L., Lipsitt, L., and Buka, S.L. (2010). Mother's affection at 8 months predicts emotional distress in adulthood. J Epidemiol Community Health.

Mason, J.B., and Choi, S.W. (2005). Effects of alcohol on folate metabolism: implications for carcinogenesis. Alcohol 35, 235-241.

Mason, J.B., Dickstein, A., Jacques, P.F., Haggarty, P., Selhub, J., Dallal, G., and Rosenberg, I.H. (2007). A temporal association between folic acid fortification and an increase in colorectal cancer rates may be illuminating important biological principles: a hypothesis. Cancer Epidemiol Biomarkers Prev 16, 1325-1329.

Matherly, L.H., and Goldman, D.I. (2003). Membrane transport of folates. Vitam Horm 66, 403-456.

Matherly, L.H., and Hou, Z. (2008). Structure and function of the reduced folate carrier a paradigm of a major facilitator superfamily mammalian nutrient transporter. Vitam Horm 79, 145-184.

Mathews, T.J., Honein, M.A., and Erickson, J.D. (2002). Spina bifida and anencephaly prevalence--United States, 1991-2001. MMWR Recomm Rep 51, 9-11.

Mattson, M.P., and Shea, T.B. (2003). Folate and homocysteine metabolism in neural plasticity and neurodegenerative disorders. Trends Neurosci 26, 137-146.

232

McCaddon, A., Davies, G., Hudson, P., Tandy, S., and Cattell, H. (1998). Total serum homocysteine in senile dementia of Alzheimer type. Int J Geriatr Psychiatry 13, 235- 239.

McGowan, P.O., Meaney, M.J., and Szyf, M. (2008). Diet and the epigenetic (re)programming of phenotypic differences in behavior. Brain Res 1237, 12-24.

McMahon, J.A., Green, T.J., Skeaff, C.M., Knight, R.G., Mann, J.I., and Williams, S.M. (2006). A controlled trial of homocysteine lowering and cognitive performance. N Engl J Med 354, 2764-2772.

McNulty, H., Dowey le, R.C., Strain, J.J., Dunne, A., Ward, M., Molloy, A.M., McAnena, L.B., Hughes, J.P., Hannon-Fletcher, M., and Scott, J.M. (2006). Riboflavin lowers homocysteine in individuals homozygous for the MTHFR 677C->T polymorphism. Circulation 113, 74-80.

Melnyk, S., Pogribna, M., Miller, B.J., Basnakian, A.G., Pogribny, I.P., and James, S.J. (1999). Uracil misincorporation, DNA strand breaks, and gene amplification are associated with tumorigenic cell transformation in folate deficient/repleted Chinese hamster ovary cells. Cancer Lett 146, 35-44.

Merlo, A., Herman, J.G., Mao, L., Lee, D.J., Gabrielson, E., Burger, P.C., Baylin, S.B., and Sidransky, D. (1995). 5' CpG island methylation is associated with transcriptional silencing of the tumour suppressor p16/CDKN2/MTS1 in human cancers. Nat Med 1, 686-692.

Metivier, R., Gallais, R., Tiffoche, C., Le Peron, C., Jurkowska, R.Z., Carmouche, R.P., Ibberson, D., Barath, P., Demay, F., Reid, G., et al. (2008). Cyclical DNA methylation of a transcriptionally active promoter. Nature 452, 45-50.

Michaud, E.J., van Vugt, M.J., Bultman, S.J., Sweet, H.O., Davisson, M.T., and Woychik, R.P. (1994). Differential expression of a new dominant agouti allele (Aiapy) is correlated with methylation state and is influenced by parental lineage. Genes Dev 8, 1463-1472.

233

Miller, C.A., Gavin, C.F., White, J.A., Parrish, R.R., Honasoge, A., Yancey, C.R., Rivera, I.M., Rubio, M.D., Rumbaugh, G., and Sweatt, J.D. (2010). Cortical DNA methylation maintains remote memory. Nat Neurosci 13, 664-666.

Miller, J.W. (1999). Homocysteine and Alzheimer's disease. Nutr Rev 57, 126-129.

Mills, J.L., McPartlin, J.M., Kirke, P.N., Lee, Y.J., Conley, M.R., Weir, D.G., and Scott, J.M. (1995). Homocysteine metabolism in pregnancies complicated by neural- tube defects. Lancet 345, 149-151.

Mills, J.L., and Signore, C. (2004). Neural tube defect rates before and after food fortification with folic acid. Birth Defects Res A Clin Mol Teratol 70, 844-845.

Mitchell, H.K., Snell, E.E., and Williams, R.J. (1941). The concentration of 'folic acid'. J Am Chem Soc 63, 2284.

Mitchell, L.E. (2005). Epidemiology of neural tube defects. Am J Med Genet C Semin Med Genet 135C, 88-94.

Molloy, A.M., Daly, S., Mills, J.L., Kirke, P.N., Whitehead, A.S., Ramsbottom, D., Conley, M.R., Weir, D.G., and Scott, J.M. (1997). Thermolabile variant of 5,10- methylenetetrahydrofolate reductase associated with low red-cell folates: implications for folate intake recommendations. Lancet 349, 1591-1593.

Moran, R.G. (1999). Roles of folylpoly-gamma-glutamate synthetase in therapeutics with tetrahydrofolate antimetabolites: an overview. Semin Oncol 26, 24-32.

Morgan, H.D., Jin, X.L., Li, A., Whitelaw, E., and O'Neill, C. (2008). The culture of zygotes to the blastocyst stage changes the postnatal expression of an epigentically labile allele, agouti viable yellow, in mice. Biol Reprod 79, 618-623.

Morgan, H.D., Sutherland, H.G., Martin, D.I., and Whitelaw, E. (1999). Epigenetic inheritance at the agouti locus in the mouse. Nat Genet 23, 314-318.

Morison, I.M., Ramsay, J.P., and Spencer, H.G. (2005). A census of mammalian imprinting. Trends Genet 21, 457-465.

234

Morita, H., Taguchi, J., Kurihara, H., Kitaoka, M., Kaneda, H., Kurihara, Y., Maemura, K., Shindo, T., Minamino, T., Ohno, M., et al. (1997). Genetic polymorphism of 5,10- methylenetetrahydrofolate reductase (MTHFR) as a risk factor for coronary artery disease. Circulation 95, 2032-2036.

Morris, K.V., Chan, S.W., Jacobsen, S.E., and Looney, D.J. (2004). Small interfering RNA-induced transcriptional gene silencing in human cells. Science 305, 1289-1292.

MRC Vitamin Study Research Group (1991). Prevention of neural tube defects: results of the Medical Research Council Vitamin Study. Lancet 338, 131-137.

Muntjewerff, J.W., Kahn, R.S., Blom, H.J., and den Heijer, M. (2006). Homocysteine, methylenetetrahydrofolate reductase and risk of schizophrenia: a meta-analysis. Mol Psychiatry 11, 143-149.

Murray, C.J., and Lopez, A.D. (1997). Mortality by cause for eight regions of the world: Global Burden of Disease Study. Lancet 349, 1269-1276.

Nadon, R., and Shoemaker, J. (2002). Statistical issues with microarrays: processing and analysis. Trends Genet 18, 265-271.

Nakayama, J., Rice, J.C., Strahl, B.D., Allis, C.D., and Grewal, S.I. (2001). Role of histone H3 lysine 9 methylation in epigenetic control of heterochromatin assembly. Science 292, 110-113.

Narayan, A., Ji, W., Zhang, X.Y., Marrogi, A., Graff, J.R., Baylin, S.B., and Ehrlich, M. (1998). Hypomethylation of pericentromeric DNA in breast adenocarcinomas. Int J Cancer 77, 833-838.

Narayanan, S., McConnell, J., Little, J., Sharp, L., Piyathilake, C.J., Powers, H., Basten, G., and Duthie, S.J. (2004). Associations between two common variants C677T and A1298C in the methylenetetrahydrofolate reductase gene and measures of folate metabolism and DNA stability (strand breaks, misincorporated uracil, and DNA methylation status) in human lymphocytes in vivo. Cancer Epidemiol Biomarkers Prev 13, 1436-1443.

235

Neidhart, M., Rethage, J., Kuchen, S., Kunzler, P., Crowl, R.M., Billingham, M.E., Gay, R.E., and Gay, S. (2000). Retrotransposable L1 elements expressed in rheumatoid arthritis synovial tissue: association with genomic DNA hypomethylation and influence on gene expression. Arthritis Rheum 43, 2634-2647.

New Zealand Ministry of Health (2010). Folate - Nutrition. Available at http://www.moh.govt.nz/moh.nsf/indexmh/nutrition-folate#folate3.

Nicholls, R. D., Saitoh, S., and Horsthemke, B. (1998). Imprinting in Prader-Willi and Angelman syndromes. Trends in genetics : TIG 14, 194–200.

Niculescu, M.D., Craciunescu, C.N., and Zeisel, S.H. (2005). Gene expression profiling of choline-deprived neural precursor cells isolated from mouse brain. Molecular Brain Research 134, 309-322.

Nijhout, H.F., Reed, M.C., Budu, P., and Ulrich, C.M. (2004). A mathematical model of the folate cycle: new insights into folate homeostasis. J Biol Chem 279, 55008-55016.

Nijhout, H.F., Reed, M.C., and Ulrich, C.M. (2008). Mathematical models of folate- mediated one-carbon metabolism. Vitam Horm 79, 45-82.

Nile, C.J., Read, R.C., Akil, M., Duff, G.W., and Wilson, A.G. (2008). Methylation status of a single CpG site in the IL6 promoter is related to IL6 messenger RNA levels and rheumatoid arthritis. Arthritis Rheum 58, 2686-2693.

Nilsson, K., Gustafson, L., and Hultberg, B. (2009). Plasma homocysteine levels and different forms of vascular disease in patients with dementia and other psychogeriatric diseases. Dement Geriatr Cogn Disord 27, 88-95.

Noma, K., Allis, C.D., and Grewal, S.I. (2001). Transitions in distinct histone H3 methylation patterns at the heterochromatin domain boundaries. Science 293, 1150- 1155.

Novak, P., Jensen, T., Oshiro, M.M., Wozniak, R.J., Nouzova, M., Watts, G.S., Klimecki, W.T., Kim, C., and Futscher, B.W. (2006). Epigenetic inactivation of the HOXA gene cluster in breast cancer. Cancer Res 66, 10664-10670.

236

Ntaios, G., Savopoulos, C., and Hatzitolios, A. (2008). Lowering homocysteine with B vitamins in patients with coronary artery disease. JAMA 300, 2852-2853; author reply 2853-2854.

NZPA (2009). Govt reviews folic acid decision. In The Press (Christchurch, Fairfax Media).

O'Broin, S., and Kelleher, B. (1992). Microbiological assay on microtitre plates of folate in serum and red cells. J Clin Pathol 45, 344-347.

Oakeley, E.J., Podesta, A., and Jost, J.P. (1997). Developmental changes in DNA methylation of the two tobacco pollen nuclei during maturation. Proc Natl Acad Sci U S A 94, 11721-11725.

Oakeley, E.J., Schmitt, F., and Jost, J.P. (1999). Quantification of 5-methylcytosine in DNA by the chloroacetaldehyde reaction. Biotechniques 27, 744-746, 748-750, 752.

Oakes, C.C., La Salle, S., Robaire, B., and Trasler, J.M. (2006). Evaluation of a quantitative DNA methylation analysis technique using methylation- sensitive/dependent restriction enzymes and real-time PCR. Epigenetics 1, 146-152.

Oakes, C.C., La Salle, S., Trasler, J.M., and Robaire, B. (2009). Restriction digestion and real-time PCR (qAMP). Methods Mol Biol 507, 271-280.

Oakley, G.P., Jr., Bell, K.N., and Weber, M.B. (2004a). Recommendations for accelerating global action to prevent folic acid-preventable birth defects and other folate-deficiency diseases: meeting of experts on preventing folic acid-preventable neural tube defects. Birth Defects Res A Clin Mol Teratol 70, 835-837.

Oakley, G.P., Jr., and Johnston, R.B., Jr. (2004). Balancing benefits and harms in public health prevention programmes mandated by governments. BMJ 329, 41-43; discussion 43-44.

Oakley, G.P., Jr., Weber, M.B., Bell, K.N., and Colditz, P. (2004b). Scientific evidence supporting folic acid fortification of flour in Australia and New Zealand. Birth Defects Res A Clin Mol Teratol 70, 838-841.

237

Oakley, G.P., and Mandel, J.S. (2004). Folic acid fortification remains an urgent health priority. BMJ 329, 1376.

O’Donnell, W. T., and Warren, S. T. (2002). A decade of molecular studies of fragile X syndrome. Annual review of neuroscience, 25, 315-38.

Olinski, R., Jurgowiak, M., and Zaremba, T. (2010). Uracil in DNA-Its biological significance. Mutat Res.

Ongenaert, M., Van Neste, L., De Meyer, T., Menschaert, G., Bekaert, S., and Van Criekinge, W. (2008). PubMeth: a cancer methylation database combining text-mining and expert annotation. Nucleic Acids Res 36, D842-846.

Ouko, L.A., Shantikumar, K., Knezovich, J., Haycock, P., Schnugh, D.J., and Ramsay, M. (2009). Effect of alcohol consumption on CpG methylation in the differentially methylated regions of H19 and IG-DMR in male gametes: implications for fetal alcohol spectrum disorders. Alcohol Clin Exp Res 33, 1615-1627.

Pajares, M.A., and Perez-Sala, D. (2006). Betaine homocysteine S-methyltransferase: just a regulator of homocysteine metabolism? Cell Mol Life Sci 63, 2792-2803.

Palmgren, G., Mattsson, O., and Okkels, F.T. (1990). Employment of hydrolytic enzymes in the study of the level of DNA methylation. Biochim Biophys Acta 1049, 293-297.

Pandey, R., Guru, R.K., and Mount, D.W. (2004). Pathway Miner: extracting gene association networks from molecular pathways for predicting the biological significance of gene expression microarray data. Bioinformatics 20, 2156-2158.

Panning, B. (2008). X-chromosome inactivation: the molecular basis of silencing. J Biol 7, 30.

Passeri, M., Cucinotta, D., Abate, G., Senin, U., Ventura, A., Stramba Badiale, M., Diana, R., La Greca, P., and Le Grazie, C. (1993). Oral 5'-methyltetrahydrofolic acid in senile organic mental disorders with depression: results of a double-blind multicenter study. Aging (Milano) 5, 63-71.

238

Paulsen, M., Tierling, S., and Walter, J. (2008). DNA methylation and the mammalian genome. In Epigenetics, J. Tost, ed. (Norfolk, Caister Academic Press), pp. 1-22.

Pejchal, R., Campbell, E., Guenther, B.D., Lennon, B.W., Matthews, R.G., and Ludwig, M.L. (2006). Structural perturbations in the Ala --> Val polymorphism of methylenetetrahydrofolate reductase: how binding of folates may protect against inactivation. Biochemistry 45, 4808-4818.

Pembrey, M.E., Bygren, L.O., Kaati, G., Edvinsson, S., Northstone, K., Sjostrom, M., and Golding, J. (2006). Sex-specific, male-line transgenerational responses in humans. Eur J Hum Genet 14, 159-166.

Pfeiffer, C.M., Caudill, S.P., Gunter, E.W., Osterloh, J., and Sampson, E.J. (2005). Biochemical indicators of B vitamin status in the US population after folic acid fortification: results from the National Health and Nutrition Examination Survey 1999- 2000. Am J Clin Nutr 82, 442-450.

Pitkin, R.M. (2007). Folate and neural tube defects. Am J Clin Nutr 85, 285S-288S.

Pogribny, I., Yi, P., and James, S.J. (1999). A sensitive new method for rapid detection of abnormal methylation patterns in global DNA and within CpG islands. Biochem Biophys Res Commun 262, 624-628.

Pogribny, I.P., and James, S.J. (2002). De novo methylation of the p16INK4A gene in early preneoplastic liver and tumors induced by folate/methyl deficiency in rats. Cancer Lett 187, 69-75.

Pogribny, I.P., James, S.J., Jernigan, S., and Pogribna, M. (2004). Genomic hypomethylation is specific for preneoplastic liver in folate/methyl deficient rats and does not occur in non-target tissues. Mutat Res 548, 53-59.

Pogribny, I.P., Miller, B.J., and James, S.J. (1997). Alterations in hepatic p53 gene methylation patterns during tumor progression with folate/methyl deficiency in the rat. Cancer Lett 115, 31-38.

Pomraning, K.R., Smith, K.M., and Freitag, M. (2009). Genome-wide high throughput analysis of DNA methylation in eukaryotes. Methods 47, 142-150.

239

Posfai, J., Bhagwat, A.S., Posfai, G., and Roberts, R.J. (1989). Predictive motifs derived from cytosine methyltransferases. Nucleic Acids Res 17, 2421-2435.

Pradhan, S., Bacolla, A., Wells, R.D., and Roberts, R.J. (1999). Recombinant human DNA (cytosine-5) methyltransferase. I. Expression, purification, and comparison of de novo and maintenance methylation. J Biol Chem 274, 33002-33010.

Prosser, J., Frommer, M., Paul, C., and Vincent, P.C. (1986). Sequence relationships of three human satellite DNAs. J Mol Biol 187, 145-155.

Qiu, A., Jansen, M., Sakaris, A., Min, S.H., Chattopadhyay, S., Tsai, E., Sandoval, C., Zhao, R., Akabas, M.H., and Goldman, I.D. (2006). Identification of an intestinal folate transporter and the molecular basis for hereditary folate malabsorption. Cell 127, 917- 928.

Qu, G., Dubeau, L., Narayan, A., Yu, M.C., and Ehrlich, M. (1999). Satellite DNA hypomethylation vs. overall genomic hypomethylation in ovarian epithelial tumors of different malignant potential. Mutat Res 423, 91-101.

Quackenbush, J. (2002). Microarray data normalization and transformation. Nat Genet 32 Suppl, 496-501.

Quere, I., Perneger, T.V., Zittoun, J., Bellet, H., Gris, J.C., Daures, J.P., Schved, J.F., Mercier, E., Laroche, J.P., Dauzat, M., et al. (2002). Red blood cell methylfolate and plasma homocysteine as risk factors for venous thromboembolism: a matched case- control study. Lancet 359, 747-752.

Quinlivan, E.P., and Gregory, J.F., 3rd (2003). Effect of food fortification on folic acid intake in the United States. Am J Clin Nutr 77, 221-225.

Quinlivan, E.P., and Gregory, J.F., 3rd (2008). DNA methylation determination by liquid chromatography-tandem mass spectrometry using novel biosynthetic [U- 15N]deoxycytidine and [U-15N]methyldeoxycytidine internal standards. Nucleic Acids Res 36, e119.

Rachmilewitz, J., Goshen, R., Ariel, I., Schneider, T., de Groot, N., and Hochberg, A. (1992). Parental imprinting of the human H19 gene. FEBS Lett 309, 25-28.

240

RajBhandary, U.L. (1994). Initiator transfer RNAs. J Bacteriol 176, 547-552.

Rajkovic, A., Catalano, P.M., and Malinow, M.R. (1997). Elevated homocyst(e)ine levels with preeclampsia. Obstet Gynecol 90, 168-171.

Ramakrishnan, S., Sulochana, K.N., Lakshmi, S., Selvi, R., and Angayarkanni, N. (2006). Biochemistry of homocysteine in health and diseases. Indian J Biochem Biophys 43, 275-283.

Rampersaud, G.C., Kauwell, G.P., Hutson, A.D., Cerda, J.J., and Bailey, L.B. (2000). Genomic DNA methylation decreases in response to moderate folate depletion in elderly women. Am J Clin Nutr 72, 998-1003.

Ramsahoye, B.H. (2002). Measurement of genome wide DNA methylation by reversed- phase high-performance liquid chromatography. Methods 27, 156-161.

Rasouli, M.L., Nasir, K., Blumenthal, R.S., Park, R., Aziz, D.C., and Budoff, M.J. (2005). Plasma homocysteine predicts progression of atherosclerosis. Atherosclerosis 181, 159-165.

Refsum, H., and Ueland, P.M. (1998). Recent data are not in conflict with homocysteine as a cardiovascular risk factor. Curr Opin Lipidol 9, 533-539.

Refsum, H., Ueland, P.M., Nygard, O., and Vollset, S.E. (1998). Homocysteine and cardiovascular disease. Annu Rev Med 49, 31-62.

Reik, W., and Dean, W. (2001). DNA methylation and mammalian epigenetics. Electrophoresis 22, 2838-2843.

Reik, W., Dean, W., and Walter, J. (2001). Epigenetic reprogramming in mammalian development. Science 293, 1089-1093.

Rice, J.C., and Allis, C.D. (2001). Histone methylation versus histone acetylation: new insights into epigenetic regulation. Curr Opin Cell Biol 13, 263-273.

Richardson, B. (2003). Impact of aging on DNA methylation. Ageing Res Rev 2, 245- 261.

241

Robinson, K., Arheart, K., Refsum, H., Brattstrom, L., Boers, G., Ueland, P., Rubba, P., Palma-Reis, R., Meleady, R., Daly, L., et al. (1998). Low circulating folate and vitamin B6 concentrations: risk factors for stroke, peripheral vascular disease, and coronary artery disease. European COMAC Group. Circulation 97, 437-443.

Rollins, R.A., Haghighi, F., Edwards, J.R., Das, R., Zhang, M.Q., Ju, J., and Bestor, T.H. (2006). Large-scale structure of genomic methylation patterns. Genome Res 16, 157-163.

Rong, N., Selhub, J., Goldin, B.R., and Rosenberg, I.H. (1991). Bacterially synthesized folate in rat large intestine is incorporated into host tissue folyl polyglutamates. J Nutr 121, 1955-1959.

Rush, L.J., and Plass, C. (2002). Restriction landmark genomic scanning for DNA methylation in cancer: past, present, and future applications. Anal Biochem 307, 191- 201.

Russell, D., Parnell, W., and Wilson, N. (1999). NZ food: NZ people - key results of the 1997 National Nutrition Survey (Wellington, New Zealand Ministry of Health).

Said, H.M., Chatterjee, N., Haq, R.U., Subramanian, V.S., Ortiz, A., Matherly, L.H., Sirotnak, F.M., Halsted, C., and Rubin, S.A. (2000). Adaptive regulation of intestinal folate uptake: effect of dietary folate deficiency. Am J Physiol Cell Physiol 279, C1889- 1895.

Sakai, T., Toguchida, J., Ohtani, N., Yandell, D.W., Rapaport, J.M., and Dryja, T.P. (1991). Allele-specific hypermethylation of the retinoblastoma tumor-suppressor gene. Am J Hum Genet 48, 880-888.

Sanderson, P., McNulty, H., Mastroiacovo, P., McDowell, I.F., Melse-Boonstra, A., Finglas, P.M., and Gregory, J.F., 3rd (2003). Folate bioavailability: UK Food Standards Agency workshop report. Br J Nutr 90, 473-479.

Sandovici, I., Smith, N.H., Ozanne, S.E., and Constancia, M. (2008). The dynamic epigenome: the impact of the environment on epigenetic regulation of gene expression and developmental programming. In Epigenetics, J. Tost, ed. (Norfolk, Caister Academic Press), pp. 343-370.

242

Sanjoaquin, M.A., Allen, N., Couto, E., Roddam, A.W., and Key, T.J. (2005). Folate intake and colorectal cancer risk: a meta-analytical approach. Int J Cancer 113, 825-828.

Santos-Rosa, H., Schneider, R., Bannister, A.J., Sherriff, J., Bernstein, B.E., Emre, N.C., Schreiber, S.L., Mellor, J., and Kouzarides, T. (2002). Active genes are tri- methylated at K4 of histone H3. Nature 419, 407-411.

Sarge, K.D., and Park-Sarge, O.K. (2005). Gene bookmarking: keeping the pages open. Trends Biochem Sci 30, 605-610.

Sauls, D.L., Arnold, E.K., Bell, C.W., Allen, J.C., and Hoffman, M. (2007). Pro- thrombotic and pro-oxidant effects of diet-induced hyperhomocysteinemia. Thromb Res 120, 117-126.

Schernhammer, E.S., Giovannucci, E., Kawasaki, T., Rosner, B., Fuchs, C.S., and Ogino, S. (2010). Dietary folate, alcohol and B vitamins in relation to LINE-1 hypomethylation in colon cancer. Gut 59, 794-799.

Schirch, V., and Strong, W.B. (1989). Interaction of folylpolyglutamates with enzymes in one-carbon metabolism. Arch Biochem Biophys 269, 371-380.

Schneider, E., Pliushch, G., El Hajj, N., Galetzka, D., Puhl, A., Schorsch, M., Frauenknecht, K., Riepert, T., Tresch, A., Muller, A.M., et al. (2010). Spatial, temporal and interindividual epigenetic variation of functionally important DNA methylation patterns. Nucleic Acids Res.

Schwarzacher-Robinson, T., Cram, L.S., Meyne, J., and Moyzis, R.K. (1988). Characterization of human heterochromatin by in situ hybridization with satellite DNA clones. Cytogenet Cell Genet 47, 192-196.

Selhub, J., Bagley, L.C., Miller, J., and Rosenberg, I.H. (2000). B vitamins, homocysteine, and neurocognitive function in the elderly. Am J Clin Nutr 71, 614S- 620S.

Sengupta, A.K., Ohhata, T., and Wutz, A. (2008). X chromosome inactivation. In Epigenetics, J. Tost, ed. (Norfolk, Caister Academic Press), pp. 273-301.

243

Seshadri, S., Beiser, A., Selhub, J., Jacques, P.F., Rosenberg, I.H., D'Agostino, R.B., Wilson, P.W., and Wolf, P.A. (2002). Plasma homocysteine as a risk factor for dementia and Alzheimer's disease. N Engl J Med 346, 476-483.

Shafa, M., Krawetz, R., and Rancourt, D.E. (2010). Returning to the stem state: Epigenetics of recapitulating pre-differentiation chromatin structure. Bioessays.

Shane, B. (1995). Folate chemistry and metabolism. In Folate in health and disease, L.B. Bailey, ed. (New York, M. Dekker), pp. xiv, 469 p.

Shane, B. (2003). Folate fortification: enough already? Am J Clin Nutr 77, 8-9.

Sharp, L., and Little, J. (2004). Polymorphisms in genes involved in folate metabolism and colorectal neoplasia: a HuGE review. Am J Epidemiol 159, 423-443.

Sharp, L., Little, J., Brockton, N.T., Cotton, S.C., Masson, L.F., Haites, N.E., and Cassidy, J. (2008). Polymorphisms in the methylenetetrahydrofolate reductase (MTHFR) gene, intakes of folate and related B vitamins and colorectal cancer: a case- control study in a population with relatively low folate intake. Br J Nutr 99, 379-389.

Shelnutt, K.P., Kauwell, G.P., Chapman, C.M., Gregory, J.F., 3rd, Maneval, D.R., Browdy, A.A., Theriaque, D.W., and Bailey, L.B. (2003). Folate status response to controlled folate intake is affected by the methylenetetrahydrofolate reductase 677C-- >T polymorphism in young women. J Nutr 133, 4107-4111.

Shelnutt, K.P., Kauwell, G.P., Gregory, J.F., 3rd, Maneval, D.R., Quinlivan, E.P., Theriaque, D.W., Henderson, G.N., and Bailey, L.B. (2004). Methylenetetrahydrofolate reductase 677C-->T polymorphism affects DNA methylation in response to controlled folate intake in young women. J Nutr Biochem 15, 554-560.

Shen, L., Kondo, Y., Guo, Y., Zhang, J., Zhang, L., Ahmed, S., Shu, J., Chen, X., Waterland, R.A., and Issa, J.P. (2007). Genome-wide profiling of DNA methylation reveals a class of normally methylated CpG island promoters. PLoS Genet 3, 2023- 2036.

Shields, D.C., Kirke, P.N., Mills, J.L., Ramsbottom, D., Molloy, A.M., Burke, H., Weir, D.G., Scott, J.M., and Whitehead, A.S. (1999). The "thermolabile" variant of

244 methylenetetrahydrofolate reductase and neural tube defects: An evaluation of genetic risk and the relative importance of the genotypes of the embryo and the mother. Am J Hum Genet 64, 1045-1055.

Shiota, K. (2004). DNA methylation profiles of CpG islands for cellular differentiation and development in mammals. Cytogenet Genome Res 105, 325-334.

Sibani, S., Christensen, B., O'Ferrall, E., Saadi, I., Hiou-Tim, F., Rosenblatt, D.S., and Rozen, R. (2000). Characterization of six novel mutations in the methylenetetrahydrofolate reductase (MTHFR) gene in patients with homocystinuria. Hum Mutat 15, 280-287.

Simon, R.M., and Dobbin, K. (2003). Experimental design of DNA microarray experiments. Biotechniques Suppl, 16-21.

Simpson, R.C., and Brown, P.R. (1986). High-performance liquid chromatographic profiling of nucleic acid components in physiological samples. J Chromatogr 379, 269- 311.

Sims, J.K., Magazinnik, T., Houston, S.I., Wu, S., and Rice, J.C. (2008). Histone Modification and Epigenetics. In Epigenetics, J. Tost, ed. (Norfolk, Caister Academic Press), pp. 105-126.

Sinclair, K.D., Allegrucci, C., Singh, R., Gardner, D.S., Sebastian, S., Bispham, J., Thurston, A., Huntley, J.F., Rees, W.D., Maloney, C.A., et al. (2007). DNA methylation, insulin resistance, and blood pressure in offspring determined by maternal periconceptional B vitamin and methionine status. Proc Natl Acad Sci U S A 104, 19351-19356.

Sirotnak, F.M., and Tolner, B. (1999). Carrier-mediated membrane transport of folates in mammalian cells. Annu Rev Nutr 19, 91-122.

Skibola, C.F., Smith, M.T., Kane, E., Roman, E., Rollinson, S., Cartwright, R.A., and Morgan, G. (1999). Polymorphisms in the methylenetetrahydrofolate reductase gene are associated with susceptibility to acute leukemia in adults. Proc Natl Acad Sci U S A 96, 12810-12815.

245

Skinner, M.K., and Guerrero-Bosagna, C. (2009). Environmental signals and transgenerational epigenetics. Epigenomics 1, 111-117.

Slattery, M.L., Potter, J.D., Samowitz, W., Schaffer, D., and Leppert, M. (1999). Methylenetetrahydrofolate reductase, diet, and risk of colon cancer. Cancer Epidemiol Biomarkers Prev 8, 513-518.

Slotkin, R.K., and Martienssen, R. (2007). Transposable elements and the epigenetic regulation of the genome. Nat Rev Genet 8, 272-285.

Smithells, R.W., Sheppard, S., and Schorah, C.J. (1976). Vitamin dificiencies and neural tube defects. Arch Dis Child 51, 944-950.

Smithells, R.W., Sheppard, S., Schorah, C.J., Seller, M.J., Nevin, N.C., Harris, R., Read, A.P., and Fielding, D.W. (1981). Apparent prevention of neural tube defects by periconceptional vitamin supplementation. Arch Dis Child 56, 911-918.

Sohn, K.J., Jang, H., Campan, M., Weisenberger, D.J., Dickhout, J., Wang, Y.C., Cho, R.C., Yates, Z., Lucock, M., Chiang, E.P., et al. (2009). The methylenetetrahydrofolate reductase C677T mutation induces cell-specific changes in genomic DNA methylation and uracil misincorporation: a possible molecular basis for the site-specific cancer risk modification. Int J Cancer 124, 1999-2005.

Solomons, N.W. (2007). Food fortification with folic acid: has the other shoe dropped? Nutr Rev 65, 512-515.

Song, J., Medline, A., Mason, J.B., Gallinger, S., and Kim, Y.I. (2000a). Effects of dietary folate on intestinal tumorigenesis in the apcMin mouse. Cancer Res 60, 5434- 5440.

Song, J., Sohn, K.J., Medline, A., Ash, C., Gallinger, S., and Kim, Y.I. (2000b). Chemopreventive effects of dietary folate on intestinal polyps in Apc+/-Msh2-/- mice. Cancer Res 60, 3191-3199.

Song, L., James, S.R., Kazim, L., and Karpf, A.R. (2005). Specific method for the determination of genomic DNA methylation by liquid chromatography-electrospray ionization tandem mass spectrometry. Anal Chem 77, 504-510.

246

Spiegelstein, O., Mitchell, L.E., Merriweather, M.Y., Wicker, N.J., Zhang, Q., Lammer, E.J., and Finnell, R.H. (2004). Embryonic development of folate binding protein-1 (Folbp1) knockout mice: Effects of the chemical form, dose, and timing of maternal folate supplementation. Dev Dyn 231, 221-231.

Steegers-Theunissen, R.P., Boers, G.H., Blom, H.J., Nijhuis, J.G., Thomas, C.M., Borm, G.F., and Eskes, T.K. (1995). Neural tube defects and elevated homocysteine levels in amniotic fluid. Am J Obstet Gynecol 172, 1436-1441.

Steegers-Theunissen, R.P., Boers, G.H., Trijbels, F.J., Finkelstein, J.D., Blom, H.J., Thomas, C.M., Borm, G.F., Wouters, M.G., and Eskes, T.K. (1994). Maternal hyperhomocysteinemia: a risk factor for neural-tube defects? Metabolism 43, 1475- 1480.

Steegers-Theunissen, R.P., Obermann-Borst, S.A., Kremer, D., Lindemans, J., Siebel, C., Steegers, E.A., Slagboom, P.E., and Heijmans, B.T. (2009). Periconceptional maternal folic acid use of 400 microg per day is related to increased methylation of the IGF2 gene in the very young child. PLoS One 4, e7845.

Stempak, J.M., Sohn, K.J., Chiang, E.P., Shane, B., and Kim, Y.I. (2005). Cell and stage of transformation-specific effects of folate deficiency on methionine cycle intermediates and DNA methylation in an in vitro model. Carcinogenesis 26, 981-990.

Stern, L.L., Mason, J.B., Selhub, J., and Choi, S.W. (2000). Genomic DNA hypomethylation, a characteristic of most cancers, is present in peripheral leukocytes of individuals who are homozygous for the C677T polymorphism in the methylenetetrahydrofolate reductase gene. Cancer Epidemiol Biomarkers Prev 9, 849- 853.

Stevens, V.L., Rodriguez, C., Pavluck, A.L., McCullough, M.L., Thun, M.J., and Calle, E.E. (2006). Folate nutrition and prostate cancer incidence in a large cohort of US men. Am J Epidemiol 163, 989-996.

Stirzaker, C., Millar, D.S., Paul, C.L., Warnecke, P.M., Harrison, J., Vincent, P.C., Frommer, M., and Clark, S.J. (1997). Extensive DNA methylation spanning the Rb promoter in retinoblastoma tumors. Cancer Res 57, 2229-2237.

247

Stocker, R., and Keaney, J.F., Jr. (2004). Role of oxidative modifications in atherosclerosis. Physiol Rev 84, 1381-1478.

Stokstad, E.L., and Manning, P.D. (1938). A New Dietary Water Soluble Factor Required by Chicks. Science 88, 35-36.

Stolzenberg-Solomon, R.Z., Chang, S.C., Leitzmann, M.F., Johnson, K.A., Johnson, C., Buys, S.S., Hoover, R.N., and Ziegler, R.G. (2006). Folate intake, alcohol use, and postmenopausal breast cancer risk in the Prostate, Lung, Colorectal, and Ovarian Cancer Screening Trial. Am J Clin Nutr 83, 895-904.

Stover, P., Kruschwitz, H., and Schirch, V. (1993). Evidence that 5- formyltetrahydropteroylglutamate has a metabolic role in one-carbon metabolism. Adv Exp Med Biol 338, 679-685.

Stover, P.J. (2004). Physiology of folate and vitamin B12 in health and disease. Nutr Rev 62, S3-12; discussion S13.

Suh, J.R., Herbig, A.K., and Stover, P.J. (2001). New perspectives on folate catabolism. Annu Rev Nutr 21, 255-282.

Sweeney, M.R., McPartlin, J., and Scott, J. (2007). Folic acid fortification and public health: report on threshold doses above which unmetabolised folic acid appear in serum. BMC Public Health 7, 41.

Tada, T., Obata, Y., Tada, M., Goto, Y., Nakatsuji, N., Tan, S., Kono, T., and Takagi, N. (2000). Imprint switching for non-random X-chromosome inactivation during mouse oocyte growth. Development 127, 3101-3105.

Tagami, H., Ray-Gallet, D., Almouzni, G., and Nakatani, Y. (2004). Histone H3.1 and H3.3 complexes mediate nucleosome assembly pathways dependent or independent of DNA synthesis. Cell 116, 51-61.

Tagarro, I., Fernandez-Peralta, A.M., and Gonzalez-Aguilera, J.J. (1994). Chromosomal localization of human satellites 2 and 3 by a FISH method using oligonucleotides as probes. Hum Genet 93, 383-388.

248

Takai, D., Gonzales, F.A., Tsai, Y.C., Thayer, M.J., and Jones, P.A. (2001). Large scale mapping of methylcytosines in CTCF-binding sites in the human H19 promoter and aberrant hypomethylation in human bladder cancer. Hum Mol Genet 10, 2619-2626.

Takai, D., and Jones, P.A. (2002). Comprehensive analysis of CpG islands in human chromosomes 21 and 22. Proc Natl Acad Sci U S A 99, 3740-3745.

Tang, S.H., Yang, D.H., Huang, W., Zhou, H.K., Lu, X.H., and Ye, G. (2006). Hypomethylated P4 promoter induces expression of the insulin-like growth factor-II gene in hepatocellular carcinoma in a Chinese population. Clin Cancer Res 12, 4171- 4177.

Tchantchou, F., and Shea, T.B. (2008). Folate deprivation, the methionine cycle, and Alzheimer's disease. Vitam Horm 79, 83-97.

Teitz, T., Wei, T., Valentine, M.B., Vanin, E.F., Grenet, J., Valentine, V.A., Behm, F.G., Look, A.T., Lahti, J.M., and Kidd, V.J. (2000). Caspase 8 is deleted or silenced preferentially in childhood neuroblastomas with amplification of MYCN. Nat Med 6, 529-535.

Thomas, P., and Fenech, M. (2008). Methylenetetrahydrofolate reductase, common polymorphisms, and relation to disease. Vitam Horm 79, 375-392.

Tobi, E.W., Lumey, L.H., Talens, R.P., Kremer, D., Putter, H., Stein, A.D., Slagboom, P.E., and Heijmans, B.T. (2009). DNA methylation differences after exposure to prenatal famine are common and timing- and sex-specific. Hum Mol Genet 18, 4046- 4053.

Tohgi, H., Utsugisawa, K., Nagane, Y., Yoshimura, M., Genda, Y., and Ukitsu, M. (1999). Reduction with age in methylcytosine in the promoter region -224 approximately -101 of the amyloid precursor protein gene in autopsy human cortex. Brain Res Mol Brain Res 70, 288-292.

Tost, J. (2008). Epigenetics (Norfolk, Caister Academic Press).

Tran, P., Leclerc, D., Chan, M., Pai, A., Hiou-Tim, F., Wu, Q., Goyette, P., Artigas, C., Milos, R., and Rozen, R. (2002). Multiple transcription start sites and alternative

249 splicing in the methylenetetrahydrofolate reductase gene result in two enzyme isoforms. Mamm Genome 13, 483-492.

Tusher, V.G., Tibshirani, R., and Chu, G. (2001). Significance analysis of microarrays applied to the ionizing radiation response. Proc Natl Acad Sci U S A 98, 5116-5121.

Ubbink, J.B., Vermaak, W.J., van der Merwe, A., and Becker, P.J. (1993). Vitamin B- 12, vitamin B-6, and folate nutritional status in men with hyperhomocysteinemia. Am J Clin Nutr 57, 47-53.

Ueland, P.M., Hustad, S., Schneede, J., Refsum, H., and Vollset, S.E. (2001). Biological and clinical implications of the MTHFR C677T polymorphism. Trends Pharmacol Sci 22, 195-201.

Uhl, J., Klan, N., Rose, M., Entian, K.D., Werz, O., and Steinhilber, D. (2002). The 5- lipoxygenase promoter is regulated by DNA methylation. J Biol Chem 277, 4374-4379.

Ulrich, C.M., and Potter, J.D. (2007). Folate and cancer--timing is everything. JAMA 297, 2408-2409.

US Department of Health and Human Services, Food and Drug Administration (1996). Food standard: amendment of the standards of identity for enriched grain products to require addition of folic acid. Fed Regist 61, 8781-8809. van den Donk, M., van Engeland, M., Pellis, L., Witteman, B.J., Kok, F.J., Keijer, J., and Kampman, E. (2007). Dietary folate intake in combination with MTHFR C677T genotype and promoter methylation of tumor suppressor and DNA repair genes in sporadic colorectal adenomas. Cancer Epidemiol Biomarkers Prev 16, 327-333.

Van den Veyver, I.B. (2002). Genetic effects of methylation diets. Annu Rev Nutr 22, 255-282. van der Put, N.M., Steegers-Theunissen, R.P., Frosst, P., Trijbels, F.J., Eskes, T.K., van den Heuvel, L.P., Mariman, E.C., den Heyer, M., Rozen, R., and Blom, H.J. (1995). Mutated methylenetetrahydrofolate reductase as a risk factor for spina bifida. Lancet 346, 1070-1071.

250

Van Emburgh, B.O., and Robertson, K.D. (2008). DNA methyltransferases and methyl- CpG binding proteins as multifunctional regulators of chromatin structure and development in mammalian cells. In Epigenetics, J. Tost, ed. (Norfolk, Caister Academic Press), pp. 23-61.

Van Guelpen, B., Hultdin, J., Johansson, I., Hallmans, G., Stenling, R., Riboli, E., Winkvist, A., and Palmqvist, R. (2006). Low folate levels may protect against colorectal cancer. Gut 55, 1461-1466.

Voet, D., and Voet, J.G. (1995). Biochemistry, 2nd edn (New York, J. Wiley & Sons).

Vollset, S.E., Igland, J., Jenab, M., Fredriksen, A., Meyer, K., Eussen, S., Gjessing, H.K., Ueland, P.M., Pera, G., Sala, N., et al. (2007). The association of gastric cancer risk with plasma folate, cobalamin, and methylenetetrahydrofolate reductase polymorphisms in the European Prospective Investigation into Cancer and Nutrition. Cancer Epidemiol Biomarkers Prev 16, 2416-2424.

Wagner, C. (1995). Biochemical role of folate in cellular metabolism. In Folate in health and disease, L.B. Bailey, ed. (New York, M. Dekker), pp. 23-42.

Wald, D.S., Law, M., and Morris, J.K. (2002). Homocysteine and cardiovascular disease: evidence on causality from a meta-analysis. BMJ 325, 1202.

Wang, Y., Nimec, Z., Ryan, T.J., Dias, J.A., and Galivan, J. (1993). The properties of the secreted gamma-glutamyl hydrolases from H35 hepatoma cells. Biochim Biophys Acta 1164, 227-235.

Wasson, G.R., McGlynn, A.P., McNulty, H., O'Reilly, S.L., McKelvey-Martin, V.J., McKerr, G., Strain, J.J., Scott, J., and Downes, C.S. (2006). Global DNA and p53 region-specific hypomethylation in human colonic cells is induced by folate depletion and reversed by folate supplementation. J Nutr 136, 2748-2753.

Waterland, R.A., Dolinoy, D.C., Lin, J.R., Smith, C.A., Shi, X., and Tahiliani, K.G. (2006). Maternal methyl supplements increase offspring DNA methylation at Axin Fused. Genesis 44, 401-406.

251

Waterland, R.A., and Jirtle, R.L. (2003). Transposable elements: targets for early nutritional effects on epigenetic gene regulation. Mol Cell Biol 23, 5293-5300.

Waterland, R.A., and Jirtle, R.L. (2004). Early nutrition, epigenetic changes at transposons and imprinted genes, and enhanced susceptibility to adult chronic diseases. Nutrition 20, 63-68.

Watts, G.S., Futscher, B.W., Holtan, N., Degeest, K., Domann, F.E., and Rose, S.L. (2008). DNA methylation changes in ovarian cancer are cumulative with disease progression and identify tumor stage. BMC Med Genomics 1, 47.

Weaver, I.C., Cervoni, N., Champagne, F.A., D'Alessio, A.C., Sharma, S., Seckl, J.R., Dymov, S., Szyf, M., and Meaney, M.J. (2004). Epigenetic programming by maternal behavior. Nat Neurosci 7, 847-854.

Weaver, I.C., Champagne, F.A., Brown, S.E., Dymov, S., Sharma, S., Meaney, M.J., and Szyf, M. (2005). Reversal of maternal programming of stress responses in adult offspring through methyl supplementation: altering epigenetic marking later in life. J Neurosci 25, 11045-11054.

Weaver, I.C., Meaney, M.J., and Szyf, M. (2006). Maternal care effects on the hippocampal transcriptome and anxiety-mediated behaviors in the offspring that are reversible in adulthood. Proc Natl Acad Sci U S A 103, 3480-3485.

Weaving, L. S., Ellaway, C. J., Gécz, J., and Christodoulou, J. (2005). Rett syndrome: clinical review and genetic update. Journal of medical genetics 42, 1–7.

Webber, A.L., Ingram, R.S., Levorse, J.M., and Tilghman, S.M. (1998). Location of enhancers is essential for the imprinting of H19 and Igf2 genes. Nature 391, 711-715.

Weber, M., Davies, J.J., Wittig, D., Oakeley, E.J., Haase, M., Lam, W.L., and Schubeler, D. (2005). Chromosome-wide and promoter-specific analyses identify sites of differential DNA methylation in normal and transformed human cells. Nat Genet 37, 853-862.

252

Weber, M., Hellmann, I., Stadler, M.B., Ramos, L., Paabo, S., Rebhan, M., and Schubeler, D. (2007). Distribution, silencing potential and evolutionary impact of promoter DNA methylation in the human genome. Nat Genet 39, 457-466.

Weisenberger, D.J., Campan, M., Long, T.I., Kim, M., Woods, C., Fiala, E., Ehrlich, M., and Laird, P.W. (2005). Analysis of repetitive element DNA methylation by MethyLight. Nucleic Acids Res 33, 6823-6836.

Weiss, N., Keller, C., Hoffmann, U., and Loscalzo, J. (2002). Endothelial dysfunction and atherothrombosis in mild hyperhomocysteinemia. Vasc Med 7, 227-239.

Weksberg, R., Smith, A.C., Squire, J., and Sadowski, P. (2003). Beckwith-Wiedemann syndrome demonstrates a role for epigenetic control of normal development. Hum Mol Genet 12 Spec No 1, R61-68.

West, R.L., Lee, J.M., and Maroun, L.E. (1995). Hypomethylation of the amyloid precursor protein gene in the brain of an Alzheimer's disease patient. J Mol Neurosci 6, 141-146.

Whetstine, J.R., Flatley, R.M., and Matherly, L.H. (2002). The human reduced folate carrier gene is ubiquitously and differentially expressed in normal human tissues: identification of seven non-coding exons and characterization of a novel promoter. Biochem J 367, 629-640.

Whitehead, A.S., Gallagher, P., Mills, J.L., Kirke, P.N., Burke, H., Molloy, A.M., Weir, D.G., Shields, D.C., and Scott, J.M. (1995). A genetic defect in 5,10 methylenetetrahydrofolate reductase in neural tube defects. QJM 88, 763-766.

Widschwendter, M., Jiang, G., Woods, C., Muller, H.M., Fiegl, H., Goebel, G., Marth, C., Muller-Holzner, E., Zeimet, A.G., Laird, P.W., et al. (2004). DNA hypomethylation and ovarian cancer biology. Cancer Res 64, 4472-4480.

Wiemels, J.L., Smith, R.N., Taylor, G.M., Eden, O.B., Alexander, F.E., and Greaves, M.F. (2001). Methylenetetrahydrofolate reductase (MTHFR) polymorphisms and risk of molecularly defined subtypes of childhood acute leukemia. Proc Natl Acad Sci U S A 98, 4004-4009.

253

Wilcken, B., Bamforth, F., Li, Z., Zhu, H., Ritvanen, A., Renlund, M., Stoll, C., Alembik, Y., Dott, B., Czeizel, A.E., et al. (2003). Geographical and ethnic variation of the 677C>T allele of 5,10 methylenetetrahydrofolate reductase (MTHFR): findings from over 7000 newborns from 16 areas world wide. J Med Genet 40, 619-625.

Williams, W.J. (1990). Hematology, 4th edn (New York, McGraw-Hill Information Services Co., Health Professions Division).

Wilson, A.S., Power, B.E., and Molloy, P.L. (2007). DNA hypomethylation and human diseases. Biochim Biophys Acta 1775, 138-162.

Wolff, G.L., Kodell, R.L., Moore, S.R., and Cooney, C.A. (1998). Maternal epigenetics and methyl supplements affect agouti gene expression in Avy/a mice. FASEB J 12, 949- 957.

Wolff, G.L., Roberts, D.W., and Mountjoy, K.G. (1999). Physiological consequences of ectopic agouti gene expression: the yellow obese mouse syndrome. Physiol Genomics 1, 151-163.

Wong, N.C., Wong, L.H., Quach, J.M., Canham, P., Craig, J.M., Song, J.Z., Clark, S.J., and Choo, K.H. (2006). Permissive transcriptional activity at the centromere through pockets of DNA hypomethylation. PLoS Genet 2, e17.

Wong, S.C., Zhang, L., Proefke, S.A., Hukku, B., and Matherly, L.H. (1998). Gene amplification and increased expression of the reduced folate carrier in transport elevated K562 cells. Biochem Pharmacol 55, 1135-1138.

Wright, A.J., Dainty, J.R., and Finglas, P.M. (2007). Folic acid metabolism in human subjects revisited: potential implications for proposed mandatory folic acid fortification in the UK. Br J Nutr 98, 667-675.

Wu, K., Platz, E.A., Willett, W.C., Fuchs, C.S., Selhub, J., Rosner, B.A., Hunter, D.J., and Giovannucci, E. (2009). A randomized trial on folic acid supplementation and risk of recurrent colorectal adenoma. Am J Clin Nutr 90, 1623-1631.

Wutz, A., and Jaenisch, R. (2000). A shift from reversible to irreversible X inactivation is triggered during ES cell differentiation. Mol Cell 5, 695-705.

254

Xu, G. L., Bestor, T. H., Bourc’his, D., Hsieh, C. L., Tommerup, N., Bugge, M., Hulten, M., Qu, X., Russo, J. J., and Viegas-Péquignot, E. (1999). Chromosome instability and immunodeficiency syndrome caused by mutations in a DNA methyltransferase gene. Nature 402, 187–191.

Yamada, K., Chen, Z., Rozen, R., and Matthews, R.G. (2001). Effects of common polymorphisms on the properties of recombinant human methylenetetrahydrofolate reductase. Proc Natl Acad Sci U S A 98, 14853-14858.

Yamada, K., Strahler, J.R., Andrews, P.C., and Matthews, R.G. (2005). Regulation of human methylenetetrahydrofolate reductase by phosphorylation. Proc Natl Acad Sci U S A 102, 10454-10459.

Yamada, Y., Watanabe, H., Miura, F., Soejima, H., Uchiyama, M., Iwasaka, T., Mukai, T., Sakaki, Y., and Ito, T. (2004). A comprehensive analysis of allelic methylation status of CpG islands on human chromosome 21q. Genome Res 14, 247-266.

Yan, P.S., Chen, C.M., Shi, H., Rahmatpanah, F., Wei, S.H., Caldwell, C.W., and Huang, T.H. (2001). Dissecting complex epigenetic alterations in breast cancer using CpG island microarrays. Cancer Res 61, 8375-8380.

Yan, P.S., Chen, C.M., Shi, H., Rahmatpanah, F., Wei, S.H., and Huang, T.H. (2002). Applications of CpG island microarrays for high-throughput analysis of DNA methylation. J Nutr 132, 2430S-2434S.

Yang, H.H., Hu, N., Wang, C., Ding, T., Dunn, B.K., Goldstein, A.M., Taylor, P.R., and Lee, M.P. (2010). Influence of genetic background and tissue types on global DNA methylation patterns. PLoS One 5, e9355.

Yang, Q., Botto, L.D., Erickson, J.D., Berry, R.J., Sambell, C., Johansen, H., and Friedman, J.M. (2006). Improvement in stroke mortality in Canada and the United States, 1990 to 2002. Circulation 113, 1335-1343.

Yang, Y.H., Dudoit, S., Luu, P., Lin, D.M., Peng, V., Ngai, J., and Speed, T.P. (2002). Normalization for cDNA microarray data: a robust composite method addressing single and multiple slide systematic variation. Nucleic Acids Res 30, e15.

255

Yazdy, M.M., Honein, M.A., and Xing, J. (2007). Reduction in orofacial clefts following folic acid fortification of the U.S. grain supply. Birth Defects Res A Clin Mol Teratol 79, 16-23.

Yi, P., Melnyk, S., Pogribna, M., Pogribny, I.P., Hine, R.J., and James, S.J. (2000). Increase in plasma homocysteine associated with parallel increases in plasma S- adenosylhomocysteine and lymphocyte DNA hypomethylation. J Biol Chem 275, 29318-29323.

Yideng, J., Jianzhong, Z., Ying, H., Juan, S., Jinge, Z., Shenglan, W., Xiaoqun, H., and Shuren, W. (2007). Homocysteine-mediated expression of SAHH, DNMTs, MBD2, and DNA hypomethylation potential pathogenic mechanism in VSMCs. DNA Cell Biol 26, 603-611.

Yoder, J.A., Walsh, C.P., and Bestor, T.H. (1997). Cytosine methylation and the ecology of intragenomic parasites. Trends Genet 13, 335-340.

You, Q., Karrow, N.A., Cao, H., Rodriguez, A., Mallard, B.A., and Boermans, H.J. (2008). Variation in the ovine cortisol response to systemic bacterial endotoxin challenge is predominantly determined by signalling within the hypothalamic-pituitary- adrenal axis. Toxicology and Applied Pharmacology 230, 1-8.

Yu, Z. (2008). Non-coding RNAs in Gene Regulation. In Epigenetics, J. Tost, ed. (Norfolk, Caister Academic Press), pp. 171-186.

Yuen, R.K., Avila, L., Penaherrera, M.S., von Dadelszen, P., Lefebvre, L., Kobor, M.S., and Robinson, W.P. (2009). Human placental-specific epipolymorphism and its association with adverse pregnancy outcomes. PLoS One 4, e7389.

Zetterberg, H., Regland, B., Palmer, M., Ricksten, A., Palmqvist, L., Rymo, L., Arvanitis, D.A., Spandidos, D.A., and Blennow, K. (2002). Increased frequency of combined methylenetetrahydrofolate reductase C677T and A1298C mutated alleles in spontaneously aborted embryos. Eur J Hum Genet 10, 113-118.

Zhang, S., Hunter, D.J., Hankinson, S.E., Giovannucci, E.L., Rosner, B.A., Colditz, G.A., Speizer, F.E., and Willett, W.C. (1999). A prospective study of folate intake and the risk of breast cancer. JAMA 281, 1632-1637.

256

Zhang, S.M., Cook, N.R., Albert, C.M., Gaziano, J.M., Buring, J.E., and Manson, J.E. (2008). Effect of combined folic acid, vitamin B6, and vitamin B12 on cancer risk in women: a randomized trial. JAMA 300, 2012-2021.

Zhang, S.M., Willett, W.C., Selhub, J., Hunter, D.J., Giovannucci, E.L., Holmes, M.D., Colditz, G.A., and Hankinson, S.E. (2003). Plasma folate, vitamin B6, vitamin B12, homocysteine, and risk of breast cancer. J Natl Cancer Inst 95, 373-380.

Zhang, Y., and Tycko, B. (1992). Monoallelic expression of the human H19 gene. Nat Genet 1, 40-44.

Zhao, L., and Funk, C.D. (2004). Lipoxygenase pathways in atherogenesis. Trends Cardiovasc Med 14, 191-195.

Zilberman, D., Gehring, M., Tran, R.K., Ballinger, T., and Henikoff, S. (2007). Genome-wide analysis of Arabidopsis thaliana DNA methylation uncovers an interdependence between methylation and transcription. Nat Genet 39, 61-69.

Zilberman, D., and Henikoff, S. (2007). Genome-wide analysis of DNA methylation patterns. Development 134, 3959-3965.