THE ROLE OF HORMONAL AND VASCULAR GENES IN MIGRAINE

Transcription

1 THE ROLE OF HORMONAL AND VASCULAR GENES IN MIGRAINE By Natalie Colson BHSc (Hons) A thesis submitted as fulfillment for the degree of Doctor of Philosophy (PhD) in the School of Medical Science, Griffith University, Gold Coast, Queensland.

2

3 Abstract Migraine is a frequent debilitating neurological disorder that is considered to be genetically complex with a multifactorial mode of inheritance. It has a high prevalence with approximately 18% of women and 6% of men suffering from the disorder. Migraine is characterized by severe head pain with associated nausea, emesis, photophobia, phonophobia, and neurological disturbances. The International Headache Society (IHS) has classified various types of migraine according to their clinical features. The two main subtypes of migraine are migraine without aura (MO), occurring in ~70-75% of migraineurs, and migraine with aura (MA) which occurs in ~25% of migraineurs. Some people experience both types of attack in their lives. While the precise pathogenesis of migraine is unknown, it is widely accepted that short-term alterations in neuronal activity occur in relation to the attack, along with temporary changes in the cerebral vasculature. Trigeminal nerve activation is also considered pivotal to progression of a migraine attack. Neurotransmitters, especially serotonin (5-hydroxytryptamine, 5-HT), platelet activation and sympathetic hyperactivity all appear to play a part, whether as part of the primary triggering event, or as a response mechanism. Migraine imparts a significant burden on society, both socially and financially. The World Health Organization has identified migraine among the world's top 20 leading causes of disability, with an impact that extends far beyond individual suffering. There is significant evidence from family and twin studies to indicate a strong genetic component to migraine. The current understanding of migraine is that it is a polygenic multifactorial disorder. It has been postulated that genetic factors set the individual migraine threshold, with environmental influences playing a modulating role. It is likely that many genes may provide an important although moderate contribution to an individual s migraine susceptibility. The identification of migraine susceptibility genes has been the focus of substantial research to date and could eventually lead to improved treatments and greater understanding of the disorder. Several loci have shown promise, although these need to be followed up by both replication and functional studies to determine a definitive causative role. i

4 This research investigated the role of both hormonal and vascular related genes as candidate genes that may play a role in migraine susceptibility due to the well-known role of hormones and vascular changes in some migraineurs. The estrogen receptor (ESR) and progesterone receptor (PGR) genes are potential migraine candidates due to the recognized hormonal influence on migraine susceptibility. Migraines in women frequently occur during the childbearing years and are often influenced by significant hormonal milestones. The fluctuating hormone levels of the menstrual cycle have been implicated in migraine but a definitive role is yet to be established. It has been suggested that factors additional to circulating hormone levels may be at play. This research considered that variation in the ESR 1 and PGR genes may confer an increased migraine risk. To investigate the potential role of these genes in migraine, association studies investigating variants in ESR 1 and PGR were undertaken in two independent casecontrol cohorts. This was followed up by mutation screening and gene expression analysis in an effort to elucidate a functional role for these genes in the pathogenesis of migraine. Vascular genes also represent likely migraine candidates as alterations in both vascular function and cerebral blood flow are well known in migraine. Furthermore, cortical spreading depression (CSD), a depolarization wave that propagates across the brain cortex and has been speculated to cause the neurological symptoms that present in MA, has also been linked to vascular dysfunction. The methylenetetrahydrofolate reductase (MTHFR) and methionine synthase (MTRR) genes both play a role in vascular functioning and were thus considered potential migraine candidates for this study. Both are involved in the pathway of homocysteine metabolism. Impaired activity of these enzymes can lead to mild hyperhomocysteinemia which is believed to lead to oxidative arterial damage. This may in turn impact on migraine susceptibility, possibly through the activation of trigeminal fibres. The MTHFR 677T allele results in an amino acid change in the catalytic domain of the enzyme leading to mild hyperhomocysteinemia. This particular variant has been implicated in migraine in four separate studies. One of these studies also suggested a role for the MTHFR 1298C allele in migraine. This allele also results in an amino acid change and reduced enzyme activity. Similarly, the MTRR 66G allele results in an amino acid change and has been associated with reduced activity of MTRR and increased plasma homocysteine concentration. ii

5 To investigate the role of the ESR 1, PGR, MTHFR and MTRR genes in migraine, samples from two large independent case control cohorts were investigated. Cohort 1 was comprised of 275 migraineur samples and 275 age, sex and ethnicity matched controls while cohort 2 comprised 300 cases and 300 matched controls. All individuals were collected at the Genomics Research Centre with migraine diagnosis undertaken by IHS criteria and migraine affected individuals designated as MO or MA. Results of analysis of ESR 1 indicated a positive association with migraine in the two large independent cohorts for the exon 8 G594A polymorphism (P = 0.003; P = 8x10-6 ). Similarly, the PGR analysis showed a positive association with migraine for the PROGINS allele (P = 0.02; P = 0.003). Results also showed that individuals with both ESR 1 and PGR susceptibility alleles were 3.2 times more likely to suffer migraine those those with no susceptibility alleles. As the ESR 1 variant is synonymous, a mutation analysis was undertaken in a small sub-sample of individuals carrying the susceptibility allele, but no mutations were detected in these particular samples. Detailed mutation analysis of ESR 1 in a larger study group may be warranted. An ESR 1 and PGR expression analysis by RT-PCR was undertaken to examine if there were any notable expression level changes in migraineurs versus controls and additionally whether the susceptibility genotypes influenced gene expression. Altered expression levels may point to a functional change in the gene. Although results did not show any significant difference in expression levels in the case/control group, nor any influence in gene expression conferred by the specific susceptibility genotypes, ESR 1 expression did appear to be down-regulated in the migraine group and more specifically in the migraine susceptibility genotype subgroup. A larger study group may therefore be warranted to detect any potential genuine changes in gene expression. Overall, these results suggested that these hormonal genes appear to play a role in migraine susceptibility, although further studies are needed to define this. Results of the MTHFR 677 analysis showed that the TT genotype was significantly associated with the MA subgroup in a joint analysis of the two independent cohorts (P = 0.004). Results of analysis of MTHFR 1298, which is tightly linked to the 677 locus, showed a significant association in female migraineurs (P = 0.009). Similarly, results of the MTRR analysis also showed a significant difference between the female case and iii

6 control groups with the G allele over-represented in female migraineurs (P = 0.022) These results may indicate that a significant gender effect appears in this locus as well as the MTHFR 1298 locus although results may also be due to a larger number of female migraineurs conferring increased statistical power to the gender subgroup. Interaction analysis of the MTHFR 1298 locus and the MTRR locus showed that females who carried both variants under a recessive model were 5 times more likely to suffer migraine those those with no susceptibility genotypes. Overall these results indicated that these vascular genes appear to play a role in migraine susceptibility. The final study focused on 6 genetic variants that had shown a positive association with migraine and/or MA in the same large association population analysed in this research. The aim of this study was to provide preliminary data on the potential role of genetic profiling in migraine. Using the genotypic data to create vascular and hormonal risk profiles based on positive association and interaction of MTHFR 677 T and ACE D alleles, and MTHFR 1298 AA and MTRR GG genotypes as vascular variants; and positive association and interaction of ESR A and PGR PROGINS as hormonal variants, this study was able to demonstrate the relevance of genetic risk profiling to migraine. Results showed a significantly higher proportion of individuals with at least one genetic risk profile in the migraine group compared to those in the control group (P = 6 x 10-6 ). Individuals who possessed either the vascular and/or hormonal genetic risk profile were 8.6 times more likely to suffer from migraine than those who possessed a no risk profile. This indicated a greater effect than the individual effect of each of these variants. Furthermore individuals who possessed a vascular or both risk profiles were more likely to suffer nausea, emesis, phonophobia and photophobia, and have a mother who also suffered migraine. Overall, the genetic profiling approach provided interesting preliminary data on migraine susceptibility and indicated that such an approach may prove very useful for migraine diagnosis, particularly when all migraine genes have been identified. In conclusion this study provided the first indication that hormone receptor genes play a role in migraine susceptibility. Hormones have long been considered to play a role in the disorder but this study has provided the first molecular evidence to support this premise. In addition, this study showed that vascular related genes also play a role in migraine susceptibility. Finally, this study has clearly shown that migraine is a complex disorder iv

7 involving multiple genes. Although a number of studies have implicated neurotransmitter related genes in the disorder, the present study is the first to show that both vascular and hormonal genes also play a role in migraine susceptibility. Thus there now appear to be three classes of genes that affect migraine susceptibility and although this study has implicated new variants, the preliminary genetic profiling study has shown that not all predisposing variants involved in the disorder have been defined. v

8 Acknowledgements There are many people whom I would like to thank for the important role that they played in this work. Firstly, I wish to thank my supervisor, Professor Lyn Griffiths, who gave me the tremendous opportunity to undertake my PhD at the Genomics Research Centre. Lyn took a chance on me and I am extremely grateful for her willingness to take me on as an unknown entity. Lyn has been an excellent supervisor and mentor and she has given me some amazing and valuable opportunities. I will forever be grateful for the valuable role that Lyn has played in this work and in my future. Secondly I would like to thank my husband Richard. Richard has been an incredible support to me throughout this journey, from making me many many of cups of tea and packed lunches to ease the burden of my (oftentimes) very demanding workload, to calmly accepting my decision to go to university for the first time in my life just before my 40 th birthday. Without Richard s unwavering support, I have no doubt that I would not be here writing this acknowledgement. I would also like to thank my family for their support, particularly my parents, who have given me incredible encouragement every step of the way. They never doubted me, even though I frequently doubted myself. A big thankyou must also go out to everyone at the Genomics Research Centre past and present. There are too many people to mention here, but I must give a special thanks to Dr Rod Lea, my second supervisor, for his patience and willingness to help and guide me along the way. Thanks also to Sharon, Mel, Larisa, Merilyn, the two Roberts, Micky, Lotti, Claire, Rachel, Michael, Attila, Kate, Gunter, Linda, Matt, Hannah, Jess, Sherin, Pam, and Angela. You have all played an important role in one way or another. An extra special thanks must go to Laura and Francesca. You have both helped me in so many ways, least of all providing the frequent bursts of comic relief amongst the seriousness of it all. I consider myself very fortunate to have your friendship. vi

9 Statement of Originality The material presented in this report has not been previously submitted for a degree or diploma in any university, and to the best of my knowledge contains no material previously published or written by another person except where due acknowledgement is made in the thesis itself. Natalie Colson vii

10 Table of Contents Introduction General Introduction Aim Significance... 3 Chapter Migraine... 5 Background Migraine Classification and Diagnosis Migraine Subtypes Diagnosis of MO Diagnosis of MA Underdiagnosis of Migraine Migraine Clinical Manifestation Migraine Prevalence Migraine Comorbidity Migraine Pathophysiology The Primary Cause The Vascular Hypothesis The Theory of Cortical Spreading Depression The Brainstem Generator The Trigeminovascular System and Neurogenic Inflammation An Integrated Neurovascular Model Migraine Treatment viii

11 Chapter Migraine Genetic studies Introduction to Complex Genetic Disorders Migraine as a Complex Genetic Disorder Relative Risk Heritibility Mode of Transmission Approaches to Gene Discovery in Complex Disorders Linkage Analysis Association Studies Functional Genomics Molecular Genetics Studies of Migraine FHM FHM FHM FHM Genotype-Phenotype Correlations FHM Loci in Common Migraine Other Loci Linked to Migraine Candidate Gene Studies in Migraine Neurotransmitter Function Dopamine Related Genes Serotonin Related Genes Vascular Function ix

12 Other Implicated Genes Hormone Related Genes Gene Expression Studies in Migraine Summary of Migraine Genetic Studies Chapter Methodology and Research Background Ethical Approval Association Analysis Study Design Population Demographics Clinical Characteristics of the Migraine Group Migraine Treatment Used in the Study Group Sample Collection DNA Purification Genotyping Overview Polymerase Chain Reaction Primer design and preparation Restriction Enzyme Digestion Agarose Gel Electrophoresis Quality Control Statistical Analysis Power Analysis Hardy Weinberg Equilibrium Linkage Disequilibrium Analysis Mutation Screening x

13 3.4 Gene Expression Analysis Study Design Control for Endogenous Variation in Expression Levels Study Population Sample Collection and Storage DNA Extraction RNA Extraction RNA Integrity cdna Synthesis Detection of Possible Genomic Contamination Real-Time Reverse Transcriptase Polymerase Chain Reaction Primer design RT-PCR Chemistry Assay Optimisation Rotor-Gene Reference Gene Selection PCR Efficiencies Quality Control Statistical Analysis of Results Summary of Research Plan Chapter Association Studies of Hormone Receptor Genes Migraine and Hormones The Estrogen Receptor Gene as a Migraine Candidate Gene The Progesterone Receptor as a Migraine Candidate Gene xi

14 4.2 Association Analysis of Estrogen Receptor 1 Regions Estrogen receptor 1 PvuII variant Genotyping Results of PvuII variant analysis Estrogen receptor 1 C325G Variant Genotyping Results of C325G variant analysis Estrogen Receptor 1 G594A Variant Genotyping Results of G594A variant analysis Linkage Disequilibrium Analysis of Estrogen Receptor 1 regions Association Analysis of the Progesterone Receptor PROGINS variant Genotyping Results of PGR PROGINS analysis Interaction Analysis of Estrogen Receptor 1 and Progesterone Receptor Variants Methodology Results of ESR 1 and PGR Interaction Analysis Mutation Screening of the Estrogen Receptor Gene Methodology and Sample Selection Primers Assay Design and Sample Preparation Sequencing Results Summary and Discussion of Association Analysis of Hormone Receptor Genes xii

15 Chapter Expression Analysis of Hormone Receptor Genes Introduction Results of RNA Quality and Integrity Check Results of Genomic Contamination Check Reference Gene Selection Serum Hormone Levels in Study Group Genotyping of the ESR 1 G594A and PGR PROGINS Variants Estrogen Receptor 1 Expression Primers Assay Conditions for ESR cdna concentration Primer Concentration Assay Protocol and Cycling Parameters Results of ESR 1 Expression Analysis Migraine versus Non-migraine Migraine ESR 594A versus ESR 594G Progesterone Receptor Expression Primers Assay Conditions for PGR cdna Concentration Primer Concentration Assay Protocol and Cycling Parameters Results of PGR Expression Analysis Migraine versus Non-migraine xiii

16 Migraine PGR PROGINS versus PGR non-progins Summary and Discussion of Hormone Receptor Gene Expression Analysis 163 Chapter Association Studies of Vascular Genes Migraine and Vascular Function The Methylenetetrahydrofolate Reductase (MTHFR) Gene as a Migraine Candidate Gene The Methylenetetrahydrofolate Reductase Synthase (MTRR) Gene as a Migraine Candidate Gene Association Analysis of MTHFR Regions MTHFR C677T Polymorphism Genotyping Results MTHFR A1298C Polymorphism Genotyping Results Linkage Disequilibrium Analysis of the MTHFR C677T and A1298C Variants Association Analysis of MTRR MTRR A66G Polymorphism Genotyping Results Interaction analysis of MTRR A66G and MTHFR A1298C Methodology Results of MTRR and MTHFR 1298 Interaction Analysis xiv

17 6.7 Summary and Discussion of Association Analyses of Vascular Genes Chapter Genetic Risk Profiling in Migraine Introduction Migraine Risk Profiling Classification Statistical Analysis Results of Genetic Risk Profiling Analysis Endophenotype Profiles Summary and Discussion of Migraine Genetic Risk Profiling Analysis Chapter Discussion and Future Directions Introduction Hormonal Genes and Migraine Vascular Genes and Migraine Genetic Risk Profiling in Migraine Conclusion References Appendices xv

18 List of Figures Page Figure 1.1 The trigeminal nerve and cranial vasculature 17 Figure 1.2 Patterns of medication use among migraineurs 20 Figure 2.1 A model of multifactorial inheritance 25 Figure 3.1 Longest duration of migraine attacks 54 Figure 3.2a Usual duration of migraine attacks 55 Figure 3.2b Usual duration of migraine attacks in females and males 56 Figure 3.3 Frequncy of migraine attacks 56 Figure 3.4 Age of onset 56 Figure 3.5 Total number of attacks 57 Figure 3.6 Individuals who suffer nausea 58 Figure 3.7 Individuals who suffer emesis 58 Figure 3.8 Individuals who suffer phonophobia 59 Figure 3.9 Individuals who suffer photophobia 59 Figure 3.10 Individuals who suffer pulsating head pain 60 Figure 3.11 Individuals who suffer unilateral head pain 60 Figure 3.12 Individuals who report stress as a trigger 61 Figure 3.13 Individuals who report holiday/relaxation as a trigger 62 Figure 3.14 Individuals who report weather changes as a trigger 62 Figure 3.15 Individuals who report red wine as a trigger 63 Figure 3.16 Individuals who report other alcohol as a trigger 63 Figure 3.17 Individuals who report chocolate as a trigger 64 Figure 3.18 Sample chromatogram from an automated sequencing reaction 73 Figure 3.19 Diagram of steroid synthesis from cholesterol 76 Figure 3.20 Agilent 2100 bioanalyzer scan 81 Figure 3.21 The regions that are indicative of intact eukaryote RNA 81 Figure 3.22 Agilent 2100 bioanalyzer electrophoretogram 82 Figure 3.23 Corbett Rotor-Gene 6000 internal mechanism 86 Figure 3.24 Corbett Rotor-Gene 6000 Sybr green detection system 87 Figure 3.25 Typical real-time PCR amplification plot 88 xvi

19 Figure 3.26 Template amplification curves for HPRT1, ESR 1 92 Figure 4.1 Potential sites of action of Ovarian Hormones on the CNS 99 Figure 4.2 Classical and Nonclassic genomic mechanisms of estrogen 100 Figure 4.3 Structure of the estrogen receptor gene 103 Figure 4.4 Agarose gel electrophoretogram of ESR 1 Pvu II genotypes 105 Figure 4.5 Agarose gel electrophoretogram of ESR 1 C325G genotypes 109 Figure 4.6 Agarose gel electrophoretogram of ESR 1 gene exon 8 PCR 113 Figure 4.7 Agarose gel electrophoretogram of PGR gene PCR product 121 Figure 4.8 Chromatogram of base sequencing 132 Figure 4.9 Actual base sequence calls for sequencing 132 Figure 5.1 Electrophoretogram for extracted RNA used in this study 142 Figure 5.2 Agilent 2100 bioanalyzer scan sample Figure 5.3 Electrophoretogram of CAPN1 143 Figure 5.4 Genorm expression stability 144 Figure 5.5a-g Genorm stepwise elimination of unstable reference genes Figure 5.6 Electrophoretogram showing PCR products using ESR 1 RT 150 Figure 5.7 Melt curve analysis for ESR 1 and HPRT1 151 Figure 5.8 Linear scale of ESR 1 amplification plot 152 Figure 5.9 Log scale of ESR 1 amplification plot 152 Figure 5.10 ΔCt values of ESR 1 in migraine and control 154 Figure 5.11 ΔCt values of ESR 1 594A and control 594G 155 Figure 5.12 Melt curve analysis for PGR and HPRT1 157 Figure 5.13 Linear scale of PGR amplification plot 158 Figure 5.14 Log scale of PGR amplification plot 159 Figure 5.15 ΔCt values of PGR in migraine and control 161 Figure 5.16 ΔCts of PGR PROGINS and control no-progins samples 162 Figure 6.1 The pathway of homocysteine metabolism 168 Figure 6.2 Agarose gel electrophoretogram of MTHFR 677 genotypes 172 Figure 6.3 Agarose gel electrophoretogram of MTHFR 1298 genotypes 176 Figure 6.4 Agarose gel electrophoretogram of MTRR 66 genotypes 181 Figure 7.1 Comparison of risk and no risk gene profiles 195 xvii

20 Figure 7.2 Individuals in the risk categories who suffered nausea 198 Figure 7.3 Individuals in the risk categories who suffered emesis 198 Figure 7.4 Individuals in the risk categories who suffered phonophobia 199 Figure 7.5 Individuals in the risk categories who suffered photophobia 199 Figure 7.6 Individuals who suffered pulsating head pain 200 Figure 7.7 Individuals who suffered eye movement discomfort 200 Figure 7.8 Individuals whose mother suffered migraine 201 xviii

21 List of Tables Page Table 1.1 Studies of migraine co-morbidity with psychiatric disorders 13 Table 1.2 Potential side effects of migraine preventative treatments 19 Table 2.1 List of migraine genome-wide linkage studies 44 Table 2.2 List of published association studies on migraine 45 Table 3.1 Summary of specific migraine symptoms in the study group 60 Table 3.2 Summary of reported triggers in the migraine study group 64 Table x 2 table of risk and no risk allele carriers 70 Table 3.4 Genetic association study power calculation 71 Table 3.5 cdna synthesis protocol 83 Table 3.6 Calpain PCR protocol 84 Table 3.7 Internal control genes and primer sequences 90 Table 4.1 ESR 1 Pvu II PCR protocol 104 Table 4.2 Distribution of ESR 1 Intron 1 Pvu II polymorphisms 106 Table 4.3 Results of ESR 1 Intron 1 Pvu II polymorphism 107 Table 4.4 ESR 1 C325G PCR protocol 108 Table 4.5 Distribution of ESR 1 Exon 4 Codon C325G 110 Table 4.6 Results of ESR Exon 4 Codon C325G 110 Table 4.7 ESR 1 G594A PCR protocol 112 Table 4.8 Distribution of ESR 1 Exon 8 Codon G594A in MAP Table 4.9 Results of chi-squared analysis of the ESR 1 Exon 8 G594A 114 Table 4.10 Distribution of ESR 1 Exon 8 G594A in MAP Table 4.11 Results of chi-squared analysis of the ESR 1 Exon 8 G594A 116 Table 4.12 Distribution of ESR 1 Exon 8 Codon G594A polymorphism frequencies in a meta-analysis of MAP 1 and MAP Table 4.13 Results of chi-squared analysis of the ESR Exon 8 Codon G594A polymorphism in in a meta-analysis of MAP 1 and MAP Table 4.14 Linkage disequilibrium D values 119 Table 4.15 PGR PROGINS PCR protocol 120 xix

22 Table 4.16 Distribution of PGR PROGINS in MAP Table 4.17 Results of PGR PROGINS in MAP Table 4.18 Distribution of PGR PROGINS in MAP Table 4.19 Results of PGR PROGINS in MAP Table 4.20 Distribution of PGR PROGINS in a meta-analysis of MAP 1 and MAP Table 4.21 Results of PGR PROGINS in in a meta-analysis of MAP 1 and MAP Table 4.22 MAP 1 & MAP 2 distribution for PGR PROGINS, ESR 1 594A 127 Table 4.23 ESR 1 primers used to sequences exons Table 4.24 ESR 1 sequencing PCR protocol. 130 Table 5.1 Serum-estradiol and progesterone levels for expression study sample 147 Table 5.2 ESR 1 G594A and PGR PROGINS genotype for each sample 148 Table 5.3 ESR 1 RT-PCR protocol 151 Table 5.4 ΔCt values of the migraine and control samples for ESR Table 5.5 ΔCt values of the migraine ESR 1594A and control 594G 155 Table 5.6 PGR RT-PCR protocol 158 Table 5.7 ΔCts of the migraine and control samples for PGR 160 Table 5.8 ΔCts of the migraine PGR PROGINS and control no-progins 162 Table 6.1 MTHFR 677 PCR reaction protocol 171 Table 6.2 Distribution of the MTHFR C677T in MAP Table 6.3 Results of MTHFR C677T in MAP Table 6.4 MTHFR C677T frequencies in MAP 1 & Table 6.5 Results of MTHFR C677T polymorphism MAP 1 &2 174 Table 6.6 MTHFR 1298 PCR protocol 175 Table 6.7 Distribution of MTHFR A1298C in MAP Table 6.8 Results of MTHFR A1298C in MAP Table 6.9 Frequency distribution of MTHFR 677 and Table 6.10 Linkage disequilibrium MTHFR 677 and 1298 in MAP Table 6.11 PCR reaction protocol for MTRR A66G 180 Table 6.12 Distribution of MTRR A66G in MAP xx

23 Table 6.13 Results of MTRR A66G in MAP Table 6.14 Distribution of MTHFR C677T and MTRR A66G 183 Table 6.15 Distribution of MTHFR A1298C and MTRR A66G 185 Table 7.1. Independent & interaction analysis of MTHFR & ACE 192 Table 7.2 Chi-squared analysis for Migraine Risk Profiling 194 Table 7.3 Chi-squared analysis for Migraine Risk Profiling risk v no risk 194 Table 7.4 Number and percentage of genetic risk categories 197 xxi

24 Abbreviations BP Base pair BLAST Basic Local Alignment Search Tool BSA Bovine serum albumin cdna Complementary deoxyribonucleic acid CGRP Calcitonin gene related peptide Cl Chloride CSD Cortical Spreading Depression Ct Cycle threshold DNA Deoxyribonucleic acid dntp Deoxynucleotide triphosphate DTT Dithiothreitol EDTA Ethylenediaminetetraacetic acid ESR Estrogen receptor HLA Human leukocyte antigen GABA γ-aminobutyric acid GEM Genetic Epidemiology of Migraine GRC Genomics Research Centre Hcy Homocysteine HRM High resolution melt curve analysis IHC International Headache Committee of the International Headache Society inos Inducible nitric oxide synthase K Potassium Kb Kilobase L Litre LD Linkage Disequilibrium LED Light emitting diodes MAP 1 Migraine association population 1 MAP 2 Migraine association population 2 MAPK Mitogen-Activated Protein Kinase MgCl 2 Magnesium chloride mrna Messenger RNA xxii

25 MTHFR Methylenetetrahydrofolate reductase MS Methionine synthase MTRR Methionine synthase reductase NO Nitric Oxide NOS Nitric Oxide Synthase NTC Non-template control OCP Oral contraceptive pill OR Odds ratio P Probability PCR Polymerase chain reaction PET Positron emission tomography PGR Progesterone receptor pmol Picomoles PMT Photomultiplier PPI Polyphosphoinositide RFLP Restriction fragment length polymorphism RNA Ribonucleic acid RPM Rotations per minute RT-PCR Reverse transcriptase polymerase chain reaction SD Standard deviation SEM Standard error of the mean SST Serum separator tube Taq Thermas aquaticus Tm Melting temperature Var Variance 5-HT Serotonin xxiii

26 Publications Arising from Work Described in This Thesis 1. Colson NJ, Lea RA, Quinlan S, MacMillan J, Griffiths LR. The estrogen receptor 1 G594A polymorphism is associated with migraine susceptibility in two independent case/control groups. Neurogenetics. 2004;5(2): Colson NJ, Lea RA, Quinlan S, MacMillan J, Griffiths LR. Investigation of hormone receptor genes in migraine. Neurogenetics. 2005;6(1): Colson NJ, Lea RA, Quinlan S, Griffiths LR. The role of vascular and hormonal genes in migraine susceptibility. Molecular Genetics and Metabolism. 2006;88(2): Colson NJ, Lea RA, Quinlan S, Griffiths LR. No role for estrogen receptor 1 gene intron 1 Pvu II and exon 4 C325G polymorphisms in migraine susceptibility. BMC Medical Genetics. 2006;28: Colson NJ, Fernandez F, Lea RA, Quinlan S, Griffiths LR. (2006) The Search for Migraine Genes: an overview of current knowledge. Cellular and Molecular Life Sciences. In Press. (Accepted) 6. Liu A, Colson NJ, Quinlan S, Peterson M, Lea RA, Griffiths LR. (2006) Endophenotype profiles of MTHFR and ACE gene polymorphisms in migraine susceptibility. Cephalagia. Under review. Related Publications 7. Johnson MP, Lea RA, Colson NJ, Macmillan JC, Griffiths LR. A population genomics overview of the neuronal nitric oxide synthase (nnos) gene and its relationship to migraine susceptibility. Cellular and Molecular Biology (Noisy-le-grand) Sep 5;51(3): Fernandez F, Lea RA, Colson NJ, Bellis C, Quinlan S, Griffiths LR. Association between a 19 bp deletion polymorphism at the dopamine beta-hydroxylase (DBH) locus and migraine with aura. Journal Neurological Science Nov Tajouri L, Fernandez F, Tajouri S, Detriche G, Szvetko A, Colson N, Csurhes P, Pender MP, Griffiths LR. Allelic variation investigation of the estrogen receptor within an Australian multiple sclerosis population. Journal Neurological Science Nov Szvetko AL, Fowdar J, Nelson J, Colson N, Tajouri L, Csurhes PA, Pender MP, Griffiths LR. No association between MTHFR A1298C and MTRR A66G polymorphisms, and MS in an Australian cohort. Journal Neurological Science Nov 17. xxiv

27 11. Fernandez F, Curtain R, Colson N, Lea R, Griffiths LR. Association Analysis of Chromosome 1 Migraine Candidate Genes. BMC Genetics, Under review. xxv

28 Conference Presentations Arising from Work Described in This Thesis ORAL PRESENTATIONS 2004 L.R. Griffiths, R.A. Lea, D.R. Nyholt, M.P. Johnson, N. J. Colson, R.P. Curtain, S. Quinlan, J. MacMillan, R. A. Gibson and L. C. McCarthy. Mapping the genes involved in migraine susceptibility. XI Congress of the International Headache Society/ IHC 2003, Rome, Italy, September L.R. Griffiths, R.A. Lea, M.P. Johnson, N.J. Colson, R.P. Curtain, S. Quinlan, and J. MacMillan. Migraine Susceptibility: Role for Vascular and Hormonal Gene Variants. 54th ASHG Annual Meeting, Toronto, Canada, October N. Colson, R.A. Lea, S. Quinlan, L.R. Griffiths. Genetic risk profiling in migraine: hormonal and vascular genotypes and susceptibility to migraine. XII Congress of the International Headache Society/ IHC 2005 Kyoto, Japan, October ABSTRACTS 2002 N.J. Colson, R.A. Lea, M. Ovcaric, R. Sciascia, R.P. Curtain, S. Quinlan, J. Curran and L.R. Griffiths Population association study of hormone receptor genes and migraine. Australian Society of Medical Research Student Conference, Brisbane, Australia, June N.J. Colson, R.A. Lea, M. Ovcaric, R. Sciascia, R.P. Curtain, S. Quinlan, J. Curran and L.R. Griffiths. Population Association Study of Hormone Receptor Genes and Migraine. Australian Health & Medical Research Congress, Melbourne, Australia, November, L.R. Griffiths, R.A. Lea, M.P. Johnson, N. Colson, R.P. Curtain, S. Quinlan, J.MacMillan, R.A. Gibson, L.C. McCarthy. Mapping the genes involved in migraine susceptibility. XIX International Congress of Genetics, Melbourne, Australia, July N.J. Colson, R.A. Lea, S. Quinlan, J. Curran, L.R. Griffiths. Gender specific differences in progesterone receptor gene variants. International Congress of Genetics, Melbourne, Australia, July L.R. Griffiths, R.A. Lea, D.R. Nyholt, M.P. Johnson, N.J. Colson, R.P. Curtain, S. Quinlan, and J. MacMillan. Implications of vascular and hormonal gene variants in migraine susceptibility. 4th "GeneMappers", Perth, Australia, August xxvi

29 L.R. Griffiths, R.A. Lea, N. Colson, M.P. Johnson, R.P. Curtain, S. Quinlan, J. MacMillan. Vascular and hormonal variants are implicated in migraine susceptibility. 5th HUGO Pacific Meeting & 6th Asia-Pacific Conference on Human Genetics, Singapore, November, Griffiths L.R., Colson N.J, Lea R.A., Johnson M.P., Curtain R.P., Quinlan S. and MacMillan J. The Role of Vascular and Hormonal Gene Variants in Migraine Susceptibility. European Society of Human Genomics, Prague, Czech Republic, May Griffiths L.R., Lea R.A., Colson N.J., Johnson M.P., Quinlan S, MacMillan J. Mapping the Genes Involved in Migraine Susceptibility. 11th World Congress on Pain, International Association for the Study of Pain, Sydney, Australia, August Colson N.J. Lea R.A. Quinlan S.A. Griffiths L.R. Migraine susceptibility genotypes and B vitamins as a prophylactic treatment. International Conference of Human Genetics, Brisbane, Australia, August xxvii

30 Head pounding, ears buzzing, stomach churning. all around me, the world keeps turning. But not mine, no, it's suspended in timein this painful, dreadful, so-called life of mine. I've become a prisoner in my own home; as I lie here enduring this painwhile everyone else continues to live. No one knows or understands what I feel. If they did, surely they'd rescue me from this hell! My mind is slowly slipping away from meas I cry and cry- wondering if I'll ever be free!?! All of the doctors PRETEND to try, but if they're trying so hard, then WHY???? by Sherry Letford ( This poem was written by a chronic migraine sufferer and describes the author s intense pain and despair that she suffers during a migraine attack. Her poem so descriptively conveys the misery and hopelessness felt by those who suffer from migraine. While there is much work to be done, it is sincerely hoped that the research undertaken in this work may contribute in some small way to one day helping ease the burden of suffering for the many people worldwide who endure this debilitating disorder. 1

31 Introduction 1.1 General Introduction Migraine is a common neurovascular disorder that shows strong familial aggregation. Migraine is considered to be genetically complex with a multifactorial mode of inheritance (Larsson et al. 1995; Gervil et al. 1999; Gervil et al. 1999; Ulrich et al. 1999). Migraine is characterized by severe head pain with associated nausea, emesis, photophobia, phonophobia, and neurological disturbances (HCCIHS 1988). Although there is variation in different populations, the average prevalence of migraine in the general population may be estimated at ~18% in females and 6% in males (Lipton and Stewart 1997; Lipton et al. 2001; Wang 2003), although considering the reports of migraine underdiagnosis, prevalence rates may be substantially higher (Maizels 2001; Lipton et al. 2002). Evidence suggests that migraine prevalence may be increasing (Lipton et al. 1997). Family and twin studies have indicated a significant genetic component to migraine and it appears likely that numerous susceptibility gene variants provide a small yet nonetheless significant contribution to the varied phenotypic expression of the disorder. The identification of susceptibility genes for complex traits such as migraine can be challenging. Not only is there the likely confounding influence of environmental factors, there is also the probability that multiple and potentially interacting modest effect genetic loci may contribute to the disorder. Furthermore classifying migraineurs for genetic studies can be problematic as migraine displays clinical and genetic heterogeneity and could also occur as a symptom of another undiagnosed cerebral disorder (Ducros et al. 2002). Although the diagnostic criteria set out by the International Headache Society has improved migraine diagnosis (HCCIHS 1988; Ferrari 1998), the fact that migraine diagnosis must currently rely on patient reported symptoms further confounds the problem. In recent years there has been immense research interest focusing on the molecular genetics underlying the pathophysiology of migraine. Genetic linkage and association studies have both shown some interesting results to date, however follow-up 2

32 confirmatory studies are sparse and at present the type and number of genes involved in migraine are not clear. Rigourous research is needed to determine the definitive genetic causes of the disorder. 1.2 Aim The aim of this research was to investigate the potential role of vascular and hormonal genes as candidate genes in migraine pathogenenesis. Three approaches were used. The initial approach was genetic association analysis in large case/control cohorts. The second approach aimed to further investigate candidate genes implicated in association studies by analysing their expression levels in migraineurs and unaffected individuals, and to determine if genotype differences affected expression levels. This type of analysis sought to investigate potential functional gene changes that may provide clues to the pathological features and processes involved in the disorder. The final approach was to explore migraine genetic risk profiling as a potential tool for diagnostic and therapeutic use. 1.3 Significance Migraine is a chronic and debilitating neurological disorder that affects approximately 12% of the world s population (Lipton and Stewart 1997). Apart from the significant toll in human suffering, migraine also imposes a serious economic burden on society due to the associated costs of medical care, treatment, and lost productivity. Diagnosis is problematic and misdiagnosis can lead to incorrect therapy or failure to appropriately treat the disorder. In addition, the efficacy of current treatment options is variable. The pathophysiology of migraine is not fully understood, and in the absence of any biological marker, diagnosis must be based on purely subjective criteria alone. It has been estimated that nearly one half of patients with migraine are undiagnosed (Lipton et al. 2002; Lipton et al. 2003) and less than one-half of those who met the International Headache Society Criteria for Migraine were currently using prescription medication (Celentano et al. 1992). 3

33 The World Health Organization has identified migraine among the world's top 20 leading causes of disability, with an impact that extends far beyond individual suffering (WHO 2001). The Australian Brain Foundation reports the annual cost of migraine in Australia to be ~$44 million on medical costs, and up to $670 million due to absenteeism and lost productivity (Parry 1992). The personal burden of migraine cannot be understated. Not only must sufferers contend with the pain and disability during the attack, there is also the impact of the unpredictability of future attacks causing additional anxiety for the migraineur. The identification of migraine susceptibility genes would aid in a greater understanding of the mechanisms that underlie the disorder and lead to significant therapeutic and diagnostic applications. The current understanding of migraine is that many genes may provide an important although small contribution to an individual s migraine susceptibility (Peroutka 2002). Due to the threshold nature of the illness, it is possible that targeting and treating one small aspect of one s susceptibility, if it can be identified, may control individual migraine susceptibility and greatly reduce the burden of the disorder. 4

34 Chapter 1 Migraine Background 5

35 1.1 Migraine Classification and Diagnosis Migraine is a primary headache disorder characterised by recurrent attacks of disabling head pain, which may be accompanied by nausea and emesis, and neurological disturbances (Ducros et al. 2002). The earliest classification of migraine syndrome dates back to around 80AD, when Aretaeus of Cappodocia first described "heterocrania" as a one-sided headache with blackness before the eyes, nausea, vomiting, photophobia and osmophobia (Koehler and van de Wiel 2001). Current understanding of migraine has progressed significantly, although in the absence of any biochemical or radiological marker, migraine diagnosis must be based on patient reported symptoms alone (Ducros et al. 2002). In 1988 The International Headache Society published an International Classification of Headache Disorders which provides classification and diagnosis criteria for headache and migraine. It distinguishes different varieties of migraine based on certain clinical features. This system of classification, which has recently been updated (HCCIHS 2004), is considered an important guideline for migraine diagnosis and management (Linde 2006). The two most frequent migraine sub-types are migraine with aura, previously known as classic migraine, and migraine without aura, previously known as common migraine (HCCIHS 1988). Migraine without aura, which occurs in ~75% of migraineurs (Ferrari 1998), is characterised by moderate to severe head pain that is generally unilateral and pulsating, and exacerbated by physical activity. Nausea, along with phonophobia and photophobia may occur (HCCIHS 2004). Migraine with aura, which occurs in ~33% of sufferers (Ferrari 1998), is classified by the existence of focal neurological symptoms preceding or accompanying the headache (HCCIHS 2004). Neurologic symptoms may include focal paresthesia or weakness, visual or auditory hallucination, vertigo, fainting, or a confusional episode. Some sufferers experience both types of attack during their life (Ducros et al. 2002). A confounding factor in the classification and diagnosis of migraine is the fact that recurrent migraine attacks may result as a symptom of a range of organic cerebral disorders such as cerebral arteriopathies, or other underlying pathologies (Ducros et al. 2002). 6

36 1.1.1 Migraine Subtypes In the Classification of Primary Migraine Disorders provided by The Headache Classification Subcommittee of the International Headache Society the following subtypes of migraine are recognised:- 1.1 Migraine without aura 1.2 Migraine with aura Typical aura with migraine headache Typical aura with non-migraine headache Typical aura without headache Familial Hemiplegic Migraine (FHM) Sporadic Hemiplegic Migraine Basilar Type Migraine 1.3 Childhood periodic syndromes that are common precursors of migraine Cyclical vomiting Abdominal migraine Benign paroxysmal vertigo of childhood 1.4 Retinal migraine 1.5 Complications of migraine Chronic migraine Status migrainosus Persistent aura without infarction Migraineous infarction Migraine-triggered seizure 1.6 Probable Migraine Probable Migraine without aura Probable Migraine with aura Probable chronic migraine 7

37 1.1.2 Diagnosis of MO The Headache Classification Subcommittee of the International Headache Society Classification of Primary Migraine Disorders outlines the diagnostic criteria for migraine without aura as follows:- Migraine without aura 1. At least 5 attacks fulfilling Headache attacks lasting 4-72 hours (untreated or unsuccessfully treated) 3. Headache has at least two of the following four characteristics: a. unilateral location b. pulsating quality c. moderate or severe intensity which inhibits or prohibits daily activities d. aggrevated by walking stairs or similar routine physical activity 4. During headache at least one of the two following symptoms occur: a. nausea and/or vomiting b. photophobia and phonophobia 5. The history and physical and neurologic examinations do not suggest headache due to a secondary disorder, or appropriate investigations rule out a secondary disorder. A diagnosis of migraine without aura may be made even if a secondary disorder is found, provided that the migraine attacks did not start at about the same time the secondary disorder started Diagnosis of MA The Headache Classification Subcommittee of the International Headache Society Classification of Primary Migraine Disorders outlines the diagnostic criteria for migraine with aura as follows:- Migraine With aura 1. At least two attacks fulfilling Headache has at least three of the following four characteristics: 8

38 a. one or more fully reversible aura symptoms indicating focal cerebral cortical and/or brain stem dysfunction b. at least one aura symptom develops gradually over more than 4 minutes, or tow or more symptoms occur in succession c. no aura symptom lasts more than 60 minutes; if more than one aura symptom is present, accepted duration is proportionally increased d. headache follows aura with a free interval of less than 60 minutes (it may also begin before or simultaneously with the aura) 3. The history and physical and neurologic examinations do not suggest headache due to a secondary disorder, or appropriate investigations rule out a secondary disorder. A diagnosis of migraine with aura may be made even if a secondary disorder is found, provided that the migraine attacks did not start at about the same time the secondary disorder started. (HCCIHS 2004) Underdiagnosis of Migraine Despite the wide acceptance of these diagnostic criteria in clinical practice (HCCIHS 2004), migraine has a history of underdiagnosis and undertreatment. This fact was highlighted in the American Migraine Study II, a population-based survey of 20,000 households conducted in 1999 describing the patterns of migraine diagnosis and medication use in a representative sample of the US population (Lipton et al. 2001). This study of individuals reported that approximately half (48%) of IHS-defined migraineurs (41% of males and 51% of females) reported a physician diagnosis of migraine. A number of factors have been suggested to explain why individuals with symptoms of migraine do not report a physician diagnosis. Migraineurs may not seek medical care for migraine; they may seek care but not receive a diagnosis; or they may be diagnosed but fail to remember a diagnosis. Poor patient-physician communication about migraine is reported as a barrier to appropriate care and compounded by the fact that physicians rarely see patients during a migraine and must rely on patients' retrospective descriptions of their symptoms (Lipton et al. 2001). 9

39 Lipton et al. also report that sociodemographic characteristics were associated with selfreported physician diagnosis of migraine and that low income, young age (18 to 29 years), and male gender were associated with a lower probability of being diagnosed. Clinical characteristics including severe disability as well as symptoms such as nausea, vomiting, and aura were also associated with physician diagnosis (Lipton et al. 2001). 1.2 Migraine Clinical Manifestation The migraine attack can consist of five distinct phases: the prodrome, occurring hours or days before the headache; the aura, which come immediately before the headache; the headache itself; the headache termination; and the postdrome (Blau 1980). A description of the five phases in the manifestation of migraine follows, however it should be noted that while most migraineurs experience at least one phase, no one particular phase is required for the diagnosis of migraine (Silberstein and Lipton 1994). The prodrome occurs in 20-60% of migraineurs hours or days before an attack and may include symptoms of hyperactivity or alternatively tiredness and yawning, food cravings, neck stiffness and depression. This phase allows some migraineurs to predict their migraine headache beforehand. In sufferers of migraine with aura, the aura phase which consists of focal neurological symptoms (visual, sensory, motor, language or brainstem) usually precedes or accompanies the attack. The headache phase can occur at any hour but frequently manifests as mild pain upon wakening, developing to a peak and then gradually diminishing to resolution, commonly although not always during sleep (Olesen 1978). The duration of the headache phase may be related to triggering factors and age (Linde 2006). During the resolution or postdromal phase migraineurs frequently report feeling tired, depressed or alternatively refreshed and euphoric (HCCIHS 2004; Silberstein 2004). 10

40 1.3 Migraine Prevalence Prevalence relates to the number or percentage of existing cases of a particular disease or disorder in a given population at a designated time. Although there is variation in different populations, the average prevalence of migraine in the general population has been estimated at ~12% (Lipton and Stewart 1997; Lipton et al. 2001; Wang 2003), although considering the reports of migraine underdiagnosis, it has been suggested that prevalence rates may be substantially higher (Maizels 2001; Lipton et al. 2002; Tepper et al. 2004). A 2003 survey of over 1000 random individuals in five countries showed the percentage of individuals who met the IHS criteria for migraine was 5% in France, 7% in UK, 11% in Germany, 11% in USA and 12% in Italy with an overall average prevalence of 9% (MacGregor et al. 2003). Females, at 18% are three times more likely to suffer the disorder than males at 6% (Stewart et al. 1992; Lipton and Stewart 1997; Rasmussen 1999). Prevalence increases more rapidly in girls than boys as adolescence approaches, and continues until mid adulthood (Lipton and Stewart 1997). Evidence suggests that migraine prevalence may be increasing (Sillanpaa and Anttila 1996). 1.4 Migraine Comorbidity Comorbidity refers to a greater than coincidental occurrence of two conditions in the same individual (Lipton and Silberstein 1994). There are four possible explanations for findings of comorbidity, it may arise due to coincidence, one condition may cause the other, there may be shared environmental or genetic risk factors, or these risk factors may produce a state that leads to both conditions (Lipton and Silberstein 1994). The study of comorbidity may be useful to determine treatment options, and moreover to provide clues to the mechanism behind an illness. Epidemiological studies have demonstrated that migraine is associated with stroke, epilepsy and psychiatric conditions (Lipton and Stewart 1998). The association between migraine and ischemic stroke has been shown in many studies (Tzourio et al. 1993; Carolei et al. 1996; Etminan et al. 2005) with strongest evidence shown for stroke in young women, particularly with MA and even more so for women 11

41 who smoke or use oral contraceptives (Sacco et al. 2006). However, it is interesting to note in the Genetic Epidemiology of Migraine (GEM) study on cardiovascular risk factors and migraine (Scher et al. 2005), migraineurs were more likely to smoke and to report a parental history of early myocardial infarction. Migraineurs with aura were more likely to have an unfavourable cholesterol profile, to have elevated blood pressure and to report a history of early onset coronary heart disease or stroke, while female migraineurs with aura were more likely to be using oral contraceptives (Scher et al. 2005). Nevertheless, in the absence of major cardiovascular risk factors, migraine remains associated with stroke (Diener and Kurth 2005). Thus research suggests that migraineurs may be more susceptible to stroke due to shared environmental as well as genetic risk factors. Several studies have reported a positive association between migraine and psychiatric conditions (Breslau et al. 1991; Breslau and Davis 1993; Breslau et al. 1994; Merikangas et al. 1994; Lipton et al. 2000; Swartz et al. 2000; McWilliams et al. 2004; Oedegaard et al. 2006). The following table 1.1, adapted from Radat and Swendsen (2005) and updated, summarises population based studies of migraine co-morbidity with psychiatric disorders using IHS headache diagnostic criteria. Findings of all studies illustrate that migraine sufferers are at higher risk of depression and various other psychiatric conditions (in particular panic, anxiety and phobic disorders). 12

42 Table 1.1 Population based studies of migraine comorbidity with psychiatric disorders using IHS headache diagnostic criteria Study Sample Assessment Sumary of Results Oedegaard et al, 2005 n=49205 Norway HADS Significant association between DEP, COM for female MA McWilliams et al, 2004 n=3032, USA DSM-IIIR Significant association between DEP, PA, ANX Breslau et al, 2001 and n=1696, USA DSM-IV Significant association between 2000 migraine/severe headache and PA and DEP Lipton et al, 2000 n=768, USA, UK PRIME-MD Significant association between migraine and DEP Swartz et al, 2000 n=1343, USA DSM-III/R Significant association between Migraine and DEP, PA, PHO Wang et al, 1999 n=1421, Taiwan GDS Significant association between migraine and DEP Breslau et al, 1994, 1993, 1991 n=1007, USA DSM-IIIR Significant association between migraine and DEP, ANX, PHO Merikangus et al, 1993 and 1994 n=379, Switzerland DSM-III/R Significant association between migraine and DEP, PA, PHO, ANX HADS = Hospital Anxiety and Depression Scale; DEP = depression; COM = comorbid depression with anxiety disorder; DSM = Diagnostic and Statistical Manual of Mental Disorders; PA = panic attacks/panic disorder; ANX = anxiety; PRIME-MD = The Primary Care Evaluation of Mental Disorders; PHO = phobia; Geriatric Depression Scale Further research is needed to determine whether psychiatric disorders are themselves a causal factor in the development of migraine, whether migraine is a causal factor in the development of psychiatric conditions (e.g. repeated and intense pain leads to anxiety and other behavioural or cognitive risk factors for psychiatric syndromes), or whether both share a common shared aetiological factor (such as a common genetic factor) which may explain the co-occurrence of both syndromes without a causal association between them (Radat and Swendsen 2005). 13

43 1.5 Migraine Pathophysiology The Primary Cause The primary cause of migraine is not fully understood although a large body of research has led to the proposal of various theories. The following discussion reviews recent developments in the understanding of the cause and pathogenesis of migraine and outlines the current integrated model of our understanding of migraine pathophysiology based on research to date The Vascular Hypothesis The oldest theory of migraine pathophysiology, once widely accepted, was based on the observations of Graham and Wolff (1938) and considered an underlying vascular disorder. Wolff and colleagues, observed that stimulation of the cerebral and meningeal blood vessels produced severe headaches in conscious patients while stimulation of the brain parenchyma did not (Parsons and Strijbos 2003). They hypothesised that transient ischemia resulting from intracranial vasoconstriction was responsible for the migraine aura, while rebound vasodilation and activation of perivascular nociceptive nerves resulted in the headache pain. Supporting this theory were observations that extracranial vessels became distended and pulsatile during an attack. Furthermore, vasoconstrictors improved the headache pain, while vasodilators tended to provoke an attack (Wolff 1963). However it has since been shown that the headache phase of migraine starts while blood flow is still reduced therefore the pain of migraine cannot be due simply to vasodilation (Olesen et al. 1990; Cutrer et al. 1998). In addition, vascular changes do not occur in all migraineurs (Pietrobon and Striessnig 2003) The Theory of Cortical Spreading Depression The theory of cortical spreading depression (CSD), originally proposed by Leao (1944), predicted that blood flow changes seen in migraine developed as a result of a spreading 14

44 depression, a wave of intensive neural activity followed by a wave of depolarisation that spread across the cerebral cortex (Leao 1944). This wave of depolarisation was considered to be associated with the aura of migraine. In 2002 Bolay et al provided evidence in animal models that CSD activated trigeminal afferents causing inflammatory changes (vasodilation, oedema and protein extravasation) in the pain sensitive meninges, which are heavily innervated by the trigeminal nerve (Bolay et al. 2002), lending support to Leao s original theory. While CSD has not yet been demonstrated in the human cerebral cortex, characteristics of CSD in humans have been shown (Hadjikhani et al. 2001). However, while CSD and vascular changes are able to explain some aspects of the migraine attack in some patients, neither fully explain all the events that are considered to take place during an attack, nor address the triggering mechanism that precipitates an attack (Edvinsson 1999; Parsons and Strijbos 2003) The Brainstem Generator An episodic dysfunction of the brainstem nuclei that are involved in central pain processing has been proposed as the primary cause of migraine (Pietrobon and Striessnig 2003). Support for this view is shown in a 1995 study using positron emission tomography (PET) to examine the changes in regional cerebral blood flow in 9 migraine patients as an index of neuronal activity in the human brain during migraine. During right-sided migraine attacks, increased blood flow was found in the left brainstem. Interestingly brainstem activation persisted after the injection of sumatriptan had induced complete relief from headache and phono- and photophobia (Weiller et al. 1995). These findings support the idea that the pathogenesis of migraine is related to an imbalance in activity between brain stem nuclei regulating antinociception and vascular control. Similar studies have shown activation in the dorsal rostral brainstem in a patient with MO (Bahra et al. 2001), the red nucleus and substantia nigra in a patient with MA (Welch et al. 1998) and the dorsal lateral pons in 24 migraine patients (Afridi et al. 2005). Whether these changes form part of a unique response to brain dysfunction elsewhere, or are themselves the crucial changes, remains to be determined. 15

45 1.5.2 The Trigeminovascular System and Neurogenic Inflammation It is well accepted that the trigeminovascular system plays a pivotal role in migraine. Animal studies investigating the relationship between the trigeminal nerve and the cranial vasculature have shown that trigeminovascular axons from blood vessels of the pia mater and dura mater can release vasoactive peptides producing a sterile inflammatory reaction (Moskowitz 1984; Markowitz et al. 1988). During this reaction, stimulation of the trigeminal ganglion induces the release of vasodilatory peptides such as calcitonin gene related peptide (CGRP), substance P and neurokinin A (Moskowitz 1992; Moskowitz 1993). This reaction, termed neurogenic inflammation may lower the nociceptive threshold required to stimulate meningial sensory fibres (Moskowitz 1990) and also act on vascular tissues to cause vasodilatation, plasma protein extravasation in the surrounding area, endothelial changes, platelet aggregation and subsequent release of serotonin (5-hydroxytryptamine, 5-HT) and other mediators, white cell adhesion and inflammation (Dimitriadou et al. 1991; Dimitriadou et al. 1992; Arulmozhi et al. 2005). Figure 1.1 illustrates this mechanism and the structures involved. In support of this theory, human studies have shown increased CGRP in the extracerebral circulation during migraine attacks (Goadsby et al. 1990). Further support comes from the findings that specific antimigraine drugs, including sumatriptan, are able to suppress plasma extravasation produced by antidromic stimulation of trigeminal nerve terminals in the rodent meninges (Moskowitz and Buzzi 1991). Stimulation of trigeminal nerves is regarded as a necessary step for the process that produces the pain and associated symptoms of the migraine attack. 16

46 Figure 1.1 Depicted in this image are the trigeminal nerve (yellow) and cranial vasculature which, according to current theories of migraine, are considered to be central to migraine pathophysiology. When stimulated, the trigeminal nerve releases neuropeptides into dural and meningeal blood vessels leading vasodilation and exudation of plasma into the tissues. The resulting inflammation and swelling of the blood vessels can cause distension of cranial arteries and headache pain An Integrated Neurovascular Model The current understanding of migraine based on clinical investigations, as well as the characteristic activities of the wide range of drugs known to provide relief from migraine favours an integrated neurovascular hypothesis. Cortical spreading depression, vasospasm, serotonin metabolism, platelet activation and sympathetic hyperactivity may all to play a part, whether as part of the primary triggering event, or as a response mechanism. Once the indefinable initiating migraine trigger has occurred, cerebral blood flow decreases. It is likely that this event is either followed, or preceded by a wave of cortical spreading depression (CSD) although it should be noted that at present, the role of CSD in MO is speculative (Sanchez-Del-Rio et al. 2006). Vasodilation then follows, most likely due to changes in the activity of neurones innervating cranial extracerebral large arteries (De Vries et al. 1999). Studies have demonstrated the presence of various vasodilator neurotransmitters including 5-HT, nitric oxide (NO), substance P, neurokinin A and CGRP in perivascular nerve fibres supplying intracranial blood vessels (Gulbenkian et al. 1999). Cranial vasodilation in turn stimulates the trigeminal nerve, which may also release neuropetides, thus reinforcing vasodilation and sensory nerve activity (De Vries et al. 1999). The release of neuropeptides can trigger inflammation of 17

47 the pain sensitive meninges causing the head pain associated with migraine. The meninges receive trigeminal and autonomic innervation as well as a rich vascular supply. In addition it contains resident macrophages and mast cells that participate in the inflammatory response during injury (chemical or otherwise) (Buzzi and Moskowitz 2005). Hence, our current understanding of migraine pathogenesis indicates that a number of key biological pathways are involved in the disorder. Migraine is considered to be a disorder with a threshold character (Epstein et al. 1975; MacGregor et al. 1990; Montagna 2000), therefore it may be reasonable to consider that any reduction in the threshold to activation of any of these pathways, may lead to increased susceptibility to migraine. 1.6 Migraine Treatment Migraine treatment is generally classified as prophylactic, or abortive. Abortive therapies used for acute treatment include simple analgesics, non-steroidal anti-inflammatory drugs (NSAIDs), and vasoconstrictors such as ergots and triptans. Simple analgesics are frequently the first choice for mild to moderate attacks (MacGregor et al. 2003). Ergots have been used successfully for many years, but some migraineurs report severe sideeffects making this type of treatment intolerable (Linde 2006). The triptans (selective serotonin receptor agonists) have been shown to be effective in numerous studies (Ferrari et al. 2001; Ferrari et al. 2004), but for some migraineurs only provide pain relief in some attacks (Linde 2006). Frequent use of migraine abortive medication has been associated with drug-overuse headache resulting in daily or near daily headache (Linde 2006). Preventative treatments include β-adrenergic blockers (their exact mechanism is uncertain), calcium channel antagonists (which inhibit vasospasm, and block platelet 5- HT release and aggregation), serotonin antagonists, anticonvulsants (which influence the activity of voltage-gated ion channels or γ-aminobutyric acid [GABA] receptors), and NSAIDs (Silberstein et al. 2000). Most of these treatments can produce possible side effects such as sleepiness, exercise intolerance, impotence, nightmares, dry mouth, 18

48 weight gain, tremor, hair loss, or fetal deformities (Goadsby 2006). Table 1.2, adapted from Goadsby (2006), lists some of the potential side effects of commonly used preventative treatments. Table 1.2 Potential side effects of commonly used migraine preventative treatments (Goadsby 2006) Agent Pizotifen (antihistamine) Propranolol (β blocker) Tricyclics (inhibit noradrenaline and serotonin uptake) Anticonvulsants: Valproate Topiramate Gabapentin Methysergide Flunarizine Side Effects Weight gain, drowsiness, fatigue, nausea, unusual weakness, dizziness, headache, dry mouth Reduced energy, tiredness, postural symptoms, contraindicated in asthma Drowsiness Drowsiness, weight gain, tremor, hair loss, fetal abnormalities, haematological or liver abnormalities Paraesthesiae, cognitive dysfunction, weight loss, renal stones, glaucoma Dizziness, sedation Drowsiness, leg cramps, hair loss, retroperitoneal fibrosis (1 month drug holiday required every 6 months) Drowsiness, weight gain, depression, parkinsonism In addition to the potential side effects many such treatments have limited efficacy, or may be inappropriate for some sufferers. Other treatments include riboflavin, the medicinal herb feverfew, magnesium supplementation, and Botulinum toxin type A injections (Silberstein and Goadsby 2002). Non-pharmacological treatments include massage, diet changes, accupuncture, and avoidance of known migraine triggers (Silberstein et al. 2000). Despite the high prevalence of migraine, many migraineurs are not diagnosed by a medical practitioner, and consequently resort to over the counter medications rather than prescription drugs (Silberstein et al. 2000; Lipton et al. 2001). Figure 1.2 shows patterns of medication use among International Headache Society- 19

49 defined migraineurs from the 1989 and 1999 American Migraine Studies illustrating both the reliance on over the counter medication and the likely undertreatment of migraine (Lipton et al. 1998; Lipton et al. 2001). Figure 1.2 patterns of medication use among IHS defined migraineurs from the 1989 and 1999 American Migraine Studies (Lipton et al. 1998; Lipton et al. 2001) 20

50 Chapter 2 Migraine Genetic studies 21

51 2.1 Introduction to Complex Genetic Disorders Unlike diseases or disorders that are inherited from a single genetic defect in a mendelian fashion, complex inheritance is caused by a combination of environmental factors and mutations in multiple genetic loci. Although complex inheritance shows familial aggregation, it is rarely in a mendelian fashion and often shows variable phenotypic expression. This variable expression may occur due to chance, environmental factors or interactions with other genes (Lander and Schork 1994). Further confounding factors include the possibility of incomplete penetrance, where individuals who inherit predisposing alleles may not manifest the disease; and phenocopy, where individuals who do not inherit predisposing alleles manifest the disease due to other non-genetic factors (Lander and Schork 1994). Twin and adoption studies can be undertaken to discriminate between familiality due to genetic or environmental factors (Burmeister 1999). Adoption studies determine if individuals are more concordant with their adoptive parents or birth parents. If there is more concordance with adoptive parents, it can be concluded that a common environment is more responsible for the disease. Similarly, studies can be undertaken in twins who have been reared apart to determine potential differences in disease liability and evaluate the influences of genetics and rearing environment. Furthermore, in twin studies, concordance may be compared between monozygous versus dizygous twins with the expectation that genetic influences result in higher concordance in monozygous twins (Summers 1993). 2.2 Migraine as a Complex Genetic Disorder Relative Risk The aggregation of a disease in families may be the first observable indication of an underlying genetic component. Strong familial aggregation is certainly seen in migraine. 22

52 A standard measure of familial aggregation is the disease occurrence risk ratio, defined as the risk of disease in relatives of a random individual with disease, divided by the population prevalence of the disease (Rybicki and Elston 2000). Population relative risk for first degree relatives of migraineurs has been reported at 1.9 (Russell and Olesen 1995) to >2 (Gervil et al. 2001) with familial incidence figures varying from 61% (Dalsgaard-Nielsen 1965) to 90% (Dalsgaard-Nielsen and Ulrich 1973). In a 1997 population based study Stewart et al. investigated the risk of migraine in first-degree relatives of 73 migraineurs and 72 matched controls. It was found that the risk of migraine was 50% more likely in relatives of migraine probands than in relatives of controls (Stewart et al. 1997). In the most recent familial risk study of 532 migraineurs, Stewart et al. (2006) found that the relative risk of migraine in first-degree relatives was significantly increased at 1.88 compared to controls. The relative risk of migraine was even higher for relatives of probands with onset before age 16 years (relative risk of 2.50) and with more severe pain (relative risk of 2.38). Results suggested higher levels of familial aggregation in those with earlier onset and more severe pain scores (Stewart et al. 2006) Heritibility Along with the strong familial aggregation seen in migraine, evidence for a genetic component has also been determined by family and twin studies. A genetic component in migraine was shown in an epidemiologic study comparing the risk of migraine in 44 families to the general population. Russel at al. (1996) found that first-degree relatives of probands with MO had a 1.9-fold increased risk of MO, and first-degree relatives of probands with MA had a four-fold increased risk of MA. It was concluded that both genetic and environmental factors are important in MO and that MA is determined largely by genetic factors (Russell et al. 1996). Numerous twin studies have shown a role for both genetic and environmental components, many showing higher concordance rates in monozygous twins compared to dizygous twins and furthermore, providing evidence that there is no simple inheritance of migraine since concordance does not reach 100% in monozygous twins (Larsson et al. 1995; Gervil et al. 1999; Gervil et al. 1999; Ulrich et al. 1999). An estimation of heritability (ie. the degree to which a trait is genetically 23

53 determined) may be undertaken by performing regression-correlation in close relatives and is usually expressed as the total genetic variance to genotypic variance (V G /V P ) (King and Stansfield 2002). A 2003 study looking at genetic variance across six countries reported heritability estimates ranging from 34% -57% (Mulder et al. 2003) Mode of Transmission There is contention concerning the mode of transmission of migraine. However it is generally considered to be genetically determined with a multifactorial mode of inheritance. In a complex segregation analysis of 126 families with MO and 127 families with MA it was concluded that both MO and MA have a non-mendelian multifactorial mode of inheritance (Russell et al. 1995). Danish twin studies also suggest an inheritance model combing additive genetic and environmental factors (Gervil et al. 1999; Gervil et al. 1999; Ulrich et al. 1999). However, Russel and Olesen (1993) report that autosomal dominant inheritance with reduced penetrance cannot be excluded in either sub-type of migraine (Russell and Olesen 1993). Interestingly an inheritance mode with reduced penetrance is evident in some affected pedigrees (Kalfakis et al. 1996; Wessman et al. 2002). Nevertheless, it is generally acknowledged that migraine has a complex or multifactorial mode of inheritance influenced by both genetic and environmental factors, along with variable expression of clinical symptoms (Ophoff et al. 2001). A number of modest effect genes are considered to contribute to the genetic factor and combined with environmental liability, these factors are believed to exist along a contimuum. Individuals that exceed a certain threshold are affected by the disease. This is illustrated in Figure

54 Figure 2.1 A model of multifactorial inheritance. Disease liability shows a normal distribution; individuals to the right of the threshold are affected by the disease (Gervil et al. 2001). 2.3 Approaches to Gene Discovery in Complex Disorders Linkage Analysis Linkage analysis is based on the fact that genetic loci that are physically close together (linked) will undergo less crossing over and recombination events and thus will tend to be inherited together. The aim is to find chromosomal regions that are shared among affected relatives and different between affected and unaffected individuals. These implicated chromosomal regions are expected to harbour the disease gene in question. In practice, linkage analysis requires the investigation of families in which the disease phenotype segregates by looking at a group of polymorphic DNA markers. It may then be determined the likelihood that the loci (genetic marker and disease gene) are linked by calculating the logarithm of the odds or lod score. The lod score is a ratio of 2 likelihoods: the odds that the loci are linked and the odds that the loci are not linked. A lod score of 3 or more is used as an indication of statistically significant linkage, although more stringent criteria have been recommended for genome-wide scans (Lander and Schork 1994). Two-point (ratio of the likelihoods that 2 loci are linked) and multipoint linkage (ratio of likelihoods at each location across the genome) analyses are standard analyses used in gene mapping. Once a location or set of locations suggestive of linkage are identified, fine mapping methods may be employed with a more dense set of 25

55 additional polymorphisms in the implicated genomic region (Mayeux 2005). Certainly linkage analysis has had successful results with mendelian diseases such as Cystic Fibrosis (Kerem et al. 1989; Rommens et al. 1989) and Huntingtons Disease (Gusella et al. 1983). However in complex genetic diseases, with ambiguous relationships between phenotype and genotype, linkage analysis can be problematic. In many cases, even after evidence is shown for linkage, convincing candidate genes or mutations have not been located (Botstein and Risch 2003) Association Studies Association studies are a well known approach for fine mapping complex disease. They are generally model-free with no assumption of a mode of inheritance. Association analyses determine whether or not a specific genetic variant is associated with the disease phenotype and can be conducted in a group of randomly selected patients diagnosed with the disease under analysis (cases) compared to unaffected individuals (controls). A version of the association study, the Transmission Disequilibrium Test (TDT) tests the transmission of alleles from heterozygous parents to affected offspring (Spielman et al. 1993; Cardon and Bell 2001). A significant difference in frequencies of the variant in cases compared to controls in association studies suggests that the marker is located either within the susceptibility gene or is in linkage disequilibrium (LD) with the susceptibility gene and they are inherited together (Ducros et al. 2002). Similarly, distortion of allele transmission from parents to affected offspring in the TDT implicates the variant preferentially transmitted to the affected offspring. Similar to linkage analysis, association studies have had their successes and failures. Associations between specific human leukocyte antigens (HLA) and a variety of autoimmune diseases insulindependent diabetes melitis (Qiu et al. 1997), multiple sclerosis (Kwon et al. 1999), rheumatoid arthritis (Suarez-Almazor et al. 1995) have been reported and confirmed, yet many more reported associations have failed to be replicated in later independent studies (Risch 2000). Important issues to be considered in the design of association studies include: 1. Accuracy of diagnosis for the disorder to be studied. 2. Selection of appropriate matched control subjects, especially regarding age, sex, and ethnic background. 26

56 3. Choice of study strategy, such as using a population-based, case-control study vs a family approach eg. TDT 4. The problem of multiple comparisons increasing the likelihood of a false-positive result occurring by chance because of large numbers of comparisons in the study. 5. The choice of statistical analysis and threshold for significance (Bird et al. 2001) Functional Genomics Functional genomics focuses on dynamic aspects of DNA sequences such as gene transcription, translation, and protein-protein interactions. The aim of functional genomics is to determine the genetic differences that result in disease or predisposition to disease, from the level of the DNA sequence through to protein level and the resultant role in disease pathogenesis. Functional genomics demonstrates a tight correlation between protein function and gene expression levels (Brown and Botstein 1999). Consequently gene expression studies provide a convincing basis for hypotheses concerning the functional consequences of a particular gene of interest (Mocellin et al. 2003). Many cellular processes are reflected in altered patterns of gene expression, thus the reaction of a cell or cell population to stimulus or comparison of gene profiles in different types of samples has many applications in research. Because DNA is copied into an intermediate, RNA, and RNA molecules are used to make proteins that carry out the basic processes of life (Fields et al. 1999), the detection and quantification of RNA levels are central to research on gene function and expression. Methods currently in use to quantitatively determine RNA levels are Northern blotting, in situ hybridisation and RNAse protection assays (where specific RNA sequences may be isolated by their complementation to a particular probe), microarray (in which the expression of many genes may be analysed by complementation of RNA to specific gene sequences dotted on to a chip), and reverse transcription polymerase chain reaction (RT- PCR) (Bustin 2000). RT-PCR is an invitro technique used to amplify and quantitate specific sequences of RNA by a similar method to standard polymerase chain reaction (PCR). It is believed to be the most sensitive and most flexible of all RNA quantification methods (Wang and Brown 1999). 27

57 2.4 Molecular Genetics Studies of Migraine The identification of susceptibility genes for complex traits such as migraine can be challenging, in particular due to the contribution of multiple and potentially interacting genetic loci, as well as the confounding influence of environmental factors. Furthermore classifying migraineurs for genetic studies can be problematic as migraine displays clinical heterogeneity and may also occur as a symptom of an unknown cerebral disorder (Ducros et al. 2002). In the search for genes involved in migraine, a number of researchers have focused on familial hemiplegic migraine (FHM) a rare subtype of MA that was first described by Clarke (1910) in a family in which attacks of hemicranial pain and associated hemiparesis occurred in 4 generations (Clarke 1910). FHM is distinguished from typical migraine by its association with hemiparesis and a clear autosomal dominant mode of inheritance. FHM attacks involve similar neurological symptoms to MA but must include additional motor involvement. Typically, hemiparesis occurs during the aura (Joutel et al. 1993). The Headache Classification Committee of the International Headache Society describes FHM as migraine with aura including motor weakness with at least one first-degree or second-degree relative having migraine aura including motor weakness. The diagnostic criteria are as follows: A. At least two attacks fulfilling B and C B. Aura consisting of fully reversible motor weakness and at least one of the following 1. Fully reversible visual symptoms including positive features (eg. Flickering lights, spots or lines) and/or negative features (ie. loss of vision) 2. Fully reversible sensory symptoms including positive features (pins and needles) and/or negative features (numbness) 3. Fully reversible dysphasic speech disturbance C. At least two of the following: 1. At least one aura symptom develops gradually over > 5 minutes and/or different aura symptoms occur in succession over > 5 minutes 2. Each aura symptom lasts > 5 minutes and < 24 hours 3. Headache fulfilling criteria B-D for Migraine without Aura begins during the aura or follows onset of aura within 60 minutes 28

58 D. At least one first-degree or second-degree relative has had attacks fulfilling these criteria E. Not attributed to another disorder Both hemiplegic and non-hemiplegic migraine can occur in the same family (Young et al. 1970). Furthermore, some sufferers of FHM can also experience attacks of nonhemiplegic migraine (Ophoff et al. 1994). Certain clinical features are common to both FHM and typical migraine including similarities in headache characteristics and triggers (Gardner et al. 1997). The discovery of several genetic defects responsible for FHM has provided some interesting insights into the role of genetic defects in this type of disorder FHM 1 During the study of CADASIL (cerebral autosomal dominant arteriopathy with subcortical infarcts and leukoencephalopathy), Joutel et al. (1993) noted that some of the patients experienced recurrent attacks of severe headache with various aura symptoms, including transient hemiplegia. This led Joutel and colleagues to question whether CASADIL and familial hemiplegic migraine, although clinically distinct, might stem from a similar genetic defect. Using a set of DNA markers spanning the region containing the CADASIL locus on chromosome 19q (Tournier-Lasserve et al. 1993), linkage analysis was undertaken in 2 large pedigrees with familial hemiplegic migraine (Joutel et al. 1993). From these studies, the first FHM gene was eventually localized to chromosome 19p13 (Ophoff et al. 1996) and has been identified as CACNA1A, coding for the α1a sub-unit of the P/Q type voltage-gated neuronal calcium channel CaV2.1. Mutations in this gene, capable of modulating neurotransmitter release, have also been implicated in two other neurological disorders, episodic ataxia type 2 and spinocerebellar ataxia type 6 (Ophoff et al. 1996; Zhuchenko et al. 1997). The α1 subunit of CACNA1A forms the ionic pore of CaV2.1, playing a major role in voltage sensitivity and ionic selectivity. A 2002 study expressing FHM mutations in cultured cells revealed an increased open probability for calcium influx and a decrease in density of functional calcium channels (Tottene et al. 2002). The overall impact of these 29

59 effects in the human brain is not entirely known although nervous system abnormalities in knockout mice deficient in P/Q type calcium channel genes include absence seizures, ataxia, and selective degeneration of the cerebellum if the mouse survives past weaning (Jun et al. 1999). Loss of function mutations in the α1 subunit leads to tottering, leaner (Fletcher et al. 1996), and rocker mouse (Zwingman et al. 2001). A CACNA1A knockin mouse carrying the R192Q FHM1 mutation showed increased CaV2.1 current density, enhanced neurotransmission and lowered threshold for cortical spreading depression (CSD) (van den Maagdenberg et al. 2004). There are currently 17 different mutations known in CACNA1A (De Vries et al. 2006). Of further interest, is the report of another neuronal calcium channel α-1 subunit gene, CACNA1E in the region of 1q25-31 (Diriong et al. 1995). This gene has high sequence homology to CACNA1A and may be worthy of investigation in both FHM and common migraine FHM 2 Pedigree linkage studies have implicated a second FHM susceptibility region on chromosome 1q, originally at 1q31 (Gardner et al. 1997) although subsequent and additional studies have implicated a locus in the 1q21-23 region (Ducros et al. 1999). FHM pedigrees with linkage to this locus were found to have mutations in ATP1A2 (De Fusco et al. 2003; Vanmolkot et al. 2003). This ATPase appears to be involved in Na+/Ca2+ exchange following activation of voltage gated Ca2+ channels thus resembling the effect of CACNA1A mutations and potentially facilitating CSD (Estevez and Gardner 2004). There are currently 22 different mutations known in ATP1A2 (De Vries et al. 2006) FHM 3 A third locus for FHM has recently been identified on chromosome 2q24, the implicated gene being SCN1A which encodes the α1 subunit of the neuronal voltage-gated sodium channel NaV1.1. The specific mutation is a mis-sense mutation in the so-called hinged- 30

60 lid domain of the protein, which is critical for fast inactivation of the channel. Mutations of this gene have also been associated with epilepsy (Dichgans et al. 2005). At present just one mutation has been identified FHM Genotype-Phenotype Correlations Particular FHM genotypes may in part be responsible for the clinical variability seen in FHM. Mutations in CACNA1A are present in 50 percent of families with hemiplegic migraine, including all those with cerebellar signs, although it is reported to show a broad clinical spectrum (Ducros et al. 2001). An investigation of 46 individuals from 3 families linked to the chromosome 19 locus showed a higher frequency of unconsciousness during attacks and provocation of attacks by mild head trauma than unlinked families. In one linked family patients also displayed chronic progressive cerebellar ataxia (Terwindt et al. 1996). However, clinical heterogeneity has also been seen in families with the same chromosome 19 CACNA1A T666M mutation (Kors et al. 2003) FHM Loci in Common Migraine Research into the more common forms of migraine has considered FHM loci as candidate loci as they are likely to affect neuronal excitability thresholds and potentially initiate CSD. Furthermore, as discussed earlier, there is some clinical overlap between FHM and the more common forms of migraine and thus FHM has been proposed as a suitable model for studying common migraine, with FHM representing the most severe phenotype of the migraine spectrum (Ophoff et al. 1994). In a 1998, Nyholt et al. reported a 3.4cM region around the FHM1 gene at 19p13 as implicated in familial typical migraine (which includes MA & MO) in one large multigenerational pedigree although not in the other three pedigrees tested, thus indicating the genetic heterogeneity of common migraine (Nyholt et al. 1998). These results are supported by two affected sibpair studies also implicating this genomic region at 19p13 in the common forms of migraine. The first, conducted prior to the 1998 study by Nyholt et al. used non-parametric affected sib-pair analysis in 28 unrelated families (May et al. 1995), while the second study undertaken in 2001 investigated 189 affected siblings from 36 extended families with typical migraine 31

61 (Terwindt et al. 2001). Hence, it would seem that the FHM gene (CACNA1A) is an obvious candidate for involvement in MA and MO. However, a recent linkage and association study in the Genomics Research Centre of CACNA1A intragenic markers in independent migraine populations did not show any relationship of this gene to the disorder. In this study two migraineurs from MF1 showing linkage to the 19p13 locus reported in our earlier study were screened. Screening all 47 exons of CACNA1A did not show any mutations in these individuals. In addition, family based linkage and association analysis was undertaken on 82 unrelated pedigrees and an unrelated case control cohort with no evidence for linkage or association to CACNA1A (Lea et al. 2001). An earlier study by Kim et al. (1998) also using a direct sequencing approach to screening the exons of CACNA1A had failed to reveal any mutations in individuals affected with MA or MO (Kim et al. 1998). Furthermore, other studies have failed to find a role for this chromosome 19 locus (Hovatta et al. 1994; Monari et al. 1997) although it should be noted that differences in study methodology make comparisons difficult. For eample, in a negative linkage study of 19p13 in MA performed by Noble-Topham et al. individuals who suffered MO were categorized as unaffected (Noble-Topham et al. 2002). Although the CACNA1A gene cannot be completely ruled out as contributing to the overall genetic risk of the common forms of migraine in the general population, it is possible that variations within an adjacent gene on 19p13 may confer a more substantial risk to the disorder. This idea is well supported by a recent migraine linkage study, which incorporated haplotype analysis of 16 pedigrees affected with MA. Results of this work provide evidence that a MA susceptibility gene on chromosome 19p13 may localise to a region distinct from (and telomeric to) the FHM (CACNA1A gene) locus (Jones et al. 2001). With regard to other genes in this region, McCarthy et al. (2001) reported positive association of several single nucleotide polymorphisms (SNPs) in the Insulin Receptor gene (INSR) located at 19p13 in typical migraine (McCarthy et al. 2001) although in a more recent study Kaunisto et al. (2005) found no evidence for a susceptibility region at this locus using both parametric and non-parametric linkage analysis in 72 Finnish MA families (Kaunisto et al. 2005). 32

62 The additional FHM susceptibility region on chromosome 1q has offered an alternative focus for research into typical migraine (incorporating MA and MO). In a 2002 study at the Genomics Resarch Centre, a family-based linkage and association approach to investigating the FHM susceptibility region on chromosome 1q31 for involvement in typical migraine susceptibility was undertaken in affected Australian pedigrees. Initial multipoint analysis provided strong evidence for linkage of 1q31 markers to typical migraine in a single large multigenerational pedigree. Subsequent analysis of an independent sample of 82 affected pedigrees added support to the initial findings. Utilising the independent sample of 82 pedigrees we also performed a family-based association test (FBAT). Results of this analysis indicated distortion of allele transmission at a marker on 1q31 in these pedigrees (Lea et al. 2002). A follow up study performed at the Genomics Research Centre analysing 21 affected pedigrees showed evidence for linkage at 1q23 to markers spanning the ATP1A2 gene. However, testing of the known FHM ATP1A2 gene mutations in migraine probands of pedigrees showing excess allele sharing was negative. Sequencing of the entire coding areas of the gene through all the three MA affected individuals from an implicated pedigree was also negative for mutations (Curtain et al. 2005). As other candidate genes may lie within the 1q23 and 1q31 regions, further studies are needed for clarification Other Loci Linked to Migraine As outlined in 2.2.1, linkage analysis is often used to detect genomic regions thought to harbour susceptibility genes. Several loci have been identified in genome-wide scans, a summary of which appears in Table 2.1. The first genome wide linkage analysis was conducted in 2002 in 50 Finnish families with MA, implicating 4q24 (Wessman et al. 2002). A region close-by, 4q21, was implicated in 103 Icelandic families with MO (Bjornsson et al. 2003). In a Swedish family with MA and MO there was significant linkage to 6p (Carlsson et al. 2002). There was also significant linkage to 14q in an Italian pedigree with MO (Soragna et al. 2003), and 11q24 in 43 Canadian families with MA (Cader et al. 2003). Lea et al. (2005) has shown significant linkage to 18p11 and 3q-tel for a more severe heritable form of migraine. Excess allele sharing was also seen in previously implicated regions at 1q23 and 14q22 (Lea et al. 2005). More recently Nyholt et al. (2005) performed a genome-wide linkage analysis on 33

63 756 twin families. Results showed significant linkage for latent class analysis (LCA) derived migraine to 5q21 and replication of previously reported susceptibility loci at 6p12.2-p21.1 and 1q21-q23 (Nyholt et al. 2005). Nyholt et al. had also previously implicated a region on Xq24-28 after scanning the X chromosome in 3 large Australian pedigrees (Nyholt et al. 1998; Nyholt et al. 2000). Russo et al. (2005) recently investigated and implicated a region at 15q11-q13 in 10 families with MA. This genomic region contains three γ-aminobutyric acid (GABA) A receptor subunit genes, products of which play an important role in regulating neuronal inhibition (Russo et al. 2005). GABA is the major inhibitory neurotransmitter of the brain, occurring in 30-40% of all synapses in regions as cerebral cortex, hippocampus, thalamus, basal ganglia, cerebellum, hypothalamus and brainstem (McCormick 1989). Notably, studies have reported higher GABA levels in blood platelets (Kowa et al. 1992) and saliva (Marukawa et al. 1996) in patients suffering migraine compared to controls. Recently, Vieira and colleagues reported an increase of GABA levels in the cerebrospinal fluid in depressed patients during headache attacks (Vieira et al. 2006). These findings suggest that GABA could be involved in the pathophysiology of migraine. In addition, GABAergic drugs can modulate biochemical and physiological events in the disorder. In fact cortical events that underlie MA may be suppressed by an increase of inhibitory GABAergic neurotransmission induced for example, by valproate (Cutrer et al. 1997). In addition, the concept of migraine as a result of CNS hyperexcitability (Aurora et al. 1998) has led to the use of GABAergic anticonvulsant medications as efficient therapy for prevention of migraine (Silberstein 2004). Thus further investigation of chromosomal regions harbouring GABA related genes may be worthwhile Candidate Gene Studies in Migraine A further approach in complex disease gene mapping is targeted candidate gene analysis using association studies as outlined in Selection of genetic markers for this type of analysis is based on the hypothesis of the marker being functionally relevant, or in linkage disequilibrium with a causal variant (Risch 2000). This type of approach has been employed by numerous research groups and tests for differences in allele frequencies between affected individuals (cases) and migraine unaffected individuals (controls). 34

64 Genetic association studies have suggested a role of numerous candidate genes in migraine susceptibility, although follow-up confirmatory studies are limited and none have yet been functionally linked. As case-control analyses have a history of spurious and controversial results, independent replication is now considered a key factor in evaluating the validity of significant results (Malhotra and Goldman 1999). Table 2.2 summarises results of published migraine association studies in the past 10 years. Positive results of association studies to date are discussed in broad sub-categories that are either wellknown migraine triggers, or implicated in migraine pathophysiology Neurotransmitter Function There is some evidence that the neurotransmitter system plays a key role in migraine pathophysiology. During migraine attacks, activation of the cerebral structures of the thalamus in response to excessive afferent stimulation, or of the hypothalamus in response to changes in the internal environment are believed to occur (Coppola et al. 2005; Montagna 2006). These responses involve modulation of the intracranial perivascular nerve, initiating neurogenic inflammation causing an increase in the diameter of the meningeal blood vessels. The vasodilatation in turn, allows propagation of nociceptor factors (liberated in blood circulation), causing headache pain. Several neurobiological systems are part of these events, including the serotoninergic, acetylcholinergic and catecholaminergic systems. Norepinephrine can mediate many important functions in the central and peripheral nervous systems. It is the major neurotransmitter of the sympathetic division of the autonomous nervous system, where its release regulates vascular tone and cardiac contractility among other vital functions. In the central nervous system, norepinephrine is localised within several neurones in the hindbrain and midbrain and participates in the regulation of several vital physiological functions such as cardiac rhythm, awake-sleep cycle, and cognition (Foote et al. 1983). Disturbances in the noradrenergic system can contribute to pathologies such as hypertension (Esler et al. 2001), hyperactivity sleep disorders (Mitler et al. 1993) and migraine (Holroyd et al. 1991). In some migraineurs, the level of plasma noradrenaline has been measured as significantly lower in patients compared to controls (Martinez et al. 1993), indicating a sympathetic dysfunction. 35

65 Dopamine Related Genes A role for central dopamine hypersensitivity in migraine has also been proposed. A lower threshold for dopamine receptor activation (Blin et al. 1991; Cerbo et al. 1997) and increased expression of certain dopamine receptors on lymphocytes (Barbanti et al. 2000) has been found in migraineurs compared to controls. Furthermore, platelet levels of dopamine in migraine patients have been found to be higher than those found in healthy control subjects (D'Andrea et al. 1989) Investigations into genes involved in dopaminergic pathways, have shown interesting, although at times conflicting results. Peroutka et al. (1997) showed a role for the Noc I allele of the D 2 dopamine receptor (DRD2) gene in MA susceptibility (Peroutka et al. 1997) although Dichgans et al. (1998) was unable to confirm this result in a smaller German group (Dichgans et al. 1998), as did Maude et al. testing a different polymorphism in the same gene (Maude et al. 2001). Del Zompo et al. (1998) tested a different intragenic polymorphism within the gene in a subgroup of migraineurs with specific dopaminergic symptoms (yawning and nausea during migraine) from 50 families by transmission disequilibrium testing (TDT), providing evidence for a role of the DRD2 gene in MO (Del Zompo et al. 1998). The TDT is quite robust to population stratification (Spielman et al. 1993). Population stratification is a limitation of traditional case-control analysis if the study groups are not carefully selected to be ethnically homogenous. Therefore follow-up studies of this gene in a large study group may be warranted. More recently, Mochi et al. (2003b) showed association of the D 4 dopamine receptor (DRD4) gene with MO (Mochi et al. 2003). Several functional polymorphisms have been reported for the dopamine beta hydroxylase (DBH) gene. The first association between DBH alleles of a short tandem repeat (STR), and the DBH plasma concentration was observed in a unrelated British population (Wei et al. 1997). This functional polymorphism in DBH (STR) has been confirmed by Cubells et al. (1998), who have also shown that an individual with the deletion of both alleles has only half of the mean plasma enzyme activity than a homozygote with the insertion/insertion genotype (Cubells et al. 1998). A 19 base pair insertion/deletion locus, located also in the DBH promoter (Porter et al. 1992), was associated with phenotypic variation in DBH activity in plasma (Cubells et al. 2000). Lea and colleagues have examined the prevalence of different alleles of both markers, the DBH STR and DBH 36

66 insertion/deletion in an association study with 177 unrelated migraineurs and 182 controls plus a TDT analysis of 296 subjects (263 affected) from 82 families of migraineurs. The results showed a distortion of allele transmission of the microsatellite (STR) marker in individuals suffering from both migraine with or without aura (Lea et al. 2000). More recently the DBH insertion/deletion has been associated with migraine in a large sudy group of 275 cases and 275 controls (Fernandez et al. 2006). Both of theses genetic markers of DBH have been reported to be associated with differences in plasma and CSF levels of DBH (Cubells et al. 1998; Zabetian et al. 2003). A single 10-kb block beginning from the indel marker in the DBH gene has been recently reported as being highly associated with the phenotype of the enzyme (Zabetian et al. 2003). This block contains in particular the single nucleotide polymorphism (SNP) 1021 C T polymorphism, which accounts for 35%-52% of the total phenotypic variance in plasma DBH activity in samples from three different populations (Zabetian et al. 2003). This functional polymorphism has been associated with Parkinson s disease (Healy et al. 2004). A new intragenic non-synonymous SNP polymorphism (+1603 C T) has also been recently identified in exon 11 of the DBH gene and has also associated with the phenotype of the enzyme (Tang et al. 2005). Interestingly significant differences in serum DBH have been observed in migraine patients compared with healthy control subjects (Gotoh et al. 1976; Gallai et al. 1992) and during a migraine attack (Anthony 1981). DBH plays an important role in the noradrenergic system. This enzyme is localised within the membrane fraction of norepinephrine and epinephrine producing neurons and neurosecretory cells, where it catalyses the conversion of dopamine to norepinephrine (Cimarusti et al. 1979; Kemper et al. 1987). DBH may be as important as dopamine and norepinephrine in dysfunction of the catecholamine pathway, notably in nervous and mental disorders Serotonin Related Genes Many studies have implicated the brain neurotransmitter 5-hydroxytryptamine (5-HT; serotonin), as being involved in the pathophysiology of migraine supporting a proposed serotonergic theory of migraine. Documented evidence of increased serotonin activity associated with a corresponding decrease in nociception and decreased serotonin activity 37

67 levels with a subsequent increase in nociception is indicative of the role of serotonin in pain modulation (Anthony et al. 1967). Changes in 5-HT serum levels during patient migraine attacks (Ferrari et al. 1989) and disruptions within the synthesis of serotonin (Chugani et al. 1999) all substantiate the pathophysiological role of serotonin in migraine pathogenesis. Hypermetabolism within the brain stem region of serotonergic raphe nuclei has also been documented (Weiller et al. 1995). The role of serotonin in migraine is further suggested by its involvement in therapy. The triptans, which are selective serotonin 5-HT1B/1D agonists, are very effective acute migraine drugs (Ferrari et al. 2001; Ferrari et al. 2004). Although conflicting results have been reported about the use of SSRIs (selective serotonin reuptake inhibitors) in migraine prevention (Adly et al. 1992; Saper et al. 1994; d'amato et al. 1999; Landy et al. 1999), trycyclic antidepressants such as amitriptyline are currently employed in migraine prophylaxis with evidence supporting their effectiveness for use (Oguzhanoglu et al. 1999; Krymchantowski et al. 2002). Thus genes in the serotonergic system have been investigated as potential candidates to mediate susceptibility to migraine. One such gene is the human serotonin transporter gene (5-HTT) located on chromosome 17q11.1-q12. This gene contains a 44 bp insertion/deletion functional polymorphism in the promoter region, which regulates the expression and function of the 5-HT transporter (Heils et al. 1996). Another polymorphism has been described, a variable-numbertandem-repeats (VNTR) of 17-bp sequence in the second intron, and has several alleles; STin 2.7, STin 2.9, STin 2.10, STin 2.11 and STin 2.12 (Lesch et al. 1996). The function of the VTNR polymorphism is not well known. Ogilvie et al. (1998) investigated the VNTR polymorphism in 266 individuals with migraine, and 133 unaffected controls. Results suggested a role for this marker in migraine (Ogilvie et al. 1998). Yilmaz and colleagues (2001) have also reported that the presence of the STin 2.10 allele in this marker may increase the risk of migraine (Yilmaz et al. 2001). Juhasz et al (2003) investigated a functional polymorphism in the upstream regulatory region of the serotonin transporter gene in the Hungarian female population. This marker was analysed in 126 migraine sufferers and 101 unrelated healthy controls. A borderline association between the short allele and migraine was found (Juhasz et al. 2003). Marziniak et al. (2005) performed an association study in 197 migraineurs and 115 controls investigating the same functional serotonin transporter gene promoter polymorphism. Results showed that the frequency of the less active short allele was increased in MA sufferers but not in 38

68 MO sufferers in comparison with the control population (Marziniak et al. 2005). More recently a significantly higher frequency of the short allele was found in MA patients, and was also related to anxiety (Gonda et al. 2006). The serotonin transporter catalyses a high-affinity sodium chloride dependent process to transport serotonin from the extracellular space into serotonergic neurons and blood platelets, thereby terminating its actions (Lesch et al. 1996). This key element of serotonergic transmission might be an indicator of genetically driven vulnerability to migraine. Other important components of the serotonergic system are the serotonin receptors of which there are many subtypes throughout the brain. For example, the 5-HT2C receptor subtype is a metabotropic receptor, able to stimulate several phospholipases (C and A2) and channels (K+, Cl-) where it is expressed nearly exclusively in the brain (Abramowski et al. 1995). Two lines of evidence point to the importance of the 5-HT2 receptor in migraine: several anti-migraine drugs are 5-HT2 receptor antagonists (Schmuck et al. 1996; Porter et al. 1999; Schaerlinger et al. 2003) and the 5-HT2 receptor agonist metachlorophenylpiperazine has been shown to induce a migraine attack (Brewerton et al. 1988). Nonetheless, several studies investigating genes encoding various 5-HT receptor subtypes have shown insufficient evidence for a role in migraine. Buchwalder et al. (1996) failed to find a role for 5-HT2A and 5-HT2C in migraine by linkage analysis in 18 pedigrees (Buchwalder et al. 1996). Burnet et al. (1997) failed to find an association of a codon 23 variant of 5-HT2C and migraine in an association study (Burnet et al. 1997). Similarly Johnson et al. (2003) found no association with migraine of 5-HT2C using both a linkage and association approach (Johnson et al. 2003). Racchi et al. (2004) found no evidence for involvement of 5-HT1B/1D and 5-HT2C polymorphisms in migraine with aura (Racchi et al. 2004). In 1996 Nyholt et al. tested two markers in the 5-HT2A by linkage and association analysis and found no evidence for a role of these loci in migraine (Nyholt et al. 1996). A positive association with the 102T/C polymorphism of 5-HT2A and MA was reported by Erdal et al. (2001) in a small study group (n=105) (Erdal et al. 2001). However Juhasz et al (2003) found no association with migraine of this particular locus in a larger study sample (Juhasz et al. 2003). 39

69 Vascular Function Alterations in vascular function have been noted in migraineurs (Silvestrini et al. 1995; Rosengarten et al. 2003; Yetkin et al. 2006). Thus genes involved in vascular functioning have also been explored as likely migraine candidates. Of particular note is the methylenetetrahydrofolate reductase (MTHFR) gene, mutations of which have been implicated in mild hyperhomocysteinemia, a condition which is understood to exert deleterious effects on the vascular endothelium through oxidative damage (Das 2003). Notably both 677T and 1298C mutations of MTHFR have been implicated in reduced enzymatic capacity (Frosst et al. 1995; Weisberg et al. 2001). 5, 10- methylenetetrahydrofolate reductase (MTHFR) is a key enzyme in the metabolism of methionine and homocysteine is an intermediate of this pathway. Plasma homocysteine levels are determined by both genetic factors (such as mutations in genes involved in the metabolism pathway) and non-genetic factors (eg. nutritional factors, sex, age). Several studies have investigated plasma homocysteine levels as a cause of vascular disease (Brattstrom et al. 1984; Malinow et al. 1989; den Heijer et al. 1996), although it has been suggested that hyperhomocysteinemia may in fact be a secondary effect of the disease (Brattstrom 1997). Nevertheless, recent meta-analyses have confirmed a significant link between hyperhomocysteinemia and vascular disease and further concluded that individuals with the MTHFR 677TT genotype have a modest but significant increased risk of vascular disease (Kelly et al. 2002; Hackam and Anand 2003). Kowa et al. (2000) provided the first evidence for a role of the T allele of the MTHFR C677T mutation in migraine, in particular MA susceptibility. Kara et al. (2003) confirmed these results in an independent study, as well as the C allele of the A1298C mutation in the same gene. Lea et al. (2004) also confirmed a role in MA of the T allele in the C677T mutation (Lea et al. 2004). Oterino et al. (2004) was unable to demonstrate a significant difference in frequency of the MTHFR 677 genotypes in a Spanish cohort, however they did report a significant difference in the frequency of TT homozygosis between MA and MO with the T allele occurring more frequently in MA (Oterino et al. 2004). More recently Sher et al. (2006) showed an association of the TT genotype in migraine in a large study group (n = 1625) (Scher et al. 2006). 40

70 Other vascular genes implicated in migraine are the Angiotensin I-converting enzyme (ACE) gene and the endothelin type A receptor (ETA) gene. The Angiotensin I- converting enzyme is one of the key enzymes in the rennin-angiotensin-aldosterone system and plays an important role in blood pressure regulation (McCaa 1980), while the ETA receptor is involved in vasoconstriction (Mickley et al. 1997). The ACE I/D polymorphism is considered to influence serum enzyme levels (Rigat et al. 1990). Paterna et al. (1997, 2000) has shown a role for the D allele of this polymorphism in MO in two independent studies (Paterna et al. 1997; Paterna et al. 2000). Kowa et al. (2005) has also shown a role for the D allele of the ACE gene in MA (Kowa et al. 2005). Interestingly, Lea et al. (2005) has shown an interacting role for MTHFR gene 677T and ACE gene D variants, in migraine, with the greatest effect in MA (Kaunisto et al. 2005). With regards to the ETA gene, in a population based study in France, Tzourio et al. (2001) found that the A allele in the ETA -231 A/G polymorphism was associated with migraine and that the association was stronger in participants with a family history of severe headaches than in those without (Tzourio et al. 2001). Interestingly the potent vasoconstrictor endothelin 1, whose effect is mediated via endothelin type A and B receptors, has been shown as a potent inducer of CSD (Dreier et al. 2002) Other Implicated Genes Additional genes that have been implicated in migraine susceptibility by association analysis and do not fall into the broad categories above include the insulin receptor gene (INSR), tumour necrosis factor (TNF) and low density lipo-protein receptor (LDLR) genes (McCarthy et al. 2001; Trabace et al. 2002; Mochi et al. 2003; Rainero et al. 2004). Plausible hypotheses can be proposed for a role of these genes in migraine. INSR is one of the key players in glucose metabolism. Fasting is a well known migraine trigger (Robbins 1994). TNFα is a proinflammatory cytokine that has been shown to play a role in nociception (Wagner and Myers 1996; Sachs et al. 2002). Transitory increases in the levels of TNF-α have been observed in the plasma of migraineurs (22 MO, 4 MA) (Perini et al. 2005) and in the internal jugular blood of migraineurs with aura (Sarchielli et al. 2006) during an attack. LDLR is located in the migraine implicated region 19p13 and has a role in cholesterol homeostasis. Cholesterol levels have been shown to influence 41

71 platelet behaviour in migraine patients (Kozubski and Stanczyk 1986) and migraineurs with aura have been reported to have an increased likelihood of an unfavourable cholesterol profile (Scher et al. 2005). Interestingly in a study performed in our laboratory, a role for the same LDLR variant studied by Mochi et al (2003a) could not be found (Curtain et al. 2004). Thus a potential role for this gene in migraine susceptibility remains unclear. As discussed earlier, functional analyses of the INSR gene variant (at 19p13) associated with migraine were unable to show altered functioning of the receptor when tested in mononuclear cells. However as considered by the authors, limitations of the functional analysis leave open the possibility that INSR function could be altered in migrainerelevant cell types such as in neurons or other central nervous system cell types (McCarthy et al. 2001). With regard to TNF genes, Trabace et al. (2002) analysed TNFβ polymorphisms in 47 patients with MO, 32 patients with MA, and 101 controls. Results suggested a role for TNFβ in MO but not in MA. Rainero et al. (2004) investigated a polymorphism in TNFα in 299 migraineurs and 306 controls also showing an association with MO but not MA. Interestingly, local expression of TNF-α has been shown in the rat brain after induction of CSD (Jander et al. 2001) Hormone Related Genes Female sex hormones have long been considered to play a role in migraine (MacGregor et al. 1990; MacGregor 1996; Silberstein and Merriam 2000; Couturier et al. 2003). There is no gender difference in migraine prevalence prior to puberty. However after puberty prevalence in women increases and exceeds that in men by approximately threefold (Silberstein and Merriam 2000). Furthermore, the different stages in a woman s life where changes in female hormone levels occur (such as puberty, pregnancy, and menopause) are commonly coupled with concurrent changes in migraine frequency and severity (MacGregor et al. 1990; Gupta et al. 2006). Many female migraineurs report attacks during the perimenstrual period, which the IHS has addressed by including menstrual and menstrually related migraine in their revised classification of headache disorders (HCCIHS 2004). In fact, more than half of female migraineurs report an 42

72 association between migraine and menstruation (Couturier et al. 2003; Gupta et al. 2006). Several studies have assessed the hypothesis that the fall in hormone levels prior to menstruation may trigger an attack. However treatment involving stabilization of hormone levels has not been effective in all cases (MacGregor et al. 1990). Certainly the correlation between sex steroid hormone levels and migraine is not merely a positive or negative one. Epstein et al. (1975) has suggested a role for hormonal variation in all women with migraine, but also a role for factors additional to the hormonal environment (Epstein et al. 1975). It is possible that genetic factors may play this role. Nyholt et al. have shown that an X-linked dominant gene may be involved in some families (Nyholt et al. 1998; Nyholt et al. 2000), but it is unlikely that this locus would completely account for the large sex difference in migraine prevalence (Gupta et al. 2006), nor explain the influence of hormonal milestones on migraine in women. Various hypotheses may be proposed whereby genetic factors could interact with sex steroid hormones to impact on migraine susceptibility. Steroid hormones regulate a wide range of biological functions either through genomic (transcription dependent) or nongenomic (non-transcription dependent) mechanisms. Most importantly they can modulate the central nervous system and vascular tone (Gupta et al. 2006). The predominant biological effects of steroid hormones are mediated through their cognate receptors, which are expressed in a wide range of target tissues, notably many regions of the CNS including cerebral blood vessels and serotonin neurons. Steroid hormone receptors have been shown to be involved in both genomic and non-genomic signalling mechanisms. Mutations in steroid hormone receptors have been implicated in a number of human disease states including breast cancer, endometrial and ovarian cancer, and psychiatric disease (Agoulnik et al. 2004; Herynk and Fuqua 2004). Interestingly however, a number of mutations have shown little or no observable effect on receptor function (Herynk and Fuqua 2004). The impact of subtle effects on receptor function is uncertain, but could potentially play a role in disorders such as migraine due to their interaction with factors known to be involved in the pathophysiology of the disorder. At the commencement of the work detailed in this thesis no-one had published research investigating the possible role of hormone related genes in migraine. Thus one of the aims of this research was to investigate genes involved in hormonal function for a potential role in migraine susceptibility. Chapter 4 describes cross-sectional association 43

73 analysis investigating the estrogen and progesterone receptor genes as migraine candidate genes in two large independent cohorts and demonstrates that variation in both estrogen and progesterone receptor genes do indeed appear to play a role in migraine susceptibility. 44

74 Table 2.1 List of migraine genome-wide linkage studies showing significant LOD scores Authors No of families/patients studied Phenotype Region detected Nyholt, Morley et al families LCA diagnosis 5q21 Lea, Nyholt et al pedigrees MA MO 18p11, 3qtel in severe heritable form Bjornsson, Gudmundsson et al patients MO 4q21 Cader, Noble-Topham et al families MA 11q24 Soragna, Vettori et al large pedigree MO 14p12-21 Carlsson, Forsgren et al large pedigree MA MO 6p12-21 Wessman, Kallela et al pedigrees MA 4q24 LCA = Latent Class Analysis, MA = migraine with aura, MO = migraine without aura 45

75 Table 2.2. List of published association studies on migraine in the past 10 years Authors Gene/s analysed Chromosomal Location n MIG/MA/MO Results Gonda et al STG 17q11.1-q12 45MA/52C Significant association (p = 0.035) Todt et al STG 17q11.1-q12 472MA/506C No association Fernandez et al DBH 9q34 275M/275C Significant association (p = 0.01) Yang et al HT1A 5q M/93C No association Schwaag et al NOTCH3 19p13.2-p M/97C Significant association (p = <0.05) Rainero et al HLA-DRB1 6p M/325C Significant association (p = <0.05) Oterino et al ESR 1 6q M/232C Significant association (p = 0.008) Scher et al MTHFR 1p MA/226MO 1212C Significant association (p = 0.006) Kowa, Fusayasu et al ACE 17q23 54MA/122MO 248C Significant association MA (p <0.01) Kaunisto, Tikka et al ACE, MTHFR 17q23, 1p MA/MO 270C ACE, MTHFR interact to increase MA risk (p = 0.018) Oterino, Valle et al MTHFR, TS, MS, MTHFD1 1p36.3, 18p11.32, 138MA/191MO 237C MTHFR, TS, MTHFD1 interact to increase migraine risk 5p15.3-p15.2, 14q24 Marziniak, Mossner et al STG 17q11.1-q12 197MA/MO 115C Significant association MA (p < 0.001) Filic, Vladic et al MAO-A Xp MA/80MO 150C Weak association in male MO MAO-B Xp MA/80MO 150C No association Marziniak, Mossner et al MAO-A Xp MIG 229C No association Curtain, Lea et al LDLR 19p MA/MO 244C No association Rainero, Grimaldi et al TNF alpha 6p MA/MO 306C Significant Association (p < 0.001) Otterino et al MTHFR 1p MA/152MO 204C No overall association but higher allele frequency in MA Lea, Ovcaric et al MTHFR 1p MA/MO 275C Significant association (p = 0.017) 46

76 Racchi et al HT1B/1D, 5-HT2C 6q13, Xq24 44MA 33C No association Johnson, Lea et al HT2C receptor Xq24 275MA/MO 275C No association Juhasz, Zsombok et al HT2A receptor 13q14-q21 126MA/MO 101C No association STG 17q11.1-q12 126MA/MO 101C Borderline association Kusumi, Ishizaki et al GST 11q13 174MA/MO/TT 372C Significant Association (p < 0.01) Kara, Sazci et al MTHFR 1p MA 70MO 136C Significant association (p <0.01) Mochi, Cevoli et al LDLR 19p MA 220MO 200C Significant association (p = 0.017) Mochi, Cevoli et al DRD4 DAT DBH 11p15.5, 5p15.3, 9q34 93MA 101MO 117C Significant association DRD4 in MO (p = ) Rainero, Grimaldi et al APOE 19q MA/MO 587C No association Shepherd, Lea et al DRD1 DRD3 DRD5 5q35.1, 3q13.3, 4p MA/MO 275C No association Trabace, Brioli et al TNF 6p MA 47MO 101C Significant association TNFB in MO (p = 0.004) McCarthy, Hosford et al INSR 19p13.3-p MA/MO 765C Significant association (p < 0.05) Yilmaz et al STG 17q11.1-q12 52MA/MO 80 C Significant Association Stin 2.10 (p = 0.01) Lea, Curtain et al inos 17q11.2-q12 262MA/MO 252C No association Lea, Curtain et al CACNA1A 19p13 177MA/MO 182C No association CACNA1A 81 families (TDT) No association Terwindt, Ophoff et al CACNA1A 189 sibpairs Increased allele sharing Maude, Curtin et al DRD2 11q23 200MA/MO 464C No association Erdal, Herken et al HT2A receptor 13q14-q21 61MA/MO 41C Significant association codon 102 in MA (p=0.02) Tzourio, El Amrani et al endothelin receptors 13q22 140MA/MO Significant association ETA (p < 0.001) Kowa, Yasui et al MTHFR 1p MA 52MO 261C Significant association (p < ) Lea, Dohy et al DBH 9q34 177MA/MO 182C Significant association (p = 0.019) DBH 82 families (TDT) Disequilibrium Paterna, Di Pasquale et al ACE 17q23 302MO 201C Significant association (p < 0.05) Peroutka et al DRD2 11q23 52MA 121C Significant association (p < 0.005) 47

77 Del Zompo, Cherchi et al DRD2 DRD3 DRD4 11q23, 3q13.3, 11p families MO (TDT) Disequilibrium in DRD2 Dichgans, Forderreuther et al DRD2 11q23 47MA/55MO 145C No association Ogilvie, Russell et al STG 17q11.1-q12 173MO/94MA 133C Significant association (p = 0.025) Griffiths, Nyholt et al enos 7q36 91MA/MO 85C No association Burnet, Harrison et al HT2C receptor Xq24 73MA 169MO 129C No association codon 23 Peroutka, Wilhoit et al DRD2 11q23 129MA/MO 121C Significant association MA (p < 0.005) Paterna, Di Pasquale et al ACE 17q23 191MO 201C Significant association (p < 0.05) Nyholt, Curtain et al HT2A receptor 13q14-q21 96MA/MO 91C No association M=migraine (no MA/MO specified), MA=migraine with aura, MO=migraine without aura, TT=tension type headache, C=control, AR=androgen receptor gene, PGR=progesterone receptor gene, ACE=angiotensin converting enzyme gene, MTHFR=5,10 methylenetetrahydrofolate reductase gene, TS=thymidylate synthase gene, MS=methionine synthase gene, MTHFD1=methylenetetrahydrofolate dehydrogenase, methylenetetrahydrofolate cyclohydrolase formyltetrahydrofolate synthetase gene, MAO=monoamine oxidase gene, TNF=tumour necrosis factor gene, STG=serotonin transporter gene, GST= glutathione S-transferase, LDLR=low density lipoprotein gene, DRD=dopamine receptor gene, APOE=apolipoprotein E gene, INSR=insulin receptor gene, inos=inducible NOS gene, CACNA1A=calcium channel subunit gene, TDT=transmission disequilibrium test, DBH=dopamine beta hydroxylase gene, enos=endothelial nitric oxide synthase gene, DAT=dopamine transporter gene 48

78 2.3.8 Gene Expression Studies in Migraine Few human studies have been published that examine gene expression profiles in migraine. In neurological disorders such as migraine, a limitation for tissue specific gene expression studies is the difficulty in obtaining brain tissue. Thus, the few published studies in humans have relied on gene expression in peripheral blood cells. The first study, published in 1998, looked at expression of G protein mrna under the hypothesis of abnormalities in signal transduction in migraine and cluster headache, specifically polyphosphoinositide (PPI) responsiveness, that show specificity for the headache phenotype. Findings showed a consistent downregulation of Giα mrna (which inhibits adenyl cyclase) in all migraine groups whether quiescent or acute, with aura or without (Gardiner et al. 1998). The authors propose that this may arise from endocrine, neurotransmitter, or glucocorticoid effects, all of which are known to influence the expression specifically of Gi mrna, and all of which have been implicated in migraine. A study published in 2004 examined the expression levels of genes in migraineurs, individuals with chronic migraine (defined as having >15 headache days per month) and healthy controls using microarray techniques. Results showed that many of the genes upregulated in migraine and chronic migraine were those expressed primarily in platelets (Hershey et al. 2004). The most recent study investigated NF-κB activity, a transcription factor which plays an important role in inos (inducable nitric oxide synthase) induction, and inos expression in monocytes from the internal jugular blood of migraine without aura patients during attacks. Nitric oxide (NO) has been implicated in migraine and is produced by different isoforms of NOS. Results suggested an increased production of NO by monocytes of MO patients during attacks possibly due to a transient increase in the activity of NF-κB (Sarchielli et al. 2006) Summary of Migraine Genetic Studies In summary, numerous lines of evidence point to a role for genetic factors in migraine. However the number and type of genes involved remain unclear. Considering the probable polygenic, multifactorial nature of common migraine, is appears likely that a 49

79 number of modest effect common genetic variants that also occur in the non-migraine population interact with environmental factors to impact on the individual migraine threshold. At present, the most plausible evidence exists for genes that can be classified as being involved in neuronal or vascular function, although a number of other variants have also been implicated. Genes that are involved in vascular function through their role in homocysteine metabolism have been implicated in migraine in numerous populations. In light of the reported association of the MTHFR 677 variant with MA at the GRC in an Australian population, a follow-up independent study in a similar population is clearly warranted. Follow-up confirmatory studies are now considered crucial for verification of positive associations. Investigation of other genes involved in this pathway may also shed more light on the role of these genes in migraine susceptibility. Hence, this research aimed to investigate variants in the MTHFR and MTRR genes as candidate genes for a potential role in migraine by cross-sectional association analysis in a large Australian study group. Several lines of evidence also support a role for hormones in migraine, yet a potential role for genes involved in hormone function in migraine susceptibility has not yet been investigated. Therefore this research also aimed to investigate hormonally related genes, specifically ESR 1 and PGR, as these genes play a major role in mediating the effects of estrogen and progesterone. ESR 1 and PGR were investigated by cross-sectional association analysis in two large Australian cohorts. Functional analyses investigating gene expression by real-time reverse transcription polymerase chain reaction were also undertaken to determine if gene expression levels were different in migraineurs versus controls and if specific gene variations impacted on expression levels. Finally, to address the issue that a number of modest effect common genetic variants are most likely involved in migraine susceptibility, this study aimed to investigate the impact of multiple and potentially interacting genetic variants on migraine by creating specific genetic risk profiles and analysing their compound impact instead of focusing on a single causative gene. 50

80 Chapter 3 Methodology and Research Background 51

81 3.1 Ethical Approval Prior to commencement of this study, research was approved by the Griffith University Ethics Committee for experimentation on human subjects (App. No. MSC/05/05/HREC). 3.2 Association Analysis Association analysis is often used to investigate whether certain variants in candidate genes are involved in disease susceptibility. In a complex disorder such a migraine, where a number of genes are likely to be involved in susceptibility to the disorder, association analysis may be a worthwhile method to detect particular genetic influences of modest effect. Migraine pathogenesis is considered to involve several systems and pathways and furthermore, there are numerous well known triggers common to many sufferers. Thus there exists an extensive range of suitable candidate genes that may be worthy of investigation for a potential role in migraine. This research analysed two genes involved in hormonal function, the estrogen receptor (ESR) and progesterone receptor (PGR) genes, along with two genes involved in vascular function, the methylenetetrahydrofolate reductase gene (MTHFR) and the methylenetetrahydrofolate reductase synthase gene (MTRR). Alerations in both hormonal and vascular function have been implicated in migraine. Thus this study considered these genes suitable candidates for investigation for a role in migraine Study Design This study investigated migraine candidate genes by independent cross-sectional association analysis to detect if genotype/allele frequencies of specific genetic loci in candidate genes occured at altered frequencies between a large group of migraineurs compared to a large carefully age, sex and ethnicity matched group of individuals who did not suffer from migraine. Statistically significant differences in frequencies would 52

82 implicate the marker under analysis or a marker in linkage disequilibrium with the marker under analysis as playing a role in migraine susceptibility Population Demographics There were 1150 individuals recruited for this study. All 1150 participants gave informed consent prior to participation. All participants were interviewed, and completed a detailed questionnaire providing information including personal and family medical history, and if a migraine sufferer, details of migraine symptoms, age of onset, frequency, severity and treatment. This questionnaire revealed that 78% of individuals in the migraine group had a known family history of migraine. Migraineurs were diagnosed by a clinical neurologist as having either migraine with aura (MA) or migraine without aura (MO) based strictly on the widely accepted criteria specified by the International Headache Society (HCCIHS 1988). The study group of 1150 participants comprised two separate association populations which were considered representative sub-populations of migraineurs and non-migraineurs in Australia. The first study population, migraine association population 1 (MAP 1) comprised 275 migraineurs and 275 unrelated control individuals. The controls were matched for sex, age (+/- 5 years), and ethnicity (Caucasian) to avoid the potential bias of population stratification, were recruited in parallel at a similar time and geographical location (East Coast of Australia) as the case group. The second study population, migraine association population 2 (MAP 2) consists of 300 migraineurs similarly recruited and diagnosed, matched with 300 control individuals. This second migraine association population will be used for replication of positive associations Clinical Characteristics of the Migraine Group In the migraine study group (MAP 1 and MAP 2 together) there were 462 females (79%) and 120 males (21%). There were 339 migraineurs who suffered MO (58%), 226 who suffered MA (39%) and 17 individuals who suffered both types of attacks (3%). This distribution was similar in the male and female subgroups. 53

83 Of the 558 migraineurs who provided information on the longest duration of their migraine attacks, 33% reported that the longest duration of an attack was between 1-3 days, while 18% reported the longest duration of an attack as hours. There were 19% of migraineurs who had experienced attacks lasting more than 4 days. See Figure 3.1 below. This distribution was similar in the male and female subgroups. Longest Duration Frequency <1hr 1-3hrs 3-<4hrs 4- <12hrs days 3-4 days 4-6 days >6 days <24hrs Figure 3.1 Longest duration of migraine attacks in the migraine study group Of the 494 migraineurs who provided information on the usual duration of an attack, 36% reported the usual duration of a migraine attack as 1-3 days, 21% reported the usual duration of an attack as hours and 21% reported the usual duration of an attack as 4-12 hours. See Figure 3.2a below. The distribution in the female subgroup was similar to that seen in the total study group with the 39% of women reporting their usual migraine attack duration as 1-3 days, 20% reported the usual duration of an attack as hours and 20% reported the usual duration of an attack as 4-12 hours. The distribution in the male group was notably different with the most frequent attack duration of hours (29%) followed by 4-12 hours (27%), then 1-3 days (22%). See Figure 3.2b. Thus it appears that the usual duration of a migraine attack is shorter for males than it is for females in this large study group. 54

84 Usual Duration Frequency <1hr 1-3hrs 3-<4hrs 4- <12hrs 12- <24hrs 1-3 days UsualDuration 3-4 days 4-6 days >6 days Figure 3.2a Usual duration of migraine attacks in the migraine study group UsualDuration UsualDuration Female 20 Male Frequency 100 Frequency <1hr 1-3hrs 3-<4hrs <12hrs <24hrs days days UsualDuration 4-6 days >6 days hrs 3-<4hrs 4-<12hrs 12- <24hrs 1-3 days 3-4 days 4-6 days >6 days UsualDuration Figure 3.2b Usual duration of migraine attacks in the female and male migraine study groups Of the 559 migraineurs who provided information on attack frequency, 31% reported the frequency of their migraine attacks as <1 per month, 34% reported the frequency of their migraine attacks as 1-2 per month, 15% reported the frequency of their migraine attacks as 3-4 per month and 19% reported the frequency of their migraine attacks as >4 per month. See Figure 3.3 below. This distribution was similar in the male and female subgroups. 55

85 Frequency Frequency <1/ month 1-2/month 3-4/ month >4/month Frequency Figure 3.3 Frequncy of migraine attacks in the migraine study group Of the 458 migraineurs who provided information on the age of onset of their attacks, 37% reported that their attacks first occurred between the ages of This percentage was similar in both males and females. See Figure 3.4 below. Onset Age Frequency Figure 3.4 Age of onset in the migraine study group O ta 56

86 Of the 392 migraineurs who provided information on their total number of attacks, 87% reported that they had suffered over 20 attacks at time of participation in the study. See Figure 3.5 below. This distribution was similar in the male and female subgroups. Total Attacks Frequency >20 Figure 3.5 Total number of attacks in the migraine study group When migraine specific symptoms were investigated in all migraineurs and the male and female subgroups, it was found that 87% of all migraineurs suffered nausea (77% of males, 89% of females), 65% suffered emesis (54% of males, 68% of females), 79% suffered phonophobia (78% of males, 80% of females) and 92% suffered photophobia (81% of males, 93% of females). Additionally, 85% of all migraineurs suffered pulsating headpain (81% of males, 86% of females) while 81% suffered unilateral headpain (69% of males, 83% of females). See Figures below and Table 3.1 for a summary of migraine specific symptoms. 57

87 Nausea Frequency yes Nausea no Figure 3.6 Individuals who suffer nausea in the migraine study group Emesis Frequency yes Emesis no Figure 3.7 Individuals who suffer emesis in the migraine study group 58

88 Phonophobia Frequency yes Phonophobia no Figure 3.8 Individuals who suffer phonophobia in the migraine study group Photophobia Frequency yes Photophobia no Figure 3.9 Individuals who suffer photophobia in the migraine study group 59

89 Pulsating Head Pain Frequency yes HeadpainPulsating no Figure 3.10 Individuals who suffer pulsating head pain in the migraine study group Unilateral Head Pain Frequency yes HeadpainUnilateral no Figure 3.11 Individuals who suffer unilateral head pain in the migraine study group 60

90 Table 3.1 Summary of specific migraine symptoms in the study group Migraine Symptom Yes No Nausea 430 (87%) 63 (13%) Emesis 322 (65%) 170 (35%) Phonophobia 387 (79%) 106 (21%) Photophobia 452 (92%) 42 (8%) Pulsating Head Pain 419 (85%) 72 (15%) Unilateral Head Pain 380 (69%) 91 (31%) When migraine specific triggers were investigated in the study group it was found that 71% of all migraineurs reported stress as a specific migraine trigger, while 24% reported holidays/relaxation and weather changes as triggers for their migraines. Food and beverage triggers were red wine (29%), other alcohol (24%), chocolate (32%), oranges (16%) and ripe cheese (15%). These triggers were similar in the male and female subgroups. See Figures and Table 3.2 for a summary of specific migraine triggers investigated in the study group. Stress Frequency yes Stress no Figure 3.12 Individuals who report stress as a trigger in the migraine study group 61

91 Holidays Relaxation Frequency yes HolidayRelaxation no Figure 3.13 Individuals who report holiday/relaxation as a trigger in the migraine study group Weather Changes Frequency yes WeatherChanges no Figure 3.14 Individuals who report weather changes as a trigger in the migraine study group 62

92 Red Wine Frequency yes RedWine no Figure 3.15 Individuals who report red wine as a trigger in the migraine study group Other Alcohol Frequency missing data yes no OtherAlcohol Figure 3.16 Individuals who report other alcohol as a trigger in the migraine study group 63

93 Chocolate Frequency missing data yes no Chocolate Figure 3.17 Individuals who report chocolate as a trigger in the migraine study group Table 3.2 Summary of reported triggers in the migraine study group Migraine Trigger Yes No Stress 342 (71%) 142 (29%) Holidays and relaxation 113 (24%) 357 (76%) Weather changes 112 (24%) 361 (76%) Red wine 135 (29%) 337 (71%) Other alcohol 112 (24%) 361 (76%) Chocolate 151 (32%) 323 (68%) In all migraineurs 19% suffered other pain, 28% suffered chronic neck pain, 21% suffered depression, 16% suffered high blood pressure and 11% suffered chronic back pain. There were no notable differences between male and female migraineurs. 64

94 Migraine Treatment Used in the Study Group In the migraine study group 57% reported that they used medication prescribed by a medical practitioner. This was an interesting observation as the American Migraine Studies of 1989 and 1999 reported that 37% (in 1989) and 41% (in 1999) of migraineurs in the USA used medication prescribed by a medical practitioner (Lipton et al. 2001). Although these reports were generated in a different population, the similarities between USA and Australian populations suggest that there is increased use of prescribed medication by migraineurs in the study population. This may be due to the improved efficacy of available migraine treatments since the studies by Lipton et al. (2001) or an increase in migraineurs seeking medical attention. Of the group that used medication prescribed by a medical practitioner, 64% used a treatment that acted on the serotonergic system. In the group of migraineurs who used prescribed medication, 62% reported that the treatment was effective. Of all migraineurs, 85% used painkillers, 25% used a natural remedy, 36% used medication for nausea and 43% used other medications. The distribution for treatments used was similar in the male and female subgroups, except for use of a natural remedy. There were 28% of female migraineurs who used a natural remedy compared to only 9% of male migraineurs Sample Collection For the association and mutation analysis, all study participants provided 24mL of whole blood which was collected into BD vacutainers containing EDTA to prevent clotting. All blood not used for immediate DNA isolation was centrifuged for 15 minutes at 2000 rpm to separate the plasma and formed elements which were stored separately at -80ºC for future DNA extraction and purification. 65

95 3.2.4 DNA Purification The DNA was isolated from leucocytes by a modification of the salting out method used by Miller et al. (1988) and is detailed in Appendix A (Miller et al. 1988). The extracted DNA was precipitated in -20 C ethanol and resuspended in 100μL of sterile milli Q water. It was quantitated by spectrophotometric analysis using the QuantaGene RNA/DNA analyser (The Australian Chromatography Company, Australia) determining the optical density at a wavelength of 260nm, with an optical density (OD) of 1.0 representing 50μg/mL of double stranded DNA, and adjusted to a working concentration of 20ng/ul. The DNA was stored at 4ºC until used Genotyping Overview Polymerase Chain Reaction All samples were subjected to the polymerase chain reaction (PCR) to amplify the particular genomic regions of interest. PCR allows amplification of a specific target DNA sequence so that it may be used for further analysis. PCR requires specific oligonucleotide primers flanking the region of interest, PCR buffers, magnesium chloride (MgCl 2 ), deoxynucleotide triphosphates (dntps) and Thermus aquaticus (Taq) polymerase. Exponential growth of the region of interest is achieved by cycles of denaturation, annealing of the primers, and extension. Taq polymerase is the enzyme of choice due to its thermostable quality, avoiding the need to add more polymerase at each cycle (Mullis 1990) Primer design and preparation Primer sequences were either obtained from existing published assays or designed using the Primer3 program (available Sequences were checked for homology and specificity to the genomic region of interest by entering the sequences into Basic Local Alignment 66

96 Search Tool (BLAST) (available All oligonucleotide primers were obtained from either Proligo (New South Wales, Australia) or Geneworks (South Australia). All primers were provided in sequencing quality either as a lypholysed pellet or in a concentrated liquid form. Primers were diluted to a working concentration of 5μM with sterile water Restriction Enzyme Digestion Restriction enzymes are enzymes naturally occurring in bacteria that defend against incorporation of foreign DNA. Their method of action is cleavage of DNA at particular sequence recognition sites providing an identifying landmark that may differentiate it from other DNA samples. When used at a polymorphic site, differentiation in the DNA samples can be obtained due to success or failure of cleavage by the enzyme, resulting in fragment lengths of different sizes. The different fragments sizes may then be detected using gel or capillary electrophoresis. These particular sites are termed restriction fragment length polymorphisms (RFLPs). RFLP techniques were used to differentiate between wildtype and variant alleles in the estrogen receptor gene (ESR 1), and the methylenetetrahydrofolate reductase gene (MTHFR) and methionine synthase reductase (MTRR) gene. All restriction enzymes were supplied by New England Biolabs Inc. (Ipswich, MA, USA) and included an appropriate buffer. Restriction enzyme digestion reaction conditions were optimised using the appropriate buffer at 1x concentration and sufficient amount of enzyme to obtain complete exonuclease activity within the recommended time Agarose Gel Electrophoresis Gel electrophoresis allows the separation and identification of DNA fragments of various sizes. DNA is loaded into wells in a supporting matrix of agarose, and exposed to an electrical field. The negatively charged DNA molecules migrate through the pores in the agarose matrix to the positive anode, their velocity decreased as their fragment size is increased. The location of the DNA within the gel is determined by staining with a 67

97 fluorescent intercalating dye such as ethidium bromide, which is used in this study, and viewed under a UV light source. The DNA molecules form distinctive bands in the agarose gel according to their fragment sizes. This technique was used for analysis of all PCR products in this study Quality Control To reduce the likelihood of genotyping error and contamination producing spurious results, random repeat samples and positive and negative controls were included in all assays. Hardy Weinberg equilibrium as described in was used as a tool to further check for genotyping errors. Genotype calls were confirmed by an independent researcher who was blinded to all details of the sample. The genotyping error rate was estimated as <5% Statistical Analysis Genotype data and allele frequencies were compared between the migraine and the control groups, male and female, and MO and MA subgroups using standard chi-squared analysis (X 2 ). Pearson s chi squared statistic is a non-parametric test of statistical significance used to assess two types of comparisons, tests for goodness of fit and tests for independence. The chi squared computational formula is as follows. X 2 = Σ (Observed frequency Expected frequency) 2 /Observed frequency All analyses in this research used the chi-squared statistic to examine independence assessing whether paired observations on two variables (in a 2 x 3 contingency table for genotype frequencies and a 2 x 2 contingency table for allele frequencies) were independent of each other. These contingency tables were analysed directly using an observed vs expected test statistic, on one degree of freedom using SPSS version A standard alpha level of 0.05 was set. 68

98 For all contingency tables greater than 2 x 3, for example when the effect of two genotypes was assessed, the Clump program was used (Sham and Curtis 1995). Clump analysis is most useful when there are small frequencies in some cells. The Clump program is designed to assess the significance of the departure of observed values from the expected values in a contingency table conditional on the marginal totals. The significance is assessed using a Monte Carlo approach, by performing repeated simulations to generate tables having the same marginal totals as the one under consideration, and counting the number of times that a chi-squared value associated with the real table is achieved by the randomly simulated data. An original feature of clump is a novel chi-squared value which it derives. This is produced by clumping columns together into a new two-by-two table in a manner which is designed to maximise the chisquared value (Sham and Curtis 1995). Clump results were presented as the T 1 statistic (chi-square statistic generated from the raw 2 x n table) and T4 statistic (analysis of a 2- by-2 table obtained by clumping the columns of the original table to maximise the chisquared value) along with the p values. Odds ratios (OR) and 95% confidence intervals were also calculated using an on-line program available at (Bland and Altman 2000). The odds ratio for disease is the ratio of risk allele carriers to non-carriers in cases compared with controls and gives the increase in disease risk for carriers of risk alleles compared to non-carriers. To illustrate how the odds ratio is calculated, Table 3.3 shows a hypothetical 2 x 2 table of frequencies of carriers of a risk allele compared to no risk alleles in migraine and control groups. The odds ratio is calculated by the following formula. (100/50)/(40/60) = 3 Therefore migraineurs are 3 times more likely to carry a risk allele than non-migraineurs. Alternatively it may be calculated as follows. (100/40)/(50/60) = 3 Therefore it also may be stated that subjects who carry a risk allele are 3 times more likely to suffer migraine than those who do not carry a risk allele. 69

99 Table 3.3 Hypothetical 2 x 2 table of frequencies of risk and no risk allele carriers in migraine and control groups Risk Allele No Risk Allele Migraine Control In cases of multiple testing, the Bonferroni correction for multiple testing was applied. This is considered a conservative multiple-comparison correction and is applied by adjusting the alpha level by the number of comparisons run in the analysis Power Analysis Power is an important consideration in genetic association studies. Power calculations address the concerns of whether a study is designed to maximise success in accurately identifying an effect, while minimising the use of resources (Schork 2002). Apriori power was estimated on all association studies performed in this research using CaTS, a multi-platform interface for carrying out power calculations for large genetic association studies (Skol et al. 2006). CaTS was downloaded from The Michigan University Centre for Statistical Genetics and is available at Table 3.4 illustrates the calculated power estimates based on the minimum study group size of 275 cases and 275 controls (MAP 1) setting an alpha level of 0.05 and a disease prevalence of The power calculation assumes both multiplicative (where the joint effect of two or more factors is the product of their effects) and additive (the presence or absence of contribution to a trait by different alleles) genetic models. Results show that all 70

100 association studies have >85% power to detect a genotype relative risk of 1.5 considering both multiplicative and additive genetic models with disease allele frequencies ranging from Table 3.4 Genetic association study power calculation based on 275 cases and controls considering α = 0.05 and disease allele frequencies ranging from Genotype Relative Risk M A M A M A M A Disease Allele Frequency M = multiplicative genetic model; A = additive genetic model Hardy Weinberg Equilibrium The Hardy Weinberg Law provides a quantitative relationship between phenotype, genotype and allele frequencies within populations. For a marker with two alleles, it uses the binomial equation of p 2 + 2pq + q 2 = 1, where p represents the frequency of one allele and q represents the frequency of the other, and p 2, 2pq, and q 2 represents the frequencies of two homozygous and one heterozygous genotypes. Hardy Weinberg equilibrium implies that the genotype frequencies can be determined directly from the allele frequencies and provides a check to ensure that genotyping errors do not explain observed results (Yonan et al. 2006). All association study results were checked to ensure that they fell within the Hardy Weinberg equilibrium by the chi-square test for goodness of fit using Microsoft Excel. In the case of a significant difference between genotype and/or allele distributions between the migraine and control groups, the Hardy Weinberg 71

101 equilibrium was applied to the control group only. In instances where a SNP has a true genetic effect, the distribution in the migraine group may not be in Hardy Weinberg equilibrium (Lewis 2002) Linkage Disequilibrium Analysis Linkage disequilibrium (LD) refers to correlations among neighbouring alleles, reflecting 'haplotypes' descended from single, ancestral chromosomes (Reich et al. 2001). A statistic that measures pairwise LD between two genetic markers is D which denotes the difference between the probabilities of observing alleles in two polymorphic markers independently in the population. The measure of D can be represented by the following equation: D = f(a 1 B 1 ) f(a 1 )f(b 1 ) where A and B refer to two genetic markers and f is their frequency. D' is obtained from D/Dmax and a value of 0.0 suggests independent assortment, whereas 1.0 means that all copies of the rarer allele occur exclusively with one of the possible alleles at the other marker (Lewontin 1964; Pittman et al. 2005). A program, called 2LD (2 locus linkage disequilibrium), has been developed by Zhao (2004) to enable LD calculation for 2 markers giving the D estimate and standard error (Zhao 2004). In order to undertake pairwise linkage disequilibrium analysis between the ESR 1 intron 1, exon 4 and exon 8 polymorphisms, and also in the MTHFR C677T and A1298C mutations tested in the same study population, the programs EH (estimating haplotypes) (Terwilliger and Ott 1993) and 2LD (Zhao 2004) were used. The EH program was used to estimate haplotype frequencies for each marker and these data were entered into the 2LD program for LD analysis. 72

102 3.3 Mutation Screening To detect possible mutations in DNA sequences, the base sequence of specific DNA samples can be determined by automated sequencing methods and compared to known wild-type sequences available on public databases such as Genbank (Benson et al. 2006). This study used the chain termination method of DNA sequence determination where a short primer complementary to the region of interest is annealed and extended by DNA polymerase. Low concentrations of specific dideoxy nucleotides, in addition to the normal deoxynucleotides are added. Dideoxynucleotides have no 3' hydroxyl group, and once incorporated, block further chain extension. This method has been automated using fluorescent dyes for the detection of the electrophoretically resolved DNA fragments. Different dyes are attached to each of the the terminating dideoxynucleoside triphosphates so that only a single extension reaction is required for each template (Rosenblum et al. 1997). Therefore, each reaction accumulates a mixture of differentially labelled chains of lengths determined by the template sequence. In this study, automated DNA sequencing was undertaken on the Applied Biosystems (Foster City, California USA) AB3730xl, a capillary electrophoresis based automated sequencer which provides base-call and output results in the form of a chromatogram according to Figure This service was provided by the Australian Genome Research Facility (Brisbane). Figure Sample chromatogram from an automated sequencing reaction. The resultant chromatograms were analysed using the Applied Biosystems Sequence Scanner TM v1.0 software which is comprehensive software designed to display, edit, trim, 73

103 print, export and generate reports for sequencing sample files from Applied Biosystems genetic analysis instruments. 3.4 Gene Expression Analysis Study Design Reverse transcriptase polymerase chain reaction (RT-PCR) involves isolation of RNA, which is converted to cdna by reverse transcriptase and amplified using PCR techniques. Real-time reverse transcriptase polymerase chain reaction evaluates the PCR data in real time by analysing the accumulation of amplicon during the exponential phase of the PCR reaction. This technique allows for the quantification of RNA and thus provides a method for measuring gene expression in various tissues and samples. This study used a two-step procedure, whereby mrna was first transcribed into cdna, and the genes of interest were later amplified in a separate reaction. This study investigated expression of the ESR 1 and PGR genes by relative real-time reverse transcriptase polymerase chain reaction in a sample of 6 female migraineurs who suffer from MA and 6 non-affected females. The MA phenotype is a more severe phenotype, and hence should provide less scope for diagnostic error. The study investigated whether expression levels of the ESR 1 and PGR genes in leukocytes were altered in migraineurs compared to non-migraineous individuals. Furthermore, this study investigated whether specific ESR1 and PGR genotypes impact on expression levels. The quantitation of gene expression in leukocytes is considered a good surrogate for direct tissue sampling, as leucocyte expression levels mimic expression levels in many other tissues within the individual (Whitney et al. 2003). Therefore altered expression levels may provide clues to the biological mechanism behind the role of hormones in migraine susceptibility. Relative gene expression analysis was performed using the Corbett Rotor-Gene 6000, which uses fluorescent techniques to amplify, detect, and analyse the genes of interest. This technique involves the binding of the fluorescent dye SYBR green to the doublestranded cdna during the polymerisation step, which is then measured at the end of each 74

104 elongation cycle to monitor the increasing amount of product (Bustin 2000). To normalize errors as a result of variation in the amount of starting material between the samples, a gene which is expressed at a constant level in the tissue of interest and is unaffected by disease state, was used as an internal reference against which the gene expression of ESR 1 and PGR was normalised (Bustin 2000) Control for Endogenous Variation in Expression Levels An important consideration in analysing the expression of hormone receptor genes was to ensure that samples were collected under similar endogenous conditions. This was undertaken to ensure that any potential changes in gene expression had not occurred solely due to differences in hormone receptor gene expression levels arising from cyclical hormone levels. The primary source of circulating sex steroid hormones is the gonads although it is well known that non-reproductive tissues can also synthesise hormones (Simpson 2003; Krause et al. 2006). There are three estrogens in the circulation, estradiol, estrone and estriol. Estradiol binds strongly to target tissues and is the most potent estrogen. Estrone has lower potency, and estriol is the weakest estrogen. A major factor in the potency differences is the length of time the estrogen receptor complex occupies the nucleus (Simpson 2003). Due to the long half-life of the estradiol hormone-receptor complex, only small amounts of estrogens need to be present in the circulation to cause an effect (Speroff and Lobo 1994). Like estrogen, progesterone is synthesized from pregnenolone, a derivative of cholesterol. Figure 3.19 illustrates a simplified diagram of the major pathways of steroid synthesis from cholesterol. In males steroid hormone levels are relatively stable over time, however in menstruating females, the menstrual cycle requires a carefully coordinated sequence of events controlled in part by changing cyclical estrogen and progesterone levels (Silberstein and Merriam 2000). 75

105 Figure 3.19 Simplified diagram of the major pathways of steroid synthesis from cholesterol From: Nussey and Whitehead Endocrinology: An Integrated Approach, BIOS Scientific Publishers Ltd, Oxford. Available: Accessed Estrogen increases target tissue responsiveness to itself, androgens and progestins by increasing receptor concentration. Progesterone on the other hand limits tissue response to estrogen by decreasing the concentration of estrogen receptors (Speroff and Lobo 1994). Progesterone receptor expression is upregulated by estrogen and down-regulated by progesterone in most target tissues (Bouchard 1999). Consequently, cyclic hormonal conditions could result in a range of circulating leukocyte populations that have developed under different hormonal influences. It is therefore likely that leukocyte 76

106 expression of ESR and PGR may alter according to cyclic hormonal conditions (Molero et al. 2002; Slayden and Brenner 2004). To ensure that blood samples that had developed under similar hormonal influences were used for this study, all samples were collected during one of the first 4 days of the ovarian cycle, the early follicular phase (day 1-4 of menstruation). Plasma estradiol and progesterone levels were measured in all participants to ensure that any potential differences in expression levels did not occur due to cyclical hormonal differences between the participants. If any samples fell outside of the normal range for circulating estradiol and progesterone levels during the first week of the ovarian cycle, they were excluded from the study. Mean estradiol and progesterone levels were determined in the migraine and control groups and a Student s t test was performed to ensure that there were no significant differences in hormone levels in the migraine and control groups under analysis Study Population Research was approved by the Griffith University Ethics Committee for experimentation on human subjects. All participants of the study were asked to provide written consent prior to participation. All participants were interviewed, and asked to complete a detailed questionnaire providing information including personal and family medical history. In migraineur participants, detailed information was obtained concerning migraine symptoms, age of onset, frequency, severity and treatment, and details of hormonal cycle and triggers was obtained. All migraineurs were diagnosed based strictly on the widely accepted criteria specified by the International Headache Society (HCCIHS 1988) as having migraine with aura. The study group consisted of 6 female migraineurs who suffered MA and 6 female controls with no history of migraine. The study group ranged in age from All participants reported regular monthly menstrual cycles. All migraineurs reported that their migraines were at times related to their menstrual cycle. This characteristic was considered important to enrich the study group to detect a potential effect. 77

107 3.4.3 Sample Collection and Storage All study participants provided 10ml whole blood of which 8 ml was collected in a serum separator collection tube (SST) for determination of plasma hormone levels, and 2 ml was collected into a Paxgene Blood RNA Tube (PreAnalytiX, Switzerland). This method of collection has been shown to reduce RNA degradation and inhibit or eliminate gene induction in phlebotomy whole blood samples (Rainen et al. 2002). Participants also provided a saliva sample collected in an Oragene DNA saliva kit (DNA Genotech Inc. Ontario, Canada). The SST tube was gently inverted 5 times, allowed to clot 30 minutes, then centrifuged for minutes at RCF to separate the plasma from the formed elements. The plasma was removed and transferred to a separate 10mL Eppendorf tube and stored at -80ºC for later analysis to determine circulating estradiol and progesterone levels. Circulating estradiol and progesterone levels were performed by Queensland Medical Laboratories, Brisbane. The Paxgene Blood RNA tubes were stored overnight at -20ºC then transferred to -80ºC for storage until RNA extraction was performed. The Oragene saliva kits were also stored at -80ºC until DNA extraction DNA Extraction DNA was isolated from each Oragene saliva sample according to the recommended method as follows: 1. 20µL Oragene Purifier added to 500µL Oragene saliva mix 2. Incubated on ice for 10 minutes 3. Centrifuged 3 minutes 13K rpm 4. Pipeted clear supernatant to fresh tube, discarded pellet 5. Added 500µL 95% ethanol to supernatant and mix 6. Incubate for 10 minutes to allow DNA precipitation 7. Centrifuged 1 minutes 13K rpm 8. Discarded supernatant 9. Dissolved pellet in TE (10mM Tris-HCl, 1mM EDTA, ph 8.0) The DNA was quantitated and diluted to a working concentration of 20ng/ µl. 78

108 Each sample was checked for purity by spectrophotometry measuring the ratio of the absorbance at the wavelengths of 260nm to 280 nm. DNA absorbs UV light at 260 nm, and protein absorbs UV light at 280 nm. A pure sample of DNA has the 260/280 ratio of A DNA preparation that is contaminated with protein will have a 260/280 ratio lower than 1.8 (Teare et al. 1997). Results of spectrophotometric analysis indicated that all samples fell within a 260/280 ratio of 1.8 to 2 indicating pure DNA RNA Extraction RNA was extracted from the Paxgene collected samples using the standard recommended method provided in the Paxgene Blood RNA Kit handbook. Purification involved an initial centrifugation step to pellet nucleic acids in the PAXgene Blood RNA Tube. The pellet was then washed and resuspended, and incubated in optimized buffers together with proteinase K to bring about protein digestion. An additional centrifugation through the PAXgene Shredder spin column was carried out to homogenize the cell lysate and remove residual cell debris. Ethanol was added. The lysate was applied to a PAXgene RNA spin column. During a brief centrifugation, RNA was selectively bound to the PAXgene silica-gel membrane as contaminants pass through. Remaining contaminants were removed in several wash steps. Between the first and second wash steps, the membrane was treated with DNase 1 to remove trace amounts of bound DNA. After the wash steps, the RNA was eluted in elution buffer and heat-denatured. A portion of the RNA was converted to cdna while the remainder was stored at -80ºC RNA Integrity As RNA is extremely fragile once it has been removed from its cellular environment, its purification is more problematic than that of DNA. Degradation of RNA is a common 79

109 problem after purification. Assessment of RNA integrity is an important consideration to ensure that it is a suitable template for RT-PCR (Bustin and Nolan 2004). The assessment of RNA integrity was undertaken in all samples by inspection of the 28S and 18S ribosomal RNA bands using an Agilent (Palo Alto, CA) 2100 Bioanalyzer. The Agilent 2100 Bioanalyzer uses a combination of microfluidics, capillary electrophoresis, and fluorescent dyes that bind to nucleic acid to evaluate RNA concentration and integrity. The output is a scan of mass vs. size as seen in Figure 3.20 and an electrophoretogram as seen in Figure Strong peaks with little background noise on the scan are indicative of intact RNA. The specific regions that are indicative of eukaryote RNA quality are indicated in Figure 3.21 (Mueller et al. 2004). Crisp 28S and 18S rrna bands (of approximately 5 kb and 2 kb in size) on an electrophoretogram are also indicative of intact RNA. The electrophoretogram in Figure 3.22 illustrates RNA integrity analysis with progressive degradation. The first 4 samples show crisp 28S and 18S rrna bands with a shift towards shorter fragment sizes with progressing degradation (Mueller et al. 2004). All samples were analysed for RNA quality and integrity using this method. 80

110 Figure Agilent 2100 bioanalyzer scan of mass vs. size for extracted RNA. The top scan indicates intact RNA. Partially degraded RNA is seen in the middle scan and degraded RNA is seen in the bottom scan (Mueller et al. 2004). Figure 3.21 The regions that are indicative of intact eukaryote RNA (Mueller et al. 2004). 81

111 Figure 3.22 Agilent 2100 bioanalyzer electrophoretogram for extracted RNA showing progressive degradation from left to right (Mueller et al. 2004) cdna Synthesis RNA cannot function as a template for PCR, therefore RNA must be converted into cdna by reverse transcription. In this study RNA was transcribed into cdna using a two-step reverse transcription-pcr protocol using random hexamer primers which maximise the number of mrna molecules that can be analysed from a small sample of RNA (Bustin 2000). Random hexamers prime RNA at multiple points along the transcript. The method is non-specific, but yields the most cdna and is useful for transcripts with significant secondary structure (Bustin and Nolan 2004). A reported advantage of random priming is that it generates the least bias in the resulting cdna (Ginzinger 2002). Due to potential problems arising from reported experimental variation in reverse transcription-pcr attributable to the reverse transcription step (Stahlberg et al. 2004), all samples were transcribed into cdna using the same reverse transcription methodology and reaction conditions. Reverse transcription in all samples was carried out over a two week period and all reagents used were from the same batch. The protocol used for all samples is in Table 3.5 below. 82

112 Table 3.5 cdna synthesis protocol Reagent Final One Reaction Concentration 20mM dntps 2mM 2μL (New England Biolabs) 5x 1st strand buffer* 1x 4μL (Invitrogen) 0.1mM DTT 0.08mM 4μL (Invitrogen) Random Hexamer primers 0.5ng 1μL (500ng/ul) (Invitrogen) Sterile Water To 20 μl RNA 2ug TOTAL 20μL *250mM Tris-HCl (ph 8.3 at RT, 375 mm KCl, 15mM MgCl 2 ) The reaction was incubated for 5 mins at 42 C. 200 units of Superscript III (Invitrogen) was added. The reaction was incubated for 1 hr at 42 C. 200 units of Superscript III (Invitrogen) was added. The reaction was incubated for a further 1 hr at 42 C. The resultant cdna was stored at 4 C Detection of Possible Genomic Contamination PCR cannot discriminate between cdna targets synthesized by reverse transcription and genomic DNA, therefore genomic DNA contamination in the cdna samples has the potential to cause false positive and spurious results. To ensure that all samples were free from genomic DNA contamination a PCR assay was designed that would differentially amplify both genomic and cdna using the Calpain (CAPN1) gene. Primers were designed to amplify a 116bp fragment in an exonic region of CAPN1 in cdna and a 261 bp intron spanning fragment in genomic DNA. The primers were as follows: 83

113 CAPNI F 5 AAACAGTTCGACACTGACCGAT 3 CAPNI R 5 AGTAGCGTCGGATGATCATGTT 3 All cdna samples were amplified using the protocol and thermocycler (Corbett Research) conditions outlined in Table 3.6: Table 3.6 Calpain PCR protocol Reagent Final One Reaction Concentration 5mM dntps 0.2mM 0.8μL (New England Biolabs) 10x PCR Buffer 1x 2μL (Applied Biosystems) 25mM MgCl 2 2mM 1.6μL (Roche) 5μM Forward Primer 0.2μM 0.4μl 5μM Reverse Primer 0.2μM 0.4μl Taq Polymerase 5U/μL 1Unit 0.2μl (New England Biolabs) Sterile Water 12.6 μl DNA 20ng/μL or cdna 40ng 2 μl The thermocycler conditions were 94 C for 4 minutes, 35 cycles of 94 C for 1 minute, 59 C for 1 minute and 72 C for 1 minute, with a final step of 72 C for 4 minutes Real-Time Reverse Transcriptase Polymerase Chain Reaction Primer design Oligonucleotide primer sequences were either obtained from an existing published assay or designed using the Primer3 program (available Primers were selected according to the following parameters: length between 18 and 24 bases (optimal bases); melting temperature 84

114 (T m ) between 55 and 60 C (optimal Tm 58 C); G+C content between 40 and 60% (optimal 50%). Where possible, primers that bound to separate exons were used. This strategy is a further technique that may be used to avoid amplification of contaminating DNA as genomic DNA will be evident by larger PCR products including the intronic sequence between the exons. Shorter amplicons amplify more efficiently that longer ones as they are more tolerant of PCR reaction conditions (Bustin 2000), therefore primers were designed to produce amplicons that were <250bp in size. Primer sequences were checked for homology and specificity to the genomic region of interest by entering the sequences into Basic Local Alignment Search Tool (BLAST) (available All oligonucleotide primers were obtained from Proligo (New South Wales, Australia). Primers were provided in sequencing quality in a concentrated liquid form and were diluted to a working concentration of 5μM with sterile water RT-PCR Chemistry To detect amplified PCR product during RT-PCR, SYBR Green, a fluorescent DNA intercalating dye was used. SYBR Green exhibits little fluorescence in solution but binds to double stranded DNA during the elongation step of PCR. In real time, an increase in fluorescence can be measured at the end of each elongation step of each cycle to monitor the increasing amounts of amplified product. Product verification can be undertaken by slowly increasing the temperature at the end of the PCR cycle above the T m of the amplicon and plotting fluorescence as a function of temperature (Bustin 2000) Assay Optimisation Each assay was optimised for primer concentration to ensure maximal possible reaction efficiency with minimal formation of primer dimers in the reaction. Melt curves were analysed to ensure that only one single peak occurred in sample reactions with no additional peaks at lower melt temperatures to indicate primer dimers due to excessive primer concentrations in the reaction mix. 85

115 In order to determine the amount of cdna required for each PCR reaction, experiments were performed with each gene of interest and the reference gene using serial dilutions of a pooled cdna preparation with all other conditions being identical. The Ct value was then plotted against the log concentration of cdna. These experiments provided information about the range of template concentrations that yielded similar amplification efficiency (Dussault and Pouliot 2006). Any template concentration falling within the linear range was considered suitable to run Rotor-Gene 6000 Real-time PCR was performed on the Rotor-Gene 6000 by Corbett Research Pty Ltd (Sydney, Australia). The Rotor-Gene is a real-time thermal cycling system that can be used for DNA amplification by PCR in real time. During a run on the Rotor-Gene 6000, a rotor spins at around 500 rpm as the tubes are thermally cycled. As illustrated in Figures 3.23 and 3.24 as the sample tubes spin, they pass a detection module where a LED source irradiates the tube and a photomultiplier (PMT) detects the fluorescent energy of SYBR green. Figure 3.23 Corbett Rotor-Gene 6000 internal mechanism showing rotor and PMT detector 86

116 Figure 3.24 Corbett Rotor-Gene 6000 Sybr green detection system In PCR reactions, PCR product is produced exponentially. It takes several cycles for enough products to be readily detectable in a real time PCR reaction, after which the early cycles are characterized by the exponential increase in target amplification. As reaction components become limiting, the rate of target amplification decreases until a plateau is reached and there is little or no net increase in PCR product. Therefore the plot of fluorescence vs. cycle in RT-PCR reactions has a sigmoidal appearance. Figure 3.25 represents a typical real-time PCR reaction. The amplification plot is the plot of fluorescence signal versus cycle number. In the initial cycles of PCR, there is little change in fluorescence signal. This defines the baseline for the amplification plot. An increase in fluorescence above the baseline indicates the detection of accumulated PCR product. A fixed fluorescence threshold can be set above the baseline. The parameter Ct (threshold cycle) is defined as the fractional cycle number at which the fluorescence passes the fixed threshold. This can best be seen in the linear scale in the Rotor-Gene quantitation window. This should be set where the rate of amplification is greatest during the exponential phase (Corbett 2004). 87

117 Figure 3.25 Representation of a typical real-time PCR amplification plot. (From PE Biosystems, Real-Time Quantitative PCR Guide) Reference Gene Selection In gene expression analysis, several variables must be controlled for. These include the amount of starting material, differences in efficiencies of the assays and differences in tissues between disease and non-disease state. The accepted method for controlling for sample to sample variation is the use of internal control genes, frequently called reference genes or house-keeping genes. The ideal reference gene should be expressed at a constant level amongst the different tissues, and should be unaffected by the disease state (Bustin 2000). Commonly used reference genes have included 18S or 28S rrna, and the mrna specified genes glyceraldehyde-3-phosphate dehydrogenase (GAPDH) and β-actin (Bustin 2000). However, the literature shows that reference gene expression can vary considerably in different samples and disease states (Thellin et al. 1999; Suzuki et al. 2000; Warrington et al. 2000), therefore a more accurate method for selection of an appropriate reference gene should be used. This study used the method described by Vandesompele et el. (2002) which compares a set of prospective reference genes to determine those that are both stably expressed and show identical expression patterns throughout all samples (Vandesompele et al. 2002). The authors initially established a measure to assess the variation in expression of various 88

118 control genes, and subsequently defined a gene-stability measure (M) as the average pairwise variation between a particular gene and all other control genes. M values <7 are considered satisfactory and an indication that the most variable reference genes have been eliminated (Vandesompele et al. 2002). The authors went on to develop an applet programmed in Visual Basic for Applications (VBA) called genorm (available to allow determination of the most stable reference gene from a set of tested genes. The eight potential reference genes that were analysed in this study were tested on 3 migraineur and 3 control samples and are as outlined in Table

119 Table 3.7 Internal control genes and primer sequences used to determine the most stable reference gene for this study Gene Symbol Accession Number Gene Name Primer Sequence ACTB NM_ Beta actin F 5 CTGGAACGGTGAAGGTGACA R 5 AAGGGACTTCCTGTAACAATGCA GAPD NM_ Glyceraldehyde-3-phosphate dehydrogenase F 5 TGCACCACCAACTGCTTAGC R 5 GGCATGGACTGTGGTCATGAG B2M NM_ Beta-2-microglobulin F 5 TGCTGTCTCCATGTTTGATGTATCT R 5 TCTCTGCTCCCCACCTCTAAGT RPL13A NM_ Ribosomal protein L13a F 5 CCTGGAGGAGAAGAGGAAAGAGA R 5 TTGAGGACCTCTGTGTATTTGTCAA HMBS NM_ Hydroxymethyl-bilane synthase F 5 GGCAATGCGGCTGCAA R 5 GGGTACCCACGCGAATCAC PRKG1 NM_ Protein kinase, cgmp-dependent, type I F 5 TGGGCTATTCCCTTTCTTCA R 5 CCAACAGATGTGTGGTCCTCT TBP NM_ TATA box binding protein F 5 ATGTTTTTCCCCATGAACCA R 5 TGCAATACTGGAGAGGTGGA HPRT1 NM_ Hypoxanthine phosphoribosyl-transferase F 5 TGACACTGGCAAAACAATGCA R 5 GGTCCTTTTCACCAGCAAGCT 90

120 PCR Efficiencies PCR reaction efficiencies are a further important consideration in RT-PCR. In the case of 100% reaction efficiency, there will be a doubling of the amount of DNA at each PCR amplification cycle, for 90% the amount of DNA will increase from 1 to 1.9 at each cycle, so the factor is 1.9 for each cycle, and similarly for 80% and 70% it will be a factor of 1.8 and 1.7 respectively. Therefore, a small difference in the PCR reaction efficiency of different assays can potentially make a big difference on the amount of final product in the two assays. If reaction efficiencies differ between the target and reference genes, these differences should be accounted for in the final analysis of relative expression, otherwise spurious results can occur. For this study, PCR reaction efficiencies were calculated for each sample by the Rotor- Gene 6000 software and averaged for each gene. Samples with reaction efficiencies that deviated too far from the mean (+/- 1 standard deviation) were excluded from the analysis. The average PCR reaction efficiency for each gene was used in the final analysis of relative expression. The Rotor-Gene software calculates reaction efficiency as amplification (λ) for each individual sample (verbal communication, Corbett Research Melbourne). The sample s amplification value λ is the average amplification taken from the 4 readings following the Take-Off point t. The Take-Off cycle is the last point before which the signal emerges from the noise level. The average amplification for each sample is then calculated Quality Control All cdna samples were run in triplicate and each run included a genomic DNA sample and a non-template control (NTC) to check for possible contamination. While the NTC may result in a fluorescence reading due to the indiscriminate binding of SYBR Green to any double-stranded DNA, such as primer dimers, these amplification curves should occur in later amplification cycles (>5 Ct different to template samples) and be distinguishable from template amplification curves (Bustin and Nolan 2004). 91

121 Figure 3.26 illustrates template amplification curves for HPRT1 (red arrow), ESR 1 (blue arrow) and fluorescent readings occurring in later cycles by two the non-template controls (black arrows). Figure 3.26 Corbett Rotor-Gene amplification curves for HPRT1 (red arrow), ESR 1 (blue arrow) and fluorescent readings occurring in later cycles by two the non-template controls (black arrows) Statistical Analysis of Results To analyse the relative expression of the estrogen receptor and progesterone receptor genes compared to a reference gene in the study groups, the Relative Expression Software Tool 2005 (REST 2005) was used. REST 2005 was designed by M. Herrmann (Corbett Research) and M. Pfaffl (Technical University Munich) and is available for download at The purpose of REST 2005 is to determine whether there is a significant difference between two groups, while taking into account issues of reaction efficiency and reference gene normalisation. REST 2005 uses randomization techniques in contrast to traditional statistical tests. Randomization tests differ from parametric tests in that there are no requirements that random samples are selected from one or more populations, therefore there is no assumption of normality or homoscedasticity. The hypothesis test P (H1) indicated in the results table of REST 2005 represents the probability of the alternate hypothesis that the difference between sample and control groups is due only to chance. To devise a strong randomisation test, REST 2005 uses the following randomisation scenario: "If any perceived variation between samples and controls is 92

122 due only to chance, then values between the two groups could be randomly exchanged and a greater difference would not be seen than what is seen between the labelled groups." The hypothesis test performs 50,000 random reallocations of samples and controls between the groups, and counts the number of times the relative expression of the randomly assigned group is greater than the sample data (REST 2005 User Manual). REST 2005 was based on a mathematical model proposed by Pfaffl (2001) where crossing point (CP) deviations are used to describe the relationship between gene expression levels (Pfaffl 2001). The analysis method is described by the following formula. In this formula R describes the relative expression ratio between the target and the reference gene, E describes the reaction efficiency of the target and reference genes and ΔCP describes the CP deviation of a control minus a case (migraine) sample. The CP value equals the Ct value used in the Rotor-Gene software (Corbett 2004). In addition to this method, ΔCt values were obtained for each individual sample to give an overview of ESR 1 and PGR mrna in the samples. The HPRT1 expression value was used for normalization in each sample according to the following equation: ΔCt = mean Ct target gene mean Ct reference gene 93

123 ΔCt values were analysed in the groups using the t test which assesses if the means of two groups are statistically different. The computational formula for the t test is as follows. t = ū 1 ū 2 / (var 1 /n 1 + var 2 /n 2 ) Where ū = mean, var = variance in group, n = number in group This analysis was undertaken using SPSS version The aim of this research was to determine if there was a significant difference in ESR 1 and PGR gene expression in migraineurs versus non-migraineurs, and to determine if specific genotypes influenced gene expression. If statistically significant, relative expression levels would be calculated by first determining the ΔΔCt value using the following formula. ΔΔCt = ΔCt controls ΔCt migraine Fold difference would then be determined as E ΔΔCt (where E = amplification efficiency) Thus, groupwise camparisons in expression levels were undertaken in the migraine versus control group, and genotype specific groups. 3.5 Summary of Research Plan In summary, this research investigated variants in the hormone receptor genes ESR 1 and PGR as candidate genes in migraine susceptibility by cross-sectional association analysis in two large independent cohorts. To determine a possible functional role of variations in these genes, gene expression studies were also undertaken by RT-PCR in a group of migraineurs compared to control individuals. This research also investigated variants in two genes involved in vascular function, MTHFR and MTRR, by crosssectional association analysis in a large matched cohort. Finally, to address the issue of 94

124 multiple and potentially interacting susceptibility gene variants playing a role in migraine, genetic risk profiling studies were undertaken in an effort to elucidate the complex role of genetic factors in migraine. 95

125 Chapter 4 Association Studies of Hormone Receptor Genes 96

126 4.1 Migraine and Hormones There is considerable evidence to indicate a role for the female sex hormones in migraine (MacGregor et al. 1990; MacGregor 1996; Silberstein and Merriam 2000; Couturier et al. 2003). There is no gender difference in migraine occurrence prior to puberty. However migraine develops in 3 times as many adult women than men (Silberstein and Merriam 2000). In many women, migraine worsens around the time of menstruation, and may cease altogether after menopause or during pregnancy (MacGregor et al. 1990). The exact mechanism of the role of hormones in migraine has not yet been established. Estrogen withdrawal has been suggested, however treatment involving stabilization of estrogen levels has not been effective in all cases (MacGregor et al. 1990). Epstein et al. (1975) has suggested a role of hormonal variation in all women with migraine, but also a role for factors additional to the hormonal environment (Epstein et al. 1975). Sex steroids are produced in the ovaries and adrenal glands. Steroid hormones are able to passively diffuse through the blood brain barrier with studies showing similar levels in the brain to circulating levels (Martin and Behbehani 2006). The human brain produces several enzymes necessary for the production of steroids such as cytochrome P450 SCC, aromatase, 5alpha-reductase, 3alpha-hydoxysteroid dehydrogenase, 17 betahydroxysteriod dehydrogenase, and various other steroidogenic enzymes necessary for the production of estrogens, progesterone and androgens from cholesterol (Stoffel- Wagner 2003). The brain also contains additional steroid metabolizing enzymes, including sulfotransferases and sulfohydrolases, which convert classic steroid hormones to a variety of neuroactive compounds. Therefore it is understood that the brain can synthesize neurosteroids, which may act via their cognate receptor or other receptor systems (Compagnone and Mellon 2000) and can have significant effects on neurotransmission within the brain. More specifically steroid hormones may have a considerable effect on serotonergic, noradrenergic, glutamatergic, GABAergic and opiatergic systems (Martin and Behbehani 2006). Estrogen appears to play a role in serotonin synthesis and uptake (Pecins-Thompson et al. 1996; Pecins-Thompson et al. 1998), while evidence exists for 97

127 both progesterone and estrogen effects on serotonin degradation (Smith et al. 2004). Estrogen treatment has been shown to upregulate gene expression of tyrosine hydroxylase (involved in the production of noradrenaline) in animal studies (Herbison et al. 2000; Pau et al. 2000) and both estrogen and progesterone may affect certain subtypes of adrenoreceptors (Karkanias and Etgen 1993; Karkanias et al. 1995; Karkanias et al. 1996). Glutamic acid is the major excitatory neurotransmitter in the brain while GABA is the major inhibitory neurotransmitter in the brain. Several studies have shown the effect of estrogen alone and the addition of progesterone on the glutamatergic system (Martin and Behbehani 2006). GABAergic neurons are strongly modulated by ovarian hormones with studies showing an effect of estrogen and progesterone and its metabolites on GABA receptors (Kelly et al. 1992; Twyman and Macdonald 1992; Shughrue and Merchenthaler 2000) as well as cortical GABA levels (Epperson et al. 2002). It is evident that the cerebral blood vessels are targets for steroid hormones as they express specific receptors and metabolic enzymes for gonadel steroids (Krause et al. 2006). Martin and Behbehani (2006) reviewed the potential role of steroid hormones in structures/pathways that could increase or decrease the frequency, severity or duration of migraine headache and have shown that ovarian hormones could potentially affect numerous loci within the trigeminal vascular pain pathways, the autonomic nervous system, the brainstem, and the meningeal artery. Figure 4.1, from Martin and Behbehani (2006), summarises the potential sites of action of ovarian hormones in the CNS. 98

128 Figure 4.1 Potential sites of action of Ovarian Hormones on the CNS (Martin and Behbehani 2006) The Estrogen Receptor Gene as a Migraine Candidate Gene Estrogen is a key regulator of growth, differentiation and function in a wide range of tissues including the male and female reproductive systems, cardiovascular and skeletal systems, as well as the CNS. The predominant biological effects of estrogens are mediated through estrogen receptors, which are expressed in a wide range of target tissues (Hall et al. 2001). Estrogen receptors (ESRs) are classically known as ligand activated transcription factors binding with high affinity to estrogen responsive genes when activated by hormone (Silberstein and Merriam 2000). As well as hormonemediated activation, ESR function can also be modulated by extracellular signals in the absence of estrogen (Cenni and Picard 1999; Hall et al. 2001). Non-genomic mechanisms of ESR signalling are also well documented. Estrogens have been 99

129 demonstrated to act swiftly at the plasma membrane via non-genomic mechanisms through membrane receptors (Kelly and Levin 2001). Estrogen has recently been shown to activate the Mitogen-Activated Protein Kinase (MAPK) signalling pathway through ESR, resulting in rapid effects on the cardiovascular system. (Mendelsohn 2000; Russell et al. 2000; Prorock et al. 2003). Other non-genomic signalling mechanisms include coupling to G proteins and generation of second messengers. Estrogens can induce Ca 2+ mobilization, and activate several kinases including protein kinase C, and phosphatidylinositol-3-oh kinase (Simoncini et al. 2000; Kelly and Levin 2001). Figure 4.2 illustrates genomic and non-genomic mechanisms of estrogen (McEwen 2001). Figure 4.2 Classical and Nonclassic genomic mechanisms of estrogen. In the classical genomic mechanism: (1) estrogen binds to an intracellular estrogen receptor (ER), (2) ER dimerizes with another ER, and (3) ER complex translocates to the nucleus and then binds to the estrogen response element (ERE) leading to transcription. In the nonclassical system: (1) estrogen binds to a membrane bound ER, (2) second messenger systems are activated, and (3) molecules from second messenger systems bind to DNA regulatory regions such as camp response element (CRE) and serum response element (SRE) regions to activate transcription. PKA, protein kinase A; PKC, protein kinase C; MAPK/ERK, mitogen-activated protein kinase/extracellular signal-regulated kinase, CREB, camp response element binding protein (McEwen 2001). 100

130 There are two known receptors for estrogen, ESR 1 (ESR alpha) and ESR 2 (ESR beta). ESR 1 was first cloned in 1986 (Green et al. 1986), and believed to mediate all of the physiological effects of estrogen until 1995, when ESR 2 was cloned from rat prostate (Kuiper et al. 1996). While less is known of ESR 2, ESR 1 has been shown to be expressed in tissues that may be related to migraine pathogenesis such as smooth muscle and endothelial cells of cerebral vessels (Stirone et al. 2003). ESR 1 is located on chromosome 6q25.1. It has 8 exons (Iwase et al. 1996), and is over 295 kilobases in size (Ensembl). ESR 1 is expressed in various human brain regions including the hypothalamus, limbic system, hippocampus, cortices of the temporal lobe and the brainstem (Osterlund et al. 2000). It is expressed in serotonin neurons of some species (Bethea et al. 2002). In addition to alternative splicing mechanisms, different promoters are used to regulate ESR 1 in distinct neuronal populations (Osterlund et al. 2000). Along with its role in target gene transcription, ligand activated ESR 1 has rapid effects on neuronal excitability via second messenger systems, resulting in a range of cellular effects including changes in Ca 2+ currents and activation of endothelial nitric oxide synthase (Cenni and Picard 1999; Chen et al. 1999; Luconi et al. 2002). In fact, it has been hypothesized that ESR 1 may be fundamental for estrogen modulation of cerebral vascular reactivity (Dan et al. 2003). As changes in neuronal excitability and vasoreactivity have been implicated in migraine pathogenesis, genetic variation in the ESR 1 may impact on gene expression or function, in turn influencing migraine susceptibility. This study considered the hypothesis that ESR 1 variation may play a role in migraine susceptibility, and investigated this proposition by genetic association analysis initially, with follow up of implicated variants by mutation screening and gene expression analysis The Progesterone Receptor as a Migraine Candidate Gene The physiological effects of progestins (progesterone and its metabolites such as 5αdihydroprogesterone and allopregnanolone) are mediated primarily by intracellular progesterone receptors. There are two known isoforms of progesterone receptors, PGR-A and PGR-B. Their only difference is an additional 164 amino acids at the N 101

131 terminal of PGR-B and they are both products of PGR. Similar to the estrogen receptor, the progesterone receptor (PGR) plays a complex role in the central nervous system. PGR is found in various regions of the human brain including serotonin neurons (Bethea et al. 2002). In fact, PGRs have been localized in many of the same brain areas as estrogen receptors, including the amygdala, hippocampus, cortex, basal forebrain, cerebellum, locus coeruleus, midbrain rafe nuclei, glial cells and central grey matter (Martin and Behbehani 2006; Pluchino et al. 2006) and there is considerable biological evidence for cross-talk between the estrogen and progestin hormone receptor signaling pathways (Kraus et al. 1995). Similar to estrogen receptors, progesterone receptors can undergo ligand-independent activation and are involved in various intracellular signalling pathways (Cenni and Picard 1999). The human PGR gene is located on chromosome 11q22. The PROGINS polymorphic Alu insertion is a 306 base pair insertion that occurs within intron 7 of PGR in some individuals (Rowe et al. 1995). In exon 4 there is a G to T substitution, causing a Valine to Leucine change in the hinge region of the receptor. This substitution, along with a synonymous C to T substitution in exon 5 is closely linked to the Alu insertion. This complex of PGR gene polymorphisms is designated PROGINS (Runnebaum et al. 2001; Tong et al. 2001). The PROGINS polymorphism has been suggested to have a deleterious effect on progesterone receptor expression, through recombination or missplicing (Rowe et al. 1995; Donaldson et al. 2002). However functional characterization of the PGR protein with PROGINS revealed a greater transcriptional activity compared with the wild-type receptor although hormone binding and hormone dissociation rates were similar in both receptor proteins (Agoulnik et al. 2004). Due to the complex role of progesterone and PGR in the CNS, this study examined the PROGINS variant for a possible association with migraine. 102

132 4.2 Association Analysis of Estrogen Receptor 1 Regions In this study, three polymorphic markers spanning ESR 1 were analysed for a potential role in migraine susceptibility. Figure 4.3 shows a representation of the ESR 1 gene showing a cdna sequence contains 6322 nucleotides encoding a 66Kda protein with 6 conserved domains. The polymorphic markers under analysis are indicated. C325G G594A PvuII Figure 4.3. Representation of the structure of the estrogen receptor gene and cdna region showing transactivation domain (A/B) which contains a ligand-independent transcriptional activation function, DNA binding domain (C), hinge region (D), and ligand binding domain (E) containing a hormone-dependent transcriptional activation function. The function of the F domain is currently unknown. The polymorphic markers under analysis are indicated. Adapted from (Luconi et al. 2002) Estrogen receptor 1 PvuII variant The Pvu II C/T SNP (rs ) is in intron 1 of ESR 1. The Pvu II allele has been associated with variation in estradiol levels in post menopausal women and with an increased risk of stroke in men (Shearman et al. 2005). Interestingly both estrogen withdrawal and high estrogen concentrations have been implicated in migraine susceptibility in women (MacGregor 2004), and there is evidence for an increased risk of stroke in MA sufferers (Diener and Kurth 2005). 103

133 Genotyping For the ESR 1 Pvu II marker primers used were those previously described by Lai et al. 2002, and were as follows: ESR 1 Pvu II F 5 CTGCCACCCTATCTGTATCTTTTCCTATTCTCC 3 ESR 1 Pvu II R 5 TCTTTCTCTGCCACCCTGGCGTCGATTATCTGA 3 PCR resulted in a 239bp fragment (Lai et al. 2002). The PCR protocol is described in Table 4.1 Table 4.1 ESR 1 Pvu II PCR protocol Reagent Final One Reaction Concentration 5mM dntps 0.2mM 0.8μL (New England Biolabs) 10x PCR Buffer* 1x 2μL (Applied Biosystems) 25mM MgCl 2 2mM 1.6μL (Roche) 5μM Forward Primer 0.2μM 0.8μL 5μM Reverse Primer 0.2μM 0.8μL Taq Polymerase 5U/μL 1 Unit 0.2μL (New England Biolabs) Sterile Water 8.8 μl DNA 20ng/μL 40ng 2μL TOTAL 20μL *PCR Buffer contents:- 500mM KCl, 100mM Tris-HCl, ph 9.0, 1.0% Triton X-100 Thermocycler conditions were 94 C for 2 minutes, 35 cycles of 94 C for 30 seconds, 55 C for 1 minute and 72 C for 1 minute, with a final step of 72 C for 5 minutes. 104

134 The C allele introduces a restriction site for the Pvu II enzyme, resulting in fragments of 140 and 99 base pairs. Following amplification, 10 µl of product was digested with Pvu II overnight at 37 C. After digestion, the product was loaded into a 2% highresolution agarose gel stained with ethidium bromide and electrophoresed at 90V for 30 minutes. An undigested sample indicated presence of the T allele. An electrophoretogram of the digested PCR product illustrating all genotypes appears in Figure 4.4. Figure 4.4. Agarose gel electrophoretogram of ESR 1 Pvu II genotypes. Lane 1 shows the 100bp ladder. Lanes 2 & 5 show 239bp fragments representing TT genotypes. Lanes 3 & 6 show 99 and faint 39 bp fragments representing CC genotypes. Lane 4 shows 239, 99 and faint 39 bp fragments representing TC genotypes Results of PvuII variant analysis In total 231 samples in the migraine group and 202 samples in the control group of MAP 1 provided satisfactory results for genotyping. Genotype data and allele frequencies were compared between the migraine case and control groups using standard chi-square analysis. Due to multiple testing, the Bonferroni correction for 5 tests was applied, which set the level of significance at 0.01 (ie. 0.05/5). Statistical analysis of the Pvu II marker as summarized in Table 4.3, revealed no significant difference between genotyped migraineurs and the matched control group with regard to genotype frequencies (X 2 = 1.94, P = 0.38) and allele frequencies (X 2 = 1.02, P = 0.31). Furthermore, no significant difference was seen when the migraine population was subdivided into MA (genotype frequencies X 2 = 3.53, P = 0.17, allele frequencies X 2 = , P = 0.57) and MO (genotype frequencies X 2 = 1.66, P = 0.44, 105

135 allele frequencies X 2 = 1.52, P = 0.22), although the increased frequency of the CC genotype in MO (30%) compared to MA (20%) may warrant follow-up in a larger study group. There was no statistically significant difference in the migraine and control groups with regard to males (genotype frequencies X 2 = 2.36, P = 0.31, allele frequencies X 2 = 2.36, P = 0.12) and females (genotype frequencies X 2 = 0.65, P = 0.72, allele frequencies X 2 = 0.03, P = 0.86). With regard to male frequencies, it was interesting to note that there was a higher frequency of the CC genotype in male migraineurs (27%) compared to the male control group (19%). While this analysis did not reach statistical significance due to small numbers in the male subgroup, it may also warrant further investigation in a larger study group, particularly in view of the previously reported role of the CC genotype in increased stroke risk in males (Shearman et al. 2005) and the potential relationship between migraine and stroke (Merikangas et al. 1997; Donaghy et al. 2002; Schwaag et al. 2003; Diener and Kurth 2005; Etminan et al. 2005). Allele frequencies did not deviate from Hardy Weinberg Equilibrium in both case and control groups (P = 0.4; P = 0.6) and were similar to previously reported frequencies (Sasaki et al. 2003). Frequency distribution appears in Table 4.2. Table 4.2 Distribution of ESR Intron 1 Pvu II polymorphism frequencies in migraineurs and controls, and the MA, MO, male and female subgroups. Genotypes Alleles CC CT TT n alleles C T Migraine 55 (24%) 122 (53%) 54 (23%) (50%) 230 (50%) MA 29 (20%) 84 (58%) 32 (22%) (49%) 148 (51%) MO 22 (30%) 33 (45%) 18 (25%) (53%) 69 (47%) MA & MO (50%) 13 (50%) Male 17 (27%) 31 (48%) 16 (25%) (51%) 63 (49%) Female 38 (23%) 91 (54%) 38 (23%) (50%) 167 (50%) Control 46 (23%) 97 (48%) 59 (29%) (47%) 215 (53%) Male 12 (19%) 27 (44%) 23 (37%) (41%) 73 (59%) Female 34 (24%) 70 (50%) 36 (26%) (49%) 142 (51%) 106

136 Table 4.3. Results of chi-squared analysis of the ESR Intron 1 Pvu II polymorphism in migraineurs and controls, and the MA, MO, Male and female subgroups. Analysis Genotype frequencies Allele Frequencies Case vs Control X 2 = 1.94, P = 0.38 X 2 = 1.02, P = 0.31 MA Case vs Control X 2 = 3.53, P = 0.17 X 2 = , P = 0.57 MO Case vs Control X 2 = 1.66, P = 0.44 X 2 = 1.52, P = 0.22 Male Case vs Control X 2 = 2.36, P = 0.31 X 2 = 2.36, P = 0.12 Female Case vs Control X 2 =0.65, P = 0.72 X 2 = 0.03, P = 0.86 Hardy Weinberg Equilibrium Case, P = 0.4; Cont P = Estrogen receptor 1 C325G Variant The C325G synonymous SNP (rs ) is in exon 4 of ESR 1, located in the hormone binding region. This allele has been reported to play a role in calcium metabolism (Hoshino et al. 2000) and susceptibility to breast cancer (Vasconcelos et al. 2002; Hsiao et al. 2004) a disease in which hormones play a role Genotyping The ESR 1 C325G marker is a HinfI restriction fragment length polymorphism (RFLP) (Curran et al. 2001). Primers used were those previously described (Iwase et al. 1996) with minor modifications, and were as follows: ESR 1 C325G F 5 AGC CCG CTC ATG ATC AAA CG 3 ESR 1 C325G R 5 GGA TCA TAC TCG GAA TAG AGA AT 3 They resulting in a 120 base pair fragment following PCR. The 20 µl PCR reaction mix is described in Table

137 Table 4.4 ESR 1 C325G PCR protocol Reagent Final One Reaction Concentration 5mM dntps 0.2mM 0.8μL (New England Biolabs) 10x PCR Buffer* 1x 2μL (Applied Biosystems) 25mM MgCl mM 1.8μL (Roche) 5μM Forward Primer 0.3μM 1.2μL 5μM Reverse Primer 0.3μM 1.2μL Taq Polymerase 5U/μL 1 Unit 0.2μL (New England Biolabs) Sterile Water 7.8 μl DNA 20ng/μL 50ng 2.5μL TOTAL 20μL *PCR Buffer contents:- 500mM KCl, 100mM Tris-HCl, ph 9.0, 1.0% Triton X-100 Thermocycler conditions were 94 C for 5 minutes, 30 cycles of 94 C for 30 seconds, 62 C for 1 minute and 72 C for 1 minute, with a final step of 72 C for 5 minutes. The G allele at codon 325 in the ESR 1 gene introduces a restriction site for the HinfI enzyme, resulting in fragments of 99 and 21 base pairs. Following amplification, 10 µl of product was digested with HinfI overnight at 37 C. After digestion, the product was loaded into a 5% ultra high-resolution agarose gel stained with ethidium bromide and electrophoresed at 90V for 60 minutes. An undigested sample indicated presence of the 325C allele. An electrophoretogram of the digested PCR product illustrating all genotypes appears in Figure

138 Figure 4.5. Agarose gel electrophoretogram of ESR 1 exon 4 C325G genotypes. Lanes 1 & 2 show 99 bp fragments representing GG genotypes (NB. the 21bp fragment is not visible). Lanes 3 & 4 show 120bp fragments representing the CC genotypes. Lane 5 shows the 100bp ladder. Lanes 6 & 7 show the 120bp & 99bp fragments representing GC heterozygotes Results of C325G variant analysis In total 231 samples in the migraine group and 249 samples in the control group of MAP 1 provided satisfactory results for genotyping. Genotype data and allele frequencies were compared between the migraine case and control groups using standard chi-square analysis. As summarized in Table 4.6 statistical analysis revealed no significant difference between genotyped migraineurs and the matched control group with regard to genotype frequencies (X 2 = 4.19, P = 0.12) and allele frequencies (X 2 = 0.86, P = 0.36). Furthermore, no significant difference was seen when the migraine population was subdivided into MA (genotype frequencies X 2 = 5.26, P = 0.07, allele frequencies X 2 = 1.67, P = 0.20) and MO (genotype frequencies X 2 = 1.15 P = 0.56, allele frequencies X 2 = 0.02, P = 0.90), and males (genotype frequencies X 2 = 2.54, P = 0.28, allele frequencies X 2 = 0.07, P = 0.79) and females (genotype frequencies X 2 = 3.05, P = 0.22, allele frequencies X 2 = 1.53, P = 0.22). Allele frequencies did not deviate from Hardy Weinberg Equilibrium in both case and control groups (at P = 0.1; P = 0.3) and were similar to frequencies previously reported in an Australian study group (Curran et al. 2001). Frequency distribution appears in Table

139 Table 4.5 Distribution of ESR 1 Exon 4 Codon C325G Polymorphism frequencies in migraineurs and controls, and the MA, MO, male and female subgroups. Genotypes Alleles CC CG GG n alleles C G Migraine 133 (58%) 90 (39%) 8 (3%) (77%) 106 (23%) MA 77 (55%) 59 (42%) 5 (3%) (76%) 69 (24%) MO 47 (62%) 26 (35%) 2 (3%) (80%) 30 (20%) MA & MO (77%) 7 (23%) Male 39 (61%) 24 (37%) 1 (2%) (80%) 26 (20%) Female 94 (56%) 66 (40%) 7 (4%) (76%) 80 (24%) Control 160 (64%) 76 (31%) 13 (5%) (79%) 102 (21%) Male 38 (63%) 18 (30%) 4 (7%) (78%) 26 (22%) Female 122 (64%) 58 (31%) 9 (5%) (80%) 76 (20%) Table 4.6 Results of chi-squared analysis of the ESR 1 Exon 4 Codon C325G polymorphism in migraineurs and controls, and the MA, MO, Male and female subgroups. Analysis Genotype frequencies Allele Frequencies Case vs Control X 2 = 4.19, P =0.12 X 2 = 0.86, P = 0.36 MA Case vs Control X 2 = 5.26, P = 0.07 X 2 = 1.67, P = 0.20 MO Case vs Control X 2 = 1.15 P = 0.56 X 2 = 0.02, P = 0.90 Male Case vs Control X 2 = 2.54, P = 0.28 X 2 = 0.07, P =0.79 Female Case vs Control X 2 = 3.05, P = 0.22 X 2 = 1.53, P = 0.22 Hardy Weinberg Equilibrium Case, P = 0.1; Cont P =

140 4.2.3 Estrogen Receptor 1 G594A Variant This ESR 1 synonymous polymorphism, first described by Roodi et al., (1985), occurs in codon 594 of exon 8 and consists of a guanine to adenine change at nucleotide 2014 (rs ) (Roodi et al. 1995). This polymorphism has previously shown an association with breast cancer, a disease in which complex hormonal influences are considered to play a role (Curran et al. 2001) Genotyping Genotyping for the ESR 1 G594A marker was undertaken by PCR and restriction enzyme digestion. Oligonucleotide primers used were those previously described by Curran et al (2001), and were as follows: ESR 1 G594A F 5 GAG GAG ACG GAC CAA AGC CAC 3 ESR 1 G594A R 5 GCC ATT GGT GTT GGA TGC ATG C3 A 227 base pair fragment resulted following PCR (Curran et al. 2001). The 20 µl PCR reaction mix is described in Table

141 Table 4.7 ESR 1 G594A PCR protocol Reagent Final One Reaction Concentration 5mM dntps 0.2mM 0.8μL (New England Biolabs) 10x PCR Buffer* 1x 2μL (Applied Biosystems) 25mM MgCl mM 3μL (Roche) 5μM Forward Primer 0.25μM 1μL 5μM Reverse Primer 0.25μM 1μL Taq Polymerase 5U/μL 1 Unit 0.2μL (New England Biolabs) Sterile Water 9.5 μl DNA 20ng/μL 50ng 2.5μL TOTAL 20μL *PCR Buffer contents:- 500mM KCl, 100mM Tris-HCl, ph 9.0, 1.0% Triton X-100 Thermocycler conditions were 94 C for 2 minutes 30 seconds, 5 cycles of 94 C for 45 seconds, 69 C for 1minute, and 72 C for 2 minutes, followed by 30 cycles of 94 C for 30 seconds, 67 C for 30 seconds and 72 C for 45 seconds, with a final step of 72 C for 5 minutes. The G allele at codon 594 in the ESR 1 gene introduces a restriction site for the BtgI enzyme, resulting in fragments of 129 and 98 base pairs. Following amplification, 10 µl of product was digested with BtgI overnight at 37 C. After digestion, the product was loaded into a 5% Agarose gel stained with ethidium bromide and electrophoresed at 90V for 60 minutes. An undigested sample indicated presence of the 594A allele. An electrophoretogram of the digested PCR product illustrating all genotypes appears in Figure

142 Figure 4.6. Agarose gel electrophoretogram of ESR 1 gene exon 8 PCR product after digestion with BtgI in 8 migraineur samples. Lane 1 shows the 100 base pair ladder. Lanes 2, 3, 4, 6, 7, and 9 show 129 and 98 base pair fragments of a homozygote for the G allele. Lane 5 shows a 227 base pair fragment of a homozygote for the A allele. Lane 8 shows 227, 129, and 98 base pair fragments of a heterozygote Results of G594A variant analysis In total 224 samples in the migraine group and 224 samples in the control group of MAP 1 provided satisfactory results for genotyping. Statistical analysis revealed a significant difference between genotyped migraineurs and the matched control group with regard to allele frequencies (X 2 = 9.77, P = 0.008) and genotype frequencies (X 2 = 8.56, P = 0.008). Results of comparisons between male case and control groups (allele frequency X 2 = 4.47, P = 0.034; genotype frequency X 2 = 6.16, P = 0.046), and female case and control groups (allele frequency X 2 = 4.63, P = 0.032; genotype frequency X 2 = 5.48, P = 0.064) indicated that no significant gender effect was evident. Furthermore, the association was seen in both subgroups, MA (allele frequency X 2 = 6.17, P = 0.013; genotype frequency X 2 = 7.36, P = 0.025) and MO (allele frequency X 2 = 5.51, P = 0.019; genotype frequency X 2 = 9.85, P = 0.007). Consequently, the significant association seen in the case-control analysis occurred similarly in both males and females, and in the MA and MO subgroups. The frequency distribution is displayed in Table 4.8 and results are displayed in Table

143 Table 4.8 Distribution of ESR 1 Exon 8 Codon G594A polymorphism frequencies in migraineurs and controls, and the MA, MO, male and female subgroups of MAP 1 Genotypes Alleles GG GA AA n G A alleles Migraine 81 (36%) 120 (54%) 23 (10%) (63%) 166 (37%) MA 55 (40%) 66 (47%) 18 (13%) (63%) 102 (37%) MO 26 (31%) 54 (63%) 5 (6%) (62%) 64 (38%) Male 18 (32%) 33 (58%) 6 (10%) (61%) 45 (39%) Female 63 (38%) 87 (52%) 17 (10%) (64%) 121 (36%) Control 112 (50%) 99 (44%) 13 (6%) (72%) 125 (28%) Male 28 (49%) 28 (49%) 1 (2%) (74%) 30 (26%) Female 84 (50%) 71 (43%) 12 (7%) (72%) 95 (28%) Table 4.9 Results of chi-squared analysis of the ESR 1 Exon 8 Codon G594A polymorphism in migraineurs and controls, and the MA, MO, Male and female subgroups of MAP 1. Analysis Genotype frequencies Allele Frequencies Case vs Control X 2 = 9.77, P = X 2 = 8.56, P = MA Case vs Control X 2 = 7.36, P = X 2 = 6.17, P = MO Case vs Control X 2 = 9.85, P = X 2 = 5.51, P = Male Case vs Control X 2 = 6.16, P = X 2 = 4.47, P = Female Case vs Control X 2 = 7.36, P = X 2 = 4.63, P = Hardy Weinberg Equilibrium Cont P = 0.14 Due to past problems with non-replication of positive associations, an additional study was performed on the independent population based cohort referred to as MAP 2 using the same marker. In total 270 samples in the migraine group and 270 samples in the control group of MAP 2 provided satisfactory results for genotyping. Results in this follow-up independent study as illustrated in Table 4.11 also revealed a significant difference between genotyped migraineurs and the matched control group with regard to allele frequencies (X 2 = 19.95, P = 8x10-6 ) and genotype frequencies (X 2 = 20.26, P = 4x10-5 ). 114

144 This significant association occurred in females (allele frequency X 2 = 21.81, P = 3x10-6 ; genotype frequency X 2 = 21.86, P = 2x10-5 ), and in the MA subgroup (allele frequency X 2 = 23.33, P = 1x10-6 ; genotype frequency X 2 = 23.81, P = 7x10-6 ). Although the association did not occur in males (allele frequency X 2 = 0.13, P = 0.717; genotype frequency X 2 = 4.13, P = 0.127) and the MO subgroup (allele frequency X 2 = 0.40, P = 0.529; genotype frequency X 2 = 0.40, P = 0.818), this may be due to small numbers in these subgroups (males n = 36, MO n = 39). Alternatively, estrogen and its receptor may play a lesser role in male migraineurs. The frequency distribution is displayed in Table As greater power can be obtained with a larger data-set, a meta-analysis of MAP 1 and MAP 2 was also performed. The frequency distribution appears in Table As illustrated in Table 4.13, the association has improved with a combined analysis of both sub-sets into one large cohort, showing a significant difference between migraineurs and the matched control group (allele frequency X 2 = 27.58, P = 2x10-7 ; genotype frequency X 2 = 29.5, P = 4x10-7 ), migraine vs control females (allele frequency X 2 = 24.24, P = 9x10-7 ; genotype frequency X 2 = 25.19, P = 3x10-6 ), the MA subgroup vs control (allele frequency X 2 = 27.40, P = 2x10-7 ; genotype frequency X 2 = 28.12, P = 8x10-7 ), and the MO subgroup vs control (allele frequency X 2 = 7.44, P = 0.006; genotype frequency X 2 = 11.29, P = 0.004). The association did not occur in males in this meta-analysis (allele frequency X 2 = 3.57, P = 0.059; genotype frequency X 2 = 4.92, P = 0.085) suggesting that indeed estrogen and its receptor may play a lesser role in male migraineurs. 115

145 Table 4.10 Distribution of ESR 1 Exon 8 Codon G594A polymorphism frequencies in migraineurs and controls, and the MA, MO, male and female subgroups of MAP 2 Genotypes Alleles GG GA AA n alleles G A Migraine 103 (40%) 125 (48%) 32 (12%) (64%) 189 (36%) MA 82 (37%) 110 (50%) 29 (13%) (62%) 168 (38%) MO 21 (54%) 15 (38%) 3 (8%) (73%) 21 (27%) Male 15 (42%) 19 (53%) 2 (5%) (68%) 23 (32%) Female 88 (39%) 106 (47%) 30 (14%) (63%) 166 (37%) Control 152 (58%) 93 (36%) 15 (6%) (76%) 123 (24%) Male 20 (55%) 11 (31%) 5 (14%) (71%) 21 (29%) Female 132 (59%) 82 (37%) 10 (4%) (77%) 102 (23%) Table 4.11 Results of chi-squared analysis of the ESR 1 Exon 8 Codon G594A polymorphism in migraineurs and controls, and the MA, MO, Male and female subgroups of MAP 2. Analysis Genotype frequencies Allele Frequencies Case vs Control X 2 = 20.26, P = 4x10-5 X 2 = P = 8x10-6 MA Case vs Control X 2 = 23.81, P = 7x10-6 X 2 = 23.33, P = 1x10-6 MO Case vs Control X 2 = 0.40, P = X 2 = 0.40, P = Male Case vs Control X 2 = 4.13, P = X 2 = 0.13, P = Female Case vs Control X 2 = 21.86, P = 2x10-5 X 2 = 21.81, P = 3x10-6 Hardy Weinberg Equilibrium Cont P =

146 Table 4.12 Distribution of ESR 1 Exon 8 Codon G594A polymorphism frequencies in a metaanalysis of MAP 1 and MAP 2 GG GA AA n alleles G A Migraine 184 (38%) 245 (51%) 55 (11%) (63%) 355 (37%) MA 137 (38%) 176 (49%) 47 (13%) (63%) 270 (37%) MO 47 (54%) 69 (38%) 8 (8%) (66%) 85 (34%) Male 33 (35%) 52 (56%) 8 (9%) (63%) 68 (37%) Female 151 (39%) 193 (47%) 47 (14%) (63%) 267 (37%) Control 264 (55%) 192 (40%) 28 (5%) (74%) 248 (26%) Male 48 (52%) 39 (42%) 6 (6%) (73%) 51 (27%) Female 216 (59%) 153 (37%) 22 (4%) (75%) 197 (25%) Table 4.13 Results of chi-squared analysis of the ESR 1 Exon 8 Codon G594A polymorphism in in a meta-analysis of MAP 1 and MAP 2 Analysis Genotype frequencies Allele Frequencies Case vs Control X 2 = 29.50, P = 4x10-7 X 2 = P = 2x10-7 MA Case vs Control X 2 = 28.12, P = 8x10-7 X 2 = 27.40, P = 2x10-7 MO Case vs Control X 2 = 11.29, P = X 2 = 7.44, P = Male Case vs Control X 2 = 4.92, P = X 2 = 3.57, P = Female Case vs Control X 2 = 25.19, P = 3x10-6 X 2 = 24.24, P = 9x10-7 Allele frequencies in both study populations did not deviate from Hardy-Weinberg equilibrium (P = 0.14; P = 0.88), and each independent sample cohort showed similar frequencies in the case and control groups. Internal controls using random repeat samples and negative controls were used to confirm genotypes and to exclude the potential for genotyping errors. Odds ratios were calculated based upon the Mantel Haenszel method of combining the datasets (Mantel and Haenszel 1959), comparing the G/G genotypes with the G/A and A/A genotype frequencies together under a dominant model. Results indicated that individuals who carried the 594A allele were 2 times more likely to suffer from migraine (OR = 1.96, 95% CI = ) than those who did not carry this allele. Similarly, odds ratios were calculated on the subgroups comparing G/G genotypes with the G/A and A/A genotype frequencies together. 117

147 Results were as follows: MA subgroup OR = 1.97, (95% CI = ); MO subgroup OR = 1.80 (95% CI = 1.10 to 2.94); males OR = 1.95 (95% CI = 0.95 to 3.98); females OR = 1.96 (95% CI = 1.39 to 2.78). In summary, association analyses were performed in two independent study populations. Results of these studies, which remained significant after a Bonferroni correction for multiple testing (α = 0.01) was applied, provide evidence for association of the Estrogen Receptor 1 G594A polymorphism with migraine susceptibility. 4.3 Linkage Disequilibrium Analysis of Estrogen Receptor 1 regions As this analysis of ESR 1 variants involved the study of three polymorphisms spanning the gene, with one marker at the 3 end of the gene showing an association with migraine, linkage disequilibrium analysis was undertaken to determine the extent of pairwise linkage disequilibrium between the three markers. Linkage disequilibrium between the ESR 1 intron 1, exon 4 and exon 8 polymorphisms was analysed using the EH and 2LD programs (Terwilliger and Ott 1993; Zhao 2004). LD results are presented as D and P values. Results indicated that here was no evidence for pairwise linkage disequilibrium between the exon 4 and exon 8 markers and similarly between the intron 1 and exon 8 markers. There was however, evidence for linkage disequilibrium between the intron 1 and exon 4 markers (D = 0.268, P = ). Table 4.14 shows D and P values generated by the LD analysis as well as the physical distance between the markers. The distance calculations were performed using information on genomic location of the relevant SNP provided by Ensembl v.34, Oct 2005 ( This information suggests that the migraine susceptibility haplotype may be specific to the 3 region of the ESR 1 gene. 118

148 Table 4.14 Linkage disequilibrium D values (upper right hand side) and distance in bases between markers (lower left hand side). Marker PvuII C325G G594A PvuII - D =0.268 P= D =0.016 P=0.52 C325G bases - D =0.060 P=0.71 G594A bases bases Association Analysis of the Progesterone Receptor PROGINS variant Genotyping For the PGR gene PROGINS variant, primers used were those previously described (Lancaster et al. 1998) and were as follows: PGR PROGINS F 5 GGC AGA AAG CAA AAT AAA AAG A 3 PGR PROGINS R 5 AAA GTA TTT TCT TGC TAA ATG TC 3 The 20 µl PCR reaction mix is described in Table

149 Table 4.15 PGR PROGINS PCR protocol Reagent Final One Reaction Concentration 5mM dntps 0.2mM 0.8μL (New England Biolabs) 10x PCR Buffer* 1x 1μL (Applied Biosystems) 25mM MgCl 2 1.5mM 1.2μL (Roche) 5μM Forward Primer 0.25μM 1μL 5μM Reverse Primer 0.25μM 1μL Taq Polymerase 5U/μL 1 Unit 0.2μL (New England Biolabs) Sterile Water 13.3μL DNA 20ng/μL 30ng 1.5μL TOTAL 20μL *PCR Buffer contents:- 500mM KCl, 100mM Tris-HCl, ph 9.0, 1.0% Triton X-100 Thermocycler conditions were 94 C for 4 minutes, 30 cycles of 94 C for 30 seconds, 51 C for 30 seconds and 72 C for 45 seconds, with a final step of 72 C for 2 minutes. The PCR product was loaded on a 2% agarose gel using a 100 base pair ladder for comparison. A fragment size of 173 base pairs indicated an allele that did not contain the PROGINS insertion. A fragment size of 479 base pairs indicated an allele that contained the PROGINS insertion. A heterozygote had both sized fragments. A gel electrophoretogram of 6 migraineur samples and a negative control appears in Figure

150 Figure 4.7. Agarose gel electrophoretogram of PGR gene PCR product in 6 migraineur samples. Lane 1 shows the 100 base pair ladder. Lanes 3 and 6 show the 173 base pair fragment of a homozygote for no PROGINS insert. Lanes 4 and 7 show the 479 base pair fragment of a homozygote for the PROGINS insert. Lanes 2 and 5 show the 173 and 479 base pair fragments of a heterozygote. Lane 8 shows a negative control Results of PGR PROGINS analysis In total 232 samples in the migraine group and 216 samples in the control group of MAP 1 provided satisfactory results for genotyping. Genotype data and allele frequencies were compared between the migraine case and control groups using standard chi-square analysis. As illustrated in Table 4.17, results showed that the PROGINS allele was overrepresented in the migraine group compared to healthy controls (genotype frequencies X 2 = 6.50 P = 0.04, allele frequencies X 2 = 5.65, P = 0.02). Results of the subgroup analysis showed a significant difference in the MO (genotype frequencies X 2 = P = 0.001, allele frequencies X 2 = 7.06, P = 0.008) and the female subgroups (genotype frequencies X 2 = P = 0.005, allele frequencies X 2 = 8.1, P = 0.004), but not the MA (genotype frequencies X 2 = 2.25 P = 0.33, allele frequencies X 2 = 2.47, P = 0.12) and male subgroups (genotype frequencies X 2 = 0.41 P = 0.82, allele frequencies X 2 = 0.27, P = 0.60). Frequency distribution appears in Table

151 Table 4.16 Distribution of PGR PROGINS Polymorphism frequencies in migraineurs and controls, and the MA, MO, male and female subgroups of MAP 1. Genotypes Alleles PROGINS -/- PROGINS -/+ PROGINS +/+ n alleles PROGINS - PROGINS + Migraine 173 (75%) 55 (23%) 4 (2%) (86%) 63 (14%) MA 113 (80%) 27 (19%) 4 (3%) (88%) 35 (12%) MO 60 (68%) 28 (32%) 0 (0%) (84%) 28 (16%) Male 43 (64%) 22 (33%) 2 (3%) (81%) 26 (19%) Female 130 (79%) 33 (20%) 2 (1%) (92%) 38 (8%) Control 182 (84%) 31 (15%) 3 (1%) (91%) 37 (9%) Male 44 (68%) 20 (31%) 1 (1%) (83%) 22 (17%) Female 138 (91%) 11 (7%) 2 (2%) (95%) 15 (5%) Table 4.17 Results of chi-squared analysis of the PGR PROGINS polymorphism in migraineurs and controls, and the MA, MO, Male and female subgroups of MAP 1. Analysis Genotype frequencies Allele Frequencies Case vs Control X 2 = 6.50 P = 0.04 X 2 = 5.65, P = 0.02 MA Case vs Control X 2 = 2.25 P = 0.33 X 2 = 2.47, P = 0.12 MO Case vs Control X 2 = 0.40, P = X 2 = 0.40, P = Male Case vs Control X 2 = 0.41, P = 0.82 X 2 = 0.27, P = 0.60 Female Case vs Control X 2 = 10.64, P = X 2 = 8.1, P = Hardy Weinberg Equilibrium Cont P = 0.22 Due to past problems with non-replication of positive associations, an additional study was performed on the independent population based cohort referred to as Map 2 using 122

152 the same marker. As detailed in Table 4.19 results showed a significant difference in genotype (X 2 = 7.92, P = 0.019) and allele frequencies (X 2 = 8.78, P = 0.003) in the total group analysis, and in the MA subgroup (genotype frequencies X 2 = 7.28 P = 0.026, allele frequencies X 2 = 7.91, P = 0.005). Similar results were seen in both male (genotype frequencies X 2 = 5.27 P = 0.07, allele frequencies X 2 = 5.87, P = 0.02) and female subgroups (genotype frequencies X 2 = 4.81 P = 0.09, allele frequencies X 2 = 4.31, P = 0.04), although they did not reach statistical significance. Analysis of the MO subgroup did not reach statistical significance (genotype frequencies X 2 = 3.53 P = 0.17, allele frequencies X 2 = 3.11, P = 0.08). Frequency distribution appears in Table As greater power can be obtained with a larger data-set, a meta-analysis of MAP 1 and MAP 2 was also performed. The frequency distribution appears in Table As illustrated in Table 4.21, the association has improved with a combined analysis of both sub-sets into one large cohort, showing a significant difference between migraineurs and the matched control group (allele frequency X 2 = 14.37, P = ; genotype frequency X 2 = 14.01, P = ), migraine vs control females (allele frequency X 2 = 11.91, P = ; genotype frequency X 2 = 14.42, P = ), the MA subgroup vs control (allele frequency X 2 = 9.86, P = ; genotype frequency X 2 = 8.93, P = 0.013), and the MO subgroup vs control (allele frequency X 2 = 12.42, P = ; genotype frequency X 2 = 14.86, P = ). Similar to ESR 1 the association did not occur in males suggesting that hormonal influences may indeed play a lesser role in male migraineurs. 123

153 Table 4.18 Distribution of PGR PROGINS Polymorphism frequencies in migraineurs and controls, and the MA, MO, male and female subgroups of MAP 2. Genotypes Alleles PROGINS -/- PROGINS -/+ PROGINS +/+ n alleles PROGINS - PROGINS + Migraine 215(78%) 54 (19%) 8 (3%) (87%) 70 (13%) MA 176 (77%) 45 (20%) 6 (3%) (87%) 57 (13%) MO 39 (78%) 9 (18%) 2 (4%) (87%) 13 (13%) Male 27 (69%) 8 (20%) 4 (11%) (79%) 16 (21%) Female 188 (79%) 46(19%) 4 (2%) (89%) 54 (11%) Control 228 (87%) 32 (12%) 3 (1%) (93%) 38 (7%) Male 35 (85%) 6 (15%) (93%) 6 (7%) Female 193 (87%) 26 (12%) 3 (1%) (93%) 32 (7%) Table 4.19 Results of chi-squared analysis of the PGR PROGINS polymorphism in migraineurs and controls, and the MA, MO, Male and female subgroups of MAP 2. Analysis Genotype frequencies Allele Frequencies Case vs Control X 2 = 7.92, P = X 2 = 8.78, P = MA Case vs Control X 2 = 7.28, P = X 2 = 7.91, P = MO Case vs Control X 2 = 3.53, P = 0.17 X 2 = 3.11, P = 0.08 Male Case vs Control X 2 = 5.27, P = 0.07 X 2 = 5.87, P = 0.02 Female Case vs Control X 2 = 4.81, P = 0.09 X 2 = 4.31, P = 0.04 Hardy Weinberg Equilibrium Cont P =

154 Table 4.20 Distribution of PGR PROGINS Polymorphism frequencies in a meta-analysis of MAP 1 and MAP 2. Genotypes Alleles PROGINS -/- PROGINS -/+ PROGINS +/+ n alleles PROGINS - PROGINS + Migraine 388(76%) 109 (22%) 12 (2%) (87%) 133 (13%) MA 289 (78%) 72 (19%) 10 (3%) (88%) 92 (12%) MO 99 (72%) 37 (27%) 2 (1%) (85%) 41 (15%) Male 70 (66%) 30 (28%) 6 (6%) (80%) 42 (20%) Female 318 (79%) 79 (20%) 6 (1%) (89%) 91 (11%) Control 410 (86%) 63 (13%) 6 (1%) (92%) 75 (8%) Male 79 (74%) 26 (25%) 1 (1%) (87%) 28 (13%) Female 331 (89%) 37 (10%) 5 (1%) (94%) 47 (6%) Table 4.21 Results of chi-squared analysis of the PGR PROGINS polymorphism in a meta-analysis of MAP 1 and MAP 2. Analysis Genotype frequencies Allele Frequencies Case vs Control X 2 = 14.01, P = X 2 = 14.37, P = MA Case vs Control X 2 = 8.97, P = X 2 = 9.86, P = MO Case vs Control X 2 = 14.86, P = X 2 = 12.42, P = Male Case vs Control X 2 = 4.40, P = X 2 = 3.35, P = Female Case vs Control X 2 = 14.42, P = X 2 = 11.91, P = Allele frequencies in both study populations did not deviate from Hardy Weinberg Equilibrium at P = 0.22 and P = 0.13 respectively. Published allele frequencies vary somewhat but a recent analysis of this variant in 21 diverse human populations reported an average allele frequency of PROGINS of 11% and a heterozygosity of (Donaldson et al. 2002). 125

155 In order to analyse whether the PROGINS variant exerted a dominant or recessive effect on migraine susceptibility, the effect of the genotype risk groups was investigated (-/+, +/+ versus -/- only) with results indicating that the -/+ and +/+ genotypes was significantly over-represented in the total migraine subgroup of both populations (24%) compared to the total control subgroup (14%) (X 2 = 13.94, P = 2x10-4 ). Odds ratios were calculated using the Mantel Haenszel method of combining the datasets (Mantel and Haenszel 1959), comparing those who carried the PROGINS allele and those who did not under the hypothesis of a dominant model. Results indicated that those who carried the PROGINS allele were 1.8 times more likely to suffer from migraine than those who did not carry this allele (OR = 1.77, 95% CI = ). In summary, association analyses were performed in two independent study populations. Results of these studies, which have remained significant after the Bonferroni correction for multiple testing was applied (α = 0.01), provide evidence for association of the PROGINS insert in PGR with migraine susceptibility. 4.5 Interaction Analysis of Estrogen Receptor 1 and Progesterone Receptor Variants Methodology Due to the positive associations of the PGR PROGINS insert, as well as an association of the ESR 1 G594A polymorphism with migraine in the same study population (Colson et al. 2004; Colson et al. 2005), interaction analysis was performed to determine if possession of both risk genotypes would confer an increased risk of migraine. To determine the magnitude of the increased risk of migraine conferred specifically by the risk alleles from both ESR 1 and PGR, odds ratios were calculated after dichotomising the genotype frequency data into risk (possessing at least 1 copy of 126

156 the risk allele from each gene under a dominant model) and no risk (possessing zero copies of the risk alleles) groups Results of ESR 1 and PGR Interaction Analysis Results showed that 30% of all migraineurs carried at least one copy of the risk allele from both ESR 1 and PGR compared to only 12% of controls (X 2 = 20.53, P = 3x10-5 ). Comparing the total migraine group against controls (MAP1 and MAP2 together), this grouping scheme produced an OR of 3.2 with a 95% CI of Therefore, it appears that the PROGINS allele of PGR acts synergistically with the 594A allele of ESR 1 to increase the risk of migraine. That is, these alleles act in combination to increase the risk of migraine by a factor of 3, which is greater than the independent effects of these genetic variants on disease susceptibility. Table 4.18 shows the distribution of individuals who carried none/both risk alleles. Table 4.22 Distribution of individuals in MAP 1 and MAP 2 who carried none/both risk alleles for PGR PROGINS and ESR 1 594A Migraine Diagnosis No risk alleles At least one risk allele from each gene Migraine 132 (70%) 57 (30%) Control 191 (88%) 26 (12%) X 2 = 20.53, P = 3x

157 4.6 Mutation Screening of the Estrogen Receptor Gene Methodology and Sample Selection As association studies of ESR 1 implicated the G594A locus at exon 8 for a role in migraine susceptibility, and LD analysis showed no evidence for linkage of the exon 8 marker with either the exon 4 or the intron 1 markers, mutation screening was undertaken in selected samples analyzing exon 5-8 regions including ~300 bp of the 3 untranslated region. For this analysis DNA was screened from five individuals who were homozygous for the ESR 1 594A allele. To enable sequencing of ESR 1 exons 5-8 and the 3 untranslated region, primers were needed to flank each exon including splice sites. Primer sequences were chosen according to a research paper published by Iwase et al. 1996, who sequenced all exons of ESR 1 for potential mutations in breast cancer samples (Iwase et al. 1996). The primer sequences were checked for homology and specificity to the genomic region of interest by entering the sequences into Basic Local Alignment Search Tool (BLAST) (available All oligonucleotide primers were obtained from Geneworks (South Australia). All primers were provided in sequencing quality as a lypholysed pellet. They were diluted to a working concentration of 5μM with sterile water. 128

158 4.6.2 Primers Table 4.19 details the primers used to sequence ESR 1 exons 5-8 including the 3 UTR and splice sites. Table 4.23 ESR 1 primers used to sequences exons 5-8 Exon Primer Product 5 F 5 CCAGTAATGAGTCTTTTTCATTTGA bp 5 R 5 CCAATGCACTCTTTTGTTAAGTAAA 3 6 F 5 CCCTTTCATGTCTTGTGGAAG bp 6 R 5 AGTGGGTAGATCGTATCTGGTT 3 7 F 5 GAGCTTCTCTCTCTCACTCTCTC bp 7 R 5 TTATGTCTCTCCTGTAGGAAGC 3 8 F 5 TGGCTCTAAAGTAGTCCTTTC bp 8 R 5 ATCTGAACCGTGTGGGAGC 3 8 and 3 UTR F 5 GTGGAGGAGACGGACCAAAG bp 8 and 3 UTR R 5 AGGCAAAATGTCTACTCTCCAGG Assay Design and Sample Preparation To sequence exons 5, 6, 7 and 8 the 20 µl PCR reaction mix is described in Table

159 Table 4.24 ESR 1 sequencing PCR protocol. Reagent Final One Reaction Concentration 5mM dntps 0.2mM 0.8μL (New England Biolabs) 10x PCR Buffer 1x 1μL (Applied Biosystems) 25mM MgCl 2 0.2mM 1.6μL (Roche) 5μM Forward Primer 0.2μM 0.8μL 5μM Reverse Primer 0.2μM 0.8μL Taq Polymerase 5U/μL 1 Unit 0.2μL (New England Biolabs) Sterile Water 12.3μL DNA 20ng/μL 50ng 2.5μL TOTAL 20μL *PCR Buffer contents:- 500mM KCl, 100mM Tris-HCl, ph 9.0, 1.0% Triton X-100 To amplify exons 5 and 6 the thermocycler conditions were 94 C for 2 minutes, 35 cycles of 94 C for 30 seconds, 56 C for 30 seconds and 72 C for 30 seconds, with a final step of 72 C for 5 minutes. For exon 8, the annealing temperature was 58 C. For the 3 untranslated region of ESR 1, the annealing temperature was 62 C. After PCR, the product was loaded into a 2% agarose gel stained with ethidium bromide and electrophoresed at 90V for 30 minutes to ensure that the PCR was successful and that there was no contamination in the negative control. After successful PCR amplification, the products were cleaned using Exosapit (Amersham Pharmacia, Piscataway USA) which is a mix of Exonuclease 1 and Shrimp Alkaline Phosphatase. Exonuclease 1 removes residual primers while Shrimp Alkaline Phosphatase removes the remaining dntps from the PCR mixture. Briefly, 1 μl of Exosapit was added to 2-9 μl of PCR product (depending on strength of PCR product as visualized on an electrophoreogram) and water was added to a final volume of 10 μl. The reaction mix was incubated at 37ºC for 30 minutes, and then inactivated at 80ºC for 15 minutes. The PCR product was then quantitated using a Nanodrop Spectrophotometer (Nanodrop Technologies, Delaware USA) and prepared for dye 130

160 termination using the BigDye Terminator v3.1 Cycle Sequencing Kit by Applied Biosystems, (Foster City, California USA) by adding 10ng of cleaned DNA to 6.4 pmol of each forward and reverse primer. The samples were then forwarded to the Australian Genome Research Facility (Brisbane) for sequencing on their Applied Biosystems (Foster City, California USA) AB3730xl sequencer. Results were provided in the form of chromatograms which were analysed using the Applied Biosystems Sequence Scanner TM v1.0 software. Each result provided a chromatogram showing peak height and base quality bars. Each forward and reverse sequence was compared to the known sequence published on Ensembl and GenBank (Ensembl; Benson et al. 2006) Sequencing Results Figure 4.8 shows a chromatogram of an analysed data run on the Applied Biosystems AB3730xl sequencer viewed through the Applied Biosystems Sequence Scanner TM v1.0 software for base sequencing of the reverse sequence of exon 6. The base quality bars, as indicated by the red arrow denote high quality base calls for this sample. Figure 4.9 shows the actual sequence reading provided by this sample. As indicated, lavender coloured base calls denote high quality base calls, pale lemon denote medium quality and pink denote low quality. As expected, the first base calls were of poorer quality (Almira et al. 2003). DNA from 5 migraineurs who possessed the ESR 1 594AA genotype was analysed to detect possible coding mutations in the gene. ESR 1 exons 5 8 and splice sites along with ~300 bp of the 3 untranslated were sequenced in all five samples. Each sample produced a chromatogram similar to Figure 4.8 for the forward and reverse sequence of each exon. Detailed analysis was undertaken by comparing the base call results to published sequences. In the case of base calls that were not of high quality, detailed inspection of the chromatogram was undertaken to ensure that genuine base sequence changes were not overlooked. 131

161 Figure 4.8 Chromatogram of base sequencing of the partial reverse sequence of exon 6 Figure 4.9 Actual base sequence calls for sequencing the reverse sequence of exon 6. Results did not reveal any new mutations in the ESR 1 coding regions under analysis. As expected, all samples showed homozygosity for the ESR 1 594A allele. All samples also showed homozygosity for the A allele of rs in exon 5. Additionally all samples showed homozygosity for two other SNPs in exon 8, the A allele for rs and the G allele for rs As all are synonymous SNPs it is unlikely that these markers would play a functional role in migraine susceptibility. Thus, in the five samples analysed in this study, no mutations could be found in the 3 region of ESR

162 4.7 Summary and Discussion of Association Analysis of Hormone Receptor Genes There is significant evidence to indicate that the fluctuating hormones of the ovarian cycle are specific migraine triggers (MacGregor 1997; MacGregor 2004), although the precise role of hormones in the pathogenesis of migraine is yet to be established. To date, studies have focused on the hormonal milieu, in particular estrogen. The currently recognised mechanism of hormonally triggered migraine is estrogen withdrawal, however this theory is based on empirical evidence that attacks can be prevented by artificially stabilizing hormone levels (Massiou and MacGregor 2000). Steroid hormones exert their effects via a cognate receptor. The classic mode of action of steroid hormone receptors is as ligand activated transcription factors, regulating gene expression via interaction with hormone response elements in the promoter region of sensitive genes. Recent research has also demonstrated non-genomic effects of steroid hormone receptors, resulting in rapid effects activated via intracellular second messenger systems (Kelly et al. 2002). Furthermore there is evidence to indicate that steroid hormone receptors can be activated in the absence of hormones (Cenni and Picard 1999). The human ESR 1 is widely expressed in a broad range of tissues including CNS areas such as the hypothalamus, limbic system, hippocampus, cortices of the temporal lobe and the brainstem (Osterlund et al. 2000). Numerous studies have demonstrated the multifunctional role of the ESR, particularly in the CNS. It is understood to play a role in neuroprotection via activation of the MAPK pathway (Mize et al. 2003), as well as in cognition, mood, and memory (McEwen 2002). ESR can be activated by neurotransmitters and growth factors, in particular, dopamine (Power et al. 1991). Estrogens can induce Ca 2+ mobilization, and activate several kinases including protein kinase C, and phosphatidylinositol-3-oh kinase (Simoncini et al. 2000; Kelly and Levin 2001). Estrogen deficiency has been implicated in pathological and degenerative processes in the CNS, while elevated levels have been involved in the development and progression of tumours (Diel 2002). According to animal models, steroid hormones may cause changes in regions involved in the neurovascular headache pathway (Marcus 1995; Martin and Behbehani 2006). For example estradiol can exert behavioural and electrophysiological effects by binding to neuronal membranes, in 133

163 particular the central serotonergic and opioid neurons. This mechanism may lead to a disturbance of pain perception (Silberstein and Merriam 2000). Estrogen changes in the rat have been associated with altered mrna expression in sensory neurons (Sohrabji et al. 1994), while estrogen injections appear to alter the size of the receptive area of the trigeminal mechanoreceptors in rats (Bereiter and Barker 1980). Considering the complex role of estrogen and its cognate receptor in the CNS, this research considered ESR 1 as a suitable candidate in migraine susceptibility. Three polymorphisms spanning the ESR 1 gene were analysed by case-control association analysis to investigate the gene as a candidate in migraine susceptibility. Association analysis is considered useful for investigating complex diseases such as migraine (Xu et al. 1998) and for detecting susceptibility genes of modest effect (Risch 2000). The study tested large carefully matched case-control populations for the ESR 1 intron 1 PvuII polymorphism, and two common synonymous SNPs in exon 4 and exon 8. Results showed no association with migraine in the case-control groups for both the intron 1 Pvu II marker and the exon 4 C325G marker. Notably, a recent report of an association of the exon 4 C325G polymorphism with migraine in women in a large Spanish cohort has been reported (Oterino et al. 2006). Although this analysis did not demonstrate such an association, there was an interesting trend in the female and MA subgroups which may warrant further investigation in a larger study group. Results of this present study showed a positive association of the ESR 1 exon 8 polymorphism with migraine in MAP 1 and an independent follow up population, MAP 2. This positive association was seen equally in all subgroups in the initial study group and in the female and MA subgroups in the follow-up group. Lack of association in males may have been related to the limited number of males in the second population, however a meta-analysis of both cohorts showing no association in the male subgroup combined with the fact that hormonal factors play a different role in the two genders in migraine, sugest that estrogen and its receptor most likely play a lesser role in male migraineurs. Although this marker showed significant differences in genotype and allele frequencies in the two migraine versus control populations, association studies by virtue of their design do not determine whether it is the marker itself or a variant in linkage 134

164 disequilibrium that is responsible for the association. As this particular marker is a non-synonymous SNP, it is more likely that a different marker in the coding region of ESR 1 that is in strong LD with the G594A marker, may be responsible for the association found in migraineurs. In an effort to refine the migraine susceptibility region in ESR 1, pairwise LD analysis was undertaken on the three SNPs analysed in this research. Although there was evidence for linkage disequilibrium between the intron 1 and exon 4 markers (D = 0.268, P = ), results showed no evidence for pairwise linkage disequilibrium between the exon 4 and exon 8 markers and the intron 1 and exon 8 markers, suggesting that a migraine susceptibility locus may lie closer to the 3 end of the gene. In an effort to locate a potential functional variant that may be responsible for the role of ESR 1 in migraine susceptibility, all coding regions from exon 5 to 8 inclusive were sequenced in 5 migraineurs who possessed the 594 AA susceptibility genotype. This analysis did not reveal any coding mutations in these 5 individuals however it should be noted that the possibility of a functional variant existing in other samples should not be ruled out. Time and resources permitting, it may be worthwhile in the future to perform a detailed mutation analysis on all 55 subjects who possessed the 594 AA susceptibility genotype. This analysis should include all exons, splice sites and the promoter region of ESR 1. Although LD generally decays with physical distance, other factors such as population history, recurrent mutations and the gene under analysis may impact on the correlation between markers (Eberle et al. 2006). Therefore, although more likely, a potential mutation may not necessarily lie in the 3 region of ESR 1. Furthermore, a potential mutation may have a minor frequency so that, for example, all those with the mutation possess the susceptibility allele, but not all those with the susceptibility allele possess the mutation. Therefore further analysis would appear worthwhile. Certainly several hypotheses may be proposed concerning a potential role of ESR 1 in migraine. In human studies, serotonin synthesis, degradation and neuronal firing appear to be influenced by estrogen receptor mediated mechanisms. This has been shown in a small clinical study of women with OCP triggered migraine lasting longer than 72 hours (status migrainosus). Women with a history of status migrainosus experienced significantly impaired neuroendocrine responses indicated by a 135

165 derangement of prolactin release and a lack of cortisol response after 5-HT agonist treatment. The neuroendocrine response was restored when women received estrogen treatment (Nappi et al. 2005). The possible link between estrogen and serotonergic signalling has added significance because 5-HT receptors are important therapeutic targets in the acute treatment of migraine. A further possible explanation for the role of ESR 1 in migraine is its influence on calcium channels (Kelly and Levin 2001). Due to the discovery of mutations in the CACNA1A calcium channel gene in familial hemiplegic migraine, a rare subtype of migraine, the role of calcium channels and calcium homeostasis in the pathogenesis of common migraine is possible. Calcium and other ion channels are significant factors in the mechanism of neurotransmitter release and cortical spreading depression (Edvinsson 1999), thus impaired function of calcium channels and calcium homoeostasis could trigger an attack. Furthermore, an altered density of calcium channels could result in excitation of the periaqueductal grey, raphe nuclei or locus coeruleus neurons that are considered to be in the region responsible for initiation of migraine attacks (Edvinsson 1999). L-type Ca 2+ channels are one of the main pathways of intercellular calcium entry in the brain. Johnson et al. (1997) demonstrated in an animal model that the density of cardiac L-type Ca 2+ channels is regulated by the estrogen receptor (Johnson et al. 1997), while Mermelstein et al. (1996) demonstrated that estradiol reduces calcium currents in rat neostriatal neurons via a membrane receptor (Mermelstein et al. 1996). A further consideration is that ESR can mediate changes in vascular tone. Chen et al. (1998) demonstrated that ESR 1 mediates the non-genomic activation of endothelial nitric oxide synthase, causing the rapid dilation of blood vessels. Furthermore, the observed response was evident at concentrations well below those found in normal cycling women (Chen et al. 1999). Obviously these hypotheses would require extensive analysis, as hormone/receptor action is understood to be tissue specific, and may even differ from region to region in similar cell types. Nevertheless they demonstrate potential mechanisms whereby genetic variation in the estrogen receptor gene could have an impact on mechanisms that may be crucial to migraine pathogenesis. 136

166 Of further interest was the significant association of the PGR PROGINS insert with migraine susceptibility. Statistical significance was reached in the first study population of 275 migraineurs compared to healthy controls (MAP 1), and also in the independent follow-up group of 300 migraineurs compared to controls (MAP 2). Furthermore, analysis of genotype risk groups showed that the PROGINS insert allele was significantly over-represented in migraineurs, and those who carried this allele were 1.8 times more likely to suffer migraine. Although sub-group analysis showed dissimilar results in the MA/MO analyses in the two independent populations, small numbers in the MO sub-group would have reduced statistical power, and may have contributed to this anomaly. Under the hypothesis of similar genetic etiology in the migraine subgroups, it would be expected that the PROGINS insert would confer a risk in both subgroups. Meta-analysis of both cohorts indicated that PROGINS indeed appears to play a role in both MA and MO with a significant association seen in both subgroups. The meta-analysis also revealed that, similar to ESR 1, PROGINS does not appear to play a role in male migraineurs. As it has been suggested that the PROGINS variant has an impact on gene expression, detailed functional analyses may clarify how this variant could plays a direct role in migraine susceptibility. A potential role of PGR in migraine may be explained by the complex role of progesterone particularly in the CNS, mediated mainly through its cognate receptor. Similar to ESR, human PGRs have been shown to mediate rapid effects via the activation of Src/Ras/Raf/MAPK signalling pathway (Pluchino et al. 2006). In most target tissues progesterone receptor expression is up-regulated by estrogen and downregulated by progesterone (Bouchard 1999). Progesterone and its derivatives (such as the neuro-active compounds 5α-dihydroprogesterone and allopregnanolone) are synthesized de novo by astrocytes and oligodendrocites starting from cholesterol (Genazzani, Stomati et al. 2000). Interactions exist between progesterone and its metabolites and the γ-aminobutyric acid (GABA) receptor subtype A (Genazzani et al. 2000). Activation of the GABA A receptor, utilized by barbiturates and benzodiazepines, makes the cell membrane permeable to chloride ions and produces strong sedating effects. Complete anaesthesia has been induced by administration of progesterone in mice (Hogskilde et al. 1987) while in progesterone-rich pregnant women, 30 50% less barbiturates (Backstrom et al. 1990) are required for abdominal surgery compared to non-pregnant women (Gruber and Huber 2003). Progesterone is 137

167 able to modulate a GABA A receptor located outside the blood brain barrier suppressing neurogenic and substance P-induced plasma extravasation within the meninges in animal models (Limmroth et al. 1996). Plasma protein extravasation within the meninges is a well known element of migraine pathogenesis. Other interactions between progesterone and neurotransmitter systems include stimulation of dopamine release in striatal tissue (Petitclerc et al. 1995), and inhibition of opioid receptor binding and activation (Sinchak and Micevych 2001). Whether these effects are mediated by PGRs via signalling pathways or cross-talk, or by direct binding of progestins to other receptor sites is currently unclear. PGR knock-out mice still exhibit anaesthesia after treatment with progesterone, most likely due to conversion to allopregnanolone, which is an allosteric positive modulator of GABA A receptors (Reddy and Apanites 2005). However another study has suggested a role for PGR in this mechanism by showing that at physiological concentrations, neuroactive progestins regulate neuronal function via their direct effects on PGR regulated gene expression as well as through transmitter-gated ion channels (Rupprecht et al. 1993). Thus is appears that neuroactive progestins can regulate gene expression via PGR mediated mechanisms. Cross-talk between PGR and ESR has certainly been shown. As an example, progesterone opposes the action of estrogen on the vascular endothelium and causes vasoconstriction probably through decreased release of nitric oxide (Miller and Vanhoutte 1991). This action is considered to be due to cross-talk between ESR and PGR. While ESR 1 turns on target gene expression and functions as a regulator of ligand-activated transcription in estrogen responsive tissues (Cooke et al. 1998), progesterone attenuates cell sensitivity to estrogen by decreasing ESR 1 levels (Clarke and Sutherland 1990). It has been shown that nuclear ESR 1 levels decrease in the rat uterus as serum progesterone levels increase (Okulicz 1989), and that progesterone decreases sensitivity of cells to estrogens by inhibiting ESR 1 mediated transactivation via direct interactions of ligand bound PGR and ESR 1 (Kraus et al. 1995). Repression of ESR mediated transcriptional activity by liganded PGR at gene level is considered an important process in the interplay between the estrogen and progestin signaling systems (Kraus et al. 1995). 138

168 Activation of PGR can also occur by the action of cyclic nucleotides that increase intracellular kinase activity, as well as extracellular compounds that interact with cell membrane receptors and stimulate intracellular phosphorylation pathways (Denner et al. 1990), including growth factors and the neurotransmitter dopamine (Pluchino et al. 2006), thus a potential role of PGR in migraine may not be solely due to progesterone mediated mechanisms. 139

169 Chapter 5 Expression Analysis of Hormone Receptor Genes 140

170 5.1 Introduction In order to further investigate the potential functional role of ESR 1 and PGR in migraine, this study investigated the expression of the ESR 1 and PGR genes in a group of female migraineurs compared to a group of healthy control females. In addition, this study investigated if susceptibility genotypes for each gene variant implicated in migraine susceptibility in association studies (as outlined in Chapter 4) impacted on gene expression. The underlying hypothesis of this analysis was that altered expression of ESR 1 and/or PGR in migraineurs compared to controls would implicate the gene for a potential functional role in migraine susceptibility. 5.2 Results of RNA Quality and Integrity Check RNA samples were taken from 6 migraineurs with aura and 6 control females. The methodology for RNA preparation and gene expression analysis is outlined in Chapter 3. All RNA samples were analysed for RNA quality and integrity on an Agilent (Palo Alto, CA) 2100 Bioanalyzer. The output was a scan of mass vs. size as seen in Figure 3.14 and an electrophoretogram as seen in Figure Figure 5.1 shows the electrophoretogram generated by the Agilent 2100 Bioanalyzer for all RNA samples extracted in ths study. All samples except three produced crisp 28S and 18S rrna bands (of approximately 5 kb and 2 kb in size) on the electrophoretogram. Similarly, all but three samples produced strong peaks with little background noise on the scans as illustrated in Figure 5.2 indicative of intact RNA. The sample that produced very poor results indicating heavily degraded RNA was discarded. The samples that indicated some RNA degradation (samples 7 and 8) but still showed 28S and 18S bands were included in the study. The reason for inclusion was two-fold. Samples were reverse transcribed immediately but not analysed immediately and transportation and thawing may have resulted in some degradation before analysis. Secondly, samples met 3 out of the five criteria to be of suitable quality for gene expression analysis as outlined by Haller et al The ratio of 28S to 18S peaks is equal to or greater than The area under the 28S and 18S peaks combined is equal to or greater than 30% of the total area. 141

171 3. The widths of the 18S and 28S peaks are each less than or equal to 4 seconds. 4. There are no distinct peaks between the 28S and 18S peaks or between the 18S peak and the lower marker peak. 5. The area under the degradation peaks is less than the combined areas of the 28S and 18S peaks (Haller et al. 2006). Figure 5.1. Agilent 2100 bioanalyzer electrophoretogram for extracted RNA used in this study, indicating intact RNA in all samples except for sample 9. Figure 5.2a. Agilent 2100 bioanalyzer scan of mass vs. size for extracted RNA, indicating intact RNA for sample 1. Similar scans were produced for samples 2-6, and

172 5.3 Results of Genomic Contamination Check All samples were checked for genomic contamination with the assay described in using the Calpain (CAPN1) gene. Primers were designed to amplify a 116 bp fragment in an exonic region of CAPN1 in cdna and a 261 bp intron spanning fragment in genomic DNA. In the top row, lanes 2-7 in the gel electrophoretogram in Figure 5.3 below, the 6 migraineur samples show 116 bp fragments denoting no genomic DNA contamination. Similarly, in the bottom row, lanes 5-10 the 6 non-migraineur samples show 116 bp fragments denoting no genomic DNA contamination. Non-template controls are in the top row lane 9 and the bottom row lane 12. Genomic DNA samples are in the top row lanes and the bottom row lanes As expected, these samples show bands at 261 bp. Therefore, it was concluded that all cdna samples were free from contaminating genomic DNA. Figure 5.3 Electrophoretogram showing PCR products using CAPN1 primers of cdna in row 1, lanes 2-7 and row 2 lanes 5-10 compared to gdna in row 1 lanes and row 2 lanes

173 5.4 Reference Gene Selection Using the method described by Vandesomple (2002), 8 reference genes were tested in 3 migraineur samples and 3 control samples to determine the most stable reference gene for normalisation of these studies of relative gene expression. Raw expression levels were input into the genorm applet and genes with the highest M value were eliminated in a stepwise manner as described by the authors (Vandesompele et al. 2002), until the two most stable reference genes remained. An M value of <0.7 was considered satisfactory and an indication that the most variable reference genes had been eliminated (Vandesompele et al. 2002). Figure 5.1 shows the average expression stability of the control genes under analysis with the least stable gene at the left and ranked according to increasing expression stability. This figure shows that both HPRT1 and TBP showed stable expression levels. Figures 5.5a 5.5g show the stepwise elimination of unstable reference genes until 2 genes remained with similar expression stability. From these two genes (HPRT1 and TBP), HPRT1 was chosen as the reference gene to be used in the experiments. Figure 5.4 Genorm average expression stability of the control genes with the least stable gene at the left and ranked according to increasing expression stability. 144

174 Figure 5.5a Genorm stepwise elimination of unstable reference genes. Figure 5.5b Genorm stepwise elimination of unstable reference genes. Figure 5.5c Genorm stepwise elimination of unstable reference genes. Figure 5.5d Genorm stepwise elimination of unstable reference genes. 145

175 Figure 5.5e Genorm stepwise elimination of unstable reference genes. Figure 5.5f Genorm stepwise elimination of unstable reference genes. Figure 5.5g Genorm stepwise elimination of unstable reference genes. 5.5 Serum Hormone Levels in Study Group An important consideration of this study was to ensure that the samples were collected during similar cycling hormonal conditions in all participants. Hence, samples were collected during day 1-4 of the menstrual cycle. As a double check, serum estrogen and progesterone levels were tested in all samples to ascertain that they fell within the expected range for the follicular phase of the menstrual cycle. All samples were tested by Queensland Medical Laboratories Pathology, Brisbane. As advised by Queensland Medical Laboratories Pathology (QML), mean serum estradiol levels oscillate between pmol/l in the follicular phase. Mean serum progesterone levels are below 5 pmol/l in the follicular phase. Results showed that for all samples levels for 146

176 progesterone fell within the expected reference range for the follicular phase of the menstrual cycle. Notably, estradiol levels fell below the QML reference range for samples MN004 and CN051 however levels as low as 30 pmol/l may occur in the early follicular phase of the menstrual cycle while levels as high as 400 pmol/l may occur during the late follicular phase of the menstrual cycle (Anderson and Cockayne 2003). Therefore all samples were considered within the normal range for the follicular phase of the menstrual cycle and were included in the study. Table 5.1 details estradiol and progesterone levels for each sample. Table 5.1 Serum-estradiol and progesterone levels for each expression study sample Migraine ID Estradiol pmol/l Progesterone pmol/l Control ID Estradiol pmol/l Progesterone pmol/l MN CN MN CN MN CN MN CN MN CN CN Mean Mean SEM SEM To compare the means of serum estradiol levels in the migraine versus control groups, an independent samples t test was run using SPSS v12.0. Results of this analysis indicated that there was no significant difference between the two groups (P = 0.99) for serum estradiol levels therefore, it was assumed that the cycling hormonal conditions were similar in each group. Similarly, an independent samples t test was run to compare serum progesterone levels using SPSS v12.0. Results of this analysis indicated that there was also no significant difference between the two groups (P = 0.31) therefore it was further assumed that the cycling hormonal conditions were similar in each group. 147

177 5.6 Genotyping of the ESR 1 G594A and PGR PROGINS Variants All samples were genotyped for the ESR 1 G594A variant according to the protocol outlined in Similarly, all samples were genotyped for the PGR PROGINS variant according to the protocol outlined in These specific variants were chosen because of the association found with migraine susceptibility in both MAP 1 and MAP 2. Genotyping results are in Table 5.2 Table 5.2 ESR 1 G594A and PGR PROGINS genotype for each sample Migraine ID G594A PGR PROGINS Control ID G594A PGR PROGINS MN004 GG -/+ CN051 AA -/- MN005 GA +/+ CN052 GG -/- MN006 GA -/- CN053 GA -/- MN007 GG -/- CN054 GG -/+ MN008 GA -/+ CN055 GG +/+ CN056 GA -/+ 5.7 Estrogen Receptor 1 Expression Primers To analyse expression of ESR 1, primers used were as previously described (Nancy and Berrih-Aknin 2005), and were as follows: ESR 1 RT F 5 AGACTCGCTACTGTGC 3 ESR 1 RT R 5 CCCTATGCTTTTCTGGCT 3 148

178 The forward primer bound to a region in exon 2 of ESR 1 and the reverse primer bound to a region in exon 4 of ESR 1 amplifying a 236bp fragment of the coding region of ESR1. In genomic DNA samples, these primers produced a range of products as illustrated in Figure 5.6. Thus, this assay provided an additional verification that samples were not contaminated with genomic DNA. Figure 5.6 Electrophoretogram showing PCR products using ESR 1 RT primers to amplify cdna samples in lanes 2-5 compared to gdna in lane Assay Conditions for ESR cdna concentration A dilution series of a cdna pool was performed to determine the amount of cdna required for the PCR reaction, with all other conditions being identical. The dilution series from the cdna pool was run in duplicate (0.1, 0.05, 0.025, 0.01, 0.005, ) and log concentration was plotted against means Ct values for both ESR 1 and the reference gene HPRT1. In these experiments, linearity was in the order of r 2 =0.98 for ESR 1 and r 2 =0.99 for HPRT1. From these results it was determined that a cdna concentration of was suitable to run in the analysis. 149

179 Primer Concentration The primer concentration that minimised primer dimer but showed satisfactory amplification was 150nM. Figure 5.7 shows results of melt curve analysis for ESR 1 and HPRT1 using 150nM primer concentration for ESR 1 and 250nM primer concentration for HPRT1. One peak for each gene indicated the absence of primer dimers. Figure 5.7 Results of melt curve analysis for ESR 1 and HPRT1 using 150nM primer concentration for ESR 1 and 250nM primer concentration for HPRT Assay Protocol and Cycling Parameters The reaction protocol used to amplify ESR 1 is described in Table

180 Table 5.3 ESR 1 RT-PCR protocol Reagent Final Concentration One Reaction 2x iq SYBR Green 1x 10μL Supermix* (Biorad) 5μM Forward Primer 0.15μM 0.6μL 5μM Reverse Primer 0.15μM 0.6μL Sterile Water cdna diluted 1:200 TOTAL 6.8μL 2μL 20μL *100 mm KCl, 40 mm Tris-HCI, ph 8.4, 0.4 mm of each dntp (datp, dctp, dgtp and dttp), itaq DNA polymerase, 50 units/ml, 6 mm MgCl 2, SYBR Green I, 20 nm fluoresein, and stabilizers. Cycling parameters were 95 C for 10 minutes, 60 cycles of 95 C for 15 seconds, 56 C for 30 seconds and 72 C for 30 seconds, with ramping from 72 C to 95 C rising by 1 C each step for melt-curve analysis. Figures 5.8 and 5.9 show the linear and log scale windows for the amplification plot of ESR 1. The Ct was set at a fluorescence of 10^2 after viewing the linear scale window. Figure 5.8 Linear scale of ESR 1 amplification plot. 151

181 Figure 5.9 Log scale of ESR 1 amplification plot Results of ESR 1 Expression Analysis Migraine versus Non-migraine To analyse the expression of ESR 1 normalised to HPRT 1 in the migraine versus control group, the Relative Expression Software Tool 2005 (REST 2005) was used. The purpose of REST 2005 is to determine whether there is a significant difference between two groups, while taking into account reaction efficiency and reference gene normalisation. To analyse the data using REST 2005, raw Ct values were entered in triplicate into the REST program for each sample. In REST 2005, amplification efficiencies must be entered as a value out of 1, therefore amplification determined by the Rotor-Gene 6000 software must be halved (verbal communication, Corbett Research, Melbourne). For ESR 1, the average amplification determined by the Rotor- Gene 6000 software was 1.71/2 which was converted to 0.85/1 and entered into REST as the amplification efficiency. For HPRT 1, the average amplification determined by the Rotor-Gene software was 1.76/2 which was converted to 0.88/1 and also entered into REST as the amplification efficiency. REST performed 50,000 random reallocations of the migraine and control samples using the following randomisation 152

182 scenario: "If any perceived variation between samples and controls is due only to chance, then values between the two groups could be randomly exchanged and a greater difference would not be seen than what is seen between the labelled groups." The hypothesis test P (H1) indicated in the results table of REST 2005 represented the probability of the alternate hypothesis that the difference between sample and control groups was due only to chance. Results of the REST analysis showed no significant difference in ESR 1 expression between the migraine and control groups with a P value of indicating the alternate hypothesis that the difference between sample and control groups was due only to chance. To get an overview of ESR 1 mrna in the individual migraine and control samples using HPRT 1 for normalization, the ΔCt was calculated for each sample according to the following formula. ΔCt = mean Ct target gene mean Ct reference gene Each sample was run in triplicate. To determine the mean Ct value for each sample for both ESR 1 and HPRT 1, the mean and standard deviations of the triplicates were calculated. Any Ct value that was >1 standard deviation from the mean was considered an outlier and was eliminated. The mean of the remaining two values was then calculated. The ΔCt values for ESR 1 expression in the migraine and control groups, including the mean, variance (var), standard deviation (SD) and standard error of the mean (SEM) are shown in Table 5.4. Figure 5.10 is a graphical representation of these data. 153

183 Table 5.4 shows ΔCt values of the migraine and control samples for ESR 1 Migraine ID ΔCt Control ID ΔCt MN CN MN CN MN CN053 2 MN CN MN CN CN Mean 2.76 Mean 3.20 Var 0.99 Var 2.14 SD 1.0 SD 1.46 SEM 0.45 SEM Delta Ct 3 2 Migraine Control 1 0 Figure 5.10 ΔCt values of the migraine and control samples for ESR 1 To compare the means of the migraine versus control group, an independent samples t test (two tailed) was performed using SPSS v12.0. Results of this analysis also indicated that there was no significant difference between the two groups (P = 0.582) 154

184 confirming that there appears to be no difference in ESR 1 expression levels in this group of migraineurs compared to the control group Migraine ESR 594A versus ESR 594G Further analysis was carried out to investigate if there was a difference in ESR 1 mrna in the migraineur group who carried the ESR 1 594A susceptibility allele compared to control group who carried no ESR 1 594A susceptibility allele. To analyse the expression of ESR 1 normalised to HPRT 1 in these groups, the Relative Expression Software Tool 2005 (REST 2005) was used as described in Results of the REST analysis also showed no significant difference between the migraine ESR 1 594A and control ESR 1 594G groups with a P value of indicating the alternate hypothesis that the difference between sample and control groups was due only to chance. To get an overview of ESR1 mrna in the individual migraine 594A and control 594G samples using HPRT 1 for normalization, the ΔCt was calculated for each sample according to The ΔCt values for ESR 1 expression in these groups, including the mean and standard error of the mean (SEM) are shown in Table 5.5. Figure 5.11 is a graphical representation of these values. Table 5.5 ΔCt values of the migraine 594A and control 594G samples for ESR 1 Migraine 594A ΔCt Genotype Control 594G ΔCt Genotype MN GA CN GG MN GA CN GG MN GA CN GG Mean 3.30 Mean 3.59 SEM 0.45 SEM

185 6 5 Delta Ct Migraine A Control G 1 0 Figure 5.11 ΔCt values of the migraine 594A and control 594G samples for ESR 1 To compare the means of the migraine ESR 1 594A versus control ESR 1 594G group, an independent samples t test (two tailed) was performed using SPSS v12.0. Results of this analysis also indicated that there was no significant difference between the two groups (P = 0.602) confirming that there appears to be no difference in ESR 1 expression levels in this group of migraineurs compared to the control group. 5.8 Progesterone Receptor Expression Primers Primers were designed to amplify a 180 base pair fragment in exon 7 and were as follows: PGR RT F 5 GCACTGAGTGTTGAATTTCC 3 PGR RT R 5 CAAGACCTCATAATCCTGAC 3 156

186 5.8.2 Assay Conditions for PGR cdna Concentration A dilution series of a cdna pool was performed to determine the amount of cdna required for the PCR reaction, with all other conditions being identical. The dilution series from the cdna pool was run in duplicate (0.1, 0.05, 0.025, 0.01, 0.005, ) and log concentration was plotted against means Ct values for both PGR and the reference gene HPRT1. In these experiments, linearity was in the order of r 2 =0.98 for PGR and 0.99 for HPRT1. From these results it was determined that a cdna concentration of was suitable to run in the analysis Primer Concentration As indicated in Figure 5.12, the primer concentration that minimised primer dimer but showed satisfactory amplification was 250nM. Figure 5.12 Results of melt curve analysis for PGR and HPRT1 using 250nM primer concentration for PGR and 250nM primer concentration for HPRT1. 157

187 Assay Protocol and Cycling Parameters The assay protocol used to amplify PGR is described in Table 5.6. Table 5.6 PGR RT-PCR protocol Reagent Final One Reaction Concentration 2x iq SYBR Green 1x 10μL Supermix* (Biorad) 5μM Forward Primer 0.25μM 1μL 5μM Reverse Primer 0.25μM 1μL Sterile Water 6μL cdna diluted 1:200 2μL TOTAL 20μL Cycling parameters were 95 C for 10 minutes, 60 cycles of 95 C for 15 seconds, 59 C for 30 seconds and 72 C for 30 seconds, with ramping from 72 C to 95 C rising by 1 C each step to allow melt-curve analysis. Figures 5.13 and 5.14 show the linear and log scale windows for the amplification plot of PGR. The Ct was set at a fluorescence of 10^2 by viewing the linear scale window. Figure 5.13 Linear scale of PGR amplification plot. 158

188 Figure 5.14 Log scale of PGR amplification plot Results of PGR Expression Analysis Migraine versus Non-migraine To analyse the expression of PGR normalised to HPRT 1 in the migraine versus control group, the Relative Expression Software Tool 2005 (REST 2005) was used. To analyse the data using REST 2005, raw Ct values in triplicate were entered into the program for each sample. For PGR, the average amplification determined by the Rotor-Gene software was 1.73/2 which was converted to 0.86/1 and entered into REST as the reaction efficiency. For HPRT 1, the average amplification determined by the Rotor-Gene software was 1.76/2 which was converted to 0.88/1 and also entered into REST as the reaction efficiency. REST performed 50,000 random reallocations of the migraine and control samples using the following randomisation scenario: "If any perceived variation between samples and controls is due only to chance, then values between the two groups could be randomly exchanged and a greater difference would not be seen than what is seen between the labelled groups." The hypothesis test P(H1) indicated in the results table of REST 2005 represented the probability of the alternate hypothesis that the difference between sample and control groups was due only to chance. Results of the REST analysis showed no significant difference in PGR expression levels between the migraine and control groups with a P value of indicating the alternate hypothesis that the difference between sample and control groups was due only to chance. 159

189 To get an overview of PGR mrna in the individual migraine and control samples using HPRT 1 for normalization, the ΔCt was calculated for each sample according to the following formula. ΔCt = mean Ct target gene mean Ct reference gene Each sample was run in triplicate. To determine the mean Ct value for each sample for both PGR and HPRT 1, the mean and standard deviations of the triplicates were calculated. Any Ct value that was > 1 standard deviation from the mean was considered an outlier and was eliminated. The mean of the remaining two values was then calculated. The ΔCt values for PGR expression in the migraine and control groups, including the mean, variance (var), standard deviation (SD) and standard error of the mean (SEM) are shown in Table 5.7. Figure 5.15 is a graphical representation of these values. Table 5.7 ΔCt values of the migraine and control samples for PGR Migraine ID ΔCt Control ID ΔCt MN CN MN CN MN CN MN CN MN CN CN Mean 4.70 Mean 4.98 Var 0.48 Var 0.63 SD 0.69 SD 0.79 SEM 0.31 SEM

190 8 7 6 Delta Ct Migraine Control Figure 5.15 ΔCt values of the migraine and control samples for PGR To compare the means of the migraine group versus the control group, an independent samples t test (two tailed) was performed using SPSS v12.0. Results of this analysis also indicated that there was no significant difference between the two groups (P = 0.56) confirming that there appears to be no difference in PGR expression levels in this group of migraineurs compared to the control group Migraine PGR PROGINS versus PGR non-progins Further analysis was carried out to investigate if there was a difference in PGR mrna in the migraineur group who carried the PGR PROGINS susceptibility allele compared to controls who carried no PGR PROGINS susceptibility allele. To analyse the expression of PGR normalised to HPRT 1 in these groups, the Relative Expression Software Tool 2005 (REST 2005) was used as described in Results of the REST analysis also showed no significant difference between the migraine PGR PROGINS and control PGR no-progins groups with a P value of indicating the alternate hypothesis that the difference between sample and control groups was due only to chance. 161

191 To get an overview of ESR1 mrna in the individual migraine PGR PROGINS and control PGR no-progins samples using HPRT 1 for normalization, the ΔCt was calculated for each sample according to The Δ Ct values for PGR expression in these groups, including the mean and standard error of the mean (SEM) are shown in Table 5.8. Figure 5.16 is a graphical representation of these values. Table 5.8 ΔCt values of the migraine PGR PROGINS and control no-progins samples Migraine 594A ΔCt Genotype Control 594G ΔCt Genotype MN /+ CN /- MN /+ CN /- MN /+ CN /- Mean 4.89 Mean 5.04 SEM 0.07 SEM Delta Ct Migraine PROGINS Migraine No PROGINS Figure 5.16 ΔCt values of the migraine PGR PROGINS and control no-progins samples To compare the means of the migraine PGR PROGINS versus control no-progins group, an independent samples t test (two tailed) was performed using SPSS v12.0. Results of this analysis also indicated that there was no significant difference between the two groups (P = 0.832) confirming that there appears to be no difference in PGR expression levels in this group of migraineurs compared to the control group. 162

192 5.9 Summary and Discussion of Hormone Receptor Gene Expression Analysis This study investigated the expression of the ESR 1 and PGR genes in a group of female migraineurs compared to a group of healthy control females to investigate the potential functional role of ESR 1 and PGR in migraine. Results of association analyses outlined in chapter 4 had implicated both of these genes in migraine susceptibility in MAP 1 and the follow up population, MAP 2. The ESR 1 G594A variant is a synonymous coding polymorphism and therefore it is unlikely that this variant is responsible for a functional alteration in the gene. It is more likely that a variant in linkage disequilibrium with ESR 1 G594A is responsible. Mutation analysis outlined in 4.6 in 5 migraineurs with the 594 AA susceptibility genotype had failed to find a responsible mutation. Therefore the next step in this study was to investigate gene function in migraineurs and controls to determine if there was a difference in expression levels in the two groups, and furthermore if ESR genotypes influenced gene expression due to an unknown mutation elsewhere in the gene. In contrast to ESR 1 G594A, PGR PROGINS has been implicated in gene function with the presence of the PROGINS insert showing increased transcriptional activity due to increased stability, resulting in higher expression of PGR containing PROGINS (Agoulnik et al. 2004). This study investigated the expression of PGR in migraineurs and controls to determine if there was a difference in expression levels in the two groups, and furthermore to determine if the presence of PGR PROGINS influenced gene expression in this study sample. This study used real-time reverse transcriptase polymerase chain reaction to investigate these hypotheses in a group of migraineurs and controls. Several potentially confounding factors were carefully considered in the study design. For example, only study participants who suffered from MA, the more severe phenotype, were recruited. This was done to avoid any possibility of incorrect migraine diagnosis. All participants reported a hormonal influence to their migraines to enrich the study group. To reduce the possibility of circulating hormones influencing gene expression, all samples were collected during the follicular phase of the menstrual cycle (when circulating estradiol 163

193 and progesterone levels are at their lowest). Serum hormone levels were tested and analysed to ensure that there were no differences between the groups. On the whole, results did not support a role for ESR 1 and PGR expression changes in the migraine versus the control group. Furthermore, ESR 1 G594A and PGR PROGINS genotypes did not appear to influence gene expression in this study group. It should be noted however, that a limitation of this study was the small sample size and resultant lack of power to detect an effect. A post-hoc power analysis was undertaken on the sample size used in this study to analyse the power to detect a true difference in the means using an independent samples t test. The power analysis was undertaken using the on-line power calculator available at (Lenth 2006). The power analysis determined that this study had 80% power to detect a true difference of the means of >1.8, considering a SD of 1. Any difference below this was underpowered. If there are in fact, genuine changes in ESR 1 and/or PGR expression in migraine in some individuals, it may be subtle and therefore may require a larger study sample to reach statistical significance. Notably, this study showed that mean ΔCt values for ESR 1 in both the migraine group and the migraine group possessing the 594A susceptibility allele were lower than in the control group. If statistically significant, this would result in a ΔΔCt value of 0.44 for the migraine versus control group and 0.29 for the migraine 594A versus control 594G group. Using the formula to calculate relative expression levels described in , this would result in a corresponding fold reduction of 1.24 in migraine versus control and 1.17 in migraine with the 594A allele versus control with the 594G allele. A further consideration concerning this study is that hormone receptor gene expression was analysed in peripheral blood samples. Notably, various tissue specific ESR 1 mrna isoforms have been detected, all generated from alternate promotors, creating unique 5 ends in the different isoforms. Alternate promoter utilization is likely to control different properties of the receptor in different cell types (Osterlund et al. 2000). The human brain has distinct regional expression of the various isoforms and this specific regional expression is unlikely to be mirrored in peripheral blood. Therefore it is possible that variations in ESR 1 may have a greater impact on certain specific tissue isoforms than others. 164

194 Overall, results presented in this study suggest that ESR 1 and PGR expression does not appear to be altered in migraine nor influenced by specific ESR 1 G594A and PGR PROGINS genotypes in this study group. However it should be noted that this study was under-powered to detect anything but a large change in expression levels, and an increase in sample size would be recommended to provide more power to detect a subtle effect on expression. 165

195 Chapter 6 Association Studies of Vascular Genes 166

196 6.1 Migraine and Vascular Function Although the nature of vascular dysfunction in migraine is contentious, alterations in vascular function (Silvestrini et al. 1995; Rosengarten et al. 2003; Yetkin et al. 2006) and cerebral blood flow (Olesen et al. 1990; Friberg et al. 1994) have certainly been noted in some migraineurs. Cortical spreading depression (CSD) has also been linked to vascular dysfunction. Cortical spreading depression is a depolarisation wave that propagates across the brain cortex and activates the trigeminal nerve in animal models. It has been speculated to cause the neurological symptoms that present in MA (Bolay et al. 2002). While CSD has not yet been demonstrated in the human cerebral cortex, characteristics of CSD in humans have been shown (Hadjikhani et al. 2001). It has been hypothesised that CSD may be initiated by a vascular event (Parsons and Strijbos 2003), consequently alteration in vascular endothelial function may increase migraine susceptibility. The potential role of vascular dysfunction in migraine has led to the investigation of genes involved in vascular functioning as possible migraine candidates 6.2 The Methylenetetrahydrofolate Reductase (MTHFR) Gene as a Migraine Candidate Gene One such gene is the methylenetetrahydrofolate reductase (MTHFR) gene which encodes a key enzyme in the metabolism of the essential amino acid methionine. Homocysteine is an intermediate in this pathway. Homocysteine is an amino acid formed during the metabolism of methionine. Once formed homocysteine is remethylated to methionine via a folic acid, B 12 and B 2 dependent pathway, or converted to cysteine via a B 6 dependent pathway. The enzymes involved in the remethylation pathway are Methylenetetrahydrofolate reductase (MTHFR), methionine synthase (MS) and methionine synthase reductase (MTRR). These enzymes and their role in homocysteine metabolism are illustrated in Figure

197 Figure 6.1 The pathway of homocysteine metabolism detailing the enzymes and co-factors involved. Homocysteine-related dysfunction of the vascular endothelium has been demonstrated to have the potential to initiate and maintain migraine. Homocysteine has produced endothelial cell injury in animal and cell culture studies, and alters the coagulant properties of blood (Hering-Hanit et al. 2001). Storer and Goadsby (1997) demonstrated that spontaneous trigeminal cell firing was accelerated by the oxidized homocysteine derivative that mimics the effect of homocysteine on arteries (Storer and Goadsby 1997). Numerous studies have shown that plasma homocysteine levels are significantly higher in patients with coronary artery disease, vascular disease, and stroke (Brattstrom et al. 1984; den Heijer et al. 1996; Welch et al. 1997; Kelly et al. 2002). Homocysteine is understood to exert its deleterious effects through oxidative damage. Auto-oxidation of homocysteine leads to the formation of superoxide anion and hydrogen peroxide, both free radicals. The usual anti-thrombotic nature of vascular endothelium is converted to a more thrombotic phenotype by events that include reduced production of nitric oxide (NO) and prostacyclin. The ability of NO to 168

198 detoxify homocysteine via vasodilation and platelet anti-aggregation is reduced. In addition, homocysteine inhibits glutathione peroxidase activity, which usually acts by preventing inactivation of NO (Das 2003). Various factors are associated with hyperhomocysteinemia, in particular dietary deficiencies in the co-factors required for homocysteine metabolism ie.folic acid, B 12, B 6 and B 2, and mutations in the genes that modulate homocysteine metabolism (Das 2003). A common mutation occurring in codon 677 (C T) in the MTHFR gene, which alters an amino acid in the catalytic domain (alanine to valine) and reduces enzymatic capacity, leads to mild hyperhomocysteinemia (Frosst et al. 1995). This effect is increased in combination with low folate levels (Geisel et al. 2001). Kowa et al. (2000) provided the first evidence for a role of the T allele of the MTHFR C677T mutation in migraine, in particular MA, in a Japanase cohort of 22 individuals with MA, 52 with MO, 47 with tension headache, and 261 controls. Migraineurs at 20.3% had an increased incidence of the homozygous TT genotype in than that in controls (9.6%). The frequency of the TT genotype in individuals with MA was even higher (40.9%) (Kowa et al. 2000). Kara et al. (2003) confirmed a role for MTHFR in an independent Japanese study investigating both mutations. The genotypes 677TT and 1298CC were significantly associated with migraine while individuals with MA with 1298CC and 677CC/1298CC genotypes were even more profoundly associated with migraine risk than others (Kara et al. 2003). A study performed at the Genomics Research Centre in MAP 1 showed a significant difference (P = 0.006) between the migraine subgroup MA and control groups with regard to the mutant TT genotype (Lea et al. 2004). A recent Spanish study was unable to demonstrate a significant difference in frequency of the MTHFR 677 genotypes in migraineurs compared to controls, however did report a significant difference in the frequency of TT homozygosis between MA and MO with the T allele occurring more frequently in MA (Oterino et al. 2004). More recently Sher et al. (2006) showed an association of the TT genotype in migraine in a large study group (187 MA, 226 MO and 1212 controls) (Scher et al. 2006). In addition, metaanalyses have concluded that this particular genotype is a modest, yet significant risk factor for stroke (Fallon and Ben-Shlomo 2003). As discussed in 1.4, migraine is also associated with an increased risk of stroke, particularly in sufferers of MA (Low and Merikangas 2003). 169

199 A further mutation, an A1298C transversion in the regulatory domain, which results in the substitution of alanine for glutamine, and causes mild hyperhomocysteinemia, has also been associated with migraine susceptibility (Kara et al. 2003). Considering the apparent role of homocysteine in the circulatory system, and the apparent migraine/stroke comorbidity, variants in MTHFR may represent important genetic determinants for both neurovascular conditions. This study investigated the C677T mutation in a follow-up case control replication population, MAP 2, and the A1298C mutation in MAP 1. Furthermore, linkage disequilibrium (LD) analysis was performed to determine paiwise LD between the MTHFR 677 and 1298 SNPs in MAP The Methylenetetrahydrofolate Reductase Synthase (MTRR) Gene as a Migraine Candidate Gene Methionine synthase (MS) catalyses the cobalamin (B 12 ) dependent methylation of homocysteine to methionine. Eventually the reactive cobalamin I cofactor of MS becomes the inert cobalamin II, rendering MS inactive. Methionine synthase reductase (MTRR) is the enzyme that reactivates MS by reductive methylation using S- adenosylmethionine as the methyl donor (Wilson et al. 1999). A common polymorphism in the human MTRR gene has been identified (66A G) that leads to replacement of isoleucine with methionine at residue 22 and has an allele frequency of 0.5 (Olteanu et al. 2002). The MTRR A66G mutation has been associated with reduced enzyme activity (Zijno et al. 2003), and has been implicated in increased plasma homocysteine concentration alone (Gaughan et al. 2001), and when combined with the homozygous MTHFR 677T variant (Vaughn et al. 2004). To date there are no published studies investigating this variant for a role in migraine susceptibility. As this variant appears to be a suitable candidate for migraine studies, this research investigated the MTRR A66G polymorphism in MAP

200 6.4 Association Analysis of MTHFR Regions MTHFR C677T Polymorphism Genotyping Genotyping for this marker was undertaken by PCR and restriction enzyme digestion. The MTHFR C677T mutation is a HinfI restriction fragment length polymorphism (RFLP). Primers used were those previously described (Kowa et al. 2000), and were: MTHFR 677 F F 5 TGA AGG AGA AGG TGT CTG CGG GA 3 MTHFR 677 R R 5 AGG ACG GTG CGG TGA GAG TG 3 They resulted in a 198 base pair fragment following PCR. The 20 µl PCR reaction mix is described in Table 6.1. Table 6.1 MTHFR 677 PCR reaction protocol Reagent Final One Reaction Concentration 5mM dntps 0.2mM 0.8μL (New England Biolabs) 10x PCR Buffer 1x 2μL (Applied Biosystems) 25mM MgCl mM 1.4μL 5μM Forward Primer 0.2μM 0.8μL 5μM Reverse Primer 0.2μM 0.8μL Taq Polymerase 5U/μL 1 Unit 0.2μL (New England Biolabs) Sterile Water 11.2μL DNA 20ng/μL 40ng 2μL TOTAL 20μL 171

201 Thermocycler conditions were 94 C for 3 minutes, 35 cycles of 94 C for 1 minute, 65 C for 1 minute, and 72 C for 2 minutes, with a final step of 72 C for 5 minutes. The T allele introduces a restriction site for the Hinf I enzyme, resulting in fragments of 175 and 23 base pairs. Following amplification, 10 µl of product was digested with Hinf I overnight at 37 C. After digestion, the product was loaded into a 5% ultra highresolution agarose gel stained with ethidium bromide and electrophoresed at 90V for 60 minutes. An undigested sample indicated presence of the C allele. Figure 6.2 shows an agarose gel electrophoretogram of all possible genotypes. Figure 6.2 Agarose gel electrophoretogram of MTHFR 677 genotypes. Lane 1 shows the 100bp ladder. Lanes 2, 3, 5 & 6 show the 175 and 198bp fragments representing the CT genotype. Lane 4 shows the 175bp fragment representing the TT genotype. Lane 7 shows the 198 bp fragment representing the CC genotype Results Statistical analysis revealed no significant difference between genotyped migraineurs and the matched control group with regard to allele frequencies (X 2 = 0.31, P = 0.576) and genotype frequencies (X 2 = 0.72, P = 0.697). There was no significant difference in both subgroups, MA (allele frequency X 2 = 0.77, P = 0.381; genotype frequency X 2 = 2.12, P = 0.346) and MO (allele frequency X 2 = 0.35, P = 0.556; genotype frequency X 2 = 2.14, P = 0.315) compared to the control group, although it should be noted that all but one migraineur who possessed the TT genotype suffered MA. There was no significant difference in the male case versus control comparison (allele frequency X 2 = 0.02, P = 0.9; genotype frequency X 2 = 1.35, P = 0.509), and female case versus 172

202 control comparison (allele frequency X 2 = 0.45, P = 0.503; genotype frequency X 2 = 0.58, P = 0.749). However, although not statistically significant in MAP 2, the TT genotype appeared to be over-represented in the MA subgroup compared to the MO group. See Tables 6.2 and 6.3 for frequeny distribution and results. This is consistent with results of this marker in MAP 1 where the TT genotype was associated with MA (Lea et al. 2004). In fact, when genotype and allele frequencies of this analysis were added to genotype and allele frequencies of the MAP 1 analysis performed by Lea et al (2004) at the Genomics Research Centre (Table 6.5), results were indeed significant. As shown in Table 6.5 while the migraine versus control analysis did not quite reach statistical significance, the MA versus control and MA versus MO analyses were both significant. As the effect was stronger with regard to genotype frequencies and the TT genotype was much higher represented in migraineurs with aura it appeared that the TT genotype conferred an increased risk for MA. An odds ratio was calculated based upon the Mantel Haenszel method of combining the datasets (Mantel and Haenszel 1959), comparing the TT genotype with the CC genotype. Results indicated that individuals who carried the TT genotype were 2 times more likely to suffer from MA (OR = 2.02, 95% CI = ) than those who carried the CC genotype. Table 6.2. Distribution of the MTHFR C677T polymorphism frequencies in migraineurs and controls, and the MA, MO, male and female subgroups of MAP 2. Allele frequencies in the caucasian population for this variant have been reported as 0.25 for the T allele and 0.75 for the C allele (Ensembl). Genotypes Alleles CC CT TT n Alleles C T Migraine 99 (51%) 73 (37%) 24 (12%) (69%) 121 (31%) MA 80 (50%) 56 (35%) 23 (15%) (68%) 102 (32%) MO 19 (51%) 17 (46%) 1 (3%) (74%) 19 (26%) Male 14 (48%) 7 (24%) 8 (28%) (60%) 23 (40%) Female 85 (51%) 66 (40%) 16 (9%) (71%) 98 (29%) Control 102 (52%) 77 (39%) 19 (9%) (71%) 115 (29%) Male 15 (43%) 13 (37%) 7 (20%) (61%) 27 (39%) Female 87 (53%) 64 (39%) 12 (8%) (73%) 88 (27%) 173

203 Table 6.3 Results of chi-squared analysis of the MTHFR C677T polymorphism in migraineurs and controls, and the MA, MO, Male and female subgroups of MAP 2. Analysis Genotype frequencies Allele Frequencies Case vs Control X 2 = 0.72, P = X 2 = 0.31, P = MA Case vs Control X 2 = 2.12, P = X 2 = 0.77, P = MO Case vs Control X 2 = 2.14, P = X 2 = 0.35, P = Female Case vs Control X 2 = 0.58, P = X 2 = 0.45, P = Male Case vs Control X 2 = 1.35, P = X 2 = 0.02, P = 0.9 Hardy Weinberg Equilibrium Case P = 0.07; Cont P = 0.42 Table 6.4 Distribution of the MTHFR C677T polymorphism frequencies in migraineurs and controls, and the MA and MO subgroups of MAP 1 (Lea et al. 2004) and MAP 2. CC CT TT n Alleles C T Migraine 203 (44%) 198 (43%) 63 (13%) (65%) 324 (35%) MA 144 (44%) 128 (39%) 55 (17%) (64%) 238 (36%) MO 59 (43%) 70 (51%) 8 (6%) (69%) 86 (31%) Control 219 (52%) 206 (39%) 42 (9%) (69%) 290 (31%) Table 6.5 Results of chi-squared analysis of the MTHFR C677T polymorphism in migraineurs and controls, and the MA, MO, Male and female subgroups of MAP 1 and 2. Analysis Genotype frequencies Allele Frequencies Case vs Control X 2 = 4.96, P = X 2 = 3.15, P = MA Case vs Control X 2 = 11.11, P = X 2 = 4.95, P = MO Case vs Control X 2 = 2.74, P = X 2 = 0.1, P = MA vs MO X 2 = 11.83, P = X 2 = 2.13, P =

204 6.4.2 MTHFR A1298C Polymorphism Genotyping Genotyping for this marker was undertaken by PCR and restriction enzyme digestion. The MTHFR A1298C mutation is a MboII restriction fragment length polymorphism (RFLP). Primers used were those previously described (Kara et al. 2003), and were as follows: MTHFR 1298 F F 5 CTT TGG GGA GCT GAA GGA CTA CTA C 3 MTHFR 1298 F R 5 CAC TTT GTG ACC ATT CCG GTT TG 3 They resulted in fragments of 84, 31, 30, 18 base pairs following PCR. The 20 µl PCR reaction mix is described in Table 6.6. Table 6.6 MTHFR 1298 PCR protocol Reagent Final One Reaction Concentration 5mM dntps 0.2mM 0.8μL (New England Biolabs) 10x PCR Buffer 1x 2μL (Applied Biosystems) 25mM MgCl 2 2mM 1.6μL 5μM Forward Primer 0.2μM 0.8μL 5μM Reverse Primer 0.2μM 0.8μL Taq Polymerase 5U/μL 1 Unit 0.2μL (New England Biolabs) Sterile Water 11.3μL DNA 20ng/μL 50ng 2.5μL TOTAL 20μL Thermocycler conditions were 94 C for 3 minutes, 35 cycles of 94 C for 1 minute, 65 C for 1 minute, and 72 C for 1 minute, with a final step of 72 C for 7 minutes. 175

205 The A allele introduces a restriction site for the MboII enzyme, resulting in fragments of 56, 31, 30, 28, and 18 base pairs. Following amplification, 10 µl of product was digested with MboII overnight at 37 C. After digestion, the product was loaded into a 5% ultra high-resolution agarose gel stained with ethidium bromide and electrophoresed at 90V for 60 minutes. An undigested sample indicated presence of the C allele. Figure 6.3 shows an agarose gel electrophoretogram of all possible genotypes. Figure 6.3 Agarose gel electrophoretogram of MTHFR 1298 genotypes. Lane 1 shows the 100bp ladder. Lanes 2 & 4 show a 56 bp fragment representing the AA genotype. Lanes 3 shows 84 and 56 bp fragments representing the AC genotype. Lane 6 shows a 84 bp fragment representing the CC genotype Results As shown in Table 6.8, statistical analysis revealed no significant difference between genotyped migraineurs and the matched control group with regard to allele frequencies (X 2 = 2.31, P = 0.126) and genotype frequencies (X 2 = 2.11, P = 0.348). There was no significant difference in the MA and MO subgroups (MA allele frequency X 2 = 0.12, P = 0.726; genotype frequency X 2 = 2.53, P = 0.282), (MO allele frequency X 2 = 1.05, P = 0.306; genotype frequency X 2 = 1.07, P = 0.585) compared to the control group. There was no significant difference in the male case versus control comparison (allele frequency X 2 = 1.45, P = 0.228; genotype frequency X 2 = 2.07, P = 0.355), however a significant association was evident in the female migraine versus control group with regard to allele frequencies X 2 = 6.76, P = This result remained significant even after Bonferroni correction for multiple comparisons which conservatively sets the alpha level to 0.05/5 = The comparison of genotype frequencies between the female case versus control group (X 2 = 6.35, P = 0.042) was also significant, although 176

206 not after Bonferroni correction for multiple comparisons. Consequently, the MTHFR 1298 A allele may confer a modest risk for female migraineurs. Genotype and allele frequency distributions are displayed in Table 6.7. Allele frequencies in the caucasian population have been reported as 0.37 for the C allele and 0.63 for the A allele (Ensembl). Table 6.7 Distribution of the MTHFR A1298C polymorphism frequencies in migraineurs and controls, and the MA, MO, male and female subgroups of MAP 1. Genotypes Alleles AA AC CC n Alleles A C Migraine 127 (56%) 81 (35%) 20 (9%) (73%) 121 (27%) MA 80 (58%) 46 (33%) 12 (9%) (75%) 70 (25%) MO 41 (55%) 28 (37%) 6 (8%) (53%) 40 (47%) Both 6 (40%) 7 (47%) 2 (13%) (63%) 11 (37%) Male 29 (45%) 27 (42%) 9 (13%) (65%) 45 (35%) Female 98 (60%) 54 (33%) 11 (7%) (77%) 76 (23%) Control 117 (50%) 90 (38%) 28 (12%) (69%) 146 (31%) Male 37 (57%) 20 (31%) 8 (12%) (72%) 36 (28%) Female 80 (47%) 70 (41%) 20 (12%) (68%) 110 (32%) Table 6.8 Results of chi-squared analysis of the MTHFR A1298C polymorphism in migraineurs and controls, and the MA, MO, Male and female subgroups of MAP 1. Analysis Genotype frequencies Allele Frequencies Case vs Control X 2 = 2.11, P = X 2 = 2.31, P = MA Case vs Control X 2 = 2.53, P = X 2 = 0.12, P = MO Case vs Control X 2 = 1.07, P = X 2 = 1.05, P = Female Case vs Control X 2 = 6.35, P = X 2 = 6.76, P = Male Case vs Control X 2 = 2.07, P = X 2 = 1.45, P = Hardy Weinberg Equilibrium Case P = 0.18; Cont P =

207 6.5 Linkage Disequilibrium Analysis of the MTHFR C677T and A1298C Variants Linkage disequilibrium analysis of MTHFR 677 and 1298 variants was undertaken to determine the extent of pairwise linkage disequilibrium between the two markers. Genotyping for the MTHFR 677 variant was previously determined by Lea et al (2004) at the GRC, while genotyping for the MTHFR 1298 variant was undertaken in the same study group and detailed in of this study. Table 6.9 shows the frequency distribution of both variants in MAP 1 along with the frequency distribution of a large meta-analysis of these variants in 16 Caucasian populations (Ogino and Wilson 2003). Notably, the frequency distribution in MAP 1 appeared similar to that reported in the meta-analysis. To determine this statistically, Clump analysis was used. Results showed no significant difference between the control group and the distribution of the meta-analysis (T1 X 2 = 9.31, P = 0.26; T4 X 2 = 5.92, P = 0.24). There was also no significant difference between the case and control groups (T1 X 2 = 6.87, P = 0.31; T4 X 2 = 5.25, P = 0.27), and the MA and control groups (T1 X 2 = 7.36, P = 0.28; T4 X 2 = 5.61, P = 0.25) in this study. Linkage disequilibrium analysis was undertaken using the EH and 2LD programs as previously described (Terwilliger and Ott 1993; Zhao 2004). The results are presented as D and P values. Results indicated that there was very strong linkage disequilibrium between the MTHFR 677 and 1298 loci with a D value close to 1 (D = 0.984, P = ). Table 6.10 shows D and P values generated by the LD analysis as well as the physical distance between the markers. The distance calculations were performed using information on genomic location of the relevant SNPs provided by Ensembl v.34, Oct 2005 ( Due to the close distance between these two loci (1902 bases) strong LD was expected. LD analysis of these markers in a US study has been reported as D = 0.985, also indicating almost complete LD of these markers in the study group under analysis (Hobbs et al. 2006). However, it should be noted that LD in these markers has also been shown to be as low as D = 0.55 in African Americans. This is consistent with reports of the early origins of African populations, 178

208 in which LD tends to be lower because the number of historical recombination events is likely to be large (Shi et al. 2003). Table 6.9 Frequency distribution of MTHFR 677 and 1298 markers in MAP 1 and a large metaanalysis (Ogino and Wilson 2003). Cases MA Case Controls Meta- Analysis MTHFR 677 MTHFR 1298 CC AA 31 (14%) 23 (15%) 34 (15%) 1846 (15%) CC AC 44 (20%) 31 (21%) 38 (17%) 2826 (22%) CC CC 19 (8%) 13 (8%) 28 (12%) 1075 (8.5%) CT AA 63 (28%) 40 (27%) 54 (23%) 2789 (22%) CT AC 37 (16%) 22 (14%) 52 (22%) 2584 (20%) CT CC TT AA 32 (14%) 22 (15%) 27 (11%) 1437 (11%) TT AC TT CC Table 6.10 Linkage disequilibrium D and P values (upper right hand side) and distance in bases between markers (lower left hand side) for MTHFR 677 and 1298 markers in MAP 1. Marker (rs ) (rs ) D =0.984 P= bases - 179

209 6.6 Association Analysis of MTRR MTRR A66G Polymorphism Genotyping Genotyping for this marker was undertaken by PCR and restriction enzyme digestion. The MTRR A66G mutation is a NspI restriction fragment length polymorphism (RFLP). Primers used were those previously described (Feix et al. 2004) and were as follows: MTRR 66 F 5 GCAAAGGCCATCGCAGAAGACAT 3 MTRR 66 R 5 AAACGGTAAAATCCACTGTAACGGC 3 The PCR reaction mix is described in Table Table 6.11 PCR reaction protocol for MTRR A66G Reagent Final One Reaction Concentration 5mM dntps 0.2mM 0.8μL (New England Biolabs) 10x PCR Buffer 1x 2μL (Applied Biosystems) 25mM MgCl 2 2mM 1.6μL 5μM Forward Primer 0.2μM 0.8μL 5μM Reverse Primer 0.2μM 0.8μL Taq Polymerase 5U/μL 1 Unit 0.2μL (New England Biolabs) Sterile Water 11.1μL BSA 100x 0.2μL DNA 20ng/μL 50ng 2.5μL TOTAL 20μL 180

210 As illustrated in Figure 6.4, the G allele introduces a restriction site for the Nsp1 enzyme, resulting in fragments of 24 and 94 base pairs. Following amplification, 10 µl of product was digested with Nsp1 overnight at 37 C. After digestion, the product was loaded into a 5% ultra high-resolution agarose gel stained with ethidium bromide and electrophoresed at 90V for 60 minutes. An undigested sample indicated presence of the A allele. Figure 6.4 Agarose gel electrophoretogram of MTRR 66 genotypes. Lane 1 shows the 100bp ladder. Lanes 2, 3 & 8 show the 118 & 94 bp fragments representing AG genotypes. Lanes 6 & 7 show a 94 bp fragment representing the GG genotype. Lane 9 shows a 119 bp fragment representing the AA genotype Results As summarised in Table 6.13 and detailed in the frequency distribution in Table 6.12, statistical analysis revealed no significant difference between genotyped migraineurs and the matched control group with regard to allele frequencies (X 2 = 1.86, P = 0.173) and genotype frequencies (X 2 = 2.38, P = 0.305). There was no significant difference in both subgroups, MA (allele frequency X 2 = 1.34, P = 0.246; genotype frequency X 2 2 = 1.72, P = 0.423) and MO (allele frequency X = 0.12, P = 0.734; genotype frequency X 2 = 4.65, P = 0.1) compared to the control group. Results of comparisons between female case and control groups (allele frequency X 2 = 5.28, P = 0.022; genotype frequency X 2 = 5.21, P = 0.074) showed a significant result with regard to allele frequencies, and borderline significance with regard to genotype frequencies indicating that a significant gender effect may be evident. Notably, although the P value for allele frequencies of the female migraine versus control group was < 0.05, a Bonferroni correction for multiple comparisons would conservatively adjust the alpha 181