UMEÅ UNIVERSITY MEDICAL DISSERTATIONS

New series no 1329 ISSN: 0346-6612 ISBN: 978-91-7264-947-7

The Genetic Contribution to Stroke in Northern Sweden

Tomas Janunger

Department of Medical Biosciences, Medical and Clinical Genetics

Department of Public Health and Clinical Medicine

Umeå University 2010

From the Department of Medical Biosciences, Medical and Clinical Genetics and Department of Public Health and Clinical Medicine

Umeå University, SE-901 85 Umeå, Sweden

Copyright © by Tomas Janunger

New series no. 1329 ISSN: 0346-6612 ISBN: 978-91-7264-947-7 Cover illustration by Erika Janunger Printed by Print och Media, Umeå, 2010

Tillägnad Edit, Elmer, Valdemar och Stella

“The deepest sin against the human mind is to believe things without evidence.”

Thomas Henry Huxley

5

Abstract

Stroke is a common multi factorial cerebrovascular disease with a large impact on global health. It is a disorder primarily associated with old age but environmental factors, lifestyle choices and medical history are all important for the risk of developing the disorder. It is also known that heritability is important for predisposition to the disorder. The aim of this work has been to identify genetic variations that increase the risk of being affected by stroke in the population of northern Sweden, a population well apt for genetic studies due to well kept church and medical records together with limited genetic diversity.

In the first paper we used linkage analysis in families with early onset of stroke. By this approach we identified a region on chromosome 5q to be linked to an increased risk of developing stroke, a region previously identified as a susceptibility locus for stroke in the Icelandic population. In the second study we used genealogy to identify common ancestry and thereby identify common susceptibility to stroke. The seven families we connected showed significant linkage to the chromosome 9q31-33 region and four of the families shared a common haplotype over 2.1 megabases. In the third manuscript we investigated sequence variation of two candidate genes, TNFSF15 and TLR4. Sequencing of the TLR4 gene revealed previously identified variations in affected individuals from two of the families. Further SNP analysis showed five separate haplotypes among the investigated families and four haplotypes for TNFSF15. However none of these co-segregated with stroke among the investigated families. In the final paper we used a case-control stroke cohort to ascertain association for genetic variation in five genes and genetic regions previously suggested to be linked with stroke. Initial analyses showed association for the 9p21 chromosomal region and a variant in Factor 5 that showed protection against stroke, but after adjustments for common risk factors for stroke, the findings were no longer significant.

In conclusion, by studying selected families we have been able to show linkage to two chromosomal regions, 5p and 9q31-33, that indicate genetic predisposition for developing stroke. Further we have shown that family based studies are still an important tool in deciphering the underlying mechanisms for complex disease.

6

Contents

Original papers 8

Populärvetenskaplig sammanfattning 9

Abbreviations 11

Introduction 13

Genetics 13

Concept of 13

Molecular genetics 13

Monogenic traits 16

Complex genetics 18

Population genetics 19

Mapping of genetic diseases 22

Linkage 23

Association 25

Stroke 28

What is stroke? 28

Epidemiology 28

Classification and Pathophysiology 29

Risk factors 31

Linkage 32

7

Pathways 34

Unbiased approach 35

Material 37

Family studies 37

Case-control studies 38

Methods 40

Genotyping 40

Statistics 42

Results 44

Discussion 50

Concluding remarks 55

Acknowledgements 56

References 59

Articles and Manuscripts 70

8

Original Papers

I. Nilsson-Ardnor S*, Janunger T*, Wiklund PG, Lackovic K, Nilsson AK, Lindgren P, Escher SA, Stegmayr B, Asplund K, Holmberg D. Genome-wide linkage scan of common stroke in families from northern Sweden. Stroke. 2007;38:34-40

II. Janunger T, Nilsson-Ardnor S, Wiklund PG, Lindgren P, Escher SA, Lackovic K, Nilsson AK, Stegmayr B, Asplund K, Holmberg D. A novel stroke locus identified in a northern Sweden pedigree: Linkage to chromosome 9q31-33. Neurology. 2009;73:1767-1773

III. Janunger T, Wiklund P-G, Nilsson A.K., Stegmayr B, Asplund K, Holmberg D, Nilsson-Ardnor S, Analysis of TLR4 and TNFSF15 as candidate genes for stroke in an extended pedigree. Manuscript

IV. Janunger T, Nilsson-Ardnor S, Stegmayr B, Asplund K, Holmberg D, Wiklund P-G. Analysis of Genetic Risk Factors for Ischemic Stroke in the Northern Sweden Population. Manuscript

*Authors contributed equally to the work

Paper I is reprinted with permission from Wolters Kluwer Health. Paper II is reprinted with permission from Lippincott Williams & Wilkins.

9

Populärvetenskaplig sammanfattning Stroke, eller slaganfall som är den gängse svenska termen, är ett samlingsnamn för det medicinska tillståndet då hjärnan drabbas av syrebrist till följd av att ett blodkärl täppts till av en propp eller brustit och orsakat en inre blödning i hjärnan. Det är ett mycket allvarligt tillstånd som drabbar 30 000 svenskar och en miljon människor globalt varje år. Det är den näst vanligaste orsaken till död i västvärlden och lämnar många människor med permanenta skador och lidande, även bland anhöriga.

Stroke är en komplex sjukdom där en mängd faktorer spelar olika stor roll för huruvida en person kommer att utveckla sjukdomen eller ej. Det är fram för allt en sjukdom som drabbar de äldre, men även livsstilsfaktorer som rökning, stillasittande och diet spelar en viktig roll för att utveckla sjukdomen. Men man vet att även ärftlighet spelar en viktig roll för risken att drabbas.

I mitt arbete har jag försökt identifiera genetiska, eller ärftliga, faktorer som ökar risken för en person att drabbas av stroke. Vi har använt oss av personer som bor i Norr- och Västerbotten, en region som historiskt har präglat av snabb befolkningsökning men samtidigt liten inflyttning, något som lett till en genetiskt sett homogen population. Detta innebär att mutationer som leder till ärftliga sjukdomar ackumuleras i populationen och fler riskerar att insjukna. Men samtidigt ger det forskningen en god chans att kunna identifiera de gener som orsakar sjukdom. Dessutom finns det väl bevarade kyrkoböcker över befolkningen över flera århundraden, något som också underlättar genomförandet av genetiska studier.

Vi har dels arbetat med familjer där man haft en överrepresentation av stroke och där genomfört en helgenomanalys för att identifiera kromosomala regioner som är kopplade till stroke. Vi kunde vid analys av totalt 109 familjer se att en region på kromosom 5 var särskilt intressant för risken att drabbas av stroke, samma region som tidigare visat sig kopplad till sjukdomen i en Isländsk studie. Vi gick vidare med

10

genealogiska studier på vårt material och kunde då koppla samman sju av de tidigare analyserade familjerna till ett founder-par som levde i slutet av 1600-talet. En upprepning av helgenomanalysen visade en stark koppling för sjukdomen till ett område på kromosom 9.

I nästa steg har vi börjat undersöka kandidatgener i området på kromosom 9. Vi har sekvenserat och SNP-analyserat TLR4 och TNFSF15 för att hitta genetiska varianter som orsakar den ökade risken att drabbas men i dessa två gener kunde vi inte identifiera några mutationer som bidrar till sjukdomen.

I den avslutande studien har vi använt oss av ett fall-kontroll material av personer drabbade av ischemisk stroke, orsakade av proppar i hjärnan. Vi har analyserat gener som tidigare haft stark indikation att vara kopplade till en ökad risk för stroke i andra studier. Dock kan vi konstatera att vi i befolkningen i norr- och västerbotten inte kan finna något stöd för hypotesen att just dessa genetiska varianter vi undersökt är kopplade till en ökad insjuknanderisk.

Sammanfattningsvis har vi visat att man genom att använda familjer i en genetiskt homogen population kan identifiera kromosomala regioner där gener som bidrar till en ökad risk att insjukna finns belägna.

11

Abbreviations

ABCA1 ATP-binding cassette, sub-family A, member 1 ABI Applied Biosystems ACE Angiotensin converting enzyme AF Atrial fibrillation AIM1 Absent in melanoma 1 ALOX5AP Arachidonate 5-lipoxygenase-activating protein AMBP Alpha-1-microglobulin/bikunin precursor APP Amyloid beta (A4) precursor protein CADASIL Cerebral autosomal dominant arteriopathy with

subcortical infarcts and leukoencephalopathy CEPH Centre d'Etude du Polymorphisme Humain CHARGE Cohorts for Heart and Aging Research in Genomic

Epidemiology COL3A1 Collagen, type III, alpha 1 CT Computed tomography CysC Cystatin C F5 Factor 5 FBN1 Fibrillin 1 GWAS Genome wide association study GWS Genome wide scan HbS Sickle cell associated hemoglobin allele HDL High density lipoprotein HSDL2 Hydroxysteroid dehydrogenase like 2 IBD Identical by descent ICH Intracerebral hemorrhage IMPA2 Inosine monophosphatase 2 LCV large-scale copy number variation LDL Low density lipoprotein LPAR1 Lysophosphatidic acid receptor 1 MELAS Mitochondrial encephalopathy, lactic acidosis and

stroke-like episodes Mb Megabase, 1 million base pairs MHC Major histability complex MONICA Monitoring of Trends and Determinants in

Cardiovascular Disease MRI Magnetic resonance imaging

12

MTHFR Methylene-tetraydrofolate reductase NCBI National center for biotechnology information NHGRI National Human Genome Research Institute NINJ2 Ninjurin 2 NOTCH3 Notch homolog 3 (Drosophila) NPL Non-parametric linkage OGTT Oral glucose tolerance test ORM1/2 Orosomucoid 1 and 2 PAI1 Plasminogen activator inhibitor 1 PAPPA Pregnancy associated plasma protein A PDE4D Phosphodiesterase 4D PICH Primary intracerebral hemorrhage PITX2 Paired-like homeodomain 2 PTGR1 Prostaglandin reductase 1 SHRsp Spontaneous hypertensive stroke prone rat SMC Smooth muscle cell STRK1 Stroke 1 locus TIA Transient ischemic attack TLR4 Toll-like receptor 4 TNFSF15 Tumor necrosis factor superfamily member 15 TOAST The Trial of Org 10172 in Acute Stroke Treatment WHO World Health Organization QTL Quantitative trait locus ZFHX3 Zinc finger homeobox 3

13

Introduction

Genetics

Concept of

Genetics is the study of how traits are being transmitted over generations and how variations in the information for these traits will influence the outcome, be it eye color, crop yields or survival. The notion of heritability and its implications has been known to humans for thousands of years and has been used in improvement of domestic animals and farm crops through systematic breeding. But it was not until the mid 1800’s in the famous cross breeding experiments by Gregor Mendel that it could be proven that traits were inherited by distinct genes rather than by a continuous blending of parental attributes, which was the predominating concept of that time.

During the last 15 years, the field of genetics has been revolutionized by the rapid progress in technical developments. Previously, it took two to three days to analyze 300 to 500 base pair long stretches of DNA, whereas today, it is possible to sequence the entire genome of roughly 3 billion base pairs from a single individual in the same time. This development has led to the discovery of many afflicted genes that cause mutations in a number of familial diseases. Lately, great efforts have been made to unravel the underlying genetic contribution to common complex disorders such as diabetes, cardiovascular disease and neuropsychiatric diseases.

Molecular genetics

All information regarding heritable traits is stored by and carried from generation to generation in the chromosomes in the form of genes. The chromosomes are made up by long stretches of the DNA (deoxyribonucleic acid molecule), in the well known shape of the double helix. The DNA is in turn made up of four basic molecules, known as

14

nucleotides, the adenosine, cytosine, guanidine and thymine, abbreviated as A, C, G and T respectively. The DNA consists of two complementary strands which allows for exact replication and transmission of the DNA through the generations.

A gene is a genomic region with a specified genomic position that is associated with regulation, transcription or other function1 (figure 1). The classical definition of a gene is of a genomic region that is transcribed into messenger RNA, mRNA for short, and further translated into a protein. The transfer of information from DNA to RNA to protein is called the central dogma of molecular biology and was first postulated by sir Francis Crick in first in 1958 and revised in 19702 (figure 2). Proteins are long stretches of amino acids that are essential molecules that catalyze biochemical reactions within the cell. They are also responsible for much of the structural and mechanical functions of a cell and communication with the extra cellular environment. According to the Wellcome Trust Sanger Institute, as of February 5th 2010, the human genome consists of some 20.000 annotated protein coding genes3.

Genes vary considerably in organization and size. In some cases a gene consist of a single exon coding for proteins of tens of amino acids while the longest characterized human gene, dystrophin, spans about 2.4 million base pairs on the X-chromosome4, with 79 exons coding for 3685

Figure 1. A schematic image of the composition of a gene on a chromosome. A coding gene is composed of exons that will translate into a functioning protein while introns serve as linkers between exons and are not translated in the final product.

15

amino acids5. The vast majority of genes contains introns and undergoes splicing before the protein is translated. In the splicing process the introns are deleted from the mRNA and the exons are fused together before translation to proteins. This process allows the organism to produce a number of different protein products from the same gene by selectively splicing together different exons. This process is often tissue specific where the same gene has slightly different functional properties in different tissues6.

A gene is not necessarily only DNA that codes for the final protein but also sequences that regulate when a gene is actively transcribed or silenced. There is evidence that transcription of a gene could be influenced by regulatory regions on a different chromosome7. We also know that far from all RNA that is transcribed in a cell is translated into proteins but instead can have regulatory functions on several levels within the cell8.

When mutations of the nucleotides in the DNA occur this could lead to a subsequent shift in the amino acid sequence of the functional protein. This can in turn lead to a protein that will function more effectively, less effectively or lose its function all together which in worst case will lead to disease. If this mutation occurs in the DNA of the sex cells it can be

Figure 2. The transcription of DNA to RNA and translation into a functional protein is referred to as the central dogma of molecular biology.

16

passed on to the offspring you have a heritable trait that can be followed through the generations and studied within the field of genetics.

Single nucleotide substitutions are the most common form of genetic variations that we detect today and over 17.2 million different genetic single nucleotide polymorphisms, or SNPs, has been identified9 although only a small subset of these have been associated with disease. Repetitions or deletions of DNA can naturally vary considerably among individuals and range from pairs of two bases, as in microsatellites to larger deletions and duplications, or copy number variation (CNV), that can span from one kilobase to millions of base pairs, then known as large-scale copy number variation (LCV)10. Epigenetic modifications, defined as changes in gene expression and/or phenotype with mechanisms other than changes in DNA sequence11, such as methylation of nucleotides, are also of interest with regard to their implications in heredity and development of disease. Much work is today focused on defining the biological impact of these structural variations.

Monogenic traits

Heritable features are primarily thought of as discrete traits that can be easily traced over generations. Classic examples are the different shapes and colors of Gregor Mendel’s peas or the ability to taste phenyltiocharbamide (PTC) in humans12. These are discrete qualities that are controlled by a single gene and are examples of monogenic inheritance. Every individual carries two copies of each of the 23 chromosomes, apart from men who carry one copy each of the X and Y sex chromosomes. Hence there are two copies of every gene in every individual. Every offspring will as a result receive one copy of every chromosome and gene from each parent.

Basic monogenic inheritance is manifested over generations in one of two ways: through dominant inheritance or recessive inheritance. In a

17

A B

dominant trait a nucleotide variation in a gene, referred to as an allele, inherited from the parent carrying the allele will manifest in every individual that receive this variant. This gives every offspring a 50% risk of inheriting this allele. These manifestations can be followed over generations in families (Figure 3A) and is the case in many well known human conditions like Huntington’s disease.

In recessive inheritance sequence variations in only one of the alleles is not enough to manifest a trait, but you need to inherit the variation from both of your parents. The parents can be asymptomatic carriers for a disease that can be passed on to the offspring. In these families the trait usually turns up in children of families that have no previous history of this manifestation in on average 1 in 4 children (Figure 3B). Examples of recessive diseases are cystic fibrosis and sickle cell anemia.

Figure 3. A. Display of the classical pattern of dominant inheritance where individuals in every generation pass the trait to the next generation. B. A recessive trait will present itself in children of unaffected parents that are carriers for the trait. Squares depict men and circles women, filled symbols are affected and white are unaffected for the trait of interest. Symbols with dots represent individuals that carry mutations for a recessive trait.

Genetic variations that are located on the X-chromosome will be much more visible in men since women, who have two copies of the X-chromosome, will most often carry an unaffected variant that will rescue a normal phenotype. Men, who only carry one X-chromosome, will only

18

express the variant on the X-chromosome present in a dominant matter. This phenomenon is known as X-linked inheritance and is best known in diseases like red-green color blindness13 and hemophilia14 in humans.

A small part of the human genome is located within the mitochondria as a circular chromosome of 16.500 base pairs present in several identical copies. The mitochondria are cellular structures that are responsible for energy production and are involved in cell signaling and cellular differentiation and much of the mitochondrial genome codes for proteins that are important in the respiratory chain. Mitochondria are passed from mother to child15 via the fertilized egg, meaning that all sequence variations follow a maternal lineage and are never passed down from fathers to their offspring. Several neurological and metabolic diseases, such as retinitis pigmentosa16 and mitochondrial encephalopathy, lactic acidosis and stroke-like episodes (MELAS)17, have been connected with mutations in the mitochondrial genome.

Complex genetics

Although many discrete traits are directed by single genes we also know that often many genes interact to produce a phenotype. Continuous traits, like the height of a person, are correlated to the height of the parents of that person where tall parents tend to get tall offspring. But deviations from this occur because the height of an individual is determined not by a single gene but by several. A genetic locus that has an effect on such a trait is known as a quantitative trait locus. Recently, genetic studies in more than 63.000 individuals indicated at least 50 different chromosomal loci are involved in regulating height18 indicating the complexity of inheritance for quantitative traits. Other continuous qualities, such as longevity and other age-related traits, have also been shown to be regulated by several genetic factors19.

However, genetics alone does not explain all of the variation that we see in living organisms but most of the phenotypes are the result of

19

interactions between several genes with their environment. If an individual grows up without sufficient nutrition he or she will never grow to full potential stature in adulthood due to malnutrition regardless of genetic predisposition for height. The interaction of several loci together with different environmental factors is apparent in many common diseases such as diabetes mellitus and cardiovascular diseases, where heritability is an important risk factor but where life style choices have a significant influence on disease outcome.

Population genetics

Population genetics is the study of how genetic variations are distributed and selected for over time in a specified population. It is a fundamental concept for the understanding of evolution and the process of speciation. Although all humans are genetically identical to 99.5 – 99.8%20 these differences can provide information that will prove useful in identifying genetic mechanisms underlying disease.

A population can be defined as a group of individuals of the same species which inhabits a defined geographic region. Social and ethnic settings can also be important factors when defining populations. When an individual or a small group is settling into a new region without much or any contact with other populations the initial genetic constitution of this group is central for subsequent generations. Initially, a new population will lose much of its genetic diversity due to the limited numbers of alleles that can be carried by a small group of settlers. This is known as the founder effect (Figure 4.).

20

The founder effect can lead to that rare alleles or mutations that occur in the mother population in the forming of a daughter population can become frequent or even dominant in subsequent generations in that population as it expands. These changes in allele frequencies occur at random and are referred to as genetic drift and over time this phenomenon can lead to permanent eradication or fixation of certain alleles in the population. Drastic changes in allele frequencies can also occur in well established populations. Severe famine or diseases can cause the eradication of large fractions of a population, known as a population bottleneck, leading to diminished genetic diversity that can be identified in subsequent generations.

If genetic drift in an isolated population has led to the fixation of a single allele the only way to introduce new allelic variants is through mutation. Mutations occur by way of natural process during cell division when the genome undergoes duplication or in cells due to environmental exposure to mutagenic agents such as radiations, viruses and chemicals. The mutation rate in humans is estimated to 2.5*10-8 mutations per nucleotide site which corresponds to approximately 175 mutations for every cell division21.

Figure 4. When daughter populations diverge from a parental population chance alone can have a large impact on allele diversity. In this example the red and blue dots represent two different alleles at the same locus, comparable to A and a, and possible outcomes of allele frequency distribution.

21

Most mutations that arise are functionally and selectively neutral and the chance to stay in a population is ruled by genetic drift. However some mutations will have deleterious or beneficial effects on reproductive outcome for individuals and these alleles are therefore under selective pressure. Beneficial mutations will tend to increase in frequency and deleterious mutations go extinct.

One way to investigate whether an allele is under selective pressure is to examine if an allele is in Hardy-Weinberg equilibrium (HWE). The assumption behind HWE requires the population to be infinite in size, undergo random mating and a complete lack of new mutations or selection. In practice HWE is impossible to achieve but it is still a very useful tool when evaluating genetic information in populations. It states that two alleles (A and a) at the same locus with frequencies p (for the A allele) and q (for the a Allele) have a total frequency of 1 (p+q=1). If a population is at equilibrium the frequency of individuals homozygous for AA = p2, homozygous for aa = q2 and heterozygous Aa = 2pq which gives the total allele frequency p2 + 2pq + q2 = 1 (Figure 5).

Figure 5. Diagram showing the expected genotype frequencies for two alleles in one locus under the assumption that the criteria of HWE are met. The x-axis shows the allele frequencies p and q and the Y-axis the genotype frequencies.

22

If the observed frequencies of homozygotes (AA and aa) and heterozygotes (Aa) deviate from what is expected when looking at the allele frequencies (A and a) this is a good indication that this locus is under selective pressure in the population. One of the allele variants could be associated with an increased or decreased reproductive success within the studied population. It could also be that random mating is not in effect. Both inbreeding and assortative mating will for instance increase the frequency of homozygosity. Other possibilities are that there has been sampling bias of the population under investigation. In collecting material from populations for genetic studies it is vital that the sampling is random. An overrepresentation of individuals from a particular geographic area could result in sampling of many close relatives which increases the risk of skewed allele frequencies at different loci22

Practical use of HWE can be seen in the clinical setting. If the allele frequency of a disease is known in the population one can deduce the risk for future parents to get a child that will be affected by that disease. If one of the parents is a carrier for a recessive disease, such as cystic fibrosis, you can, by applying the HWE equation, deduce the carrier frequencies for cystic fibrosis mutations in the population which would provide the risk that the other parent is a carrier. The complied information provides an estimate of total risk involved for a couple for their future child to become a carrier of the disease or affected by it.

Mapping of genetic diseases

The aim of studying heritable traits is to identify the genetic variations that cause the phenotype of interest. This is done by analyzing genetic markers and test if they are inherited together with the studied phenotype. This gives an indication in which chromosomal region the genetic disposition of a trait is located. Genetic markers used in mapping are well defined genetic regions, known as loci, with regard to DNA sequence. Markers are polymorphic with two or more alleles at each locus and a well defined physical genomic position.

23

Linkage

The use of microsatellites has been of great importance in identifying genetic causes for familiar diseases. Microsatellite markers are short DNA stretches, 1-4 base pairs in length repeated several times. The most common form is two nucleotides repeated from two to about 100 times. Such repetitions are dispersed throughout the genome and are virtually without known function. Most microsatellites have several known alleles in a population which makes them highly informative in linkage analysis. In linkage analysis you specify a heritable trait of interest and identify families that carry this trait. You analyze between 400 and 1.000 microsatellites that are evenly dispersed throughout the entire genome for each individual and by statistical analysis calculate if any of the markers co-segregate with the trait. Since the location of the markers are well known this will give a good indication to which chromosomal region harbor the genetic factor or factors for the investigated trait (figure 6).

There are exceptions to the rule that microsatellites have no known function. In the case of tri-nucleotide repeat disorders single codon repetitions have expanded to a point where the translated proteins

Figure 6.A. theoretical example of linkage in a small pedigree. Allele 4 for marker A is inherited together with the observed trait in all affected children and none of the unaffected. This is a good indication that the gene responsible for the analyzed trait is located in the vicinity of marker A.

24

aggregate and become toxic. This is known primarily in neurodegenerative disorders, such as Huntington’s disease23 and Fragile-X syndrome24.

Linkage between two loci is tested by calculating LOD scores, a statistical analysis that compares the likelihood that loci are linked. If distinctive alleles at two loci are physically close, chances are that they will be transmitted together to the next generation. If they are distant, alleles at the two loci will be transmitted randomly. This is due to recombination that occurs at meiosis, the genetic shuffling that occurs when sex cells are formed. A recombination is unlikely to occur between physically close loci which mean that alleles at the two loci will likely follow one another over generations. Loci far apart will often experience recombination between them causing alleles to be distributed at random in the following generations. The LOD score is a measure if the observed rate of recombination between loci is more frequent than can be expected by chance. The formula for the calculation can be seen in figure 7. Convention states that a LOD score over +3 shows evidence for linkage and a LOD score less than -2 evidence against linkage and a result between +3 and -2 can neither confirm nor reject linkage in a single test. When analyzing linkage on a genome wide basis multipoint LOD score are used. This analyzes not only a number of genetically close markers together but also the allele frequencies of the markers used. Genome wide significance has been suggested to LOD scores exceeding +3.325.

Figure 7. The formula for calculating LOD scores. Θ is the recombination frequency between two loci, R stands for the number of observed recombination events and NR for non-recombination events.

25

The approach with linkage analysis has primarily been useful in detecting monogenic disorders with high penetrance. Examples in the northern Sweden population this are the detection of genes or loci for Best’s macular dystrophy26, migraine27 and loss of pain perception28. But linkage studies have also been successfully used to identify loci for complex disorders. In 2005 a Dutch/Swedish collaboration identified a susceptibility locus for affective disorder in a large pedigree from northern Sweden29 and linkage analysis was used in the Icelandic study that reported the first susceptibility locus associated to common stroke on chromosome 5q1230 and for myocardial infarction to chromosomes 1331 and 1432.

Association

Association studies are generally performed in order to detect or confirm wheter genetic variants increase the susceptibility to develop a specific trait. These studies require two groups of individuals; one that display the trait of interest; the cases, and one group that does not; the controls. The genetic variant of interest is analyzed in all samples and if there is a significant over or under representation of that variant between the groups this is a good indication that the marker is associated to the investigated trait (figure 8).

The simplest ways of determining association is by comparing two groups and determine whether a desired trait or factor is over- or underrepresented en either group. This can be done by a χ2- test that compares an observed value with a theoretical, for example, the number of chromosomes in people diagnosed with Down’s syndrome compared to a group of unaffected individuals. The theoretical or expected number of chromosomes is that observed in the control group. If there is a statistically significant deviation of the number of chromosomes among people with Downs’s syndrome the χ2- test will display this.

26

It is important that the groups resembles each other in all aspects apart from the investigated trait. If the groups for instance originate from different geographic regions there is a risk that you will find genetic differences between the studied groups that has nothing to do with the trait you study but will reflect the geographic difference. Misrepresentation of study groups is a phenomenon referred to as population stratification33.

The most common genetic marker in the genome is sequence variations of one single base pair, known as a single nucleotide polymorphism, SNP. This is also the marker that is most analyzed in large scale association studies. A SNP is a sequence variation where a nucleotide has mutated into one of the other creating two allelic variants that can be detected in the population. Of the 3 billion base pairs that make up the human genome 17.2 million positions have been identified as being polymorphic9, a number that is likely to rise as the amount of genotyping data is increasing rapidly and a greater diversity of populations are being analyzed.

The rapid development of high throughput genotyping platforms have given new possibilities of efficiently scanning the entire genome for stroke susceptibility loci and genes, so called genome wide association

A Figure 8. If a genetic variant is significantly over-represented in a group of individuals affected by an investigated trait (the black “allele” in group A) compared to a control group (B) this a good indication that the variation is associated to that trait.

B

27

studies (GWAS). This can be done without a priori hypotheses underlying pathophysiology and has the potential to reveal new genes and pathways that has previously been overlooked. In a typical GWAS study somewhere between 100.000 and 1 million SNPs are analyzed and evaluated for each participant. This usually requires multicenter collaborations, well developed bioinformatics and statistical infrastructure as well as substantial financial resources.

As of January 7th 2010, 466 publications for GWAS meeting minimal criteria set by the National Human Genome Research Institute (NHGRI) have been compiled. Together they have identified 2.189 SNPs as associated with many different traits and diseases34. The GWAS have been informative in regard to the understanding of the structure of the human genome albeit disappointing when it comes to explaining the heritability of many traits. In the example of human height variation the 54 loci that has been validated to be associated only manage to explain 5% of the total variability of the trait despite having a heritability of approximately 80%35. The initial hypothesis was “common disease, common variant” where genetic contribution for complex diseases came from common variants, but this idea have gathered little support by the combined results from the GWAS so far.

This phenomenon has come to be referred to as “missing heritability” or the “dark matter” of genetics in an analogy to astronomy. We can see that heritability exist and detect its influence, but we cannot explain what it is. One hypothesis is that the genotyping platforms used today have been produced to include SNPs that are common over many populations at the expense of SNPs that are rare. However, if a genetic variant has a deleterious effect on its carrier there will be a selective pressure to weed out that variation from the population, keeping it at a low frequency. If it turns out that rare, but more deleterious variants are responsible for the heritability of complex diseases, the approach using common variants may have caused researchers to miss rare genetic variants with a higher impact on investigated traits.

28

Stroke

What is stroke?

Stroke is a neurological disorder that is caused by the depletion of oxygen to the brain through occlusion or rupture of cerebral blood vessels. Symptoms range from visual defects and sensory disruptions to confusion, paralysis, unconsciousness and death. The severity and outcome of a stroke episode depends on the localization and severity of the stroke, as well as the age and medical history of the affected individual.

Stroke is also a cardiovascular disease. Cardiovascular diseases include disorders that affect the heart as well as the vascular system, such as myocardial infarction and hypertension36. The term cerebrovascular disease is used to describe diseases of vascular origin that affect the brain, and includes a range of diagnoses such as vascular dementia and aneurysms37.

The most commonly used definition of stroke is described by the World Health Organization (WHO) as: “rapidly developing clinical signs of focal (or global) disturbance lasting 24 hours or longer or leading to death with no apparent cause other than of vascular origin”38. This definition encompasses all classes of the disease which, in turn, is divided into three pathological subtypes: ischemic stroke, intracerebral hemorrhage (ICH) and subarachnoid hemorrhage (SAH).

Epidemiology

Stroke has a major impact on the health and well-being of millions of people around the world. It is the leading cause of adult disability in Europe and the United States. It is calculated to be the second largest cause of death on a global level39. It not only disables millions of individuals for life, but also wreaks havoc on the families of afflicted individuals and society as a whole. The annual estimated cost for health-

29

care for stroke patients in Sweden alone is estimated at 1.5 billion euro40.

Classification and Pathophysiology

Ischemic stroke causes approximately 80% of all stroke cases and is caused by physical constriction of blood flow to the brain. In populations of Caucasian heritage, it is estimated that 50% of all ischemic stroke is caused by thrombosis, the formation of physical obstacles in the blood vessels. About 25% are lacunar infarcts, occlusions of the deep penetrating arteries of the brain, and another 20% arise from emboli, i.e. when a clot that has been formed in the heart, aortic arch or the neck vessels is transported to the brain. The remaining 5% are caused by vasculitis, arterial dissection and other less common causes41. There are however ethnical differences. Hemorrhagic stroke is more common in Asian populations than in populations of Caucasian heritage42, and studies have shown that an African heritage increases the risk of being affected both with ischemic stroke and intracerebral hemorrhage compared to populations of Caucasian origin43,44.

The most common factor behind thrombotic events is the presence and development of atherosclerosis. Atherosclerosis is the process where the accumulation of lipids, immune cells, fibrous tissue and smooth muscle cells form plaques which lead to stiffening and constriction of the blood vessel walls. Plaques can become unstable and rupture, dislodging clots that travel in the bloodstream and entangle in distant vessels, causing a thrombotic event. Atherosclerosis is a response to stress on the blood vessel wall and can be induced by hypertension45 or chemical stress by oxidized low-density lipoproteins (LDL)46. It can also be induced by pathogen infection, as with Chlamydia pneumoniae47, in which case an atherosclerotic plaque becomes a site of chronic inflammation48.

30

Clinical diagnosis of ischemic stroke is done with the help of a computed tomography (CT) scan. The presence of focal symptoms in conjunction with negative findings on a CT is consistent with ischemic stroke. Intracerebral hemorrhage is diagnosed by focal signs together with a positive CT scan. Subarachnoid hemorrhage is diagnosed by an arteriovenous malformation on angiography, positive findings on a CT, or bloodstained cerebrospinal fluid49.

In order to improve diagnosis of ischemic stroke, subtype classification systems have been developed. The TOAST classification (Trial of Org 10172 in Acute Stroke Treatment) is widely used, combining several tests such as imaging (CT or magnetic resonance imaging (MRI)) and laboratory assessments together with clinical features50. TOAST further subdivides ischemic stroke into five categories; large artery atherosclerosis, cardioembolism, small vessel occlusion, stroke of other determined etiology and stroke of undetermined etiology.

Transient ischemic attacks or TIAs are defined by the American Heart Association as “brief episodes of neurological dysfunction resulting from focal cerebral ischemia not associated with permanent cerebral

infarction”51. TIAs have not been included in studies presented in this work and will not be discussed further.

The remaining ca 20% of all stroke cases in Caucasians is due to blood vessels in the brain rupturing, leading to hemorrhagic stroke. A ruptured vessel will leak blood, initiating an expanding hematoma that causes compression damage on the neighboring brain tissue. Hemorrhagic stroke is in turn divided into two subclasses: intracerebral hemorrhage and subarachnoid hemorrhage, depending on the location of the hemorrhage. Subarachnoid hemorrhages are located in the space between the arachnoid membrane and the pia mater surrounding the brain. When a vessel ruptures and forms a hematoma within the parenchymal cerebral space, it is classified as an intracerebral hemorrhage. A distinction is usually made between primary and

31

secondary intracerebral hemorrhage, whereby a primary intracerebral hemorrhage (PICH) is a spontaneous rupture of the blood vessel, whereas a secondary intracerebral hemorrhage occurs when the rupture is associated with vascular abnormalities, tumors or coagulation disorders52. Intracerebral hemorrhage in the cortex and underlying white matter is categorized as lobar and is referred to as deep if the bleeding is located in the basal ganglia, periventricular white matter or internal capsule. It is worth noting that around 30-40% of all stroke cases cannot be correctly identified using standard classification and are referred to as cryptogenic stroke53

Risk factors

There are many identified factors that have been associated with an increased risk of being affected by stroke; these can be classified by the strength of evidence and by the potential of modification54. Certain risks are non-modifiable whilst others are modifiable or potentially modifiable.

It has been well documented that age, sex and ethnicity are independent, non-modifiable risk factors for developing the disease. Age is the single strongest risk factor for both ischemic stroke and intracerebral hemorrhage. After the age of 55, the risk of being affected by stroke doubles every decade55. For men, there is an additional increase in risk earlier in life than for women. However, more women will suffer from stroke during their lifetime due to a longer expected life span56.

Well established modifiable risk factors include several medical conditions, such as hypertension57, diabetes mellitus58 and history of coronary heart disease54. It also includes risk factors in which lifestyle plays an important part, such as physical inactivity59 and smoking60.

An independent non-modifiable risk factor is heritability, or the genetic predisposition of developing stroke. It has been shown that an individual

32

who has a close relative that has been affected by stroke is at a higher risk of developing the disorder compared with the population as a whole. Several studies point to an increased risk of 1.2-4.3 times of falling ill if one or both of an individual’s parents have been affected by stroke, compared to a person without a family history of the disorder61, 62.

Three large scale studies have been performed to assess concordance for stroke among twins. Studying the concordance of disease for monozygotic compared to dizygotic twins is considered one of the most efficient ways of determining the genetic burden for various traits on an epidemiological level. Two of the studies showed an increased risk of 2.1 and 4.3 times respectively for monozygotic twins of being affected compared to dizygotic twins63, 64. The third study showed no significance in difference between monozygotic and dizygotic twins65, but since this study did not assess non-fatal stroke events, of which many are disabling but not fatal, significant data has been missed out.

Linkage

The first genetic variants that could be connected to an increased risk of developing stroke were identified by genome wide linkage studies. Families affected by heritable syndromes where stroke was a part of the clinical picture were identified and analyzed. The first disease with an identified genetic variant that conferred stroke was cerebral autosomal dominant arteriopathy with subcortical infarcts and leukoencephalopathy (CADASIL). It is a dominantly inherited disorder where ischemic stroke is inherited with frontal lobe dementia. Mutations were first described 1996 based on studies of the NOTCH3 gene in French families66. Since then, mutations in over 500 families throughout the world have been described67. NOTCH3 is part of a family of cell surface receptors that are homologues to the Notch receptors in Drosophila, where they are vital for neuronal development. In humans,

33

the NOTCH3 gene is expressed primarily in vascular smooth muscle cells68.

Examples of familial disorders that carry an increased risk of arterial dissection which lead to occlusion of blood vessels are Ehler-Danlos syndrome type 4 and Marfan syndrome. These have also been described with the help of linkage analysis, in which mutations in the collagen, typeIII, alpha-1 (COL3A1) gene69 and fibrilin1 (FBN1)70 have been identified for the respective diseases.

Hemoglobinopathies are diseases where mutations in either of the two hemoglobin genes cause structural defects of the red blood cells which, in turn, cause anemia, seizures and an increased risk of ischemic stroke. Sickle-cell anemia is the most common form of hemoglobinopathy. Although carrier frequencies vary considerably in different populations, up to 30% of the population in parts of central Africa are affected71. The life-time risk of being affected with stroke has been estimated to 25% for homozygous carriers of the sickle cell associated hemoglobin allele (HbS), which is the most common mutation for sickle-cell anemia71.

Linkage studies have also successfully identified genes increasing the risk for intracerebral hemorrhage. Hereditary cerebral amyloid angiopathies are a family of diseases where mutations lead to protein aggregation in the cerebral blood vessels, thereby weakening the cell walls72. This weakening increases the risk of cell wall ruptures and subsequent intracerebral hemorrhage. Mutations that have been coupled to hereditary forms of cerebral amyloid angiopathies have been found in the amyloid beta (A4) precursor protein (APP) gene in Dutch, Belgian and Italian73 families, and in the cystatin C gene in families from Iceland74. All described hereditary cerebral amyloid angiopathies are inherited in an autosomal dominant fashion.

Genome wide analyses have been performed, not only for syndrome like disorders, but in order to identify genetic variants that increase common

34

stroke susceptibility. In Iceland, a genome-wide linage analysis was undertaken on 479 individuals in 179 families affected by stroke; the study included all forms of stroke. Linkage was detected on chromosome 5p12, to a locus named STRK1, where subsequent fine-mapping produced a LOD score of 4.4030. Extensive sequencing and association analysis showed significant association for variation in the phosophodiesterase 4D (PDE4D) gene and an increased risk of being affected by stroke of atherosclerotic and embolic origin75. PDE4D is a protein with several tissue specific isoforms which degrade cyclic AMP, a central signal transduction molecule in multiple cell types. A second genome wide scan was performed in a northern Sweden cohort, confirming the STRK1 susceptibility locus76.

A second stroke susceptibility locus was later found on chromosome 13q12-13 in a linkage study for myocardial infarction32. In the follow-up analysis, genetic variants in the arachidonate 5-lipoxygenase-activating protein (ALOX5AP) gene showed association to both myocardial infarction and ischemic stroke, particularly for men. The ALOX5AP protein is required for leukotrine synthesis, which is a molecule that is implicated in inflammatory responses like asthma77.

Pathways

Both successful and unsuccessful attempts have been made to validate the Icelandic findings for PDE4D and ALOX5AP as stroke susceptibility genes. But the results are dispersed and recent meta analyses could not confirm the initial findings78, 79. The attempts to analyze genes and genetic variants associated with stroke have been numerous. This has primarily been done through candidate gene studies where tested genes are chosen on merit of plausible functions in biochemical pathways.

The pathways have been involved in vascular tone regulation, platelet aggregation and cell wall integrity, and the list of genes that have been tested on the basis of plausible pathways is extensive. Several genes

35

have been suggested to regulate the risk of being affected by stroke, but reported findings have proven difficult to replicate.

One way of increasing power in genetic studies is to pool results from several different studies that have investigated the same gene in different populations in meta-analysis. Casas and colleagues published a major meta-analysis for ischemic stroke where 120 separate studies were compiled. This showed that of the 51 polymorphisms in 32 genes that had been investigated, four were significantly associated after analysis; methylene-tetraydrofolate reductase (MTHFR) C677T , factor 5 (F5) Arg506Gln, prothrombin (F2) G20210A and angiotensin converting enzyme (ACE) insertion/deletion80. Candidate gene analyses in the population from northern Sweden has indicated that plasminogen activator inhibitor 1 (PAI1) 4G/5G81 is associated with ischemic stroke and that MTHFR C677T is associated to intracerebral hemorrhage82.

Unbiased approach

As previously described, the introduction of GWAS has enabled total genome analysis without a priori hypothesis of candidate genes and functional pathways. The first GWAS published on ischemic stroke showed significant association to two novel stroke susceptibility genes; inosine monophosphatase 2 (IMPA2) and absent in melanoma 1 (AIM1) and to two loci on chromosomes 7p21 and 6q21 without candidate genes83. A second published GWAS pooled four different stroke cohorts from the Cohorts for Heart and Aging Research in Genomic Epidemiology (CHARGE) consortium in order to increase power and test reproducibility of initial findings. This study showed significant results for all types of stroke for two SNPs in the vicinity of the ninjurin 2 (NINJ2) gene, a gene that is expressed in glial cells and is known to be upregulated after neurological damage84. Hazard ratios were in the range of 1.17-1.35 in the different study materials.

36

The first study has proven limited in that it covered only a small number of participants, 249 cases and 268 controls, and none of the results were significant after Bonferroni correction. Reanalysis was performed on the material focusing on SNPs on chromosome 9p21 in response to the finding that myocardial infarction was associated to this region85-88. The reexamination of data suggested that the same haplotypes that conferred risk for myocardial infarction also conferred risk for ischemic stroke89. The second study included almost 2.500 cases and over 26.000 controls and was followed by a meta-analysis to which an additional Icelandic cohort was added and inclusion criteria were narrowed to MRI-verified brain infarcts in persons without history of TIA or stroke. A total of 1.822 affected individuals fulfilled the criteria and were analyzed together with 7.579 controls. None of the SNPs registered for genome-wide significance, and even though 51 were highly suggestive, none of them resided in the NINJ2 gene region from the initial study90.

An Icelandic GWAS, which initial findings were followed up in four separate European cohorts showed that ischemic stroke, and in particular the cardioembolic subtype, was associated to chromosome 4q2591. The association was also significant for atrial fibrillation (AF), a known risk factor for cardioembolic stroke, and has since been replicated in AF cohorts in the United States and Europe92. The associated SNPs in the studies are located close to the paired-like homeodomain 2 (PITX2) gene that in mice is important for the development of the sinoatrial node which functions as a natural pacemaker for the heart93. The Icelandic researchers followed up their initial findings using additional AF cohorts from Europe and could show that a sequence variant in the zinc finger homeobox 3 (ZFHX3) gene was associated with AF as well as ischemic stroke94.

GWAS for traits that share risk factors for stroke or that are stroke risk factors in themselves have shown overlapping results for genes suggested as stroke susceptibility genes, with the exception of the 9p21 region. Analyses have shown that the MTHFR gene is significantly

37

associated with atrial fibrillation95, systolic blood pressure96, and plasma homocysteine levels97. In four separate studies, high density lipoprotein (HDL) levels were associated to the ATP-binding cassette, sub-family A, member 1 (ABCA1 gene)98-101 located in the 9q31.1 region in the vicinity of the stroke susceptibility locus reported by us in paper II.

Material

Family studies

All participating families in papers I-III were identified through the northern Sweden Monitoring of Trends and Determinants in Cardiovascular Disease (MONICA) stroke register. This is a project that was initiated by World Health Organization (WHO) and launched in January 1985 at 39 different locations worldwide. It was centrally terminated in May 1996, but has been prolonged indefinitely at the Umeå center. The register screens all medical records from the Västerbotten and Norrbotten counties in northern Sweden for stroke incidents validating all cases according to MONICA criteria102. All stroke events in the region that concern individuals between 25 and 74 years of age are registered. Validation shows that 96% of all stroke cases in the region are identified in the register103.

Questionnaires regarding family history of stroke were sent to all MONICA registrants that had survived a first ever stroke under the age of 70. Family history of stroke was defined as one additional 1st to 3rd degree relative affected before turning 75. A total of 2.054 questionnaires were sent to individuals registered between January 1st 1985 and May 31st 1996. Of these, 1.603 responded, which correlates to a response rate of 79%. Further, positive family history was reported by 654 individuals, of which 197 reported an affected sibling. In 125 of these families, the affected sibling was alive, and in 16 families more than one affected sibling was alive. Diagnoses of family members were

38

verified in the MONICA register or by hospital records. In the end, 101 families with at least two affected members participated in the study.

Additional families were identified in the MONICA register for individuals registered from June 1996 to November 2001. Questionnaires were distributed to 1.701 individuals with a response rate of 79%. Of these, 59 families with at least two affected 1st degree relatives, and an additional four families with 2nd and 3rd degree relatives willing to participate in the study were identified.

Informed consent was obtained by all participants and blood samples for DNA analysis were collected beginning on April 17th 1997. Diagnoses for all participants were verified according to WHO criteria38, and ischemic stroke cases were subtyped according to TOAST criteria50.

All participating 164 families were genealogically examined to determine geographical origin and possible kinship among families. This was done by a single genealogist using church records, publicly available through the Department of the National Archives104 and private genealogical databases. The genealogical software DISGEN was used to store information regarding participating individuals. We chose to limit the genetic studies to affected probands who were born in the Norrbotten or Västerbotten counties. Further, both parents of an affected proband had to be born in these counties too.

Case-control material

The cases for paper IV were identified using the WHO MONICA register. All cases in the MONICA stroke registry, 15.654 individuals in April 13th 2007, were screened against participants in three population based study cohorts in order to identify collected DNA samples (figure 9). As part of the MONICA register, health surveys has been distributed to randomly selected inhabitants in Norrbotten and Västerbotten at six occasions since 1996105 and today consists of some 7.500 individual biological samples with clinical baseline data.

39

The Västerbotten intervention project (VIP) is a large scale population based screening project that since 1985 has invited all residents of the Västerbotten County to participate in a health survey at their 40th, 50th and 60th birthday. The survey involved tests for blood pressure, anthropometry, glucose tolerance test and blood lipids where measured and blood was asked to be donated to the medical biobank. Today, the VIP consists of samples from more than 80.000 individuals106.

In the Mammary Screening Project (MSP), founded in 1997, all women in Västerbotten in the age of 50–69 years are invited to undergo mammography every two or three years. In conjuncture with this they have been asked to participate in medical research by answering surveys regarding health and allowing access to medical records. Participation rate for the screening has been 85% and 33% of the participants have donated blood to the Medical Biobank for research purposes107.

Samples from about 100.000 Västerbotten residents are stored in the Umeå Medical Biobank106. A total of 1.857 individuals from the MONICA stroke registry had available DNA samples from any of the three cohorts of which 1.331 were diagnosed with ischemic stroke. Controls that

Figure 9. All stroke cases in the MONICA stroke database register were screened against the MONICA health survey material, the Västerbotten Intervention Project (VIP) and the Mammary Screening Project (MSP), all stored in the Umeå Medical Biobank, to identify samples available for the association study.

40

matched the cases for sex, age and place of residence were identified in a 1:1 ratio using the three study projects.

Methods

Genotyping

DNA was purified from 10 ml whole blood or buffy coat according to standard phenol extraction protocols. In the linkage studies an array of evenly spaced microsatellites from the Applied Biosystems (ABI) Prism Linkage Mapping Set v.2.5 HD10 and HD5 was used. A total of 449 microsatellites were used in the analysis described in both paper I and paper II. Microsatellite PCR conditions were performed according to manufacturer recommendations in 96-well format with one CEPH control DNA and blank water control in each plate. In paper I an additional 242 individuals were genotyped in a follow up study using 50 of the original genome wide microsatellite markers, as well as 48 fine mapping markers that were analyzed for all samples from both materials. In paper II, an additional 30 microsatellites were selected for haplotype analysis by using available markers were selected using the Rutgers genetic map, build 34108 and the National Center for Biotechnology Information (NCBI) build 34.3 map109. PCR reactions were run on an ABI Prism 3100 Genetic Analyzer or an ABI 3730XL Genetic Analyzer. Genotypes were analyzed and scored using the GeneMapper Genotyping software version 3.7. To confirm family relationships and to detect genotyping errors all genotypes were checked for mendelian inconsistencies using the software Pedcheck110. Microsatellite marker genotypes inconsistent with mendelian inheritance were deleted.

In paper III, sequence analysis was performed using the resequencing primer kit (RSS000020991) for the coding parts of the TLR4 gene (Applied Biosystems). PCR amplification and purification was performed according to manufacturer protocol. Fragment analysis was done using

41

the ABI 3730XL Genetic Analyzer and sequences were examined using the SeqScape v2.1.1 software (Applied Biosystems).

Subsequent SNP analysis was performed using the TaqMan genotyping platform (Applied Biosystems). SNPs (table 1) were chosen using the SNPper SNP database111. All assays were performed according to protocols supplied by the manufacturer. Genotypes were read using the ABI PRISM 7900HT Sequence Detection System version 2.1 and examined using the GeneMapper Genotyping software version 3.7. All the systems have been supplied by Applied Biosystems.

Gene SNP Position TNFSF15 rs7027251 116592591 rs3810936 116592706 rs6478108 116598524 rs4263839 116606261 rs6478109 116608587 rs1407308 116610044 TLR4 rs1927914 119504546 rs1927911 119509875 rs5030710 119514542 rs4986790 119515123 rs7869402 119517853 rs1554973 119520633

Haplotypes in papers II and III were constructed manually using the Cyrillic version 1.0 software. We manually traced inheritance in all genotyped siblings and children, and in individuals married into the families, if such information was available.

In paper IV, SNPs for association analysis were chosen and ordered using the SNP-browser software from Applied Biosystems (Table 2). The SNPs were run in a SNPlex assay (Applied Biosystems) using a 384-well format according to standard laboratory protocols with each plate containing

Table 1. SNPs chosen for genotyping of the TNFSF15 and TLR4 genes. Chromosomal positions according to NCBI build 36.3.

42

four CEPH control DNA and four blank negative controls. Genotypes were scored using the GeneMapper 4.0 (Applied Biosystems) software. The SNPlex assay is a multiplex analysis that combines PCR and hybridization techniques and can analyze up to 48 SNPs per sample112.

Statistics

In papers I and II, a model-free multi-point model was applied for linkage analysis a using the Allegro software113 with the Spairs scoring function to assess identity by descent sharing among all pairs of affected individuals within families. This is the same scoring function that was used in the Icelandic stroke study. Non-parametric linkage (NPL) Z-scores were converted to allele-sharing LOD scores using the exponential model114. To obtain an overall score, we compromised between weighing the families and weighing the affected pairs equally using the “power 0.5” option in Allegro. Allele frequencies were estimated using all genotyped individuals according to the Merlin algorithm115. In paper I, three different models where used for linkage analysis. In the first model, all types of stroke were considered affected, in the second model ischemic stroke and intracerebral hemorrhage were considered affected and cases with subarachnoid hemorrhage were excluded, and in the

Gene/Region SNP Chr Position 9p21 rs10116277 9 22071397 9p21 rs10757274 9 22086055 9p21 rs1333049 9 22115503 Factor 5 rs6025 1 167785673 MTHFR rs1801131 1 11777063 MTHFR rs1801133 1 11778965 Factor 2 rs1799963 11 46717631 PAI1 rs2227631 7 100556258 PAI1 rs1799889 7 100556430

Table 2. SNPs chosen for association studies in the case-control material. Chromosomal positions according to NCBI build 36.3. Chr refers to chromosome.

43

third model only ischemic cases were analyzed. For paper II, the only model used for genome wide linkage included ischemic stroke and intracerebral hemorrhage, excluding subarachnoid hemorrhage. A separate calculation for linkage in an ischemic stroke model only was performed over the 8.4 Mb identical by descent (IBD) region, but not for the entire genome.

None of the genome wide LOD scores are corrected for multiple testing. To estimate empirical genome wide probability values, simulation studies with 1.000 randomly created datasets under null-hypothesis of no linkage were performed. Estimations showed that for paper I, a LOD score of 2.14 would be expected in one of five GWS analyses, equivalent to a significant LOD score for the studied material size. For paper II, a LOD score of 4.19 equated significant linkage.

In paper II, the difference in frequency of intracerebral hemorrhage between the studied extended pedigree and what was expected extrapolating from the background population, was calculated with Fisher’s exact test using the SPSS version 17.0 (SPSS Inc., Chicago, IL) software.

Association analysis in paper IV was performed using a conditional Cox regression analysis in a model of additive inheritance. SNPs significantly associated to ischemic stroke in the initial analysis were adjusted for known cardiovascular risk factors; elevated blood pressure, diabetes, current smoking, total cholesterol and previous myocardial infarction in a conditional multivariate Cox regression analysis. All p-values presented are nominal and uncorrected for multiple testing. All SNPs were found to be in Hardy-Weinberg equilibrium, calculated using the Haploview version 4.0 software116. The level of linkage disequilibrium and r2-values between SNPs was calculated using the Haploview software.

44

Results

In Paper I we performed a GWS on a family based cohort from northern Sweden to identify novel stroke susceptibility loci. An initial genome wide scan was performed in 56 families containing a total of 376 family members of which 129 individuals had been affected by stroke (Figure 10). An arbitrarily set LOD score of 1.2 was set for regions that merited follow up analysis, which was reached by nine chromosomal regions (table 3) for one, two or all three models used for the calculations.

In a follow up analysis, an additional 53 nuclear families with a total of 242 family members of which 129 individuals had been affected by stroke were added. Fine mapping of the nine chromosomal regions with suggestive linkage was performed in all 109 families. The combined LOD scores from the fine mapping produced LOD scores >1.2 for three regions. The highest LOD score was obtained for chromosome 5q12, a region previously investigated and published117 in response to Icelandic findings relating stroke to the STRK1 locus through linkage analysis30,

Figure 10. GWS results for the initial 53 families. The grey line indicates model I which includes all stroke cases, the dashed line model II for ischemic stroke and intracerebral hemorrhage and the black line model III, ischemic stroke only. The X-axis shows the chromosomes and the Y-axis the allele-sharing LOD scores. The 1.2 LOD score cut off is indicated by the solid black line.

45

13q32 using the model including all stroke types and for 18p11 in the ischemic only model (table 3).

When separate calculations for fine mapping for the initial 56 families was performed LOD scores for three loci (7q35, 14q32 and 20q13) dropped below 1.2. The remaining loci retained LOD scores >1.2 with 5q13 being the highest with 2.19. Calculations for the 53 families in the second cohort showed no evidence of linkage to any of the nine chromosomal regions with allele sharing LOD scores >1.2. Chromosome 13q32 reached a maximum LOD score of 0.33.

LOD score

Cytogenetic Model GWS Fine mapping Follow

up Combined position 56 fam. 53 fam. 109 fam.

1p34 I 2,08 1.32 < 0.00 0.35 II 2,10 1.27 < 0.00 0.60 III 1,25 1.35 < 0.00 0.75

5q13 I 1,22 1.67 0.12 1.46 II 2,03 2.19 0.10 1.71 III 1,82 1.87 0.00 1.08

7q35 I 1,20 1.05 < 0.00 0.43 9q22 I 1,41 1.56 < 0.00 0.12

II 1,43 1.45 < 0.00 0.08 9q34 I 1,50 1.45 < 0.00 0.56

II 1,23 1.09 < 0.00 0.30 13q32 I 1,63 1.36 0.33 1.59 14q32 II 1,25 0.67 0.17 0.79 18p11 I 1,63 1.45 0.00 0.86

II 1,47 1.27 0.06 1.03 III 2,14 1.73 0.16 1.53

20q13 II 1,44 0.98 < 0.00 0.30 III 1,64 1.14 < 0.00 0.45

Table 3. Allele-sharing LOD scores for loci >1.2 in the initial 53 family analysis. Scores are showed for the initial 53 families and follow up 53 families separate as well as both materials combined.

46

In paper II, a subset of seven of the families in paper I were connected through genealogy (figure 11). This extended family had an overrepresentation of intracerebral hemorrhage (40%) compared to observations from the background population (15%) p=0.012.

Genome wide linkage analysis showed significant linkage to chromosome 9q31-33 (figure 11), with a maximum allele-sharing LOD score of 4.66 for marker D9S1776. The model took into account both ischemic stroke and intracerebral hemorrhage, as did model 2 in paper I. No other chromosomal region reached a nominal significant genome wide LOD score. In order to test the 9q31-33 region as a susceptibility locus for the remaining 49 nuclear families of the original material, a separate linkage analysis was performed. This revealed a maximum allele-sharing LOD score of 0.53 for marker D9S164. After analyzing an additional 30 microsatellite markers covering the 26.7 cM 1-LOD drop region surrounding marker D9S1776, the allele-sharing LOD score reached a maximum value of 4.81 over 20 consecutive markers over an 8.4 Mb region (figure 12) a region which was IBD for all affected members of the seven nuclear families.

Figure 11. GWS results for the nine nuclear families in paper II. The X-axis shows the aligned chromosomes and the Y-axis the allele-sharing LOD scores.

47

Haplotypes for the seven nuclear families were constructed using the microsatellite markers analyzed in the 9q31-33 region. This revealed that two of the families (F023 and F028) shared a common haplotype of 8.1 Mb, and that two additional families (F007 and F046) also shared a region of 2.2 Mb of the larger haplotype (Table 4).

In paper III the toll-like receptor 4 (TLR4) and tumor necrosis factor superfamily member 15 (TNFSF15) genes were analyzed as potential candidate genes for stroke in the 9q31-33 region. The coding regions of the TLR4 gene were sequenced to detect functional sequence variants. Two previously known sequence variants were identified, an A>G transition at position 896, resulting in a aspartic acid to glycine shift at amino acid 299 (D299G), and a C>T transition at position 1.196 of the TLR4 gene, which gives a threonine to isoleucine shift at amino acid position 399 (T399I).

In a second step, we analyzed a total of 12 SNPs that covered the TLR4 and TNFSF15 genes, and SNP haplotypes were constructed for all members of the seven nuclear families. This showed that there were four different haplotypes for the TNFSF15 gene and five distinct haplotypes for the TLR4 gene (table 5).

Figure 12. Linkage to chromosome 9q31-33. The grey line shows the initial GWS LOD score, and the black line shows the LOD score after fine-mapping. Microsatellite markers used in genotyping are displayed on top of the figure.

48

Marker Position bp‡ F007 F023 F028

F046

D9S1677 109017145 6 4 1 4

D9S1835 109132523 1 1 1 1

D9S160 109458553 3 1 2 2

D9S1828 110412270 1 2 2 1

D9S1683 110942891 2 3 3 2

D9S1856 112022851 2 2 2 2

D9S930 112315767 4 4 4 4

D9S262 113102310 3 1 1 1

D9S279 113307249 2 3 3 1

D9S289 113495121 2 1 1 3

D9S1824 113971659 1 1 1 3

D9S1855 114590502 1 1 1 3

D9S51 114936627 1 3 3 1

D9S1776 115038988 1 6 6 3

D9S177 115539142 1 2 2 1/2*

D9S170 116146933 1 1 1 2

D9S154 116420400 5 4 4 3

D9S1802 116702926 1 1 1 1

D9S754 117150017 1 1 1 1

D9S1811 117301214 7 1 1 5

D9S1864 117566517 3 4 1 1

D9S762 118175283 1/3* 3 2 1

Table 4. The overlapping haplotype identified for families F007, F023, F028 and F046. Only the chromosomes carrying the common haplotype is displayed for each family. * - Phase is unknown

49

TNFSF15 1 2 3 4 rs3810936 C T C C rs6478108 C T T C rs4263839 G A A G rs6478109 G A A G rs1407308 G T T T TLR4 1 2 3 4 5 rs1927914 A G G G G rs1927911 G A G G A rs5030710 C C C C C rs4986790 A A G A A rs7869402 C C C C C rs1554973 T C C T T

The analysis gave no indication that any of the haplotypes were overrepresented among affected individuals, or that there was evidence of loss of heterozygosity for either of the two investigated genes. It showed instead that the TLR4 gene falls outside the 8.1 Mb common haplotype previously identified in paper II for families F023 and F028, excluding it as a gene responsible for the incidence of stroke in the extended pedigree.

In paper IV, we first ran an association analysis on nine different SNPs for four genes and one locus previously associated with ischemic stroke. This was done in a case-control material of 2.662 individuals matched one-to-one. After initial analysis, two SNPs were significantly associated to stroke, rs1333049 in the 9p21 chromosomal region and rs6025 for Factor 5.

These two SNPs were subsequently adjusted for known risk factors for ischemic stroke; elevated blood pressure (indicated as systolic blood pressure >140 mmHg and/or diastolic blood pressure >90 mmHg and/or anti hypertensive treatment), diabetes (self reported or anti diabetic treatment or by 2 hour oral glucose tolerance test >11.1 mmol/L),

Table 5. The SNP haplotypes for the TNFSF15 and TLR4 genes

50

current smoking, total cholesterol and previous myocardial infarction. After adjustment none of the SNPs were significantly associated to ischemic stroke (table 6).

after adjustment SNP P OR (95% CI) P OR (95% CI)

rs10116277 0.087 1.10 (0.98-1.24) rs10757274 0.131 1.09 (0.97-1.22) rs1333049 0.030 1.13 (1.01-1.27) 0.188 1.09 (0.93-1.22) rs6025 0.025 0.63 (0.43-0.94) 0.273 0.73 (0.76-1.57) rs1801131 0.430 0.95 (0.85-1.07) rs1801133 0.212 1.08 (0.95-1.24) rs1799963 0.284 1.23 (0.84-1.81) rs1799889 0.286 1.06 (0.95-1.19)

Discussion

Studies undertaken on inhabitants from the northern part of Sweden have previously proven successful in elucidating underlying genetic factors for hereditary diseases26, 27. The population is also well defined demographically through church records that have been kept for centuries, and it has experienced rapid expansion historically, increasing from 16.000 inhabitants in the early 18th century to over 300.000 in the beginning of the 20th century118. Since increase in population occurred within growing families while migration was low, the population is genetically homogeneous. There are well-developed medical facilities with accessible medical records for the whole region, and the population has expressed a willingness to participate in scientific research119. We have taken advantage of these conditions in an attempt to identify genetic factors that increase stroke susceptibility.

Table 6. SNPs tested for association to ischemic stroke. Only SNPs nominally significant in the initial analysis were adjusted for stroke risk factors.

51

In Paper I we investigated a genome wide analysis of a family based stroke material would reveal novel or confirm previously identified stroke susceptibility loci. In the initial step, 56 families affected by stroke were analyzed in three separate stroke models with different inclusion criteria for affected individuals. After follow up analysis with an additional fine mapping of nine chromosomal regions together with an additional 53 stroke affected families, three regions still showed allele-sharing LOD scores surpassing the initial 1.2 cut off value: chromosome 13q32 for a model including all types of stroke according to MONICA criteria, and 18p11 for ischemic stroke only. Most interesting, however, was the findings of chromosome 5p12, where the models for all types of stroke and the model excluding SAH reached allele-sharing LOD scores of 1.46 and 1.71 respectively. This is the same region that harbors the STRK1 locus first identified to be associated with common stroke in an Icelandic study in 200230, suggesting PDE4D as the causative gene75.

The addition of 53 stroke affected families did little to increase linkage for the investigated regions. Rather, none of the investigated loci reached allele-sharing LOD scores of more than 0.33, and were less than zero in ten of the 18 calculations. Since the initial 56 families were selected based on their level of informatively, there is a risk that several of the families studied in the follow up analysis showed lesser heritability.

The concept of genetic variation in PDE4D as a risk factor for stroke has been inconclusive. A few studies have been able to replicate the initial findings, but for most populations this has not been possible to replicate. An attempt to replicate the PDE4D find has been made in an material from northern Sweden, but failed to confirm association117. It could be that it is not the PDE4D gene that is responsible for the increase in stroke susceptibility but rather a second gene within or in the vicinity of the STRK1 locus. This idea is supported by the finding in the northern Swedish population whereby the maximum allele-sharing LOD

52

score was found approximately 10 cM distal of the PDE4D gene117. This is a result that should be further investigated.

By connecting families to common ancestry we hoped to identify common stroke susceptibility factors. Genealogy on the initial 56 families from paper I was performed to establish hereditary connections among participating nuclear families. This produced an extended pedigree containing seven of the 56 families connected to a common founder couple over eight generations ago, presented in paper II. A separate GWS and subsequent finemapping for this pedigree revealed a maximum allele-sharing LOD score of 4.81 and a common haplotype in four of the participating families.

This shows the detection of a novel susceptibility locus for common stroke. There are no signs of stroke related syndromes within this extended pedigree and it could be of special importance for hemorrhagic stroke cases which are significantly over represented within the pedigree. This is a small sample of families, but the linkage findings are solid and the haplotype region that was identified contains many candidate genes with plausible functions for vascular disorders. These includes hydroxysteroid dehydrogenase like 2 (HSDL2), orosomucoid 1 and 2 (ORM1, ORM2) and alpha-1-microglobulin/bikunin precursor (AMBP)120-123, proteins involved in lipid handling. TLR4 and prostaglandin reductase 1 (PTGR1) are involved in inflammatory responses122, 123. Lysophosphatidic acid receptor 1 (LPAR1), TNFSF15 and pregnancy associated plasma protein A (PAPPA) are all implicated in vascular regulation124-126. The results of this work support the notion that families and/or genetically related individuals can be an important asset in identifying susceptibility loci for complex disorders and that rare, more deleterious genetic variants that are hypothesized to confer the missing heritability35 can be identified in extended pedigrees.

Following the finds in paper II, we investigated in paper III two candidate genes that were part of a common 8.4 Mb region shared IBD

53

on chromosome 9q31-33 among affected individuals in the extended pedigree. Sequencing of the TLR4 gene and SNP haplotypes produced for the TNFSF15 and TLR4 genes did not reveal a common haplotype that could indicate a founder mutation present in the families or indicate larger deletions. The haplotypes for the TLR4 gene also excluded it from the previously identified common 8.1 Mb haplotype shared between families F023 and F028. However, this chromosomal region still contains a number of genes that are very interesting for further studies.

The most effective way to detect variations that produce stroke susceptibility for the extended pedigree would be to sequence the entire 8 Mb region that is shared IBD among the affected individuals. By doing this, it should be possible to detect one if not several genetic variants that could explain the presence of stroke. Taking it one step further, healthy children that belong to the pedigree could be invited to participate in functional studies. These would be focused on targeting the gene/genes identified by sequencing and measuring their functions in an in vitro system with tissue donated by the participants.

In paper IV the aim was to evaluate the impact of four different genes previously associated with ischemic stroke and a locus with several reports of cardiovascular disease in northern Sweden. This was performed on a material comprising 1.331 individuals affected by ischemic stroke with matched controls. All the individuals resided in the Norrbotten and Västerbotten counties.

The initial significant association to the 9p21 region and F5 could not be supported when adjusted for common risk factors. However, the increased risk initially seen for the rs1333049 SNP was in same range (OR 1.13) as reported in previous publications for cardiovascular diseases in this locus. Further, functional studies have showed that SNPs in the 9p21 region are linked to an increased risk of developing atherosclerosis127, a major risk factor for developing ischemic stroke. We

54

cannot confirm the 9p21 chromosomal region being relevant for ischemic stroke in our population, but increasing the case-control material and the density of analyzed SNPs in the region would be of great interest.

The association to F5 suggested that the rare A-allele was protective for the risk of being affected by stroke. This was an unexpected result since the A-allele was previously shown to be associated with an increased risk of developing ischemic stroke in meta-analysis, and is a confirmed risk factor for venous thromboembolism128.

However, recent studies have not been able to detect association for F5 with an increased risk of developing ischemic stroke129-131. And chances are that reports will emerge of significant association either way if there is a lack of true association. There is at present little evidence that rs6025 is a risk factor for ischemic stroke.

55

Concluding remarks

Despite much effort there is little evidence for defined genes and genetic variants that increase the risk of being affected by stroke. Several loci that confer an increased risk of being affected by other cardiovascular disorders have been successfully mapped with various methods, but initial findings have been difficult to reproduce and the few genome wide association scans that have been performed have not been able to produce consistent results. The challenge now will be to identify the heritability that can be measured but not identified.

We have shown that working with family based materials can be a successful strategy in elucidating underlying genetic mechanisms for complex disorders. We have identified two regions that are linked to an increased risk of being affected with common stroke. Further work should be carried out to identify families with an increased heritability of stroke to replicate the identified loci and attempt to identify novel susceptibility loci. We also need to closely investigate the linked regions on chromosome 5p12 and 9q31-33 to identify and verify the genes and the functionality that cause the stroke cases in the studied families.

What the data further supports is the hypothesis that much of the genetic influence on complex traits is conferred by rare variants with deleterious effects, compared to the common SNPs that have been analyzed so far. As sequencing is becoming more efficient and less costly, large-scale whole genome sequencing projects will likely become as commonplace as GWAS studies are today. In a not so distant future, association studies using whole genome sequencing association studies should be able to analyze all sequence variation in a cohort; polymorphisms as well as CNVs. This should answer questions regarding missing heritability for complex traits. The northern part of Sweden, with its well-characterized population, should be an ideal cohort for such studies.

56

Acknowledgements

For all and everyone who made this thesis possible, thank you:

First of all, my supervisor, Dan Holmberg, a true visionary that kept me along for the ride and let me finish my part of the project. Thanks to you I have traveled around the world and learnt the most amazing, absurd biochemical details. But most importantly, have given me the opportunity to be a part of and learn the importance of science and the scientific method, the greatest human achievement.

To my co-supervisors, Sofie Nilsson-Ardnor, for keeping me organized and on track to getting things done and for being a close friend. I am very grateful for everything. Per-Gunnar Wiklund for sharing his extensive knowledge of stroke, academic writing and for being a rock n’roll-doctor in general. Reunion tour of Paris any time soon?

Birgitta Stegmayr and Kjell Asplund who’s previous work enabled this project and for keeping me along when times were rough. Thank you.

Past and present members of the DH party production group

Sofia Mayans for being the model image of a well organized and structured scientist at the same time with full focus on your personal life. Åsa Larefalk, for the experience and absolute authority at the lab and for being the back bone and fundamental support for the rest of the DH lab. Ann-Sofie Johansson my newfound neighbor – even though I now will have to scratch being Dans’ last student off my CV.

Anna Karin Nilsson, long time third musketeer of the stroke group and a good friend who learnt me a lot about meticulous lab work, accuracy and orientation tactics in general.

Marie Lundholm a free spirited and spontaneous person, don’t change, Elisabeth Einarsdottir, the fiercest geneticist I’ve met, Vincius Motta for bringing samba to the group, Kurt Lackovic and Ingela Wikström best of

57

luck with Elliott, Pia Osterman I’ll see you on the high seas! Maria Lindberg a great lab companion, Susanne Haraldsson taxi supplier at times of need, Sofie Johansson good luck at SLU, Ulrika Norén-Nyström our excellent vocalist, Petter Lindgren statistics provider extraordinaire, Mikael Hedlund database provider extraordinaire, Stefan A Escher for connecting the dots and being a long time friend, Birgitta Berglund for endless support – welcome back, Mia Eriksson for absolute coolness, Anna Löfgren-Burström best of luck with the dentistry.

Other Medical Genetics crew: Monica Holmberg for supplying me with 10+ years of genetic expertise and Elin Larsson and Angelica Nordin for keeping me updated on current popular culture. If it wasn’t for you I’d never know what was on TV. Hawaii, anyone?

To the medical genetics hang arounds: Urban Hellman for supporting good taste in life, Malin Olsson for making me laugh. Alot. Martin Johansson, my connection to the dark side, Nina Gennerbäck for the lovely Östgötska – my favorite dialect, Nina Norgren with the wicked dance steps, Lisbeth Ärlestig for great travel stories, Heidi Kokkonen, I’m so sorry about the baby gym, I’ll get you another, Linda Köhn, congratulations to the Botniabanan! Stefan K Nilsson for making me think twice, Emma Andersson for almost inspiring me to pick up my running shoes, Mikael Larsson for life saving computer support, Solveig Linghult for standing your ground. Kristina Lejon, a seriously inspirational teacher and Mia Sundström for great JC discussions, Lars Bläckberg – legendary quiz master. Basically to all you at the fika-table and vetgirigligorna, I will miss you all.

Karin Nylander, Terry Persson and Åsa Lundsten for keeping track of me and my papers.

Geneticists of past and present: Lisbet Lind good luck with you new career as a Doctor-Doctor, Lena Wennerstrand my very own personal twin, Melker Häggbom-Klingberg, where ever did that Jim Beam go?,

58

Jenni Jonasson for sharing her twin experience, Kristina Cederquist for supervision when I needed it, Anna Carlsson, part of the Holmsund crew, Irina Golovleva for believing in me - and we actually had a pretty good time in Cleveland. Björn-Anders Jonsson for the squash sparring, Jan Larsson with his eye for fly eyes , Anna Birve for being a long time real friend , Anssi Saura for inspiring me to pursuit the field of genetics, and all personnel at the Clinical Genetics laboratory.

Per Stenberg and Lars Nilsson for inspirational discussions at the scientific lunches and for flying the banner of secularism and skepticism. Keep up the good work!

Gösta Holmgren, for bringing me into the field of medical genetics, in memoriam.

Karin & Staffan, KG & Marita, Lotta & Giancarlo, Sigge och Elisabeth för ert stöd och generositet och för att ni alltid ställer upp när den flygande cirkusen anländer. Evelina, Erika och Klara för hjälp, tips och idéer under arbetets gång och för att ni är världens bästa syskon.

Och till sist men allra mest, Stella, Valdemar och Elmer, mina fantastiska barn. Tack för att ni stått ut med en förvirrad och frånvarande far. Och till Edit, utan dig hade inget av det här varit möjligt. Tack för att du varit min största supporter, en klippa för hela familjen och mitt livs kärlek.

And finally, to unnamed but certainly not forgotten family and friends.

59

References

1. Pearson, H. Genetics: What is a gene? Nature 441, 398-401 (2006).

2. Crick, F. Central Dogma of Molecular Biology. Nature 227, 561-563 (1970).

3. http://vega.sanger.ac.uk/Homo_sapiens/Info/Index. (2010). 4. Tennyson, C. N., Klamut, H. J. & Worton, R. G. The human

dystrophin gene requires 16 hours to be transcribed and is cotranscriptionally spliced. Nat Genet 9, 184-190 (1995).

5. Koenig, M., Monaco, A. P. & Kunkel, L. M. The complete sequence of dystrophin predicts a rod-shaped cytoskeletal protein. Cell 53, 219-228 (1988).

6. Grosso, A. R. et al. Tissue-specific splicing factor gene expression signatures. Nucl. Acids Res. 36, 4823-4832 (2008).

7. Spilianakis, C. G., Lalioti, M. D., Town, T., Lee, G. R. & Flavell, R. A. Interchromosomal associations between alternatively expressed loci. Nature 435, 637-645 (2005).

8. Mercer, T. R., Dinger, M. E. & Mattick, J. S. Long non-coding RNAs: insights into functions. Nat Rev Genet 10, 155-159 (2009).

9. McVean, G. in 59th annual meeting of the American Society of Human Genetics (Honolulu, USA, 2009).

10. Sebat, J. et al. Large-Scale Copy Number Polymorphism in the Human Genome. Science 305, 525-528 (2004).

11. Bird, A. Perceptions of epigenetics. Nature 447, 396-398 (2007). 12. Kim, U.-k. et al. Positional Cloning of the Human Quantitative

Trait Locus Underlying Taste Sensitivity to Phenylthiocarbamide. Science 299, 1221-1225 (2003).

13. Winderickx, J. et al. Defective colour vision associated with a missense mutation in the human green visual pigment gene. Nat Genet 1, 251-256 (1992).

14. Robertson, J. H. & Trueman, R. G. Combined Hemophilia and Christmas Disease. Blood 24, 281-288 (1964).

15. Johns, D. R. Paternal transmission of mitochondrial DNA is (fortunately) rare. Annals of Neurology 54, 422-424 (2003).

16. Edmond, J. C. Mitochondrial Disorders. International Ophthalmology Clinics 49, 27-33 (2009).

60

17. Meschia, J. F., Brott, T. G. & Brown, R. D. Genetics of Cerebrovascular Disorders. Mayo Clinic Proceedings 80, 122-132 (2005).

18. Visscher, P. M. Sizing up human height variation. Nat Genet 40, 489-490 (2008).

19. Lunetta, K. et al. Genetic correlates of longevity and selected age-related phenotypes: a genome-wide association study in the Framingham Study. BMC Medical Genetics 8, S13 (2007).

20. Kidd, K. K., Pakstis, A. J., Speed, W. C. & Kidd, J. R. Understanding Human DNA Sequence Variation. J Hered 95, 406-420 (2004).

21. Nachman, M. W. & Crowell, S. L. Estimate of the mutation rate per nucleotide in humans. Genetics 156, 297-304 (2000).

22. Chakraborty, R. & Hanis, C. L. Non-random sampling in human genetics: estimation of familial correlations, model testing, and interpretation. Stat Med 6, 629-646 (1987).

23. Walker, F. O. Huntington's disease. The Lancet 369, 218-228 (2007).

24. Verkerk, A. J. M. H. et al. Identification of a gene (FMR-1) containing a CGG repeat coincident with a breakpoint cluster region exhibiting length variation in fragile X syndrome. Cell 65, 905-914 (1991).

25. Lander, E. S. & Schork, N. J. Genetic dissection of complex traits. Science 265, 2037-2048 (1994).

26. Petrukhin, K. et al. Identification of the gene responsible for Best macular dystrophy. Nat Genet 19, 241-247 (1998).

27. Carlsson, A. et al. Identification of a susceptibility locus for migraine with and without aura on 6p12.2-p21.1. Neurology 59, 1804-1807 (2002).

28. Einarsdottir, E. et al. A mutation in the nerve growth factor beta gene (NGFB) causes loss of pain perception. Hum Mol Genet 13, 799-805 (2004).

29. Venken, T. et al. Genomewide scan for affective disorder susceptibility Loci in families of a northern Swedish isolated population. Am J Hum Genet 76, 237-248 (2005).

30. Gretarsdottir, S. et al. Localization of a susceptibility gene for common forms of stroke to 5q12. Am J Hum Genet 70, 593-603 (2002).

61

31. Broeckel, U. et al. A comprehensive linkage analysis for myocardial infarction and its related risk factors. Nat Genet 30, 210-214 (2002).

32. Helgadottir, A. et al. The gene encoding 5-lipoxygenase activating protein confers risk of myocardial infarction and stroke. Nat Genet 36, 233-239 (2004).

33. Lewis, C. M. Genetic association studies: Design, analysis and interpretation. Brief Bioinform 3, 146-153 (2002).

34. Hindorff LA, J. H., Mehta JP, and Manolio TA. A Catalog of Published Genome-Wide Association Studies. Available at: www.genome.gov/gwastudies. (2010).

35. Manolio, T. A. et al. Finding the missing heritability of complex diseases. Nature 461, 747-753 (2009).

36. Mackay, J. & Mensah, G. A. The atlas of heart disease and stroke. (2004).

37. O'Brien, J. T. et al. Vascular cognitive impairment. The Lancet Neurology 2, 89-98 (2003).

38. The World Health Organization MONICA Project (monitoring trends and determinants in cardiovascular disease): a major international collaboration. WHO MONICA Project Principal Investigators. J Clin Epidemiol 41, 105-114 (1988).

39. Feigin, V. L. Stroke epidemiology in the developing world. The Lancet 365, 2160-2161 (2005).

40. Ghatnekar, O. et al. Cost of stroke in Sweden: An incidence estimate. International Journal of Technology Assessment in Health Care 20, 375-380 (2004).

41. Warlow, C., Dennis, M. S., van Gijn, J., Hankey, G. J. & Sandercock, P. A. G. Stroke: A practical guide to management. 223-300. (Oxford: Blackwell Science, 2003).

42. Zhang, L. F. et al. Proportion of different subtypes of stroke in China. Stroke 34, 2091-2096 (2003).

43. Fullerton, H. J., Wu, Y. W., Zhao, S. & Johnston, S. C. Risk of stroke in children: Ethnic and gender disparities. Neurology 61, 189-194 (2003).

44. Stansbury, J. P., Jia, H., Williams, L. S., Vogel, W. B. & Duncan, P. W. Ethnic Disparities in Stroke: Epidemiology, Acute Care, and Postacute Outcomes. Stroke 36, 374-386 (2005).

62

45. Alexander, R. W. Hypertension and the Pathogenesis of Atherosclerosis: Oxidative Stress and the Mediation of Arterial Inflammatory Response: A New Perspective. Hypertension 25, 155-161 (1995).

46. Quinn, M. T., Parthasarathy, S., Fong, L. G. & Steinberg, D. Oxidatively modified low density lipoproteins: a potential role in recruitment and retention of monocyte/macrophages during atherogenesis. Proc Natl Acad Sci U S A 84(9), 2995-2998 (1987).

47. Belland, R. J., Ouellette, S. P., Gieffers, J. & Byrne, G. I. Chlamydia pneumoniae and atherosclerosis. Cellular Microbiology 6, 117-127 (2004).

48. Ross, R. Atherosclerosis -- An Inflammatory Disease. N Engl J Med 340, 115-126 (1999).

49. Stegmayr, B., Lundberg, V. & Asplund, K. The events registration and survey procedures in the Northern Sweden MONICA Project. Scand J Public Health Suppl 31, 9-17 (2003).

50. Adams, H. P., Jr. et al. Classification of subtype of acute ischemic stroke. Definitions for use in a multicenter clinical trial. TOAST. Trial of Org 10172 in Acute Stroke Treatment. Stroke 24, 35-41 (1993).

51. Easton, J. D. et al. Definition and Evaluation of Transient Ischemic Attack: A Scientific Statement for Healthcare Professionals From the American Heart Association/American Stroke Association Stroke Council; Council on Cardiovascular Surgery and Anesthesia; Council on Cardiovascular Radiology and Intervention; Council on Cardiovascular Nursing; and the Interdisciplinary Council on Peripheral Vascular Disease: The American Academy of Neurology affirms the value of this statement as an educational tool for neurologists. Stroke 40, 2276-2293 (2009).

52. Qureshi, A. I. et al. Spontaneous intracerebral hemorrhage. N Engl J Med 344, 1450-1460 (2001).

53. Guercini, F., Acciarresi, M., Agnelli, G. & Paciaroni, M. Cryptogenic stroke: time to determine aetiology. Journal of Thrombosis and Haemostasis 6, 549-554 (2008).

54. Goldstein, L. B. et al. Primary prevention of ischemic stroke: a guideline from the American Heart Association/American Stroke Association Stroke Council: cosponsored by the Atherosclerotic

63

Peripheral Vascular Disease Interdisciplinary Working Group; Cardiovascular Nursing Council; Clinical Cardiology Council; Nutrition, Physical Activity, and Metabolism Council; and the Quality of Care and Outcomes Research Interdisciplinary Working Group. Circulation 113, e873-e923 (2006).

55. Wolfe, C. D. The impact of stroke. Br Med Bull 56, 275-286 (2000).

56. Brown, R. D., Whisnant, J. P., Sicks, J. D., O'Fallon, W. M. & Wiebers, D. O. Stroke incidence, prevalence, and survival: secular trends in Rochester, Minnesota, through 1989. Stroke 27, 373-380 (1996).

57. Chobanian, A. V. et al. The Seventh Report of the Joint National Committee on Prevention, Detection, Evaluation, and Treatment of High Blood Pressure: The JNC 7 Report. JAMA 289, 2560-2571 (2003).

58. Kannel, W. B. & McGee, D. L. Diabetes and Cardiovascular Disease: The Framingham Study. JAMA 241, 2035-2038 (1979).

59. Sacco, R. L. et al. Leisure-Time Physical Activity and Ischemic Stroke Risk: The Northern Manhattan Stroke Study. Stroke 29, 380-387 (1998).

60. Wolf, P. A., D'Agostino, R. B., Kannel, W. B., Bonita, R. & Belanger, A. J. Cigarette Smoking as a Risk Factor for Stroke: The Framingham Study. JAMA 259, 1025-1029 (1988).

61. Kiely, D. K., Wolf, P. A., Cupples, L. A., Beiser, A. S. & Myers, R. H. Familial aggregation of stroke. The Framingham Study. Stroke 24, 1366-1371 (1993).

62. Meschia, J. F. & Brown, R. D. Stroke Genetics. 53-68. (ed. Markus, H.) (Oxford University Press, 2003).

63. Bak, S., Gaist, D., Sindrup, S. H., Skytthe, A. & Christensen, K. Genetic liability in stroke: a long-term follow-up study of Danish twins. Stroke 33, 769-774 (2002).

64. Brass, L. M., Isaacsohn, J. L., Merikangas, K. R. & Robinette, C. D. A study of twins and stroke. Stroke 23, 221-223 (1992).

65. de Faire, U., Friberg, L. & Lundman, T. Concordance for mortality with special reference to ischaemic heart disease and cerebrovascular disease. A study on the Swedish Twin Registry. Prev Med 4, 509-517 (1975).

64

66. Joutel, A. et al. Notch3 mutations in CADASIL, a hereditary adult-onset condition causing stroke and dementia. Nature 383, 707-710 (1996).

67. Chabriat, H., Joutel, A., Dichgans, M., Tournier-Lasserve, E. & Bousser, M.-G. CADASIL. The Lancet Neurology 8, 643-653 (2009).

68. Joutel, A. et al. The ectodomain of the Notch3 receptor accumulates within the cerebrovasculature of CADASIL patients. The Journal of Clinical Investigation 105, 597-605 (2000).

69. Tromp, G., Kuivaniemi, H., Shikata, H. & Prockop, D. J. A single base mutation that substitutes serine for glycine 790 of the alpha 1 (III) chain of type III procollagen exposes an arginine and causes Ehlers-Danlos syndrome IV. Journal of Biological Chemistry 264, 1349-1352 (1989).

70. Dietz, H. C. et al. Marfan syndrome caused by a recurrent de novo missense mutation in the fibrillin gene. Nature 352, 337-339 (1991).

71. Serjeant, G. R. Sickle-cell disease. The Lancet 350, 725-730 (1997).

72. Castano, E. M. & Frangione, B. Human amyloidosis, Alzheimer disease and related disorders. Lab Invest 58, 122-132 (1988).

73. Haan, J. Stroke Genetics. 223-258. (ed. Markus, H.) (Oxford University Press, 2003).

74. Abrahamson, M., Jonsdottir, S., Olafsson, I., Jensson, O. & Grubb, A. Hereditary cystatin C amyloid angiopathy: identification of the disease-causing mutation and specific diagnosis by polymerase chain reaction based analysis. Hum Genet 89, 377-380 (1992).

75. Gretarsdottir, S. et al. The gene encoding phosphodiesterase 4D confers risk of ischemic stroke. Nat Genet 35, 131-138 (2003).

76. Nilsson-Ardnor, S. et al. Genome-wide linkage scan of common stroke in families from northern Sweden. Stroke 38, 34-40 (2007).

77. Holloway, J. W., Barton, S. J., Holgate, S. T., Rose-Zerilli, M. J. & Sayers, I. The role of LTA4H and ALOX5AP polymorphism in asthma and allergy susceptibility. Allergy 63, 1046-1053 (2008).

78. Anjana, M. & Subhash, K. Stroke genetics - focus on PDE4D gene. International Journal of Stroke 3, 188-192 (2008).

65

79. Zintzaras, E., Rodopoulou, P. & Sakellaridis, N. Variants of the Arachidonate 5-Lipoxygenase-Activating Protein (ALOX5AP) Gene and Risk of Stroke: A HuGE Gene-Disease Association Review and Meta-Analysis. Am. J. Epidemiol. 169, 523-532 (2009).

80. Casas, J. P., Hingorani, A. D., Bautista, L. E. & Sharma, P. Meta-analysis of genetic studies in ischemic stroke: thirty-two genes involving approximately 18,000 cases and 58,000 controls. Arch Neurol 61, 1652-1661 (2004).

81. Wiklund, P. G. et al. Plasminogen activator inhibitor-1 4G/5G polymorphism and risk of stroke: replicated findings in two nested case-control studies based on independent cohorts. Stroke 36, 1661-1665 (2005).

82. van Guelpen, B. et al. Folate, Vitamin B12, and Risk of Ischemic and Hemorrhagic Stroke: A Prospective, Nested Case-Referent Study of Plasma Concentrations and Dietary Intake. Stroke 36, 1426-1431 (2005).

83. Matarin, M. et al. A genome-wide genotyping study in patients with ischaemic stroke: initial analysis and data release. Lancet Neurol 6, 414-420 (2007).

84. Ikram, M. A. et al. Genomewide Association Studies of Stroke. N Engl J Med 360, 1718-1728 (2009).

85. Genome-wide association study of 14,000 cases of seven common diseases and 3,000 shared controls. 447, 661-678 (2007).

86. Helgadottir, A. et al. A Common Variant on Chromosome 9p21 Affects the Risk of Myocardial Infarction. Science 316, 1491-1493 (2007).

87. McPherson, R. et al. A Common Allele on Chromosome 9 Associated with Coronary Heart Disease. Science 316, 1488-1491 (2007).

88. Samani, N. J. et al. Genomewide Association Analysis of Coronary Artery Disease. N Engl J Med 357, 443-453 (2007).

89. Matarin, M. et al. Whole Genome Analyses Suggest Ischemic Stroke and Heart Disease Share an Association With Polymorphisms on Chromosome 9p21. Stroke 39, 1586-1589 (2008).

66

90. Debette, S. et al. Genome-Wide Association Studies of MRI-Defined Brain Infarcts: Meta-Analysis From the CHARGE Consortium. Stroke 41, 210-217 (2010).

91. Gretarsdottir, S. et al. Risk variants for atrial fibrillation on chromosome 4q25 associate with ischemic stroke. Annals of Neurology 64, 402-409 (2008).

92. Kääb, S. et al. Large scale replication and meta-analysis of variants on chromosome 4q25 associated with atrial fibrillation. Eur Heart J 30, 813-819 (2009).

93. Mommersteeg, M. T. M. et al. Molecular Pathway for the Localized Formation of the Sinoatrial Node. Circ Res 100, 354-362 (2007).

94. Gudbjartsson, D. F. et al. A sequence variant in ZFHX3 on 16q22 associates with atrial fibrillation and ischemic stroke. Nat Genet 41, 876-878 (2009).

95. Benjamin, E. J. et al. Variants in ZFHX3 are associated with atrial fibrillation in individuals of European ancestry. Nat Genet 41, 879-881 (2009).

96. Newton-Cheh, C. et al. Genome-wide association study identifies eight loci associated with blood pressure. Nat Genet 41, 666-676 (2009).

97. Pare, G. et al. Novel Associations of CPS1, MUT, NOX4, and DPEP1 With Plasma Homocysteine in a Healthy Population: A Genome-Wide Evaluation of 13 974 Participants in the Women's Genome Health Study. Circ Cardiovasc Genet 2, 142-150 (2009).

98. Aulchenko, Y. S. et al. Loci influencing lipid levels and coronary heart disease risk in 16 European population cohorts. Nat Genet 41, 47-55 (2009).

99. Kathiresan, S. et al. Common variants at 30 loci contribute to polygenic dyslipidemia. Nat Genet 41, 56-65 (2009).

100. Kathiresan, S. et al. Six new loci associated with blood low-density lipoprotein cholesterol, high-density lipoprotein cholesterol or triglycerides in humans. Nat Genet 40, 189-197 (2008).

101. Willer, C. J. et al. Newly identified loci that influence lipid concentrations and risk of coronary artery disease. Nat Genet 40, 161-169 (2008).

67

102. Asplund, K., Tuomilehto, J., Stegmayr, B., Wester, P. O. & Tunstall-Pedoe, H. Diagnostic criteria and quality control of the registration of stroke events in the MONICA project. Acta Med Scand Suppl 728, 26-39 (1988).

103. Stegmayr, B. & Asplund, K. Measuring stroke in the population: quality of routine statistics in comparison with a population-based stroke registry. Neuroepidemiology 11, 204-213 (1992).

104. http://www.svar.ra.se. 105. Brännström, I., Rosén, M., Wall, S. & Weinehall, L. Local health

planning and intervention--the case of a Swedish municipality. Scand J Prim Health Care Suppl. 1, 57-64 (1988).

106. http://www9.umu.se/phmed/naringsforskning/Medical_ Biobank/indexB.htm. (2010).

107. Hultén, K. et al. An incident case-referent study on plasma enterolactone and breast cancer risk. European Journal of Nutrition 41, 168-176 (2002).

108. http://compgen.rutgers.edu/maps/. (2010). 109. http://www.ncbi.nlm.nih.gov/. (2010). 110. O'Connell, J. R. & Weeks, D. E. PedCheck: a program for

identification of genotype incompatibilities in linkage analysis. Am J Hum Genet 63, 259-266 (1998).

111. http://snpper.chip.org. (2008). 112. Tobler, A. R. et al. The SNPlex genotyping system: a flexible and

scalable platform for SNP genotyping. J Biomol Tech 16, 398-406 (2005).

113. Gudbjartsson, D. F., Jonasson, K., Frigge, M. L. & Kong, A. Allegro, a new computer program for multipoint linkage analysis. Nat Genet 25, 12-13 (2000).

114. Kong, A. & Cox, N. J. Allele-sharing models: LOD scores and accurate linkage tests. Am J Hum Genet 61, 1179-1188 (1997).

115. Abecasis, G. R., Cherny, S. S., Cookson, W. O. & Cardon, L. R. Merlin--rapid analysis of dense genetic maps using sparse gene flow trees. Nat Genet 30, 97-101 (2002).

116. Barrett, J. C., Fry, B., Maller, J. & Daly, M. J. Haploview: analysis and visualization of LD and haplotype maps. Bioinformatics 21, 263-265 (2005).

68

117. Nilsson-Ardnor, S. et al. Linkage of ischemic stroke to the PDE4D region on 5q in a Swedish population. Stroke 36, 1666-1671 (2005).

118. http://www.ddb.umu.se. (2010). 119. Stegmayr, B. & Asplund, K. Informed consent for genetic

research on blood stored for more than a decade: a population based study. Bmj 325, 634-635 (2002).

120. Budai, L. et al. Investigation of genetic variants of alpha-1 acid glycoprotein by ultra-performance liquid chromatography-mass spectrometry. Anal Bioanal Chem 393, 991-998 (2009).

121. Kanayama, S. et al. Molecular structure and function analysis of bikunin on down-regulation of tumor necrosis factor-alpha expression in activated neutrophils. Cytokine 42, 191-197 (2008).

122. Di Gennaro, A. et al. Leukotriene B4-induced changes in vascular permeability are mediated by neutrophil release of heparin-binding protein (HBP/CAP37/azurocidin). Faseb J 23, 1750-1757 (2009).

123. Lin, Y. C. et al. Toll-like receptor 4 gene C119A but not Asp299Gly polymorphism is associated with ischemic stroke among ethnic Chinese in Taiwan. Atherosclerosis 180, 305-309 (2005).

124. Beaudeux, J. L. et al. Serum plasma pregnancy-associated protein A: a potential marker of echogenic carotid atherosclerotic plaques in asymptomatic hyperlipidemic subjects at high cardiovascular risk. Arterioscler Thromb Vasc Biol 23, e7-e10 (2003).

125. Hou, W., Medynski, D., Wu, S., Lin, X. & Li, L. Y. VEGI-192, a new isoform of TNFSF15, specifically eliminates tumor vascular endothelial cells and suppresses tumor growth. Clin Cancer Res 11, 5595-5602 (2005).

126. Tager, A. M. et al. The lysophosphatidic acid receptor LPA1 links pulmonary fibrosis to lung injury by mediating fibroblast recruitment and vascular leak. Nat Med 14, 45-54 (2008).

127. Liu, Y. et al. INK4/ARF transcript expression is associated with chromosome 9p21 variants linked to atherosclerosis. PLoS One 4, e5027 (2009).

69

128. Gohil, R., Peck, G. & Sharma, P. The genetics of venous thromboembolism. A meta-analysis involving approximately 120,000 cases and 180,000 controls. Thromb Haemost 102, 360-370 (2009).

129. Berger, K. et al. The glu298asp polymorphism in the nitric oxide synthase 3 gene is associated with the risk of ischemic stroke in two large independent case–control studies. Human Genetics 121, 169-178 (2007).

130. Wang, X. et al. A Meta-Analysis of Candidate Gene Polymorphisms and Ischemic Stroke in 6 Study Populations: Association of Lymphotoxin-Alpha in Nonhypertensive Patients. Stroke 40, 683-695 (2009).

131. Lalouschek, W. et al. Candidate Genetic Risk Factors of Stroke: Results of a Multilocus Genotyping Assay. Clin Chem 53, 600-605 (2007).

70

Publications and manuscripts