UvA-DARE (Digital Academic Repository)
Genetic modifiers in familial cardiac rhythm disorders
Kolder, I.C.R.M.
Publication date 2012 Document Version Final published version
Link to publication
Citation for published version (APA): Kolder, I. C. R. M. (2012). Genetic modifiers in familial cardiac rhythm disorders.
General rights It is not permitted to download or to forward/distribute the text or part of it without the consent of the author(s) and/or copyright holder(s), other than for strictly personal, individual use, unless the work is under an open content license (like Creative Commons).
Disclaimer/Complaints regulations If you believe that digital publication of certain material infringes any of your rights or (privacy) interests, please let the Library know, stating your reasons. In case of a legitimate complaint, the Library will make the material inaccessible and/or remove it from the website. Please Ask the Library: https://uba.uva.nl/en/contact, or a letter to: Library of the University of Amsterdam, Secretariat, Singel 425, 1012 WP Amsterdam, The Netherlands. You will be contacted as soon as possible.
UvA-DARE is a service provided by the library of the University of Amsterdam (https://dare.uva.nl)
Download date:01 Oct 2021
Genetic modifiers in familial cardiac rhythm disorders
-Iris Kolder - © 2012 by Iris Kolder Genetic modifiers in familial cardiac rhythm disorders Iris C.R.M. Kolder / University of Amsterdam, 2012. Thesis
Printed by Ipskamp Drukkers B.V. Cover design: Manuel Van Der Graaf ISBN: 978-94-6191-336-4
No part of this thesis may be reproduced, stored in a retrieval system, or transmitted in any form or by any means without permission of the author Genetic modifiers in familial cardiac rhythm disorders
ACADEMISCH PROEFSCHRIFT
ter verkrijging van de graad van doctor
aan de Universiteit van Amsterdam
op gezag van de Rector Magnificus
prof. dr. D.C. van den Boom
ten overstaan van een door het college voor promoties ingestelde commissie, in het openbaar te verdedigen in de Aula der Universiteit.
op Vrijdag 29 Juni 2012, te 13:00 uur
door
Iris Cornelia Roelofje Martine Kolder
geboren te Leiden Promotor: Prof. Dr. A.A.M. Wilde en Prof. Dr. A.H. Zwinderman
Copromotor: Prof. Dr. C.R. Bezzina en Dr. M.W.T. Tanck
Overige leden: Prof. Dr. F. Baas Prof.Dr. R.N.W. Hauer Prof. Dr. J.J.P. Kastelein Prof. Dr. H. Meijers-Heijboer Dr. M.M.A.M. Mannens Dr. J.J. Schott
Faculteit der Geneeskunde
The research described in this thesis was carried out in the department of Clinical Epidemiology, Biostatistics and Bioinformatics & The Heart Failure Research Center of the Academic Medical Center, Amsterdam.
Financial support by the Dutch Heart Foundation for the publication of this thesis is gratefully acknowledged.
Additional financial support was generously supplied by Stichting tot Bevordering van de Klinische Epidemiologie Amsterdam; Heart Failure Research Center; Bayer B.V.; Merck Sharpe & Dohme B.V. and the University of Amsterdam Para mis padres
Voor mijn ouders
Content
Chapter 1 General introduction 9
Chapter 2 Common genetic variation modulating cardiac 15 ECG parameters and susceptibility to Sudden cardiac death
Chapter 3 The role of renin-angiotensin-aldosterone system 45 polymorphisms in phenotypic expression of MYBPC3-related hypertrophic cardiomyopathy
Chapter 4 Genetic Modifiers of disease expression in 61 patients with Long QT Syndrome Type 2
Chapter 5 Identification of RCAN1 as a Genetic Modifier 79 of Atrio-Ventricular Conduction in the Setting of Cardiac Sodium Channel Disease
Chapter 6 Family-based genome-wide association analysis 105 for the identification of genetic modifiers of heart rate and electrocardiographic indices of conduction and repolarization in a large Dutch family with a mutation in SCN5A
Chapter 7 General discussion and future perspective 131
Summary 139
Samenvatting 145
Dankwoord 151
Chapter 1
General introduction Chapter 1
Introduction Sudden cardiac death (SCD) is one of the most prevalent causes of death in Western societies. It underlies 20% of total mortality, and 50% of cardiovascular mortality 1. In young individuals (below 40 years of age) SCD often occurs in the setting of disorders displaying Mendelian inheritance 2, with the cardiomyopathies 3 and primary electrical disorders 4 being the most prevalent. Here, the inheritance of very rare genetic variants with large effects potentially increases risk for SCD substantially5. The primary electrical disorders have been linked primarily to mutations in genes encoding ion channel subunits or their interacting proteins (Figure 1)4. On the other hand, the cardiomyopathies are caused by mutations affecting genes coding for the contractile apparatus and structural components of the cardiomyocyte such as the sarcomere and desmosomes 6 .
Genotype-phenotype studies in these disorders have clearly established that they are not spared from the phenomena of reduced penetrance and variable expression typical of Mendelian diseases 7 . For instance, in the primary arrhythmia syndromes, extensive variability in clinical manifestations is often observed among family members carrying an identical ion channel gene mutation, with some individuals exhibiting overt abnormalities on the electrocardiogram (ECG) and suffering potentially fatal arrhythmias, whereas others do not display any ECG changes and do not develop rhythm disturbances throughout life. Probands and families with these Mendelian disorders, harboring known disease-causing mutations, likely provide a permissive, genetically sensitized setting for the identification of novel genes and pathways modulating cardiac (electrical) function.
Focus of this dissertation In this thesis we employ the phenotypic variability evidenced among probands and their relatives with Mendelian cardiac disorders to identify genetic modifiers of disease expression. We focused on two distinct groups of disorders associated with increased risk of SCD, namely the primary electrical disorders (Long QT Syndrome, Brugada Syndrome, Conduction Disease) and hypertrophic cardiomyopathy (HCM). The aim of this thesis was to identify such genetic modifiers using both linkage and (family based) association analyses. Both a candidate SNP / gene approach as well as a genome-wide unbiased approach were used in the study of common genetic variants as possible modifiers of disease severity.
In chapter 2, we reviewed the available literature on the genetic and allelic architecture of SCD. In this review we focused on the common genetic variation that has been recently identified through genome-wide association studies to modulate risk of SCD and
10 General introduction
1
Figure 1 | A schematic representation of a cardiomyocyte depicting genes encoding channel subunits and interacting proteins involved in the primary electrical disorders or in cardiac electrical function. to modulate heart rate and ECG indices of conduction (PR-interval, QRS-duration) and repolarization (QTc-interval) as intermediate phenotypes of SCD.
In chapter 3 we investigated the role of five common candidate SNPs in the renin- angiotensin-aldosterone system in families with HCM who carried one of three functionally-equivalent mutations in the MYBPC3 gene. These SNPs were previously suggested to modify the extent of hypertrophy in HCM.
In chapters 4-6, we focused on genetic modifiers of primary electrical disease. In chapter 4 we studied a large set of individuals (probands and, where available, their family- members) carrying a mutation in the KCNH2 gene and presenting clinically with Long QT syndrome type 2. Here we comprehensively investigated the effect of haplotype-tagging SNPs in and around 18 candidate genes on the QTc-interval. In this analysis, for the first time we took the effect ofKCNH2 mutation type and location in our analysis for modifiers of QTc-interval.
In the last two chapters we studied a very large Dutch kindred with the SCN5A mutation 1795insD. An extensive genealogical search allowed us to trace this family back to the eighteenth century, enabling the construction of a highly extended pedigree. Individuals
11 Chapter 1 in this kindred present with manifestations of Long QT syndrome, Brugada syndrome and progressive conduction disease occurring either in isolation or in combinations thereof. In chapter 5, we performed linkage and association analysis with heart rate and ECG indices of conduction and repolarization using haplotype-tagging SNPs in and around 18 candidate genes. These genetic studies pointed us to the calcineurin/Nfat pathway as a possible modifier of the PR-interval in the setting of sodium channelopathy. We subsequently provided further insight into the possible role of this pathway by conducting a series of functional studies in mice that are knock-in for the homologousScn5a mutation (Scn5a1798insD/+ mice). Finally, we performed a genome-wide association study (GWAs) in this family (chapter6) uncovering novel interesting candidate genes that can provide insight into novel pathways regulating heart rate and the cardiac conduction and repolarization processes.
12 General introduction
Reference List 1. Myerburg,R.J. & Castellanos,A. Cardiac arrest and sudden cardiac death. in Braunwald’s Heart Disease: A Textbook of Cardiovascular Medicine (eds. Libby,P., Bonow,R.O., Mann,D.L. & Zipes,D.P.) 933- 1 974 (Elsevier, Oxford, UK, 2007). 2. van der Werf,C., van Langen,I.M., & Wilde,A.A. Sudden death in the young: what do we know about it and how to prevent? Circ. Arrhythm. Electrophysiol. 3, 96-104 (2010). 3. Watkins,H., Ashrafian,H., & Redwood,C. Inherited cardiomyopathies. N. Engl. J. Med. 364, 1643-1656 (2011). 4. Wilde,A.A. & Bezzina,C.R. Genetics of cardiac arrhythmias. Heart 91, 1352-1358 (2005). 5. Manolio,T.A. et al. Finding the missing heritability of complex diseases. Nature 461, 747-753 (2009). 6. Watkins,H., Ashrafian,H., & Redwood,C. Inherited cardiomyopathies. N. Engl. J. Med. 364, 1643-1656 (2011). 7. Scicluna,B.P., Wilde,A.A., & Bezzina,C.R. The primary arrhythmia syndromes: same mutation, different manifestations. Are we starting to understand why? J. Cardiovasc. Electrophysiol. 19, 445-452 (2008).
13
Chapter 2
Common genetic variation modulating cardiac ECG parameters and susceptibility to sudden cardiac death
Iris C.R.M. Kolder1 ,2, Michael W.T. Tanck2 and Connie R. Bezzina1
1 Heart Failure Research Center, Department of Experimental Cardiology, Academic Medical Center, The Netherlands 2 Department of Clinical Epidemiology, Biostatistics and Bioinformatics, Academic Medical Center, The Netherlands
Journal of Molecular and Cellular Cardiology 2012 Mar;52(3):620-9 Chapter 2
Abstract Sudden cardiac death (SCD) is a prevalent cause of death in Western societies. Genome- wide association studies (GWAS) conducted over the last few years have uncovered common genetic variants modulating risk of SCD. Furthermore, GWAS studies uncovered several loci impacting on heart rate and ECG indices of conduction and repolarization, as measures of cardiac electrophysiological function and likely intermediate phenotypes of SCD risk. We here review these recent developments and their implications for the identification of novel molecular pathways underlying normal electrophysiological function and susceptibility to SCD.
Introduction Sudden cardiac death (SCD) is one of the most prevalent causes of death in Western societies. It underlies 20% of total mortality, and 50% of cardiovascular mortality 1. The most common mechanism of SCD is thought to involve ventricular tachycardia degenerating first to ventricular fibrillation (VF) and subsequently to asystole 2. It arises through multiple mechanisms, depending on the affected individual’s underlying cardiac pathology 3. In young individuals (below 40 years of age) SCD often occurs in the setting of disorders displaying Mendelian inheritance 4, with the cardiomyopathies 5 and primary electrical disorders 6 being the most prevalent. Here, the inheritance of very rare genetic variants with large effects increases risk for SCD substantially. Significant progress has been accomplished over the last two decades in uncovering the genes underlying these Mendelian disorders. This has led to an increased understanding of the underlying pathophysiological mechanisms and has had a major impact on patient management. The vast majority (~80%) of SCDs in the community however occurs in individuals above 40 years of age and are primarily caused by VF in the setting of acute myocardial ischemia or acute or prior myocardial infarction (MI), sequelae of coronary artery disease (CAD) 7. Paradoxically, progress in uncovering the genetic factors that determine susceptibility to these common arrhythmias, affecting a much greater proportion of the population, has been limited and have only recently started to be explored. We here review the current strategies and progress in tracking down these genetic variants.
A heritable component in the determination of risk for SCD in the community Two landmark population-based studies published in the late 90s demonstrated increased risk of SCD in first-degree relatives of SCD victims, providing the first evidence thata genetic component is involved in the determination of risk. A population-based case- control study of SCD patients by Friedlander and co-workers carried out in the area of
16 Common genetic variation
Seattle demonstrated that family history of MI/SCD was associated with SCD (RR=1.57), after adjustment for other common risk factors and person-years at risk among (first degree) relatives 8. These investigators later re-analyzed their case-control data, this time differentiating between family history of MI and family history of SCD and reported that, after adjustment for other risk factors and family history of MI, a positive family history of early-onset SCD (<65 years) was associated with a 2.7-fold increase in risk of SCD 9. Similar data emerged from the Paris Prospective study which showed that the relative risk of SCD associated with the sudden death of one parent was 1.80; this increased to 9.44 when both parents died suddenly 10. Although as mentioned above, SCD in the general population most often occurs in the setting of coronary artery disease, these initial studies did not differentiate between the different cardiac pathologies in which SCD occurred. Our 2 group subsequently designed the Arrhythmia Genetics in the Netherlands Study (AGNES), aimed selectively at including victims of VF in the setting of a first acute MI, to enable the analysis of risk factors for this specific prevalent SCD phenotype. In this case-control study in which we compared individuals with and without (documented) VF during the early phase of a first acute MI we also identified sudden death among first degree relatives as a strong risk factor for VF (OR=2.72) 11. A Finnish study published around the same time by Kaikkonen and co-workers provided similar evidence. These investigators showed that victims of SCD with an acute coronary event and without a history of prior MI were more likely to have a family history of SCD compared to acute MI survivors (OR=1.6) and healthy controls (OR=2.2) 12.
Genetic architecture of SCD in the community The genetic architecture of SCD in the community is still largely unknown as the search for the underlying genetic variants has only just started. However, as a complex disease it is thought to be attributable in part to common allelic variants i.e. variants present in more than 1–5% of the population Figure( 1). Such common variants are expected to be associated with small effects and as such will individually confer only relatively small increments in risk (1.1–1.5-fold) and explain only a small proportion of the heritability. A role for such variants in susceptibility to SCD has recently started to be explored in genome-wide association studies (GWAS) 13-15, triggered by the advent of commercial ‘single nucleotide polymorphism (SNP) chips’ or arrays that interrogate a large part of common variation in the genome. GWAS use dense maps of hundreds of thousands of common SNPs spread throughout the genome to systematically examine whether genotype is significantly associated with differences in phenotype. This unbiased approach has the potential of identifying genomic regions previously unlinked to the trait of interest, triggering studies that will ultimately uncover novel mechanisms.
17 Chapter 2
In addition to common genetic variants of small effects, lower-frequency genetic variants (<1%) with intermediate to large effect sizes (Figure 1) are also expected to contribute to susceptibility of complex disease such as SCD16 . The search for such rare variants requires exome or genome resequencing efforts, or the use of genotyping arrays that capture such low-frequency variants which are just becoming available following recent advances in rare variant discovery through large-scale genome and exome sequencing projects (e.g. 1000 Genomes Project and the NHLBI Exome Sequencing Project). It is envisaged that the co-inheritance of multiple common and rare risk alleles will determine one’s ultimate risk for SCD.
Figure 1 | Complex traits like SCD are thought to be modulated by genetic variants of different severity (effect size) that occur in the general population at different allele frequencies. Reproduced from Manolio et al. Nature 2009;461,747-753 17.
Conducting genetic studies for SCD is however challenging. Obtaining medical ethical committee approval for the collection of DNA from victims of SCD is complicated. Furthermore, a prerequisite for successful genetic studies is the availability of rigorously- phenotyped patient sets (and appropriate controls), which is problematic for SCD. Not all cases of sudden death in the community are due to cardiac causes (e.g. stroke, aneurysm) and when due to a cardiac cause this is not necessarily VF-mediated (e.g. pump failure, sinus arrest). When sudden death is VF-mediated, the arrhythmia may occur in the setting of various cardiac pathologies (e.g. first MI, various stages of ischemic cardiomyopathy etc) and may consequently stem from different mechanisms. Autopsies, which would
18 Common genetic variation shed light on the circumstances of the sudden death, are conducted only sporadically and the retrieval of medical history of community-based SCD victims, which would provide information about the underlying cardiac substrate, entails considerable effort. With respect to controls for association studies, the use of similarly-exposed controls would increase statistical power for detection of variants and avoid confounding.
Genome-wide association studies for SCD So far three genome-wide association studies have examined the role of common genetic variation in modulation of SCD or VF risk13-15 , with two studies identifying genetic variants at the genome-wide statistical significance threshold of p < 5 ×-8 10 13, 15. This stringent 2 significance threshold, frequently used when studying samples of European ancestry, accounts for the roughly 1,000,000 independent common variant tests in the human genome 18.
One study conducted GWAS in the AGNES case-control population (mentioned above; Dekker et al. 11), a population of European descent enrolled in the Netherlands, and compared ~500 cases with documented VF in the setting of a first MI and a comparable number of similarly exposed controls (MI with no VF) 13. This study identified genetic variation at chromosome 21q21 (lead SNP rs2824292), at about 100 kilobase pairs (kb) from the CXADR gene (encoding the coxsackie virus and adenovirus receptor), associated with VF. This association signal was replicated in a smaller case-control set of consisting of patients presenting with VF in the setting of an acute MI and MI-survivor controls.
The other study reporting an association with SCD at genome-wide statistical significance consisted of a meta-analysis of various genome-wide association studies for SCD conducted in multiple US and European case-control and cohort studies of European ancestry 15. This study which in total compared 4402 cases of SCD and >30,000 controls identified a signal at chromosome 2q24.2 (lead SNP rs4665058) mapping to an intron in the BAZ2B gene which encodes bromodomain adjacent zinc finger domain 2B. In the latter study, the BAZ2B locus did not demonstrate an effect in the AGNES case-control set. Similarly, the CXADR signal detected in the AGNES GWAS was not detected in a small case-control set (90 patients with acute MI and VF and 167 MI non-VF controls) from Germany 19. This may be due to several factors, including not only insufficient statistical power and random chance, but also differences in study design and phenotype definition. With respect to the latter, for instance, the AGNES case-control set has a much narrower phenotype definition as compared to the other cases and controls used in the Arking et al. study 15. These observations stress the need for careful ascertainment of the SCD phenotype, especially with respect to the cardiac pathology associated with the SCD.
19 Chapter 2
Although the association signals from these two GWAS studies occur in the vicinity (rs2824292) or within (rs4665058) certain genes, this does not immediately imply that the effects of the identified variants are mediated through these genes. Furthermore, these two specific SNPs do not necessarily represent the causal variant. This actually underscores one of the limitations of GWAS in that it depends upon linkage disequilibrium andin essence any genetic variant that is correlated with these sentinel signals is potentially causal. For both association signals, no missense variants in neighbouring genes display a strong correlation with the sentinel SNP, suggesting that the functional variant is regulatory and that these loci may impact on disease risk through effects on gene expression rather than through functional variation of a specific protein. Future studies, entailing expression quantitative trait locus (eQTL) analysis in cardiac tissue as well as the genome-wide identification by ChiP-Seq of regulatory regions occupied by transcriptional enhancers and transcription factors 20 will hopefully aid the process of assigning causality. Nevertheless, at the chromosome 21q21 locus, the CXADR gene which encodes the coxsackie and adenovirus receptor (CAR) represents an attractive candidate for the effect. CAR which is a transmembrane protein localizing to the intercalated disk of cardiomyocytes, has a long- recognized role as viral receptor in the pathogenesis of myocarditis and has more recently been implicated in atrioventricular conduction. Its role in ventricular conduction, perhaps of greater relevance to VF 21, has not yet been investigated.
Heart rate and ECG indices of conduction and repolarization as intermediate phenotypes of SCD risk Arrhythmia is a complex phenotype and is expected to be governed by multiple interacting biological and environmental factors. As genes do not directly encode for the ultimate arrhythmia phenotype but for biological processes that subsequently impact on the process of arrhythmogenesis, one logical way to uncover the genetic underpinnings of arrhythmia is to parse this complex phenotype into tractable biological sub-processes and first attempt to identify those genes impacting on these processes. An advantage of this approach is that it enables greater statistical power for gene discovery as the association (or penetrance) of gene effects is expected to be greater at the level of these relatively more simple phenotypes that are involved in the disease. Such biological processes involved in disease are often referred to as “intermediate phenotypes”. The use of intermediate phenotypes as a strategy for reducing genetic complexity and increasing genetic effect size for facilitating gene discovery is being applied across many fields as a means of ultimately understanding common complex disease 22. The identification of
20 Common genetic variation genetic variants impacting on these intermediate traits also provides the foundation for a broader biological understanding of these phenotypes and eventually to identify new therapeutic opportunities for disease prevention.
Besides being related to the disorder, for intermediate phenotypes to be useful in gene discovery, they also need to display reasonable heritability and be measured accurately in a sufficiently large number of subjects. Heart rate and electrocardiographic (ECG) indices of conduction and repolarization are among the very few known (if not the only) intermediate phenotypes of arrhythmia with these characteristics. They display significant heritability (see summary in Table 1) and it is therefore not surprising that many groups have in recent years dedicated much effort in identifying genetic variants within the 2 general population that modulate these parameters.
Various ECG features constitute markers of SCD risk in the Mendelian rhythm disorders. For instance, prolonged or shortened QTc-interval (a measure of ventricular repolarization) in the Long QT Syndrome and the Short QT Syndrome respectively, are well established indicators of arrhythmia risk, as is QTc-prolongation in drug-induced Long QT Syndrome23 . Moreover, in the Long QT syndrome, the risk of a cardiac event increases with the extent of QTc-interval prolongation24 . Although findings are not always consistent between studies, there is some evidence that certain ECG parameters predict risk of SCD in the general population and in sub-groups of patients with different cardiovascular disorders25 .
Studies carried out in the Rotterdam study cohort have provided evidence that prolonged QTc-interval is an independent risk factor for SCD among middle-aged and older individuals from the general population26, 27. In an early small case-control study, Schwartz et al. 28 were the first to suggest that among patients with previous MI a prolonged QTc is associated with greater risk for sudden death. A recent analysis in the Oregon Sudden Unexpected Death Study (SUDS) which compared CAD patients with SCD to controls with CAD but without a history of SCD demonstrated that within this disease group, abnormally prolonged QTc-interval was associated with a marked increase in risk of SCD 29.
Several studies have reported that intraventricular conduction delay, evidenced by prolongation of QRS-duration, and left bundle branch block can be associated with an increased risk of arrhythmic death and overall mortality, in patients with congestive heart failure, independent of extent of left ventricular dysfunction 44-47. A study carried out in a wider spectrum of patients with known or suspected CAD (and normal or mildly
21 Chapter 2
Table 1 | Summary of studies investigating the heritability of ECG parameters Heritability (%)
Heart PR- P wave QRS- QTc- Population Ethnicity Reference Rate interval duration Interval Island 34 17 3 Kosrae (Micronesia) 30 population
50 Population Amish 31
30 Families Native American 32
34 34 43 40 Families Chinese 33
31 Families Mongolian 34
American (European 29 34 39 Families 35 ancestry) American (European 35 Families 36 ancestry) American (European 34 Families 37 ancestry)
41 Families Israeli 38
American (European 21 Families 39 ancestry) Sibling 18 40 33 30 Icelandic 40 pairs
46 40 52 Twins German 41
Female 55 52 British 42 Twins Female 25 British 43 Twins impaired left ventricular function) also provided evidence of increased SCD riskand mortality within this group 48. QRS conduction delay was independently associated with sustained monomorphic VT inducibility in a large cohort of patients who underwent electrophysiological evaluation 49.
The PR-interval reflects atrio-ventricular conduction and reflects conduction in the atria, atrioventricular node, and the His-Purkinje system. Longitudinal data from the Framingham Heart Study and the Atherosclerosis Risk in Communities (ARIC) study established PR- interval prolongation as a risk factor for atrial fibrillation 50, 51. Furthermore, in the
22 Common genetic variation
Framingham Heart Study, prolonged PR-interval was associated with a moderate increase in all-cause mortality. Recently, the CARISMA (Cardiac Arrhythmias and Risk Stratification After Acute Myocardial Infarction) study identified high-degree atrioventricular block as a strong determinant of all-cause mortality and cardiac death among patients with acute MI and reduced ejection fraction. A study which evaluated 12-lead ECGs recorded during the acute ischemic phase of a first ST-elevation MI, demonstrated that patients with non- anterior wall infarctions (right coronary or left circumflex coronary artery involvement) who developed ischemic VF showed longer QRS and PR intervals compared to those that did not develop VF 52. An association of high resting heart rate with risk of future SCD was identified in multiple cohort studies 53-56. 2 Common genetic variants modulating heart rate and ECG indices of conduction and repolarization Common genetic variants impacting on ECG indices have been investigated in genome- wide association studies on large samples of the general population 40, 57-67. We here summarize those SNPs that have been found to be associated with QT-interval (Table 2), conduction indices (PR-interval, QRS-duration, Table 3) and heart rate (Table 4) at a genome-wide threshold for statistical significance (p < 5 × 10-8). Genomic locations of all of these loci are also represented in Figure 2.
As for the GWAS addressing SCD (see section 4) most of the GWAS for ECG traits have so far been carried out in large cohorts of European descent, but GWAS in populations of other ancestry have started to appear 58, 65. Carrying out GWAS in different ethnic groups, who may have a different distribution of ECG indices (e.g. PR-interval 65 and differential susceptibility to arrhythmias is likely to be very informative. For instance samples of African ancestry are characterized by a greater level of genetic diversity compared to non- African samples which will enable discovery of additional loci 68. Furthermore samples of African ancestry display less linkage disequilibrium among loci which can facilitate fine- mapping of association signals 68.
GWAS signals for QTc-interval The first GWAS on an electrocardiographic trait was carried out by Arking andco- workers on KORA German community-based samples 57. This study for the first time demonstrated that genetic variation in the region of theNOS1AP gene (Table 2), encoding nitric oxide synthase 1 (neuronal) adaptor protein, is associated with the QTc-interval, immediately demonstrating the power of this genetic tool in uncovering unanticipated genetic associations. Two subsequent large studies that meta-analyzed GWAS data of QTc- interval from several population-based cohorts of European descent 61, 63 and a GWAS
23 Chapter 2
Figure 2 | Genomic location of association signals published to date at an association p-value< 5 ×-8 10 for sudden cardiac death / ventricular fibrillation, QTc-interval, QRS-duration, PR-interval and heart rate. study in European genetically isolated populations 67 all identified a signal at NOS1AP as the strongest main association signal. In these studies multiple independent signals (SNPs that display low correlation with each other) were detected at the NOS1AP locus, providing evidence for the presence of multiple QTc-modifying variants at this locus. Besides the NOS1AP locus, these studies also identified additional loci associated with the QTc-interval at genome-wide statistical significance (Table 2). Not surprisingly, some of these associations were located at genomic regions harboring cardiac ion channel genes previously implicated in the Mendelian repolarization disorders Long QT Syndrome and Short QT Syndrome. These include KCNQ1, KCNH2, and KCNJ2, all encoding repolarizing potassium channel a(pore-forming)-subunits, KCNE1 encoding a potassium channel modulatory subunit, and SCN5A encoding the a-subunit (Nav1.5) of the major depolarizing sodium channel in heart. Some of these genome-wide significant signals at these loci actually provided further support for observations previously made in small association
24 Common genetic variation studies (KCNH2-K897T 69, see section 6.1.1; KCNE1-D85N 70, see section 6.1.2; SCN5A- D1819D 70; KCNQ1 intron 1 variant rs757092, 71). The identification of common genetic variants in these genes modulating the QTc-interval in the general population is consistent with the concept that genes harbour a spectrum of variants, ranging from rare highly deleterious variants leading to Mendelian disease to common small-effect variants that may in aggregate impact on susceptibility to complex disease in the general population. Other loci identified at genome-wide significance in these studies were located in the region of genes encoding proteins with an established role in myocardial electrophysiology, namely, ATP1B1 encoding the Na+/K+ ATPase β1 subunit 72 and PLN encoding phospolamban 73. Two independent association signals were detected at the PLN locus. The other association signals identified in these studies were at chromosomal 2 loci previously unlinked to myocardial repolarization (Table 2). Regional candidate genes for the chromosome 16q21 signal are GINS3 and NDRG4. In a zebrafish genetic mutant screen the GINS3 insertional mutant displayed resistance to dofetilide-induced 2:1AV block 74, while morpholino knockdown of Ndrg4 in zebrafish embryos was associated with a constellation of cardiac developmental defects75 .
Association between a number of these loci and QTc-interval was reported by Chambers and co-workers in Indian Asians from the UK 58.
KCNH2-K897T (rs1805123) SNP rs2968863 located downstream of KCNH2 (Table 2) encoding the rapidly activating component of the delayed rectifier K+ current, is highly correlated to rs1805123 (r2=0.91), a non-synonymous coding variant in the same gene which has received a lot of attention. This polymorphism (minor allele frequency [MAF] ~24% in populations of European descent) leads to the substitution of lysine at residue 897 to threonine and was previously associated with the QTc-interval duration in small candidate studies 76. Although there was at first some debate about the direction of the effect associated with the minor (C) allele of this polymorphism 69, 77 it seems now established that carriership of this allele is associated with a shorter QTc-interval 35, 71, 78, as also substantiated by the GWAS studies61, 63. Controversy however still remains about the exact effect of this polymorphism on the biophysical properties of the encoded potassium channel, as different groups reported discrepant findings 69, 79-82. This may have to do with the different cellular expressions systems and experimental protocols used by the different groups coupled to the fact that based on the small effect of this SNP on the QTc-interval (approximately -1.5 ms per C-allele) 63 one expects very subtle effects on channel function and these may be difficult to detect accurately and reproducibly in the experimental systems used. Moreover, although it is tempting to speculate that a non-synonymous polymorphism (i.e. leading to an amino acid change) would be functional and underlies the effect observed for the
25 Chapter 2 associated allele, one should not forget that the given allele is highly correlated with other polymorphisms in the context of a haplotype. As mentioned above, in principal any other correlated SNP(s) on that haplotype may underlie the observed effect perhaps through effects on gene expression levels.
KCNE1-D85N (rs1805128) SNP rs1805128 (Table 2), also the focus of several studies, was initially related to the 70 QTC-interval in a small French study and subsequently found to be associated with this parameter at genome-wide statistical significance in the meta-analysis by Newton- Cheh and co-workers 61. It is a rare non-synonymous SNP, present in ~1% of the general population leading to a change from aspartate to asparagine at amino acid 85 (D85N) in KCNE1, a gene encoding a potassium channel modulatory subunit. It has been proposed to modulate the variable penetrance of the congenital long QT syndrome 83 and to be related to drug-induced QTc-interval prolongation and torsade de pointes arrhythmia 84, 85. The association between this SNP and the QTc-interval was however not replicated in a large set of individuals of European descent from Iceland studied by Holm et al. 40. The reason for this is unclear.
Expression studies by various investigators demonstrated that KCNE1-D85N causes loss- 83, 86 of-function effects on theKr I current encoded by KCNH2 . Inconsistent findings however have been reported with respect to impairment of the Iks current by this variant, with some investigators reporting decreased current 83, 87 and others reporting no effects 86. A decreased IKr and/or IKs current would be in line with the longer QTc-interval associated with this allele.
GWAS signals for ECG conduction indices (PR-interval, QRS-duration) Five genome-wide association studies on ECG conduction indices, three conducted in individuals of European ancestry 40, 64, 66, one in Indian Asian individuals from the UK 58 and one in African Americans 65, have appeared in the literature over the last two years. Between them these studies uncovered a total of 27 different chromosomal loci impacting on PR-interval and/or QRS-duration (Table 3). At three regions, evidence exists for the presence of multiple independent signals (Table 3).
All five studies detected signals in the region of the SCN5A and SCN10A genes (which lie within 50kb of each other on chromosome 3), as the most significant association signal (Figure 3). The association of SNPs at the SCN5A locus with PR-interval and QRS-
26 Common genetic variation duration is not surprising given the major role of Nav1.5 channels in mediating “phase 0” depolarization in cardiomyocytes and the fact that loss of function mutations in SCN5A lead to Mendelian cardiac conduction disease which may affect different cardiac compartments 89, 90. SNP rs6795970 detected as a highly robust signal for PR-interval (40, 58, 64 and QRS-duration 40, 66 is located within SCN10A, encoding Nav1.8, a sodium channel a-subunit isoform involved in nociception. Nav1.8 is primarily expressed in small sensory neurons in dorsal root ganglia where it is a major contributor to action potential upstroke and repetitive firing91 . This SNP (Figure 3) is actually a non-synonymous variant, leading to the substitution of valine at position 1073 in Nav1.8 to adenine (V1073A), raising the possibility that this amino acid change could represent the functional variant underlying the effect on PR and QRS. No studies, to our knowledge, have yet investigated 2 the functional effects of A1073V on Nav1.8 channel function. Nevertheless, this would imply (i) that Nav1.8 is expressed in cardiomyocytes, perhaps with enrichment in the conduction system, and (ii) that it contributes significantly to the action potential upstroke in these cells. Work in mice demonstrated enrichment of Scn10a mRNA levels in Purkinje cells compared to working ventricular myocytes 92. On the other hand, telemetric ECG studies in conscious mice produced conflicting results with one study reporting shorter PR-interval in Scn10 homozygous knock-out (Scn10-/-) mice 58 and another study using an Nav1.8 channel blocker (A-803467) reporting longer PR-interval and QRS-duration [52]. The latter study also reported a prolongation of the HV-interval with the drug, suggesting that the prolongation observed for QRS-duration stemmed from conduction slowing in the specialized conduction system. It has been hypothesized 66 that the more rapid conduction in Scn10a-/- mice may reflect compensatory upregulation of TTX-sensitive currents, as previously observed in Nav1.8-deficient DRG neurons from Scn10a-/- mice 93. It is clear that further studies are needed to elucidate the role Nav1.8 may play in cardiac electrophysiology, and this obviously needs to include studies at the cardiomyocyte level. Alternatively, in spite of the fact that no linkage disequilibrium exists between rs6795970 in SCN10A and the SCN5A gene, one cannot exclude the possibility that this SNP or any other SNP correlated to it modulates cardiac conduction through regulatory effects on SCN5A mRNA expression levels.
27 Chapter 2 61 63 63 63 63 61 61 67 61, 63 61, 63 61, 63 61,63# 57,67# eference(s) R in genome-wide association -8 SNP Intron 3 ‘UTR Intronic Intronic Intronic Intronic Intronic Intergenic Intergenic Intergenic Intergenic Intergenic Intergenic Location of Leading Location of Association Signal of Association ↑ ↓ ↑ ↑ ↑ ↑ ↓ ↓ ↑ ↑ ↑ ↑ ↓ of Effect Direction 0.18 0.26 0.47 0.18 0.11 0.26 0.29 0.08 0.15 0.11 0.16 0.26 0.33 MAF T/C T/C T/C T/C T/C C/T T/G C/G C/G G/C G/A A/G A/G Allele Coded / Non-coded Non-coded ASF1A SCN5A KCNH2 ATP1B1 NOS1AP Nearest gene(s) Nearest PLEKHG5, KLH21 RNF207, NPHP4, CHDS, ACOT7, RNF207, NPHP4, CHDS, ACOT7, C6orf204, SLC35F1, PLN, BRD7P3, PLN, BRD7P3, C6orf204, SLC35F1, SNP rs846111 rs4725982 rs2968863 rs4657178 rs2880058 rs10919071 rs11970286 rs12053903 rs12210810 rs16857031 rs10494366 rs12029454 rs12143842 Single nucleotide polymorphisms (SNP)found to be associated withat theQTc-interval association p-values <5×10 2 2 | 1q24.2 7q36.1 3p22.2 1q23.3 1p36.31 le b Chromosome 6q22.1-6q22.31 a T studies.
28 Common genetic variation 61 61 63 61 40 67 61 63 61, 63 61,63# eference(s) R
SNP coding Intronic Intronic Intronic Intronic Intronic Intronic Intergenic Intergenic Intergenic Location of Leading Location of Association Signal of Association 2 ↑ ↓ ↑ ↓ ↓ ↓ ↑ ↑ ↑ ↑ of Effect Direction 0.03 0.49 0.49 0.27 0.32 0.08 0.22 0.35 0.16 0.23 MAF T/C T/C T/C T/C T/C T/C T/G A/C G/A G/A Allele Coded / Non-coded Non-coded TBX5 GOT2 KCNJ2 KCNE1 KCNQ1 SUCLA2 TNFRSF17 LIG3, RFFL Nearest gene(s) Nearest LITAF, CLEC16A, SNN, ZC3H7A, SNN, ZC3H7A, CLEC16A, LITAF, CNOT1, GINS3, NDRG4, SLC38A7, SLC38A7, GINS3, NDRG4, CNOT1, SNP rs37062 rs1805128 rs2074518 rs3825214 rs2478333 rs8049607 rs2074238 rs17779747 rs12576239 rs12296050 >0.8) with the listed SNP was reported in this study. MAF = minor allele frequency. MAF was based on 1000 Genomes CEU / Hapmap CEU MAF was MAF = minor allele frequency. in this study. reported SNP was >0.8) with the listed 2 16q21 17q24.3 11p15.5 21q22.12 12q24.21 16p13.13 17q11.2-q12 Chromosome 13q12.2-q13.3 Multiple SNPs listed at a given chromosomal locus represent multiple independent association signals atrepresent givenat the locus a locus. chromosomal given Multiple #Another listed SNP SNPs which is in high linkage disequilibrium (r
29 Chapter 2 66 66 66 66 64 66 66 66 in genome- in -8 65¥,64# eference(s) R Signal Intron Intron Intron Intron Intron Intron Intron Coding Intergenic Location of Location Leading SNP of Association of Association ↓ ↑ ↑ ↑ ↑ ↑ ↓ ↓ ↓ of Effect Direction PR PR QRS QRS QRS QRS QRS QRS QRS Trait 0.13 0.37 0.36 0.23 0.46 0.30 0.48 0.03 MAF 0.46/0.48* C/T C/T C/T G/T G/T G/T A/G A/G A/G Allele Coded / Non-coded Non-coded NFIA STRN MEIS1 CRIM1 CASQ2 RNF11, HEATR5B, HEATR5B, CACNA1D C1orf185, TKT, PRKCD, PRKCD, TKT, CDKN2C, FAF1 CDKN2C, LRIG-SLC25A26 Closest gene(s) Closest SNP rs4687718 rs2242285 rs7562790 rs4074536 rs9436640 rs17020136 rs11897119 rs10865355 rs17391905 Single nucleotide polymorphisms (SNP) found to be associated with the PR-interval and/or QRS-duration at association p-values <5×10 p-values association at QRS-duration and/or PR-interval the with associated be to found (SNP) polymorphisms nucleotide Single
3 | 2p21 2p14 3p14.3 2p22.2 3p14.1 1p32.3 le 1p13.3-p11 b 1p31.3-p31.2 Chromosome a T wide association studies. wide association
30 Common genetic variation 66 66 66 66 66 65¥ 88¥ 65¥ 65¥ 40,66# 40, 64# eference(s) R 40, 58,64#,66# 3’UTR Signal Intron Intron Intron Intron Intron Intron Intron Intron Coding Intergenic Intergenic Location of Location Leading SNP of Association of Association 2 ↑ ↑ ↓ ↑ ↓ ↓ ↑ ↑ ↓ ↑ ↑/↑ ↑/↑ of Effect Direction PR PR PR PR PR QRS QRS QRS QRS QRS Trait PR/QRS PR/QRS 0.25 0.42 0.25 0.28 0.18 0.17 0.19 0.26 MAF 0.18* 0.38/0.06* 0.23/0.29* 0.26/0.68* T/C T/A C/T C/T C/T C/T A/C C/G G/A A/G A/G A/G Allele Coded / Non-coded Non-coded SCN5A, SCN10A ARHGAP24 Closest gene(s) Closest SNP rs7660702 rs6795970 rs7627552 rs6798015 rs9851724 rs6599222 rs3922844 rs2051211 rs11129795 rs11710077 rs11708996 rs10865879 4q22.1 3p22.2 Chromosome
31 Chapter 2 64 64 66 40 66 66 66 66 64 40, 66 40,66# eference(s) R Signal Intron Intron Intron Intergenic Intergenic Intergenic Intergenic Intergenic Intergenic Intergenic Intergenic Location of Location Leading SNP of Association of Association ↓ ↓ ↑ ↑ ↑ ↑ ↑ ↑ ↑ ↓ ↓ of Effect Direction PR PR PR PR QRS QRS QRS QRS QRS QRS QRS Trait 0.15 0.29 0.23 0.27 0.43 0.13 0.48 0.48 0.31 0.33 0.38 MAF C/T C/T T/G A/C G/A G/A A/G A/G A/G A/G A/G Allele Coded / Non-coded Non-coded DKK1 VTI1A TBX20 ASF1A IGFBP3 NKX2.5 SAP30L WNT11 HAND1, C5orf41, SLC35F1, SLC35F1, C6orf204, CAV1/CAV2 SOX5, BCAT1 SOX5, PLN, BRD7P3, PI16, CDKN1A Closest gene(s) Closest SNP rs251253 rs4944092 rs1733724 rs7342028 rs3807989 rs1362212 rs7784776 rs1321311 rs11047543 rs11153730 rs13165478 5q33 7q31.1 7p14.3 7p12.3 5q35.1 6p21.2 12p12.1 11q13.5 10q11.2 10q25.2 6q22.31 Chromosome
32 Common genetic variation 66 66 66 40 66 65 66 64 66 66,65#¥ eference(s) R Exon intron Signal Intron Intron Intron Intron Intron Intron Intergenic Location of Location Leading SNP of Association of Association Another SNP which is in high linkage # 2 ↑ ↑ ↑ ↓ ↓ ↓ ↑ ↑ ↓ ↑/↑ of Effect Direction PR PR QRS QRS QRS QRS QRS QRS QRS Trait PR/ QRS 0.41 0.42 0.09 0.22 0.33 0.30 0.38 0.28 0.32 MAF 0.27/0.25* Study conducted in African American samples. MAF = minor MAF allele was frequency. ¥ T/C C/T C/T C/T A/T G/C G/A G/A G/A A/G Allele Coded / Non-coded Non-coded KLF12 PRKCA GOSR2 SETBP1 SIPA1L1 TBX3, TBX5 TBX3, Closest gene(s) Closest SNP rs991014 rs883079 rs9912468 rs3825214 rs1886512 rs1896312 rs7312625 rs11848785 rs17608766 rs10850409 >0.8) with the listed SNP was reported in this study. 2 13q22 17q21 18q21.1 14q24.2 12q24.21 17q22-q23.2 Chromosome Multipleat SNPs listed a given multiple chromosomal locusrepresent independent association signals at thegiven locus. disequilibrium (r based on 1000 Genomes CEU / Hapmap CEU.
33 Chapter 2
Figure 3 | Detailed view of the chromosome 3 region that harbours multiple independent association signals that impact on PR-interval and/or QRS-duration. In this figure, one can appreciate the location of the associating SNPs with respect to the location of the SCN5A and SCN10A genes and the chromosomal recombination rates in the region. Recombination rates corresponds to those of the CEU population.
A number of association signals were detected at chromosomal regions harboring genes encoding transcription factors with an established role in heart development, namely TBX3, TBX5, TBX20, MEIS1 and NKX2.5. These findings provide further support to the notion that the transcriptional regulation orchestrating the development of theheart and the specification of the different cardiac regions from cardiac precursor cells during embryonic development also plays an essential role in the regulation of expression of ion channel and gap junction genes 94, 95. Association signals were also detected in the region of genes encoding proteins involved in calcium-handling such as CASQ2, PLN and PRKCA which may modulate conduction of the cardiac electrical impulse either by effecting directly cellular electrophysiological properties or through effects on tissue architecture. Interestingly, the CASQ2 index SNP concerns a genetic variant associated with an amino acid change (CASQ2-T66A) and the lead SNP at the PLN locus is in linkage disequilibrium (r2= 0.69) with a non-synonymous SNP (rs3734381; S137G) in PLN.
SNPs displaying genome-wide significant association for one conduction parameter (e.g. PR), often display evidence for association with the other (e.g. QRS) with concordance in direction of effect 40, 66. This is in line with the shared physiologic processes underlying
34 Common genetic variation these conduction traits and points to shared genetic underpinnings for variability in atrial and ventricular conduction in the general population.
GWAS signals for heart rate Four different studies identified common genetic variants impacting on heart rate 40, 59, 60, 67 (Table 4). One was conducted in Asian subjects recruited in Korea 59 and uncovered two loci, one (lead SNP rs12731740) close to the CD46, C1orf132 and CD34 genes on chromosome 1q and the other (lead SNP rs12110693) on chromosome 6q at around 400 kb upstream from the GJA1 gene. GJA1 encodes connexin 43, a gap junction protein involved in conduction of the cardiac electrical impulse. Association to these two loci was also detected by Eijgelsheim and co-workers who conducted a meta-analysis of 15 GWAS 2 in subjects of European ancestry from Europe and the United States 60. Association to the region on chromosome 14q harbouring MYH6 and MYH7 genes was detected in two studies 40, 60. SNP rs365990 in MYH6 is a non-synonymous coding variant leading to the substitution of alanine at position 1101 to valine. A role forMYH6 , encoding the alpha heavy chain subunit of cardiac myosin, in modulation of sinus function was brought into further focus in a recent elegant study by Holm and co-workers 96. These investigators utilized a combined strategy consisting of genome-wide SNP typing, whole-genome sequencing and imputation to detect a rare non-synonymous variant (p.Arg721Trp, MAF=0.38%) in MYH6, which appears to be specific to Icelanders and which is associated with sick sinus syndrome 96. The precise involvement of this gene however awaits clarification as the MYH6 gene hosts a miRNA, miR-208a, which has been shown to regulate cardiac conduction in mouse studies 97.
SNP rs314370 at the chromosome 7 locus is strongly correlated with rs12666989 which results in a leucine to valine substitution at position 41 in the UFSP1 gene product.
Common genetic variants identified through GWAS modulate severity of primary electrical disorders Extensive variability in ECG and arrhythmia manifestations exists among carriers of Mendelian genetic defects associated with the primary electrical disorders 98. This is perhaps best appreciated in large pedigrees harbouring founder mutations where carriers of the same primary genetic defect display wide-ranging variability in ECG manifestations and/or arrhythmia susceptibility 99, 100. There is considerable interest in the identification of factors modulating disease expression in these disorders, as this is expected to lead to improved risk stratification among mutation carriers. Genetic modifiers are thought to play a role and it is logical that common genetic variants identified in GWAS form prime candidates for these effects. Studies investigating this concept have started to appear.
35 Chapter 2 60 60 60 60 67 60 60 60 40, 60 59, 94* eference(s) 60;59, 94* R SNP Intron Intron Intron Intron Coding Another SNP which is in high linkage Intergenic Intergenic Intergenic Intergenic Intergenic Intergenic # in genome-wide association studies. association in genome-wide -8 Location of Leading Location of Association Signal of Association ↓ ↓ ↓ ↑ ↓ ↓ ↓ ↑ ↑ ↓ ↑ of Effect Direction 0.48 0.22 0.38 0.15 0.29 0.25 0.31 0.49 0.09 0.07 0.29 MAF T/C C/T C/T A/C G/A G/A G/A G/A A/G A/G A/G Allele Coded / Non-coded Non-coded GJA1 CD34 FADS1 PLXNA2 GPR133 C6orf204, CD46, CD34, SOX5, BCAT1 SOX5, NDNG, ZFHX2 SLC35F1, PLN, SLC35F1, MYH6, MYH7, Closest gene(s) Closest BRD7P3, ASF1A SLC12A9, UFSP1 SLC12A9, SNP rs281868 rs174547 rs885389 rs223116 rs365990 rs314370 rs2745967 rs9398652 rs17287293 rs12731740 rs11154022 >0.8) with the listed SNP was reported in was *GWAS conducted this in study. Asian samples. MAF = minor MAF allele was frequency. based 2 Single nucleotide polymorphisms (SNP) found to be associated with heart rate at association p-values <5×10 p-values at association rate with heart to be associated found Single nucleotide polymorphisms (SNP)
7q22 1q32 4 | 1q32.2 6q22.31 12p12.1 14q11.2 12q24.33 le 6q21-q23.2 Chromosome b 11q12.2-q13.1 a T Multipleat SNPs listed a given multiple chromosomal locusrepresent independent association signals at thegiven locus. disequilibrium (r on 1000 Genomes CEU / Hapmap CEU.
36 Common genetic variation
For instance common genetic variants in NOS1AP, modifiers of QT-interval in the general population, were shown to modulate the QT-interval duration and risk of arrhythmias in the Long QT syndrome in two studies 101, 102.
Post-GWAS functional characterization of identified loci and other perspectives for future studies Genome-wide association studies have in recent years provided us with a wealth of data on common genetic variation impacting on heart rate and ECG indices of conduction and repolarization. By identifying chromosomal loci previously unlinked to these traits, 2 they have provided the research community with several new avenues for research into mechanisms involved in cardiac electrical function and susceptibility to arrhythmia 103. However, as alluded to above, GWAS point us to the chromosomal regions involved, but tell us nothing about the causal gene or mechanism. Linking the genetic variant or haplotype to a specific gene and ultimately to an electrophysiological mechanism remains a major task 104, 105. One major obstacle along this path is the fact that our understanding of the function of the non-protein coding regions of the genome is still quite rudimentary, while most of the common DNA variants modulating complex traits have been mapped there. A systematic strategy aimed at understanding a particular GWAS locus ideally entails an initial step of fine-mapping at high resolution for refining the association signal and reducing the number of potential functional variants to be tested. Furthermore, it will necessitate the generation of a comprehensive catalog of all the genetic variation within the region (which can be obtained by targeted sequencing) followed by annotation of the variable regulatory elements (e.g. enhancers, promoters, insulators, silencers). Overlaying GWAS data with other genomic data, for instance data on the location of transcription factor and enhancer binding sites (obtained by Chip-Seq)106 may be an efficient approach in prioritizing candidate regulatory sites and likely functional variants for downstream reporter assays in vitro and/or in vivo. Expression QTL data, yet unavailable for human heart, is also expected to help in bridging the gap between functional variant and causal gene. Similarly, measurement of allelic imbalance (also called allele-specific gene expression) in individuals heterozygous for a given variant is expected to contribute107. Finally, once a transcript is associated with a risk allele, its involvement in the trait will ultimately need to be proven in functional follow-up studies relevant to the phenotype. Moreover, resequencing of genes located in the association regions in individuals at the extreme ends of the trait distribution may uncover rare variants with large effects, which may be more amenable to functional studies.
37 Chapter 2
Although ECG indices constitute highly relevant intermediate phenotypes for electrophysiological abnormalities and arrhythmia, research is still required to test the involvement of variants impacting on these traits in the actual modulation of risk for these complex clinical phenotypes. Such studies have started to appear. For instance, an analysis by Sotoodehnia and co-workers in individuals with bundle branch block or nonspecific prolongation of QRS-interval (QRS >120 ms), demonstrated that these individuals had a greater cumulative burden of QRS-prolonging alleles compared to individuals with normal conduction66 . Furthermore, some of the SNPs that displayed association with PR-interval, an established intermediate phenotype of atrial fibrillation, also showed an association with this trait 40, 64. Moreover, rs6795970 in the PR/QRS-prolonging allele at the SCN5A/ SCN10A locus was also found to be associated with pacemaker implantation [46]. These findings underscore the potential of how such genome-wide association studies on intermediate phenotypes may lead to the discovery of clinically pertinent biological processes. On the other hand, a study by Noseworthy and co-workers 108 did not identify a linear relationship between QTscore, a score reflecting the joint influence QTc-modifying variants, and the risk of SCD.
Future larger meta-analysis of GWAS studies of common variants for ECG traits will possibly uncover more common genetic variants with smaller effects. The falling costs of whole exome and genome sequencing will however likely shift the interest of the genetics community to these technologies as they hold promise for capturing fully the allelic architecture of these traits, including lower-frequency variants, and hold the potential of providing an explanation for the large amount of variance that remains unaccounted for. Genome-wide association studies have also started to appear for VF/SCD. This is however a much more challenging phenotype to study, primarily due to the heterogeneity in the underlying cardiac substrate. Also here, rare variants may also modulate risk and such variants in genes already known to be crucial to proper cardiac electrophysiological function form prime candidates for testing as already suggested by a small candidate gene study which investigated the role of functionally significant mutations and rare variants in SCN5A in SCD risk among women 109.
38 Common genetic variation
Reference List 1. Myerburg,R.J. & Castellanos,A. Cardiac 11. Dekker,L.R. et al. Familial sudden death arrest and sudden cardiac death. in is an important risk factor for primary Braunwald’s Heart Disease: A Textbook ventricular fibrillation: a case-control study of Cardiovascular Medicine (eds. Libby,P., in acute myocardial infarction patients. Bonow,R.O., Mann,D.L. & Zipes,D.P.) 933- Circulation 114, 1140-1145 (2006). 974 (Elsevier, Oxford, UK, 2007). 12. Kaikkonen,K.S., Kortelainen,M.L., Linna,E., 2. Huikuri,H.V., Castellanos,A., & & Huikuri,H.V. Family history and the risk of Myerburg,R.J. Sudden death due to cardiac sudden cardiac death as a manifestation of arrhythmias. N. Engl. J. Med. 345, 1473- an acute coronary event. Circulation 114, 1482 (2001). 1462-1467 (2006). 3. Arking,D.E., Chugh,S.S., Chakravarti,A., & 13. Bezzina,C.R. et al. Genome-wide 2 Spooner,P.M. Genomics in sudden cardiac association study identifies a susceptibility death. Circ. Res. 94, 712-723 (2004). locus at 21q21 for ventricular fibrillation 4. van der Werf,C., van Langen,I.M., & in acute myocardial infarction. Nat. Genet. Wilde,A.A. Sudden death in the young: 42, 688-691 (2010). what do we know about it and how to 14. Arking,D.E. et al. Genome-wide association prevent? Circ. Arrhythm. Electrophysiol. 3, study identifies GPC5 as a novel genetic 96-104 (2010). locus protective against sudden cardiac 5. Watkins,H., Ashrafian,H., & Redwood,C. arrest. PLoS. One. 5, e9879 (2010). Inherited cardiomyopathies. N. Engl. J. 15. Arking,D.E. et al. Identification of a sudden Med. 364, 1643-1656 (2011). cardiac death susceptibility locus at 6. Wilde,A.A. & Bezzina,C.R. Genetics of 2q24.2 through genome-wide association cardiac arrhythmias. Heart 91, 1352-1358 in European ancestry individuals. PLoS. (2005). Genet. 7, e1002158 (2011). 7. Zipes,D.P. & Wellens,H.J. Sudden cardiac 16. Asimit,J. & Zeggini,E. Rare variant death. Circulation 98, 2334-2351 (1998). association analysis methods for complex traits. Annu. Rev. Genet. 44, 293-308 8. Friedlander,Y. et al. Family history as a (2010). risk factor for primary cardiac arrest. Circulation 97, 155-160 (1998). 17. Manolio,T.A. et al. Finding the missing heritability of complex diseases. Nature 9. Friedlander,Y. et al. Sudden death and 461, 747-753 (2009). myocardial infarction in first degree relatives as predictors of primary cardiac 18. Pe’er,I., Yelensky,R., Altshuler,D., & arrest. Atherosclerosis 162, 211-216 Daly,M.J. Estimation of the multiple testing (2002). burden for genomewide association studies of nearly all common variants. 10. Jouven,X., Desnos,M., Guerot,C., & Genet. Epidemiol. 32, 381-385 (2008). Ducimetiere,P. Predicting sudden death in the population: the Paris Prospective Study I. Circulation 99, 1978-1983 (1999).
39 Chapter 2
19. Bugert,P. et al. No evidence for an 29. Chugh,S.S. et al. Determinants of prolonged association between the rs2824292 variant QT interval and their contribution to sudden at chromosome 21q21 and ventricular death risk in coronary artery disease: the fibrillation during acute myocardial Oregon Sudden Unexpected Death Study. infarction in a German population. Clin. Circulation 119, 663-670 (2009). Chem. Lab Med. 49, 1237-1239 (2011). 30. Smith,J.G. et al. Genome-wide association 20. Blow,M.J. et al. ChIP-Seq identification of study of electrocardiographic conduction weakly conserved heart enhancers. Nat. measures in an isolated founder Genet. 42, 806-810 (2010). population: Kosrae. Heart Rhythm. 6, 634- 21. Janse,M.J. & Kleber,A.G. 641 (2009). Electrophysiological changes and 31. Post,W. et al. Associations between genetic ventricular arrhythmias in the early phase variants in the NOS1AP (CAPON) gene of regional myocardial ischemia. Circ. Res. and cardiac repolarization in the old order 49, 1069-1081 (1981). Amish. Hum. Hered. 64, 214-219 (2007). 22. Teslovich,T.M. et al. Biological, clinical and 32. Melton,P.E. et al. Genetic influences on population relevance of 95 loci for blood serum bilirubin in American Indians: The lipids. Nature 466, 707-713 (2010). Strong Heart Family Study. Am. J. Hum. 23. Roden,D.M. Drug-induced prolongation of Biol. 23, 118-125 (2011). the QT interval. N. Engl. J. Med. 350, 1013- 33. Li,J. et al. Familial aggregation and 1022 ( 2004). heritability of electrocardiographic 24. Priori,S.G. et al. Risk stratification in the intervals and heart rate in a rural long-QT syndrome. N. Engl. J. Med. 348, Chinese population. Ann. Noninvasive. 1866-1874 ( 2003). Electrocardiol. 14, 147-152 (2009). 25. Chugh,S.S. Early identification of risk 34. Im,S.W. et al. Analysis of genetic and non- factors for sudden cardiac death. Nat. Rev. genetic factors that affect the QTc interval Cardiol. 7, 318-326 (2010). in a Mongolian population: the GENDISCAN study. Exp. Mol. Med. 41, 841-848 (2009). 26. Algra,A., Tijssen,J.G., Roelandt,J.R., Pool,J., & Lubsen,J. QTc prolongation measured by 35. Newton-Cheh,C. et al. Common genetic standard 12-lead electrocardiography is an variation in KCNH2 is associated with QT independent risk factor for sudden death interval duration: the Framingham Heart due to cardiac arrest. Circulation 83, 1888- Study. Circulation 116, 1128-1136 (2007). 1894 (1991). 36. Newton-Cheh,C. et al. QT interval is a 27. Straus,S.M. et al. Prolonged QTc interval heritable quantitative trait with evidence and risk of sudden cardiac death in a of linkage to chromosome 3 in a genome- population of older adults. J. Am. Coll. wide linkage analysis: The Framingham Cardiol. 47, 362-367 (2006). Heart Study. Heart Rhythm. 2, 277-284 (2005). 28. Schwartz,P.J. & Wolf,S. QT interval prolongation as predictor of sudden death 37. Hong,Y. et al. Familial aggregation of QT- in patients with myocardial infarction. interval variability in a general population: Circulation 57, 1074-1077 (1978). results from the NHLBI Family Heart Study. Clin. Genet. 59, 171-177 (2001).
40 Common genetic variation
38. Friedlander,Y., Lapidos,T., Sinnreich,R., 48. Elhendy,A., Hammill,S.C., Mahoney,D.W., & Kark,J.D. Genetic and environmental & Pellikka,P.A. Relation of QRS duration sources of QT interval variability in Israeli on the surface 12-lead electrocardiogram families: the kibbutz settlements family with mortality in patients with known or study. Clin. Genet. 56, 200-209 (1999). suspected coronary artery disease. Am. J. 39. Singh,J.P. et al. Heritability of heart rate Cardiol. 96, 1082-1088 (2005). variability: the Framingham Heart Study. 49. Horwich,T., Lee,S.J., & Saxon,L. Usefulness Circulation 99, 2251-2254 (1999). of QRS prolongation in predicting risk 40. Holm,H. et al. Several common variants of inducible monomorphic ventricular modulate heart rate, PR interval and QRS tachycardia in patients referred for duration. Nat. Genet. 42, 117-122 (2010). electrophysiologic studies. Am. J. Cardiol. 92, 804-809 (2003). 41. Busjahn,A. et al. QT interval is linked to 2 2 long-QT syndrome loci in normal subjects. 50. Cheng,S. et al. Long-term outcomes in Circulation 99, 3161-3164 (1999). individuals with prolonged PR interval or first-degree atrioventricular block. JAMA 42. Dalageorgou,C. et al. Heritability of QT 301, 2571-2577 (2009). interval: how much is explained by genes for resting heart rate? J. Cardiovasc. 51. Soliman,E.Z., Prineas,R.J., Case,L.D., Electrophysiol. 19, 386-391 (2008). Zhang,Z.M., & Goff,D.C., Jr. Ethnic distribution of ECG predictors of atrial 43. Carter,N. et al. QT interval in twins. J. Hum. fibrillation and its impact on understanding Hypertens. 14, 389-390 (2000). the ethnic distribution of ischemic stroke 44. Iuliano,S., Fisher,S.G., Karasik,P.E., in the Atherosclerosis Risk in Communities Fletcher,R.D., & Singh,S.N. QRS duration (ARIC) study. Stroke 40, 1204-1211 (2009). and mortality in patients with congestive 52. Lemmert,M.E. et al. Electrocardiographic heart failure. Am. Heart J. 143, 1085-1091 factors playing a role in ischemic ventricular (2002). fibrillation in ST elevation myocardial 45. Kashani,A. & Barold,S.S. Significance of infarction are related to the culprit artery. QRS complex duration in patients with Heart Rhythm. 5, 71-78 (2008). heart failure. J. Am. Coll. Cardiol. 46, 2183- 53. Dyer,A.R. et al. Heart rate as a prognostic 2192 (2005). factor for coronary heart disease and 46. Shamim,W. et al. Intraventricular mortality: findings in three Chicago conduction delay: a prognostic marker in epidemiologic studies. Am. J. Epidemiol. chronic heart failure. Int. J. Cardiol. 70, 112, 736-749 (1980). 171-178 (1999). 54. Jouven,X., Zureik,M., Desnos,M., 47. Zimetbaum,P.J. et al. Electrocardiographic Guerot,C., & Ducimetiere,P. Resting heart predictors of arrhythmic death and total rate as a predictive risk factor for sudden mortality in the multicenter unsustained death in middle-aged men. Cardiovasc. tachycardia trial. Circulation 110, 766-769 Res. 50, 373-378 (2001). (2004).
41 Chapter 2
55. Kannel,W.B., Kannel,C., Paffenbarger,R.S., 65. Smith,J.G. et al. Genome-wide association Jr., & Cupples,L.A. Heart rate and studies of the PR interval in African cardiovascular mortality: the Framingham Americans. PLoS. Genet. 7, e1001304 Study. Am. Heart J. 113, 1489-1494 (1987). (2011). 56. Shaper,A.G., Wannamethee,G., 66. Sotoodehnia,N. et al. Common variants in MacFarlane,P.W., & Walker,M. Heart rate, 22 loci are associated with QRS duration ischaemic heart disease, and sudden and cardiac ventricular conduction. Nat. cardiac death in middle-aged British men. Genet. 42, 1068-1076 (2010). Br. Heart J. 70, 49-55 (1993). 67. Marroni,F. et al. A genome-wide association 57. Arking,D.E. et al. A common genetic variant scan of RR and QT interval duration in 3 in the NOS1 regulator NOS1AP modulates European genetically isolated populations: cardiac repolarization.Nat. Genet. 38, 644- the EUROSPAN project. Circ. Cardiovasc. 651 (2006). Genet. 2, 322-328 (2009). 58. Chambers,J.C. et al. Genetic variation in 68. Campbell,M.C. & Tishkoff,S.A. African SCN10A influences cardiac conduction. genetic diversity: implications for human Nat. Genet. 42, 149-152 (2010). demographic history, modern human 59. Cho,Y.S. et al. A large-scale genome-wide origins, and complex disease mapping. association study of Asian populations Annu. Rev. Genomics Hum. Genet. 9, 403- uncovers genetic factors influencing eight 433 (2008). quantitative traits.Nat. Genet. 41, 527-534 69. Bezzina,C.R. et al. A common (2009). polymorphism in KCNH2 (HERG) hastens 60. Eijgelsheim,M. et al. Genome-wide cardiac repolarization. Cardiovasc. Res. 59, association analysis identifies multiple loci 27-36 (2003). related to resting heart rate. Hum. Mol. 70. Gouas,L. et al. Association of KCNQ1, Genet. 19, 3885-3894 (2010). KCNE1, KCNH2 and SCN5A polymorphisms 61. Newton-Cheh,C. et al. Common variants with QTc interval length in a healthy at ten loci influence QT interval duration in population. Eur. J. Hum. Genet. 13, 1213- the QTGEN Study. Nat. Genet. 41, 399-406 1222 (2005). (2009). 71. Pfeufer,A. et al. Common variants in 62. Nolte,I.M. et al. Common genetic variation myocardial ion channel genes modify the near the phospholamban gene is associated QT interval in the general population: with cardiac repolarisation: meta-analysis results from the KORA study. Circ. Res. 96, of three genome-wide association studies. 693-701 (2005). PLoS. One. 4, e6138 (2009). 72. Lingrel,J.B. The physiological significance 63. Pfeufer,A. et al. Common variants at ten of the cardiotonic steroid/ouabain-binding loci modulate the QT interval duration in site of the Na,K-ATPase. Annu. Rev. Physiol the QTSCD Study. Nat. Genet. 41, 407-414 72, 395-412 (2010). (2009). 73. Periasamy,M., Bhupathy,P., & Babu,G.J. 64. Pfeufer,A. et al. Genome-wide association Regulation of sarcoplasmic reticulum study of PR interval. Nat. Genet. 42, 153- Ca2+ ATPase pump expression and its 159 (2010). relevance to cardiac muscle physiology and pathology. Cardiovasc. Res. 77, 265-273 (2008).
42 Common genetic variation
74. Milan,D.J. et al. Drug-sensitized zebrafish 82. Mannikko,R. et al. Pharmacological and screen identifies multiple genes, including electrophysiological characterization of GINS3, as regulators of myocardial nine, single nucleotide polymorphisms of repolarization. Circulation 120, 553-559 the hERG-encoded potassium channel. Br. (2009). J. Pharmacol. 159, 102-114 (2010). 75. Qu,X. et al. Ndrg4 is required for normal 83. Nishio,Y. et al. D85N, a KCNE1 myocyte proliferation during early cardiac polymorphism, is a disease-causing gene development in zebrafish. Dev. Biol. 317, variant in long QT syndrome. J. Am. Coll. 486-496 (2008). Cardiol. 54, 812-819 (2009). 76. Laitinen,P. et al. Survey of the coding 84. Paulussen,A.D. et al. Genetic variations region of the HERG gene in long QT of KCNQ1, KCNH2, SCN5A, KCNE1, and syndrome reveals six novel mutations and KCNE2 in drug-induced long QT syndrome 2 an amino acid polymorphism with possible patients. J. Mol. Med. (Berl) 82, 182-188 phenotypic effects. Hum. Mutat. 15, 580- (2004). 581 (2000). 85. Kaab,S. et al. A large candidate gene survey 77. Pietila,E. et al. Association between HERG identifies the KCNE1 D85N polymorphism K897T polymorphism and QT interval in as a possible modulator of drug-induced middle-aged Finnish women. J. Am. Coll. torsades de pointes. 2011. Cardiol. 40, 511-514 (2002). 86. Nof,E. et al. LQT5 masquerading as LQT2: 78. Marjamaa,A. et al. Common candidate a dominant negative effect of KCNE1-D85N gene variants are associated with QT rare polymorphism on KCNH2 current. interval duration in the general population. Europace. 13, 1478-1483 (2011). J. Intern. Med. 265, 448-458 (2009). 87. Westenskow,P., Splawski,I., Timothy,K.W., 79. Scherer,C.R. et al. The antihistamine Keating,M.T., & Sanguinetti,M.C. fexofenadine does not affect I(Kr) currents Compound mutations: a common cause of in a case report of drug-induced cardiac severe long-QT syndrome. Circulation 109, arrhythmia. Br. J. Pharmacol. 137, 892-900 1834-1841 (2004). (2002). 88. Jeff,J.M.et al. SCN5A variation is associated 80. Paavonen,K.J. et al. Functional with electrocardiographic traits in the characterization of the common amino Jackson Heart Study. Circ. Cardiovasc. acid 897 polymorphism of the cardiac Genet. 4, 139-144 (2011). potassium channel KCNH2 (HERG). 89. Schott,J.J.et al. Cardiac conduction defects Cardiovasc. Res. 59, 603-611 (2003). associate with mutations in SCN5A. Nat. 81. Anson,B.D. et al. Molecular and Genet. 23, 20-21 (1999). functional characterization of common 90. Remme,C.A. & Bezzina,C.R. Sodium channel polymorphisms in HERG (KCNH2) (dys)function and cardiac arrhythmias. potassium channels. Am. J. Physiol Heart Cardiovasc. Ther. 28, 287-294 (2010). Circ. Physiol 286, H2434-H2441 (2004). 91. Rush,A.M., Cummins,T.R., & Waxman,S.G. Multiple sodium channels and their roles in electrogenesis within dorsal root ganglion neurons. J. Physiol 579, 1-14 (2007).
43 Chapter 2
92. Pallante,B.A. et al. Contactin-2 expression 102. Tomas,M. et al. Polymorphisms in the in the cardiac Purkinje fiber network. NOS1AP gene modulate QT interval Circ. Arrhythm. Electrophysiol. 3, 186-194 duration and risk of arrhythmias in the (2010). long QT syndrome. J. Am. Coll. Cardiol. 55, 93. Akopian,A.N. et al. The tetrodotoxin- 2745-2752 (2010). resistant sodium channel SNS has a 103. Chang,K.C. et al. CAPON modulates cardiac specialized function in pain pathways. Nat. repolarization via neuronal nitric oxide Neurosci. 2, 541-548 (1999). synthase signaling in the heart. Proc. Natl. 94. Postma,A.V., Christoffels,V.M., & Acad. Sci. U. S. A 105, 4477-4482 (2008). Bezzina,C.R. Developmental aspects of 104. Arking,D.E. & Chakravarti,A. Understanding cardiac arrhythmogenesis. Cardiovasc. Res. cardiovascular disease through the lens of 91, 243-251 (2011). genome-wide association studies. Trends 95. Stankunas,K. et al. Pbx/Meis deficiencies Genet. 25, 387-394 (2009). demonstrate multigenetic origins of 105. Freedman,M.L. et al. Principles for the congenital heart disease. Circ. Res. 103, post-GWAS functional characterization of 702-709 (2008). cancer risk loci. Nat. Genet. 43, 513-518 96. Holm,H. et al. A rare variant in MYH6 is (2011). associated with high risk of sick sinus 106. May,D. et al. Large-scale discovery of syndrome. Nat. Genet. 43, 316-320 (2011). enhancers from human heart tissue. Nat. 97. Callis,T.E. et al. MicroRNA-208a is a Genet.(2011). regulator of cardiac hypertrophy and 107. Serre,D. et al. Differential allelic expression conduction in mice. J. Clin. Invest 119, in the human genome: a robust approach 2772-2786 (2009). to identify genetic and epigenetic cis-acting 98. Scicluna,B.P., Wilde,A.A., & Bezzina,C.R. mechanisms regulating gene expression. The primary arrhythmia syndromes: same PLoS. Genet. 4, e1000006 (2008). mutation, different manifestations. Are we 108. Noseworthy,P.A. et al. Common genetic starting to understand why? J. Cardiovasc. variants, QT interval, and sudden cardiac Electrophysiol. 19, 445-452 (2008). death in a Finnish population-based study. 99. Bezzina,C. et al. A single Na(+) channel Circ. Cardiovasc. Genet. 4, 305-311 (2011). mutation causing both long-QT and 109. Albert,C.M. et al. Cardiac sodium channel Brugada syndromes. Circ. Res. 85, 1206- gene variants and sudden cardiac death in 1213 (1999). women. Circulation 117, 16-23 (2008). 100. Brink,P.A. et al. Phenotypic variability and unusual clinical severity of congenital long- QT syndrome in a founder population. Circulation 112, 2602-2610 (2005). 101. Crotti,L.et al. NOS1AP is a genetic modifier of the long-QT syndrome. Circulation 120, 1657-1663 (2009).
44 Chapter 3
The role of renin-angiotensin-aldosterone system polymorphisms in phenotypic expression of MYBPC3- related hypertrophic cardiomyopathy
Iris C.R.M. Kolder, MSc1,2*; Michelle Michels, MD, PhD3*; I. Christiaans, MD, PhD4; F.J. Ten Cate, MD, PhDc, D. Majoor-Krakauer, MD5; A.H.J. Danser, PhD6; R. H. Lekanne Deprez, PhD4; M.W.T. Tanck, PhD2; A.A.M. Wilde, MD, PhD1; C.R. Bezzina, PhD1; D. Dooijes, PhD5,7
* These authors contributed equally
Affiliations: 1 Heart Failure Research Center, Department of Clinical and Experimental Cardiology, Academic Medical Center, Amsterdam, the Netherlands 2 Department of Epidemiology, Biostatistics and Bioinformatics, Academic Medical Center, Amsterdam, the Netherlands 3 Department of Cardiology, Erasmus Medical Center, Rotterdam, the Netherlands 4 Department of Clinical Genetics, Academic Medical Center, Amsterdam, the Netherlands 5 Department of Clinical Genetics, Erasmus Medical Center, Rotterdam, the Netherlands 6 Division of Vascular Medicine and Pharmacology, Department of Internal Medicine, Erasmus Medical Center, Rotterdam, the Netherlands 7 Department. of Medical Genetics, University Medical Centre Utrecht, Utrecht, the Netherlands
European Journal of Human Genetics 2012: Accepted for publication Chapter 3
Abstract The phenotypic variability of hypertrophic cardiomyopathy (HCM) even in patients with identical pathogenic mutations suggests additional modifiers. In view of the regulatory role in cardiac function, blood pressure, and electrolyte homeostasis, polymorphisms in the renin-angiotensin-aldosterone system (RAAS) are candidates for modifying phenotypic expression. In order to investigate whether RAAS polymorphisms modulate the HCM phenotype, we selected a large cohort of carriers of one of 3 functionally-equivalent truncating mutations in the MYBPC3 gene.
Family based association analysis was performed to analyze the effects of 5 candidate RAAS polymorphisms (ACE, rs4646994; AGTR1, rs5186; CMA, rs1800875; AGT, rs699; CYP11B2, rs1799998) in 368 subjects carrying one of 3 mutations in the MYBPC3 gene. Interventricular septum (IVS) thickness and Wigle score were assessed by 2D-echocardiography. SNPs in the RAAS system were analyzed separately and combined as a pro-left ventricular hypertrophy (LVH) score for effects on the HCM phenotype. Analyzing the 5 polymorphisms separately for effects on IVS thickness and Wigle score detected 2 modest associations. Carriers of the CC genotype in the AGT gene had less pronounced IVS thickness compared to CT and TT genotype carriers. The DD polymorphism in the ACE gene was associated with a high Wigle score (p=0.01). No association was detected between the pro-LVH score and IVS thickness or Wigle score.
In conclusion, in contrast to previous studies, in our large study population of HCM patients with functionally-equivalent mutations in theMYBPC3 gene we did not find major effects of genetic variation within genes of the RAAS system on phenotypic expression of HCM.
Introduction Hypertrophic cardiomyopathy (HCM) is the most common inheritable cardiac disorder with a phenotypic prevalence of 1:500.1 It is defined by the presence of left ventricular hypertrophy (LVH) in the absence of loading conditions (hypertension, valve disease) sufficient to cause the observed abnormality.2 Hundreds of mutations scattered among at least 25 HCM susceptibility genes encoding various sarcomere, Z-disk, calcium-handling, and mitochondrial proteins are known to cause HCM in 60% of cases.3
In the Netherlands, approximately one third of all HCM cases are caused by carrier ship of anyone of 3 founder mutations in the myosin binding protein C gene (MYBPC3). All three are truncating mutations, namely c.2373dupG, c.2827C>T (p.Arg943X) and c.2864_2865delCT.4-6 Immunoblotting studies on myocardial tissue from carriers of the
46 The role of renin-angiotensin-aldosterone c.2373dupG and c.2864_2865delCT truncating mutations have demonstrated absence of the truncated MyBPC3-protein product coupled to a decreased total (full-length) MyBPC3 content, strongly suggesting a mechanism of haploinsufficiency.5 Although not functionally investigated, the c.2827C>T mutation, which introduces a premature stop codon at residue 943, encodes for a similarly C-terminally truncated protein (Figure 1), and is therefore also expected to lead to haploinsufficiency. These three founder mutations may therefore be considered functionally equivalent.
3
Figure 1 | Schematic representation of MyBPC3 structure and the effects of the 3 founder mutations on the structure of the protein. Adapted from van Dijk et al.5
Extensive phenotypic variability exists among HCM mutation carriers. Even individuals with the same underlying genetic substrate show a broad spectrum of clinical manifestations.7,8 This indicates that HCM features are not solely determined by the nature of the pathogenic mutation and that additional environmental and/or other genetic factors (genetic modifiers) play a role in the clinical expression of the disease.
Genetic variants in the renin-angiotensin-aldosterone system (RAAS) are considered candidates for these modifying effects. The RAAS system contributes to left ventricular hypertrophy (LVH) through effects mediated by circulating angiotensin as well as local activation of RAAS in the myocardium.9 Angiotensin (Ang) I, produced from angiotensinogen (AGT), is converted to Ang II predominantly by angiotensin I-converting enzyme (ACE) and possibly by chymase (CMA1)10. Ang II binds primarily to the Ang II type 1 receptor (AGTR1) to promote cell growth and hypertrophy. It also stimulates aldosterone by aldosterone (CYP11B2) synthesis, thereby increasing the release of aldosterone, which promotes fluid retention and cardiac fibrosis.11
47 Chapter 3
Previous studies suggested a role for specific genetic variants in genes encoding components of the RAAS pathway in modulation of the severity of LVH in patients with HCM.12,13,14 In particular, two studies investigated 5 candidate polymorphisms within these genes in HCM patients with an identified HCM-causing mutation. The RAAS polymorphisms tested in these studies included: (1) an insertion/deletion (I/D) polymorphism in intron 16 of the ACE gene [rs4646994] where the D-allele was considered as the pro-LVH allele, (2) an A>C polymorphism at position 1166 of theAGTR1 gene [rs5186] where the C-allele was considered as the pro-LVH allele, (3) an A>G polymorphism at position −1903 of the CMA1 gene [rs1800875] where the A-allele was considered as the pro-LVH allele, (4) a T>C (p.M235T) polymorphism in the AGT gene [rs699] where the C-allele was considered as the pro-LVH allele, and (5) a C>T polymorphism at position −344 of the CYP11B2 gene [rs1799998] where the C-allele was considered as the pro-LVH allele.
The first of these two studies was carried out in a small number (n=26) of carriers of the MYBPC3-c.2373dupG mutation from a single family. In this study, c.2373dupG mutation carriers that harbored one or more pro-LVH RAAS polymorphisms genotypes (n=16 individuals) had greater left ventricular muscle mass and interventricular septum thickness compared to those that harbored no pro-LVH genotypes (n=10).13 This study also provided some evidence for a pro-LVH effect of these polymorphisms when assessed individually.
The second study was carried out in a cohort of 389 unrelated patients with HCM, of which 63 and 54 patients respectively carried a mutation in theMYBPC3 and MYH7 HCM- associated genes.14 In this study, while the ACE-I/D polymorphism displayed no effect on any LVH parameter in the entire cohort, subset analysis of theMYBPC3 and MYH7 genetic subtypes demonstrated a pro-LVH effect of the DD-ACE genotype in the MYBPC3-HCM subtype. In this study there was some evidence that the burden of pro-LVH genotypes was associated with an increased left ventricular wall thickness, although this was only present in the group of patients that were negative for myofilament gene mutations.
The results of these previous studies addressing effects of genetic modifiers in HCM have however been difficult to interpret because of the small sample size and/or by inclusion of a genetically heterogeneous study population with respect to the primary genetic defect. In this study we used a large cohort of subjects having one of 3 functionally-equivalent truncating founder mutations in the MYBPC3 gene, to investigate whether the 5 RAAS gene polymorphisms investigated in these previous studies modulate echocardiographic features of HCM.
48 The role of renin-angiotensin-aldosterone
Material and Methods Study population In the Netherlands, genetic counseling and genetic testing is offered to all HCM patients visiting cardiogenetics outpatient clinics. Upon the identification of the causal mutation in a proband, genetic testing is extended to relatives following appropriate genetic counseling (cascade screening).15, 16 For this study all subjects, including probands and relatives, carrying one of 3 truncating founder mutations inthe MYBPC3 gene (c.2373dupG, c.2864_2865delCT, c.2827C>T) were selected from 2 university hospitals in the Netherlands; the Academic Medical Center in Amsterdam and the Erasmus Medical Center in Rotterdam. In this way 368 carriers of equally pathogenetic MYPBC3 mutations were included. All subjects were normotensive (blood pressure < 140/90 mmHg) and did not take medication known to influence the RAAS. All subjects provided written informed consent. The study complies with the declaration of Helsinki and the local review boards of the respective hospitals approved the study.
Echocardiographic evaluation 3 Echocardiography was performed in all subjects using commercially available equipment. The acquired data were digitally stored and subsequently analyzed by 2 physicians who were blinded to the clinical and genetic data. Interventricular septum (IVS) thickness was measured in diastole from the parasternal short-axis view at the level of the papillary muscles. For relatives ≥16 years a IVS thickness ≥ 13 mm was considered as abnormal.17 For subjects <16 years IVS thickness was corrected for height and weight and was considered abnormal if the z-score was >2. The extent of hypertrophy was assessed by a semi-quantitative score method developed by Wigle et al.18 A maximum of 10 points were given: 1 to 4 points for IVS thickness (1 point for IVS thickness between 15-19 mm; 2 points for IVS thickness between 20-24 mm; 3 points for IVS thickness 25-29 mm and 4 points if IVS thickness ≥ 30 mm), 2 points for extension of hypertrophy beyond the level of the papillary muscles (basal two thirds of the IVS), 2 points for extension of hypertrophy to the apex (total IVS involvement), and 2 points for extension of hypertrophy into the lateral wall.
SNP genotyping The five polymorphisms investigated in this study were the same as those studied previously by Ortlepp et al.13 and by Perkins et al.,14 that is (1) rs4646994 in the ACE gene, (2) rs5186 in AGTR1, (3) rs1800875 in CMA1 gene, (4) rs699 in AGT, and (5) rs1799998 in CYP11B2. Patient genomic DNA was extracted from peripheral blood lymphocytes using standard protocols. Genotyping was carried out as described previously. Pro-LVH
49 Chapter 3 genotypes were defined as described previously, namely as DD-ACE, CC-AGTR1, AA-CMA, CC-AGT, and CC-CYP11B2. The pro-LVH score was calculated for each patient by adding the number of pro-LVH genotypes present.13, 14
Statistical analyses Phenotypic data for probands and relatives that was normally distributed (Shapiro-Wilk statistic W>0.9) are reported as mean ± standard deviation otherwise as median with interquartile range. Pedigree information was available for all related subjects. Allele frequencies of polymorphisms tested in the study population were compared to those reported for the CEU population in the 1000 Genomes database using a Pearson chi square test.
We assumed that each polymorphism-phenotype relationship followed a recessive genetic model as previously reported.14,15. Effects on IVS thickness and the ranked Wigle score were estimated using a linear mixed model with adjustment for sex, age and proband status, whereas effects on the dichotomous variables IVS thickness ³ 13 or ³ 30 mm were estimated using a logistic regression model with adjustment for sex and age. Next to models with a single SNP or the pro-LVH score, a model with all five SNPs and their interactions was also used. To account for the relatedness of study subjects, either the linear mixed model from the Kinship package or the generalized estimation equations from the geepack package in R were used (R foundation for Statistical Computing, Vienna, Austria). P-values < 0.05 were considered significant.
Based on the sample size (n=368), our study had 90% power to detect a 0.30 mm difference in IVS thickness between the RAAS polymorphism genotype groups and a correlation coefficient of 0.167 (± 3% explained variance) between IVS thickness and the pro-LVH score (α = 0.05 two-sided).
Results Study Population DNA and echocardiography data were available for 368 carriers of one of the 3 Dutch MYBC3 founder mutations, including 100 probands and 268 relatives. The age distribution of probands and relatives was similar (Table 1). By definition all probands had anIVS thickness ≥ 13 mm. Extreme hypertrophy (IVS thickness ≥ 30 mm), a known risk factor for sudden death was present in 9 (10 %) probands19, 20. There was a male preponderance in probands compared to the relatives (64% vs. 47%, p = 0.007). LVH (defined as IVS ≥ 13 mm)16 was present in 107 (40%) of relatives and was extreme in 1 (0.4%) relative.
50 The role of renin-angiotensin-aldosterone
Table 1 | Characteristics of the HCM population studied. Probands Relatives p-value (n=100) (n=268) Age 42 ± 14 41 ± 17 0.64 Male sex 64 (64%) 127 (47%) 0.007 Septum thickness (mm) 22 ± 5 13 ± 4 9.8e-43 Septum ≥ 13 mm 100 (100%) 107 (40%) 2.2e-16 Septum ≥ 30 mm 9 (10%) 1 (0.4%) 0.0001 Wigle score 4 (2-6) 0 (0-1) 2.5e-33
Data depict mean ± standard deviation, n (%) or median (interquartile range).
The MYBPC3 founder mutations in the study population are presented in Table 2. The most common founder mutation was c.2373dupG, present in 70% of the individuals. Table 2 | MYBPC3 mutation distribution among the HCM population studied. 3 Mutation Probands Relatives c.2373dupG 70 187 c.2827C>T 18 56 c.2864_2865delCT 12 25
Influence of age, sex and proband status Proband status, age and gender had a significant effect on IVS thickness. Probands displayed a greater mean IVS thickness compared to relatives (22±5 versus 13±4 mm) (Table 1). While in the relatives group, IVS thickness was greater in older individuals; within the proband group, it was smaller in the older individuals (Figure 2). This apparent interaction effect was, however, not significant (p=0.76). On average men had a thicker IVS than women (17±6 mm versus 14±6 mm; p=0.0014). This influence of male sex is especially clear in the relatives group (Figure 2).
SNP association analyses The minor allele frequency (MAF) of 4 of the polymorphisms tested did not differ from that reported for the CEU population in the 1000 Genomes database (P>0.05) (rs5186, MAF this study = 0.31 / MAF CEU 1000 Genomes = 0.28; rs1800875, MAF=0.52/0.54; rs699, MAF=0.36/ 0.36; rs1799998, MAF=0.39/0.43). The MAF of rs4646994 (MAF=0.49 this study) was not reported in 1000 Genomes. The 5 polymorphisms were analyzed separately and combined as a pro-LVH score for (i) effects on IVS thickness, (ii) association with IVS thickness ≥13mm, (iii) association with IVS thickness ≥30 mm, and (iv) effects on the Wigle score (Tables 3-5).
51 Chapter 3
Figure 2 | Effects of sex and age on septum thickness in probands and relatives.
Table 3 | Phenotype distribution per genotype group for each polymorphism studied. ACE, I/D (rs4646994) II ID DD *P-value Septum thickness (mm), total population 15±6 (n=102) 15±6 (n=173) 16±7 (n=92) 0.30 probands 22±4 (n=29) 21±5 (n=(48) 22±7 (n=23) relatives 13±4 (n=73) 12±4 (n=125) 13±5 (n=69)
Wigle score 0 (0-3) 0 (0-3) 1 (0-4) 0.05
≥13 septum thickness 61 (60%) 90 (52%) 54 (59%) 0.49
≥30 septum thickness 2 (2%) 4 (2%) 4 (4%) 0.24 AGT, C>T (M235T, rs699) TT TC CC *P-value Septum thickness (mm), total population 16±6 (n=164) 15±6 (n=146) 15±6 (n=58) 0.02 probands 22±6 (n=41) 21±5 (n=39) 20±4 (n=20) relatives 13±5 (n=123) 12±4 (n=107) 12±4 (n=38) Wigle score 0 (0-3) 0 (0-4) 0 (0-1) 0.17 ≥13 septum thickness 102 (62%) 70 (48%) 33 (57%) 0.32
≥30 septum thickness 6 (4%) 3 (2%) 1 (2%) 0.29
52 The role of renin-angiotensin-aldosterone
AGTR1, 1166A>C (rs5186) AA AC CC *P-value Septum thickness (mm), total population 16±6 (n=175) 15±6 (n=160) 15±7 (n=33) 0.65 probands 22±5 (n=50) 21±4 (n=41) 24±8 (n=9) relatives 13±4 (n=125) 12±5 (n=119) 12±3 (n=24) Wigle score 0 (0-4) 0 (0-3) 0 (0-4) 0,13 ≥13 septum thickness 102 (58%) 85 (53%) 18 (55%) 0.30 ≥30 septum thickness 6 (3%) 2 (1%) 2 (6%) 0.43
CMA1, -1903 A>G GG AG AA *P-value (rs1800875) Septum thickness (mm), total population 15±7 (n=87) 15±6 (n=177) 15±6 (n=104) 0.33 probands 22±7 (n=26) 21±5 (n=44) 21±5 (n=30) relatives 12±4 (n=61) 13±4 (n=133) 12±4 (n=74) 3 Wigle score 0 (0-2) 0 (0-4) 0 (0-3) 0.63 ≥13 septum thickness 44 (51%) 106 (60%) 55 (53%) 0.86 ≥30 septum thickness 4 (5%) 4 (2%) 2 (2%) 0.68 CYP11B2, -344C>T TT TC CC *P-value (rs1799998) Septum thickness (mm), total population 15±6 (n=132) 15±6 (n=185) 16±6 (n=51) 0.61 probands 21±6 (n=36) 23±5 (n=43) 21±4 (n=21) relatives 13±4 (n=96) 12±5 (n=142) 12±4 (n=30) Wigle score 0 (0-3) 0 (0-2) 1 (0-4) 0.95 ≥13 septum thickness 76 (58%) 93 (50%) 36 (71%) 0.07 ≥30 septum thickness 4 (3%) 6 (3%) 0 - Data depict mean ± standard deviation, n (%) or median (interquartile range). *Association analyses wereperformed assuming a recessive model with adjustment for sex, age and proband status.
Analyzing the 5 polymorphisms separately with correction for gender, proband status and age using a recessive genetic model (Table 3) showed an association of the T>C (p.M235T) polymorphism in the AGT gene with IVS thickness (p=0.02); the CC genotype (homozygous for threonine at position 235) was associated with attenuated IVS thickness compared to the other AGT genotypes (TC, TT). As expected from the distribution of septum thickness for the three genotype groups of this polymorphism, applying a dominant genetic model resulted in a more significant association (p=2.4e-04). The DD genotype at the ACE I/D polymorphism was associated with a high Wigle score (p=0.05) as compared to the ID
53 Chapter 3 and DD genotypes. Applying an additive or dominant model for the relations between the SNPs and the phenotypes did not reveal any other significant associations. Association analysis results were unchanged when the analysis was restricted to only those patients carrying the most prevalent founder mutation (c.2373dupG, see Supplementary Table, Furthermore, no significant interactions were found between the 5 polymorphisms for any of the phenotypes tested. No association was detected between pro-LVH score and IVS thickness or Wigle score in probands or relatives (Tables 4 and 5).
Table 4 | Phenotype distribution between the pro-LVH score groups. Pro LVH score 0 1 2 3 4 P-value Septum thickness 15±6 15±7 16±7 14±5 17±3 total population (n=134) (n=153) (n=64) (n=14) (n=3) 0.41 21±5 22±6 22±6 18±3 19±2 probands (n=36) (n=35) (n=22) (n=5) (n=2) 12±4 13±5 13±4 11±3 14±0 relatives (n=98) (n=118) (n=42) (n=9) (n=1) Wigle score 0 (0-4) 0 (0-2) 1 (0-5) 0 (0-2) 6 (6-7) 0.59 ≥13 septum thickness 73 (54%) 83 (54%) 38 (59%) 8 (57%) 3 (100%) 0.17 ≥30 septum thickness 3 (2%) 5 (3%) 2 (3%) 0 0 0.82 Data depict mean ± standard deviation, n (% within thep articular pro-LVH score group) or median (interquartile range). Association analyses were performed assuming a recessive model with adjustment for sex, age and proband status.
Table 5 | Septum thickness (mm) according to pro-LVH genotype score
Pro-LVH score Probands Probands Relatives Relatives
<16 yr ≥16 yr <16 yr ≥16 yr
(n=3) (n=97) (n=19) (n=247)
0 26 (n=1) 21±5 (n=35) 9±6 (n=6) 13±4 (n=90)
1 - 22±6 (n=35) 8±3 (n=10) 13±5 (n=108)
2 30±14 (n=2) 21±4 (n=20) 9±4 (n=2) 13±4 (n=40)
3 - 18±3 (n=5) 6±0 (n=1) 12±2 (n=8)
4 - 19±2 (n=2) - 14±0 (n=1)
Total 29±10 21±5 8±4 13±4
p-value 0.84 0.52 0.44 0.93 Data depict mean ± standard deviation. Yr = years
54 The role of renin-angiotensin-aldosterone
Discussion In a large cohort of carriers of one of 3 functionally-equivalent truncating Dutch founder mutations inMYBPC3 , we found only minor effects of candidate SNPs in the RAAS system on IVS thickness and Wigle score. These effects were limited to (i) an association of the CC genotype of the AGT T>C (p.M235T) polymorphism with a smaller IVS thickness, and (ii) an association of the DD genotype of the ACE I/D polymorphism with a higher Wigle score. There was no effect of the previously described pro-LVH score on IVS thickness or Wigle score.13, 14
The T>C (p.M235T) polymorphism in the AGT gene has been described as a predisposing factor for cardiac hypertrophy in patients with hypertension, in endurance athletes and in sporadic cases of HCM.21-23 Furthermore, Ortlepp and colleagues found an association between the CC genotype of this polymorphism with increased left ventricular mass and increased interventricular septum thickness.13 The AGT gene encodes angiotensinogen and the C allele of this polymorphism is associated with elevated angiotensinogen serum concentrations.24 Although we found an association between the CC genotype of theAGT 3 T>C polymorphism (p.M235T) and IVS thickness, it is important to note that the direction of the effect in our patient population was opposite to what is predicted based on these previous studies, i.e. increased hypertrophy in the presence of elevated angiotensinogen concentrations. Since the observed association in our study was not supported bya decreased risk for an IVS thickness of ≥ 13mm or ≥ 30mm, the observed association is suggestive of a spurious association. Indeed, also no association was detected for this polymorphism with Wigle score.
Tissue levels of angiotensin converting enzyme are increased in patients with the DD-ACE genotype which is considered to be a pro-LVH genotype.25,26 Furthermore, the previous study by Perkins and co-workers showed that in 63 HCM patients with single (different) mutations in the MYBPC3 gene, the DD-ACE genotype was a significant pro-LVH modifier, being associated with an increased left ventricular wall thickness, and with extreme IVS thickness (>30mm).14 Similar findings were reported by Ortlepp and co-workers in twenty-six c.2373dupG mutation carriers from one family.13 In our study, although we found that the DD-ACE was significantly associated with the Wigle score, suggesting a pro- hypertrophic effect, we did not detect an association with IVS thickness or risk of having an IVS thickness of ≥13mm or ≥30mm.
There was no effect of the combined pro-LVH genotypes in the pro-LVH score onIVS thickness or Wigle score. This is in contrast to the report of Ortlepp et al., where among the twenty-six c.2373dupG mutation carriers extent of cardiac hypertrophy was associated with the burden of pro-LVH genotypes13 Although the same mutation was present in 257
55 Chapter 3 of the 368 patients (70%) of the current study, we failed to observe a relation between the pro-LVH score and hypertrophy in our much larger study population. Furthermore, we found no major effects of the pro-LVH genotypes when the patient subset with the c.2373dupG was studied separately. Our findings in a much larger set of patients with the same mutation suggest that cardiac hypertrophy in MYBPC3-related HCM is not influenced by the pro-LVH score / genotypes.
IVS thickness in HCM mutation carriers increases progressively with age.8,9 In our study this was the case for relatives carrying MYBPC3 mutations but not for probands. A major difference between probands and relatives is their reason for cardiac evaluation. Probands are referred because of symptoms, abnormalities at routine physical examination or electrocardiography, i.e. before non-cardiac surgery; whereas relatives are usually asymptomatic and are referred after positive presymptomatic DNA testing. Therefore a sampling or referral bias may occur; HCM patients with severe IVS thickening at young age are more likely to be referred because of symptoms than those with minimal or moderate thickening. The fact that older probands displayed less extensive IVS thickness than the younger ones might be due to the fact that probands with larger IVS thickness die at younger age (the so-called healthy survivor phenomenon). Besides this possible explanation, it is known that about 5-10% of HCM patients progress to an end-stage form, which is characterized by systolic dysfunction, dilatation of the left ventricle and wall thinning27, 28.
Males and females differ in their presentation of HCM, with cohorts usually having a predominance of males.29,30,31 In our study an effect of gender on age at onset in the relatives was observed: men were affected at younger age than women. This is may be explained by a protective role of estrogens in the hypertrophic response and the evidence that exposure of cardiac myocytes to androgen results in hypertrophy.32, 33 Furthermore the HCM phenotype is influenced by sex hormone receptor variants.34 However, in our study, the sex effect in the probands was less clear, with females tending to be more heavily affected at young age, although one must acknowledge that there was considerable overlap between males and females in this group (Figure 2). This illustrates that there are other, currently unknown modifiers of phenotypic expression in HCM.
In conclusion, we have investigated the role of genetic polymorphisms in genes of the RAAS pathway in a large and genetically homogeneous HCM population. Our findings do not provide support for a marked effect of genetic variation in the RAAS pathway on phenotypic expression of LVH in this disorder. This does not necessarily mean that HCM patients will not benefit from the prescription of drugs blocking the RAAS system35. However, small studies showing a positive effect of RAAS inhibitors on progression of
56 The role of renin-angiotensin-aldosterone hypertrophy and fibrosis in HCM need confirmation. 36, 37 Furthermore, our study was limited to 5 candidate polymorphisms and the role of other common genetic variants in explaining phenotypic variability among HCM patients, which may act in a mutation- specific way, merits investigation perhaps in future genome-wide association studies.
3
57 Chapter 3
Reference List 1. Maron BJ, Gardin JM, Flack JM, et al. 6. Michels M, Soliman OI, Kofflard MJ, et Prevalence of hypertrophic cardiomyopathy al. Diastolic abnormalities as the first in a general population of young adults. feature of hypertrophic cardiomyopathy Echocardiographic analysis of 4111 in Dutch myosin-binding protein C founder subjects in the CARDIA Study. Coronary mutations. JACC Imaging. 2009;2:58-64. Artery Risk Development in (Young) Adults. 7. Michels M, Soliman OI, Phefferkorn J, et al. Circulation. 1995;92:785-789. Disease penetrance and risk stratification 2. Task Force for Diagnosis and Treatment for sudden cardiac death in asymptomatic of Acute and Chronic Heart Failure of hypertrophic cardiomyopathy mutation European Society of C, Dickstein K, Cohen- carriers. Eur Heart J. 2009;30:2593-2598. Solal A, Filippatos G, et al. Document R. ESC 8. Christiaans I, Birnie E, van Langen IM, Guidelines for the diagnosis and treatment et al. The yield of risk stratification for of acute and chronic heart failure 2008: the sudden cardiac death in hypertrophic Task Force for the Diagnosis and Treatment cardiomyopathy myosin-binding protein C of Acute and Chronic Heart Failure 2008 gene mutation carriers: focus on predictive of the European Society of Cardiology. screening. Eur Heart J. 2010; 31:842-848. Developed in collaboration with the Heart 9. Kim S, Iwao H. Molecular and cellular Failure Association of the ESC (HFA) and mechanisms of angiotensin II-mediated endorsed by the European Society of cardiovascular and renal diseases. Intensive Care Medicine (ESICM). Eur Heart Pharmacol Rev.2000;52:11-34. J. 2008;29:2388-2442. 10. Tom B, Garrelds IM, Scalbert E, Stegman 3. Bos JM, Towbin JA, Ackerman MJ. APA, Boomsma F, Saxena PR, Danser Diagnostic, prognostic, and therapeutic AHJ. ACE- versus chymase-dependent implications of genetic testing for angiotensin II generation in human hypertrophic cardiomyopathy. J Am Coll coronary arteries. Arterioscler Thromb Cardiol. 2009;54:201-211. Vasc Biol.2003;23:251-256. 4. Alders M, Jongbloed R, Deelen W, van den 11. Chai W, Hofland J, Jansen PM, et al. Wijngaard A, et al. The 2373insG mutation Steroidogenesis vs, steroid uptake in the in the MYBPC3 gene is a founder mutation, heart: do corticosteroids mediate effects which accounts for nearly one-fourth of via cardiac mineralocorticoid receptors? J the HCM cases in the Netherlands. Eur Hypertension 2010;28:1044-53. Heart J. 2003;24:1848-1853. 12. Orenes-Piñero E, Hernández-Romero D, 5. van Dijk SJ, Dooijes D, dos Remedios Jover E, Valdés M, Lip GY, Marín F. Impact of C, et al. Cardiac myosin-binding polymorphisms in the renin-angiotensin- protein C mutations and hypertrophic aldosterone system on hypertrophic cardiomyopathy: haploinsufficiency, cardiomyopathy. J Renin Angiotensin deranged phosphorylation, and Aldosterone Syst. 2011 Apr 20. [Epub cardiomyocyte dysfunction. Circulation. ahead of print] 2009;119:1473-1483.
58 The role of renin-angiotensin-aldosterone
13. Ortlepp JR, Vosberg HP, Reith S, et al. Genetic 19. Elliott PM, Poloniecki J, Dickie S, et al. Sudden polymorphisms in the renin-angiotensin- death in hypertrophic cardiomyopathy: aldosterone system associated with identification of high risk patients. J Am expression of left ventricular hypertrophy Coll Cardiol.2000;36:2212-2218. in hypertrophic cardiomyopathy: a 20. Christiaans I, van Engelen K, van Langen study of five polymorphic genes in a IM, Birnie E, Bonsel GJ, Elliott PM, Wilde family with a disease causing mutation AA. Risk stratification for sudden cardiac in the myosin binding protein C gene. death in hypertrophic cardiomyopathy: Heart.2002;87:270-275. systematic review of clinical risk markers. 14. Perkins MJ, Van Driest SL, Ellsworth EG, Europace.2010;12:313-321. Will ML, Gersh BJ, Ommen SR, Ackerman 21. Ishanov A, Okamoto H, Yoneya K, et al. MJ. Gene-specific modifying effects of pro- Angiotensinogen gene polymorphism LVH polymorphisms involving the renin- in Japanese patients with hypertrophic angiotensin-aldosterone system among cardiomyopathy. Am Heart J. 1997;133:184- 389 unrelated patients with hypertrophic 189. cardiomyopathy. Eur Heart J.2005;26:2457- 22. Corvol P, Jeunemaitre X. Molecular 2462. 3 genetics of human hypertension: 15. Michels M, Hoedemaekers YM, Kofflard role of angiotensinogen. Endocrine MJ, et al. Familial screening and reviews.1997;18:662-677. genetic counselling in hypertrophic 23. Karjalainen J, Kujala UM, Stolt A, cardiomyopathy: the Rotterdam Mantysaari M, Viitasalo M, Kainulainen K, experience. Neth Heart J.2007;15:184-190. Kontula K. Angiotensinogen gene M235T 16. Christiaans I, Birnie E, Bonsel GJ, Wilde polymorphism predicts left ventricular AA, van Langen IM. Uptake of genetic hypertrophy in endurance athletes. J Am counselling and predictive DNA testing in Coll Cardiol.1999;34:494-499. hypertrophic cardiomyopathy. Eur J Hum 24. Danser AHJ, Derkx FHM, Hense HW, Genet. 2008;16:1201-1207. Jeunemaitre X, Riegger GAJ, Schunkert H. 17. McKenna WJ, Spirito P, Desnos M, Dubourg Angiotensinogen (M235T) and angiotensin- O, Komajda M. Experience from clinical converting enzym (I/D) polymorphims in genetics in hypertrophic cardiomyopathy: association with plasma renin and prorenin proposal for new diagnostic criteria in levels. J Hypertension 1998, 16:1879-1883. adult members of affected families. 25. Danser AHJ, Schaalekamp MADH, Bax WA, Heart.1997;77:130-132. Maassen van den Brink A, Saxena PR, Riegger 18. Wigle ED, Sasson Z, Henderson MA, GAJ, Schunkert H. Angiotensin-converting Ruddy TD, Fulop J, Rakowski H, Williams enzyme in the human heart: effect of WG. Hypertrophic cardiomyopathy. The the deletion/insertion polymorphism. importance of the site and the extent of Circulation.1995;92:1387-1388. hypertrophy. A review. Prog Cardiovasc 26. Wang JG, Staessen JA. Genetic Dis.1985;28:1-83. polymorphisms in the renin-angiotensin system: relevance for susceptibility to cardiovascular disease. Eur J Pharmacol.2000;410:289-302.
59 Chapter 3
27. Maron BJ, Casey SA, Hurrell DG, Aeppli 35. Brugts JJ, Isaacs A, de Maat MP, et al. A DM. Relation of left ventricular thickness pharmacogenetic analysis of determinants to age and gender in hypertrophic of hypertension and blood pressure cardiomyopathy. Am J Cardiol respons to angiotensin-converting 2003;91:1195-1198. enzyma inhibitor therapy in patients with 28. Maron BJ. Hypertrophic cardiomyopathy: vascular disease and healthy individuals. J. a systematic review. Jama 2002;287:1308- Hypertens. 2011, 29:509-519. 1320. 36. Yamazaki T, Suzuki J, Shimamoto R, Tsuji 29. Spirito P, Bellone P, Harris KM, Bernabo T, Ohmoto K, Nagai R. A new therapeutic P, Bruzzi P, Maron BJ. Magnitude of strategy for hypertrophic nonobstructive left ventricular hypertrophy and risk cardiomyopathy in humans. A randomized of sudden death in hypertrophic and prospective study with an Angiotensin cardiomyopathy. The New England journal II receptor blocker. Int Heart J. 2007;48:715- of medicine.2000;342:1778-1785. 24. 30. Van Driest SL, Ommen SR, Tajik AJ, Gersh 37. Penicka M, Gregor P, Kerekes R, Marek BJ, Ackerman MJ. Yield of genetic testing in D, Curila K, Krupicka J. The effects of hypertrophic cardiomyopathy. Mayo Clin candesartan on left ventricular hypertrophy Proc.2005;80:739-744. and function in nonobstructive hypertrophic cardiomyopathy: a pilot, randomized study. 31. Olivotto I, Maron MS, Selcuk Adabag A, J Mol Diagn. 2009;11:35-41. et al. Gender-related differences in the clinical presentation and outcome of hypertrophic cardiomyopathy. J Am Coll Cardiol 2005;45:480-7. 32. Marsh JD, Lehmann MH, Ritchie RH, Gwathmey JK, Green GE, Schiebinger RJ. Androgen receptors mediate hypertrophy in cardiac myocytes. Circulation.1998;98:256-261. 33. Xin HB, Senbonmatsu T, Cheng DS, et al. Oestrogen protects FKBP12.6 null mice from cardiac hypertrophy. Nature.2002;416:334-338. 34. Lind JM, Chiu C, Ingles J, Yeates L, Humphries SE, Heather AK, Semsarian C. Sex hormone receptor gene variation associated with phenotype in male hypertrophic cardiomyopathy patients. J Mol Cell Cardiol 2008;45:217-222.
60 Chapter 4
Genetic Modifiers of disease expression in patients with Long QT Syndrome Type 2
Iris C.R.M. Kolder1,2*, Michael W.T. Tanck1*, Vincent Probst5,16,17,18, Florence Kyndt5,16,17,18, Julien G. Barc2,18, Sven Zumhagen19, Anja Husemann19,20, Pieter G. Postema2, Moritz F. Sinner3,4, Tamara T. Koopmann1, Nynke Hofman, Arne Pfeufer5,6,7¸ Peter Lichtner5, Thomas Meitinger5,7, Robert J. Myerburg8,9,10, Nanette H. Bishopric8,9,10, Dan M. Roden11,12,13, Stefan Kääb3,14, Arthur A.M. Wilde2, Jean-Jacques Schott15,16,17,18,∫, Eric Schulze-Bähr19,20,∫, Connie R. Bezzina2,∫
*∫These authors contributed equally
1 Department of Clinical Epidemiology, Biostatistics and Bioinformatics, and 2 Heart Failure Research Center, Department of Clinical and Experimental Cardiology, Academic Medical Center, Amsterdam, The Netherlands 3 Department of Medicine I, 14 Munich Heart Alliance, University Hospital Grosshadern, Lüdwig-Maximillians-Universität, Munich, Germany 4 Cardiovascular Research Center, Massachusetts General Hospital, Boston, MA, USA. 5 Institute of Human Genetics, Helmholtz Zentrum Munich, Neuherberg, Germany. 6 Institute of Genetic Medicine, EURAC Research, Bolzano, Italy 7 Institute of Human Genetics, Technische Universität München, Munich, Germany. 8 Department of Medicine, 9 Department of Molecular and Cellular Pharmacology, 10 Hussmann Institute of Human Genomics, University of Miami Miller School of Medicine, Miami, Florida, USA 11 Department of Medicine, 12 Department of Molecular Physiology and Biophysics, 13 Department of Pharmacology, Vanderbilt University School of Medicine, Nashville, Tennessee, USA 15 INSERM, UMR915. l’institut du Thorax, Nantes, France 16 CNRS, ERL3147, 17 CHU Nantes, 18 Université de Nantes, Nantes, France 19 Institute for Genetics of Heart Diseases, Department of Cardiology and Angiology, University Hospital Münster, Germany, 20 Interdisciplinary Centre for Clinical Research (IZKF) of the University of Münster, Germany
To be submitted Chapter 4
Abstract Rationale Long QT syndrome type 2 (LQT2) is a cardiac repolarization disorder that is caused by mutations in theKCNH2 gene which encodes kv11.1 (HERG) underlying the Ikr repolarizing K+ current. Although mutation type and location are known to impact on the severity of clinical manifestations, the occurrence of phenotypic variability among patients harboring the same mutation points to the presence of additional modulatory genetic factors.
Objective In this study we undertook a candidate gene-based approach to identify genetic modifiers of disease expression in this disorder. A total of 1201 haplotype-tagging SNPs across 18 candidate genes were tested for association with QTc-interval in 438 patients with LQT2. In association analysis of the genetic variants with QTc, for the first time, we corrected for effects of KCNH2 mutation type and location.
Results Two SNPs, rs16847548 located immediately 5’upstream of the NOS1AP gene, and rs956642, located in the vicinity of KCNH2, were associated with the QTc-interval at p-values exceeding the Bonferroni-corrected significance threshold for association (p<4.16×10-5). Of these, rs956642 was also found to be associated with cardiac events (p=0.02). Two other SNPs in NOS1AP, rs10494366 and rs12567211, which like rs16847548 were previously found to impact on QTc in the general population, were also significantly associated with QTc in the LQT2 patients studied (p-values, 9.96×10-3 and 2.24×10-4, respectively).
Conclusion We provide the first evidence that common genetic variants at theKCNH2 locus modulate severity of clincal manifestations in the Long QT Syndrome. Furthermore, we extend previous observations that common genetic variants in NOS1AP modulate the extent of QTc-prolongation in this disorder.
Abbreviations GWAS, genome-wide association study LD, linkage disequilibrium LQTS, Long QT Syndrome LQT2, Long QT Syndrome type 2
62 Genetic Modifiers of disease expression
Introduction The Long QT syndrome (LQTS) is a congenital disorder caused by mutations in several genes primarily encoding ion channels. It is associated with an increased risk of sudden cardiac death from torsades de pointes polymorphic ventricular tachycardia and is clinically diagnosed by a prolonged QT-interval on the electrocardiogram (ECG). Mutations in the KCNH2 gene have been linked to the type 2 LQTS (LQT2) and account for ~40% of genotyped LQT patients 1. KCNH2 encodes the pore-forming subunit of the channel conducting the rapid component of the delayed rectifier repolarizing K+ current
(IKr) and mutations causing LQTS lead to a reduction of this current, thereby prolonging cardiomyocyte repolarization and prolonging the QT-interval on the ECG. 2
Shimizu and co-workers have previously demonstrated that among patients with LQT2, the type and location of the KCNH2 mutation exerted a marked effect on the extent of QT-interval prolongation and occurrence of cardiac events 3. However although this explains some of the variability in disease severity among patients, the occurrence of phenotypic variability among patients harboring similar mutations points to the presence of additional modulatory genetic factors 4, 5. In this study we undertook a candidate gene-based approach to identify genetic modifiers of disease expression in a large set of patients with LQT2. In this analysis for genetic modifiers of disease expression in the LQTS, for the first time we also took into consideration the effects of mutation type and location. 4 Material and Methods Study population The study population consisted of 438 LQT2 probands and their relatives, all harboring a mutation in the KCNH2 gene. Patients carrying >1 mutation in KCNH2 or a second mutation in another LQTS gene were not included. These subjects were drawn from the LQTS registries of four European centers, namely Amsterdam (The Netherlands), Münster (Germany), Munich (Germany), and Nantes (France). All subjects were of European descent. The study was approved by the Medical Ethical Committees at the respective centers. All subjects or their guardians provided informed consent for genetic and clinical studies.
Phenotyping Routine clinical and ECG parameters were acquired at the time of patient enrollment in each of the registries. ECGs were digitalized and analyzed using ImageJ (http://rsb.info. nih.gov/ij/) blinded to genotype. Only sinus rhythm complexes were analyzed. The QT- interval was measured manually on-screen using the tangent method 6. Lead II was used whenever possible 7. The average QT-interval from up to 3 consecutive beats with similar
63 Chapter 4 preceding RR-intervals was calculated and the QTc-interval was expressed using Bazett’s formula. A first cardiac event was defined as a first unexplained syncope, a first aborted cardiac arrest (requiring resuscitation) or a first ventricular tachycardia or fibrillation. The observation period for cardiac events started at birth and lasted to the initiation of anti- adrenergic therapy (b-blockers) or the date of the last visit.
SNP selection and genotyping Eighteen candidate genes known to be involved in cardiac electrophysiology were selected (Table 1). These included genes involved in cardiac arrhythmia syndromes, functionally important subunits of these genes, and genes significantly associated with the QTc- interval in the general population based on the results of genome-wide association studies (GWAS). Each gene was analyzed systematically using a haplotype-tagging approach. To account for genetic variation within regulatory regions of genes, we also considered genetic variation immediately upstream and downstream of the candidate genes. Single nucleotide polymorphisms (SNPs) for genotyping were selected from all HapMap SNPs (http://www.hapmap.org/) available for the CEU population within the genes and 50 kb upstream and downstream of these genes. Tag-SNPs were selected using the Tagger program8 using the following criteria: HapMap CEU population, pairwise only tagging with r2 ≥ 0.8 and a minor allele frequency (MAF) ≥ 10%. SNPs with low Illumina GoldenGate quality design scores were replaced where possible by another SNP tagging the same haplotype block. In addition, we included 38 non-synonymous SNPs with minor allele frequencies ≥1% previously reported in the 18 candidate genes. A total of 1424 SNPs were derived in this way for genotype analysis. SNP genotyping was performed using a custom assay (GoldenGate) on an Illumina-BeadStation500GX (Illumina Inc., SanDiego, USA) at the Genome Analysis Center, Helmholtz Zentrum Munich, Germany. The Illumina BeadStudio software clustering algorithm was used for initial data analysis. Thereafter, intensity plots of all variants were examined individually and manual genotype calling was performed if necessary.
SNPs and samples were subsequently checked for quality. Monomorphic SNPs and SNPs with a MAF < 1%, as well as SNPs and samples with call rates <95% were removed from further analyses. A total of 1,201 SNPs had sufficient quality for analysis. All 438 samples had call rates ≥95% and were included in the analysis.
Statistical analyses QTc-interval data were normally distributed (Shapiro-Wilk statistic W>0.90) and are reported as the mean ± standard deviation. Although all subjects were of European descent we nevertheless tested for any population stratification using principle component analysis in the GenABEL package 9 in R (http://www.rproject.org/). No such stratification was detected.
64 Genetic Modifiers of disease expression
Table 1 | Overview of candidate genes and SNPs tested. Gene Chromosome Tag SNPs passing QC Ns SNPs AKAP9 7q21-q22 14 2 ANK2 4q25-q27 103 2 CACNA1C 12p13.3 150 CASQ2 1p13.3-p11 59 CAV3 3p25 55 FKBP1B 2p23.3 12 GPD1L 3p22.3 40 KCNE1 21q22.1-q22.2 57 1 KCNE2 21q22.1 24 KCNH2 7q35-q36 58 1 KCNJ2 17q23.1-q24.2 37 KCNQ1 11p15.5 112 NOS1AP 1q23.3 110 1 RYR2 1q42.1-q43 201 1 SCN1B 19q13.1 22 SCN4A 17q23.1-q25.3 22 1 SCN4B 11q23 45 SCN5A 3p21 70 1 4 Total 1191 10 Tag SNP, haplotype-tagging SNP, Ns SNP, non-synonymous SNP
Effects of KCNH2 mutation type and/or location, effects of SNPs and effects of covariates on the QTc-interval were estimated using the linear mixed effect model function (lmekin) in the Kinship package 10 in R, thus correcting for dependency between some study subjects due to familial relatedness. The effect of SNPs on the secondary endpoint ‘age at first cardiac event’, were estimated using the cox proportional hazards function (coxme) in Kinship. For each SNP-phenotype relationship, an additive genetic model was assumed unless mentioned otherwise. The effects of the SNPs were adjusted for sex, age at ECG recording, proband status, b-blocker use at the time of the ECG, and for mutation type and location. With respect to mutation type and location, mutations were classified into 5 different classes: (1) nonsense, frameshift, large deletions and insertions, all locations, 2) missense, N-terminus, 3) missense, transmembrane S1-S4, (4) missense, transmembrane S5-loop-S6, and (5) missense, C-terminus. This annotation was based on the Uniprot database (http://www.uniprot.org/uniprot/Q12809, version January 2012). In association analysis, the Bonferroni corrected significance threshold was set at p=4.16×10-5 (0.05/1201).
65 Chapter 4
SNPs associated with QTc-interval at genome-wide statistical significance in published genome-wide association studies 11-14 were checked for LD (r2 >0.7) with genotyped SNPs using SNAP (http://www.broadinstitute.org/mpg/snap/ldsearchpw.php).
Results Study population The characteristics of the 438 LQT2 patients studied are presented in Table 2. A total of 137 (31%) patients were probands and 260 (59%) were female. The median age at ECG was 33 (IQR=30). As expected, probands had a significantly longer QTc-interval compared to relatives: 486 ± 48 vs. 459 ± 38 ms (p=5.8×10-10). Males and females had similar QTc- intervals (465 ± 47, males; 469 ± 40 ms, females; p=0.31). Beta-blocker use at the time of the ECG did not affect the QTc-interval (467 ± 44 in non-users; 470 ± 41 in users; p=0.65).
Table 2 | Characteristics of the LQT2 patients studied.
Total n = 438 Female 260 (59%) Proband 137 (31%) Median (IQR) age at ECG (years) 33 (30) b-blocker use at time of ECG 79 (18%) Mean (±SD) QTc-interval (ms) 467 ± 43 Cardiac event 149 (34%) Median (IQR) follow-up (years) 24 (29)
A total of 131 different KCNH2 mutations (Figure 1, Supplemental Table 1) were present among the 438 patients who originated from 187 different families. The number of patients per family ranged from 1 to 20. Missense mutations accounted for 56.5% (n=74) of the mutations. Frameshift mutations accounted for 31.3% (n=41), and nonsense mutations for 7.6% (n=10). In-frame deletions (n=3), large deletions (n=2) and large duplications (n=1) accounted for 2.3, 1.5 and 0.8%, respectively. Mutations were located within four different channel sub-domains. Most mutations were located within the N-terminus (n=41, 31.3%), the transmembrane “pore” region (S5-loop-S6; n=39, 29.8%) and the C-terminus (n=40, 30.5%) of the channel. Eight mutations (6.1%) were located in the transmembrane “non-pore” region (S1-S4 transmembrane segments) while the two large deletions and the large duplication (2.3%) affected multiple regions.
66 Genetic Modifiers of disease expression
Table 3 | Effects of KCNH2 mutation type and location on QTc-interval.
Location of the mutation n=426* QTc-interval (ms) N-terminus 143 (34%) 462 ± 47 transmembrane S1-S4 19 (4%) 509 ± 49 transmembrane S5-loop-S6 119 (28%) 473 ± 42 C-terminus 145 (34%) 464 ± 36 Mutation type and location n=438 QTc-interval (ms) nonsense, frameshift, large deletions and insertions, all locations 169 (39%) 465 ± 39 missense, N-terminus 113 (26%) 460 ± 46 missense, transmembrane S1-S4 15 (3%) 518 ± 49 missense, transmembrane S5-loop-S6 108 (25%) 476 ± 42 missense, C-terminus 33 (8%) 454 ± 31 *The two large KCNH2 gene deletions and the large duplication (12 individuals) were excluded as these affected multiple channel domains.
Effects of KCNH2 mutation type and location Since the type and location of the mutation may affect the extent of QTc-interval prolongation 3 we first evaluated whether there were any effects of mutation type and location on the extent of QTc-interval prolongation in our LQT2 patient sample. The type 4 of mutation (missense vs. non-missense i.e. nonsense, frameshift, large duplications/ deletions) was not associated with the extent of QTc-interval (p=0.43). When the location of missense mutation was considered, we found that carriers of a missense mutation in the transmembrane non-pore region (S1-S4) had on average a longer QTc-interval compared to individuals carrying a missense mutation in any of the other 3 locations, i.e. transmembrane pore region (S5-pore-S6), N-terminus or C-terminus (p=2.4×10-4, Table 3). When the 15 patients with S1-S4 region missense mutations were excluded (2 different mutations), patients with a missense mutation in the pore region (transmembrane S5- loop-S6) displayed a longer QTc-interval compared to patients with a non-pore missense mutation (N- or C-terminal) (476 ± 42 vs. 458 ± 43, p=0.016).
SNP effects on QTc-interval and cardiac events A total of 1201 SNPs across the 18 candidate genes were tested for association with QTc- interval in the 438 LQT2 patients. As the KCNH2 mutation type and location impacted on the QTc-interval, we adjusted for this, in addition to sex, age at ECG and proband status. The association results for all SNPs displaying association with QTc-interval at p≤0.01 are
67 Chapter 4
Figure 1 | Location of the 128 different mutations in the Kv11.1 (HERG) potassium channel encoded by the KCNH2 gene. The open circles represent individual amino acids, the blue circles indicate the location of missense mutations, and the red circles indicate the location of non-missesnce mutations. The purple circles indicates the location where both missense and non-missense mutations occur. The transmembrane S1-S4 region was defined as amino acid residues 404 to 547. The transmembrane loop region was defined as residues 548 to 659. Residues 1-403 were defined as N-terminus, while residues 660 to 1159 were defined as the C-terminus. The cylinders represent putative α-helical segments, and the bars represent putative β-sheets. These annotations are based on the Uniprot database (http://www. uniprot.org/uniprot/Q12809, version January 2012). presented in Table 4 and the results for all association tests are shown in Supplemental Table 2. Two SNPs, rs16847548 at the NOS1AP locus and rs956642 at the KCNH2 locus, passed the Bonferroni-corrected significance threshold of p<4.2×10-5. The minor allele of rs16847548 was associated with an increase in QTc-interval of ~13 ms, while the minor allele of rs956642 was associated with ~12ms shortening in QTc-interval. Figure 2 displays the regional association plots at these two loci. SNP rs16847548 is located at 4.5kb upstream of NOS1AP while rs956642 is located at 50kb downstream of KCNH2. We next tested whether any of these two SNPs was also associated with risk of cardiac events. Only rs956642 near KCNH2 displayed a trend for association with reduced risk of cardiac events (RR=0.78, 95% CI 0.60 - 1.01, p=0.06). Since it has been previously suggested that SNP effects on risk of cardiac events might be more pronounced in patients with QTc-<500 ms, as the risk of cardiac events in patients with QTc>500ms is already very high 15, we next
68 Genetic Modifiers of disease expression
A.
B. 4
Figure 2 | Regional association plots at the NOS1AP (upper panel) and KCNH2 (lower panel) loci.
69 Chapter 4 tested whether the association of rs956642 was more pronounced when we considered only patients with QTc<500ms. When this patient category was considered, rs956642 was found to be associated with cardiac events at a p-value of 0.02 (RR=0.70, 95% CI 0.53- 0.94, p=0.02). The minor allele was associated with a decreased risk of cardiac events in line with the QTc-abbreviating effect of this allele.
We next assessed whether SNPs that have been previously linked to the QTc-interval in GWAS conducted in the general population 11-14 also impacted on QTc-interval in our LQT2 patient set. For SNPs that were not directly genotyped, we considered SNPs that had an r2 > 0.7 with those found in previous GWAS; when multiple SNPs with r2 > 0.7 were present, the SNP with the strongest LD was considered. Association data with QTc- interval for these SNPs is presented in Table 5. Four SNPs from previous GWAS at the NOS1AP locus were either genotyped directly or were well-tagged. SNP rs16847548 displaying the strongest association with QTc-interval in our set (see above, p=1.28×10-5) is in fact in high LD (r2 =0.88) with rs12143842, which was the highest ranking SNP in the two recent meta-analysis of QTc-GWAS in multiple sets of the general population13, 14. SNP rs10494366 from previous GWAS was typed directly and was associated with QTc-interval at a p-value of 9.96×10-3. SNP rs2880058 was tagged by rs12567211, which was associated with QTc-interval at a p-value of 2.24×10-4. The direction of effect at all these three SNPs was concordant with the effect direction in previous GWAS; in all cases the minor allele had a prolonging effect on the QTc-interval. In aggregate, rs10494366, rs16847548 and rs12567211 explained 13.3% of the variance observed in QTc-interval among the patients, bringing the total explained variance from 15.3% (due to effects of sex, age, proband status and mutation type and location) to 28.64%. The fourth SNP, rs4657178, was typed directly. Using an additive genetic model, no association was observed with QTc-interval. However a trend for an association with QTc-interval (p=0.08) was observed when a recessive model was applied in the association analysis (AA+Aa vs. aa). Of the 4 NOS1AP GWAS SNPs that were typed or captured by proxies, only rs4657178 showed a trend for association with cardiac events (RR=1.31, 95% CI 0.99-1.72, p=0.06).The association improved when only individuals with QTc<500ms were considered (RR=1.45, 95% CI 1.07- 1.98, p=0.02). The QTc-prolonging allele was the risk-allele.
70 Genetic Modifiers of disease expression
Table 4 | SNPs associated with QTc-interval in LQT2 patients at p<0.01. SNP§ Chr Gene Position Major Minor MAF Effect on P-value allele allele QTc-interval QTc- #β±SE (ms) interval* rs16847548 1 NOS1AP 162035274 T C 0.25 13.66 ± 3.12 1.28e-05 rs956642 7 KCNH2 150596502 A G 0.39 -12.14 ± 2.86 2.02e-05 rs12567211 1 NOS1AP 162036523 G T 0.31 10.63 ± 2.89 2.24e-04 rs1935778 1 CASQ2 116394699 A G 0.42 9.18 ± 2.65 5.45e-04 rs6677529 1 NOS1AP 162263754 C A 0.17 -12.17 ± 3.8 1.29e-03 rs6660701 1 NOS1AP 162040906 G C 0.39 9.12 ± 2.9 1.61e-03 rs2968857 7 KCNH2 150662330 T C 0.40 -8.9 ± 2.91 2.06e-03 rs929492 12 CACNA1C 2670580 C T 0.14 11.94 ± 4.02 2.72e-03 rs7944321 11 SCN4B 118055801 G A 0.21 -10.77 ± 3.66 2.94e-03 rs1415262 1 NOS1AP 162046135 G C 0.38 8.46 ± 2.9 3.43e-03 rs6674675 1 RYR2 237589433 A G 0.37 8.48 ± 2.95 3.95e-03 rs4656349 1 NOS1AP 162049824 A G 0.34 8.36 ± 2.94 4.26e-03 rs12703105 7 KCNH2 150594658 C A 0.26 7.36 ± 2.6 4.31e-03 rs3918227 7 KCNH2 150700946 C A 0.08 13.85 ± 5 5.40e-03 rs4657180 1 NOS1AP 162247738 C T 0.09 -13.44 ± 4.92 5.99e-03 rs7620066 3 GPD1L 32219424 A T 0.30 -8.06 ± 2.98 6.41e-03 rs4240550 1 CASQ2 116305471 T G 0.42 7.29 ± 2.73 6.94e-03 4 rs7539281 1 NOS1AP 162008034 G A 0.29 8.06 ± 3 7.02e-03 rs10923142 1 CASQ2 116204661 C T 0.17 9.8 ± 3.66 7.37e-03 rs2834432 21 KCNE2 35644787 C G 0.12 -11.2 ± 4.23 7.41e-03 rs313956 4 ANK2 114023849 G A 0.22 8.83 ± 3.37 8.09e-03 rs10399824 1 CASQ2 116371836 A G 0.07 14.76 ± 5.64 8.16e-03 rs6677154 1 NOS1AP 162254664 A G 0.25 8.66 ± 3.31 8.22e-03 rs2021746 17 SCN4A 62133488 T A 0.12 -11.42 ± 4.36 8.44e-03 rs11767582 7 KCNH2 150523344 T G 0.30 -8.27 ± 3.22 9.34e-03 rs10754345 1 CASQ2 116275152 G A 0.33 7.27 ± 2.84 9.87e-03 rs10494366 1 NOS1AP 162085685 T G 0.39 7.47 ± 2.92 9.96e-03 *P-values passing the Bonferroni corrected p-value threshold (p=4.16×10-5) are in bold face. Chr, chromosome. MAF, minor allele frequency. # Coded allele is the minor allele in all cases. §SNPs previously associated with QTc in GWAS in the general population (rs10494366) or which are correlated to SNPs associated with QTc in these GWAS (rs16847548, rs12567211: r2>0.7; rs956642: r2=0.2) are in bold face.
71 Chapter 4
Table 5 | Effects of SNPs in LD with those found associating with QTc-interval in previous genome-wide association studies. GWAS SNP Typed proxy R2 Effect on P-value Concordance QTc-interval with GWAS β±SE (ms) (Y/N) NOS1AP rs10494366 7.47 ± 2.92 9.96×10-3 Y rs12143842 rs16847548 0.88 13.66 ± 3.12 1.28×10-5 Y rs2880058 rs12567211 0.79 10.63 ± 2.89 2.24×10-4 Y rs4657178 0.52 ± 3.14 0.87 SCN5A rs12053903 rs6793245 0.96 -0.34 ± 3.18 0.91 KCNH2 rs4725982 5.87 ± 3.51 0.09/0.02* Y rs2968863 rs1547958 0.96 2.62 ± 3.45 0.44 KCNQ1 rs12296050 rs16928297 1.00 0.76 ± 3.36 0.82 rs12576239 2.9 ± 3.81 0.44 KCNE1 rs1805128 (D85N) 1.00 2.99 ± 9.81 0.76 *Using dominant genetic model
Another 6 SNPs that were associated with QTc-interval in the general population in previous GWAS (2 at the KCNH2 locus, 2 at KCNQ1, 1 at SCN5A and 1 at KCNE1), were either directly typed or were well-tagged (Table 5). Of these, rs4725982, located 4kb downstream of KCNH2 locus, displayed a trend for association with QTc-interval (p=0.09); the association improved when a dominant model was used in association analysis (p=0.02). The minor allele was associated with a longer QTc-interval in concordance with observations in GWAS. No association with cardiac events was detected for this SNP. None of the remaining 5 SNPs displayed association with QTc or cardiac events Table( 5).
Discussion Considerable interest exists in the discovery of genetic factors modulating severity of clinical manifestations in the LQTS as the identification of these factors is expected to contribute to the refinement of individual risk stratification. We here assessed whether genetic variants in candidate genes modulate the extent of QTc-interval prolongation, a major determinant of risk in the LQTS 16, and the occurrence of cardiac events, in patients with LQT2. We undertook a comprehensive approach addressing genetic variation within
72 Genetic Modifiers of disease expression and in flanking regions of genes known to be involved in cardiac electrical function, and thereby likely harboring such variants. In this study, which for the first time also took effects of mutation type and location into account, we confirm and extend previous observations that common genetic variants in NOS1AP modulate disease severity in the Long QT Syndrome. Moreover, we provide the first evidence that common genetic variants at the KCNH2 locus also modulate severity of clinical manifestations in this disorder.
NOS1AP variants, QTc-interval and cardiac events Genome-wide association studies of QTc-interval in the general population have provided strong evidence that common genetic variants immediately upstream of the NOS1AP gene modulate the QTc-interval 11-14. NOS1AP encodes CAPON, a neuronal nitric oxide synthase (NOS1) regulator which is expressed in heart. Multiple independent signals upstream of NOS1AP have been linked with QTc-interval in the general population 13, 14. The exact genetic mechanism involved at these loci remains unknown as for instance no effects of these genetic variants on NOS1AP gene expression have as yet been reported. However, experimental evidence suggests that CAPON interacts with NOS1 to accelerate 17 cardiac repolarization through the inhibition of ICa,L and enhancement of IKr .
Following the discovery of the effects of NOS1AP variants on QTc-interval in the general population, two studies investigated their role in modulation of QTc-interval prolongation and risk of cardiac events in patients with the LQTS15, 18. Crotti and co-workers investigated 4 the effects of these SNPs in South African patients with Long QT type 1 (LQT1), harboring a founder mutation inKCNQ1 (A341V), while Tomás et al. studied a large set of LQTS cases with different causal mutations in any of the KCNQ1, KCNH2, SCN5A, KCNE1 or KCNE2 genes. Although these two studies provided us with important first insights, further studies are needed. For instance while founder mutations provide an excellent setting for carrying out such studies, the study population was rather small and per definition, this study precluded the assessment of SNP effects in other LQTS genetic subtypes. In the other study, KCNH2 mutation carriers constituted only 32% (n=243) of the patients studied and whether the NOS1AP SNP effects observed for the entire set also apply to LQT2 patients remains to be demonstrated. Furthermore, in the latter study effects of mutation type and location were not taken into account (as covariates in the analysis), while these could impact on the SNP effect size estimates.
NOS1AP rs16847548 / rs12143842 In our comprehensive analysis of genetic variation at the NOS1AP locus, rs16847548 exceeded the Bonferroni-corrected significance threshold for association with QTc and was the SNP displaying the lowest p-value in our analysis. This SNP is in high LD with rs12143842, the highest ranking SNP in the two recent meta-analysis of QTc-GWAS in
73 Chapter 4 the general population. Our data are in line with findings of Crotti et al18 and Tomás et al 15 who reported an association for this SNP with QTc-interval in LQT patients. Crotti et al also demonstrated an association with increased risk of cardiac events among theKCNQ1 - A341V mutation carriers studied.
NOS1AP rs12567211 / rs2880058 / rs4657139 SNP rs12567211 was our second independent high-ranking signal for association with QTc- interval at the NOS1AP locus. This SNP is in strong LD with rs2880058 identified in GWAS. The effect on QTc-interval that we demonstrate for this SNP is in line with the findings of Crotti et al.18, and Tomas and co-workers 15 who both demonstrated effects of rs4657139, which is in high LD (r2>0.8) with both SNPs, on both QTc-interval and risk of cardiac events.
NOS1AP rs10494366 The third independent signal at the NOS1AP locus in our study was rs10494366, which was previously found to associate with QTc in GWAS. We here demonstrate for the first time an association between this SNP and QTc-interval among patients with LQTS. This SNP tested negative in relation to cardiac events by Crotti 18 and co-workers and was consequently not tested for effects on QTc by these investigators. It was however associated with risk of cardiac events in the study of Tomás et al 15 in individuals with QTc <500ms. In the latter study, however no effect was detected on the QTc-interval. Considering our findings and the observations in the general population, one could hypothesize that the lack of association with QTc-interval in the latter study may be due to the fact that this study considered a set of LQTS subjects with different genetic causes and did not correct for effects of gene and neither for effects of mutation type and location.
NOS1AP rs4657178 In our study we detected a suggestive association between rs4657178 and cardiac events. However, although this SNP was associated with QTc-interval in GWAS in the general population, we detected no effect on QTc-interval among patients with LQT2. This SNP was not assessed in the two previous studies carried out in LQTS patients (16,19).
In contrast to the studies of Crotti et al. and Tomás et al. which provided evidence for effects of NOS1AP SNPs on the risk of cardiac events, in spite of detecting strong effects on the QTc-interval for specific NOS1AP SNPs, we did not detect any effects on risk of cardiac events.
KCNH2 variants, QTc-interval and cardiac events GWAS in the general population have previously identified two independent loci in the region of KCNH2 impacting on QTc-interval 11. We now provide the first evidence that one of these SNPs, rs4725982, also impacts on the extent of QTc-interval prolongation in patients with LQTS.
74 Genetic Modifiers of disease expression
Furthermore, we show that another SNP at the KCNH2 locus, rs956642, impacted on QTc- interval in patients with LQT2. This SNP does not seem to affect the QTc-interval in the general population 19. Its effect may therefore be specific to patients with the LQTS. One could argue that the already severely compromised repolarization in patients with LQTS provides a setting which is permissive to the effects of variants that are otherwise silent in normal subjects. This may be particularly so as the patient group studied here consists of patients harboring a causal mutation in the same gene. The fact that we here uncover effects of this SNP on both the QTc-interval and the risk of events, coupled to the fact that the allele associated with a decreased risk of cardiac events is, as one would expect, the QTc-abbreviating allele, provides support for the real involvement of this SNP. Rs956642 is located downstream of KCNH2 and is in LD with SNPs within the KCNH2 gene. None of these are associated with coding non-synonymous changes and thus the observed effect on QTc-interval (which may be mediated by any SNP(s) within the haplotype block) is likely to occur through effects on gene expression. Of note, rs956642 is in moderate LD (r2 =0.22, D’=1) with rs2968863 previously associated with QTc-interval at genome-wide significance in GWAS.
Rare non-synonymous variants Our SNP selection strategy entailed the inclusion of low frequency (MAF≥1%) non- synonymous variants in the candidate genes. Although the majority of these were excluded from the analysis at the quality control stage as they were either monomorphic 4 or displayed a very low MAF (<1%) in our patient set, some were included in the association analysis (Supplemental Table 2). One of these was rs1805128, a low frequency variant (MAF=0.03% in HapMap CEU) in KCNE1 associated with the D85N amino acid change. This SNP was initially related to the QTc-interval in a small French study20 Institut de Myologie, IFR 14, UPMC, Groupe Hospitalier Pitie-Salpetriere, Paris, France
75 Chapter 4
Mutation type and location Our findings on the effect of mutation type and location on the QTc-interval in our LQT2 paient set differ somewhat from those of Shimizu et al. In the latter study missense mutations in the pore-region were associated with a more severe phenotype as compared to missense mutations in the other (S1-S4, C-terminus, N-terminus) channel regions. Thre were 555 carriers of missense mutations in that study, of which 33 carried one of 5 different missense mutations in the S1-S4 region. Our LQT2 set included 74 patients with missense mutations, of which 15 carried one of 5 different missense mutations in S1-S4. In our study patients carrying a missense mutation in the transmembrane S1-S4 region displayed the longest QTc-iterval. When these 15 patients were excluded from the analysis, patients with a missense mutation in the pore region had the longest QTc- interval compared to patients with missense mutations in the C- and N-terminal region. While our findings provide additional evidence that missense mutations in thepore- region are associated with a a severe clinical phenotype, they call for a re-evaluation of the impact of S1-S4 region mutations in future, larger studies, which may allow the distiction between the sub-domains within this region (transmembrane vs. extra/intra- cellular loops and different transmembrane sub-regions). Nevertheless, the mutation type × location interaction observed across the studies provides a very strong rationale for correcting for these effects when searching for effects of genetic modifiers of phenotypic expression of the disease.
Conclusion We confirm and extend previous observations that common genetic variants in NOS1AP modulate disease severity in the Long QT Syndrome and provide the first evidence that common genetic variants at the KCNH2 locus also modulate severity of clinical manifestations in this disorder.
Acknowledgements This study was supported by a grant from Fondation Leducq (Trans-Atlantic Network of Excellence “Alliance Against Sudden Cardiac Death”, 05 CVD 01). Dr. Bezzina is supported by the Netherlands Heart Foundation (NHS 2005T024). Prof. Schulze-Bahr is supported by an IZKF grant.
76 Genetic Modifiers of disease expression
Reference List
1. Shimizu,W. & Horie,M. Phenotypic 10. Atkinson,B. & Therneau,T. kinship: mixed- manifestations of mutations in genes effects Cox models, sparse matrices, and encoding subunits of cardiac potassium modeling data from large pedigrees. 2009. channels. Circ. Res. 109, 97-109 (2011). Ref Type: Computer Program 2. Sanguinetti,M.C., Jiang,C., Curran,M.E., 11. Kolder,I.C., Tanck,M.W., & Bezzina,C.R. & Keating,M.T. A mechanistic link Common genetic variation modulating between an inherited and an acquired cardiac ECG parameters and susceptibility cardiac arrhythmia: HERG encodes the to sudden cardiac death. J. Mol. Cell IKr potassium channel. Cell 81, 299-307 Cardiol.(2012). (1995). 12. Arking,D.E. et al. A common genetic variant 3. Shimizu,W. et al. Genotype-phenotype in the NOS1 regulator NOS1AP modulates aspects of type 2 long QT syndrome. J. Am. cardiac repolarization.Nat. Genet. 38, 644- Coll. Cardiol. 54, 2052-2062 (2009). 651 (2006). 4. Scicluna,B.P., Wilde,A.A., & Bezzina,C.R. 13. Newton-Cheh,C. et al. Common variants The primary arrhythmia syndromes: same at ten loci influence QT interval duration in mutation, different manifestations. Are we the QTGEN Study. Nat. Genet. 41, 399-406 starting to understand why? J. Cardiovasc. (2009). Electrophysiol. 19, 445-452 (2008). 14. Pfeufer,A. et al. Common variants at ten 5. Amin,A.S. et al. Variants in the 3’ loci modulate the QT interval duration in untranslated region of the KCNQ1-encoded the QTSCD Study. Nat. Genet. 41, 407-414 Kv7.1 potassium channel modify disease (2009). 4 severity in patients with type 1 long QT 15. Tomas,M. et al. Polymorphisms in the syndrome in an allele-specific manner. Eur. NOS1AP gene modulate QT interval Heart J.(2011). duration and risk of arrhythmias in the 6. Postema,P.G., de Jong,J.S., Van der Bilt,I.A., long QT syndrome. J. Am. Coll. Cardiol. 55, & Wilde,A.A. Accurate electrocardiographic 2745-2752 (2010). assessment of the QT interval: teach the 16. Priori,S.G. et al. Risk stratification in the tangent. Heart Rhythm. 5, 1015-1018 long-QT syndrome. N. Engl. J. Med. 348, (2008). 1866-1874 2003). 7. Moennig,G. et al. Clinical value of 17. Chang,K.C. et al. CAPON modulates cardiac electrocardiographic parameters in repolarization via neuronal nitric oxide genotyped individuals with familial long QT synthase signaling in the heart. Proc. Natl. syndrome. Pacing Clin. Electrophysiol. 24, Acad. Sci. U. S. A 105, 4477-4482 (2008). 406-415 (2001). 18. Crotti,L.et al. NOS1AP is a genetic modifier 8. de Bakker,P.I. et al. Efficiency and power in of the long-QT syndrome. Circulation 120, genetic association studies.Nat. Genet. 37, 1657-1663 (2009). 1217-1223 (2005). 19. Pfeufer,A. et al. Common variants in 9. Aulchencko,Y. & Struchalin,M. GenAbel: myocardial ion channel genes modify the genome-wide SNP association analysis QT interval in the general population: R package version 1.6-4. 2010. Ref Type: results from the KORA study. Circ. Res. 96, Computer Program 693-701 (2005).
77 Chapter 4
20. Gouas,L. et al. Association of KCNQ1, KCNE1, KCNH2 and SCN5A polymorphisms with QTc interval length in a healthy population. Eur. J. Hum. Genet. 13, 1213- 1222 (2005). 21. Nishio,Y. et al. D85N, a KCNE1 polymorphism, is a disease-causing gene variant in long QT syndrome. J. Am. Coll. Cardiol. 54, 812-819 (2009). 22. Nof,E. et al. LQT5 masquerading as LQT2: a dominant negative effect of KCNE1-D85N rare polymorphism on KCNH2 current. Europace. 13, 1478-1483 (2011). 23. Westenskow,P., Splawski,I., Timothy,K.W., Keating,M.T., & Sanguinetti,M.C. Compound mutations: a common cause of severe long-QT syndrome. Circulation 109, 1834-1841 (2004). 24. Holm,H. et al. Several common variants modulate heart rate, PR interval and QRS duration. Nat. Genet. 42, 117-122 (2010). 25. Paulussen,A.D. et al. Genetic variations of KCNQ1, KCNH2, SCN5A, KCNE1, and KCNE2 in drug-induced long QT syndrome patients. J. Mol. Med. (Berl) 82, 182-188 (2004). 26. Kaab,S. et al. A Large Candidate Gene Survey Identifies the KCNE1 D85N Polymorphism as a Possible Modulator of Drug-Induced Torsades de Pointes. Circ. Cardiovasc. Genet. 5, 91-99 (2012).
78 Chapter 5
Identification of RCAN1 as a Genetic Modifier of Atrio-Ventricular Conduction in the Setting of Cardiac Sodium Channel Disease
Iris C.R.M. Kolder1,2*, Carol Ann Remme1*, John A. Jansen6*, Antonius Baartscheer1*, Arie O. Verkerk3, Pieter G. Postema1, Brendon P. Scicluna1, Roel van der Nagel6, Yuka Mizusawa1, Rianne Wolswinkel1, Julien G. Barc1, Tamara T. Koopmann1,6, Freek van den Heuvel7, J. Peter van Tintelen8, Lia Crotti10,11, Peter J. Schwartz12, Harihan Raju13, Elijah R. Behr13, Jean- Jacques Schott14,15,16, Vincent Probst15,16, Sally-Ann B. Clur4, Eline Nannenberg5, Marcel M.A. Mannens5, Moritz F. Sinner17, Stefan Kääb17, Arthur A.M. Wilde1, Maarten P. Van den Berg9, Harold V. van Rijen6, Michael W.T. Tanck2, Connie R. Bezzina1
*These authors contributed equally
1Heart Failure Research Center, Department of Clinical and Experimental Cardiology, 2Department of Epidemiology, Biostatistics and Bioinformatics,3 Department of Anatomy, Embryology & Physiology, 4Department of Pediatric Cardiology and 5Department of Clinical Genetics, Academic Medical Center, Amsterdam, The Netherlands 6Department of Medical Physiology, Division Heart & Lungs, University Medical Center Utrecht, The Netherlands 7Department of Pediatric Cardiology, 8Department of Clinical Genetics, and 9Department of Cardiology, Medical Center Groningen, Groningen, The Netherlands 10Department of Cardiology, University of Pavia IRCCS, Fondazione Policlinico San Matteo, Pavia, Italy 11Institute of Human Genetics, Helmholtz Centrum Munich, Neuherberg, Germany 12Department of Lung, Blood and Heart, Section of Cardiology, University of Pavia IRCCS, Fondazione Policlinico San Matteo, Pavia, Italy 13Division of Cardiovascular Sciences, St George’s University of London, London, UK 14Genes and Disease Program, Center for Genomic Regulation, Pompeu Fabra University, Barcelona, Spain 15INSERM, Unité Mixte de Recherche (UMR) 1087, Institut du Thorax, 16Faculty of Medicine, Université de Nantes, Nantes, France 17Department of Medicine I, University Hospital Munich, Ludwig-Maximilians University, Munich, Germany
Submitted Chapter 5
Abstract Background We exploited the variability in cardiac conduction disease among patients from an extended kindred harboring the SCN5A mutation 1795insD and a segregating population of mice harboring the homologous mutation (Scn5a1798insD/+) to search for novel genetic modifiers of cardiac conduction.
Methods and Results Association analysis (n=1308 tag-SNPs) in the kindred (100 mutation carriers, 115 non- carriers) of genetic variants within and in flanking regions of 18 candidate genes uncovered linkage (LOD=3.7) and association with PR-interval at the region of chromosome 21 harboring the KCNE1 and KCNE2 candidate genes. The SNP displaying the most significant association within this region (rs2834506, p=9.8e-08), was observed within intron 3 of the RCAN1 (Regulator of Calcineurin 1) gene, located upstream of KCNE1. This association was subsequently validated in an independent set of patients harboring different mutations in SCN5A; this also allowed for the detection of a significant interaction between rs2834506 and SCN5A mutation carriership. As the KCNE1 and KCNE2 genes were considered unlikely modulators of PR-interval, we next sought evidence for a role of RCAN1 in mediating this effect. The involvement of RCAN1 was supported by findings in Scn5a1798insD/+ F2 progeny of FVB/N and 129P2 mice displaying variable conduction disease severity. In these mice, a significant correlation was found between ventricular Rcan1 mRNA transcript levels and PR-interval (n=56 mice; r=-0.333, p=0.012). Since RCAN1 is a regulator of the pro-hypertrophic calcineurin/Nfat-pathway, we hypothesized that the Scn5a1798insD/+ mutation disrupts intracellular Ca2+-homeostasis, thereby setting the stage for calcineurin-activation which subsequently impacts on PR-interval. Indeed, elevated intracellular Na+ and diastolic Ca2+ levels were observed in cardiomyocytes of Scn5a1798insD/+ mice. Moreover, chronic activation of the calcineurin/Nfat pathway through application of Transverse Aortic Constriction (TAC) elicited extreme AV-dysfunction, AV-block and sudden death in Scn5a1798insD/+ mice, which was prevented by treatment with the Nfat- inhibitor cyclosporine-A.
Conclusion We identify genetic variation within RCAN1, a regulator of the calcineurin/Nfat-signaling pathway, as a modifier of PR-interval duration in the setting of cardiac sodium channel disease. We propose that abnormal Ca2+homeostasis as a consequence of sodium channelopathy impacts on atrio-ventricular conduction through the activation of the calcineurin/Nfat pathway.
80 Identification of RCAN1
Introduction The genetic basis of the Mendelian cardiac arrhythmia syndromes associated with sudden cardiac death (SCD) has been brought into focus over the last 15 years and a large spectrum of mutations, primarily in genes encoding components of cardiac ion channels, has been reported1. Genotype-phenotype studies in these disorders have clearly established that they are not spared from the phenomena of reduced penetrance and variable expression typical of Mendelian disorders2, 3 . Thus, extensive variability in clinical manifestations is often observed even among family members carrying an identical ion channel gene mutation, with some individuals exhibiting overt abnormalities on the electrocardiogram (ECG) and suffering potentially fatal arrhythmias, whereas other mutation carriers display normal ECGs and remain symptom-free throughout life. While evidence points to a role for genetic background4-7, genetic modifiers of phenotypic variability in arrhythmia syndromes remain largely unknown3.
Mutations in SCN5A, which encodes the major sodium channel isoform in heart, are associated with various arrhythmia disorders. Gain-of-function mutations prolong cardiomyocyte repolarization and cause the Long-QT syndrome type 3 (LQT3), while loss- of-function mutations reduce the action potential upstroke velocity and cause Brugada syndrome (BrS) and/or cardiac conduction disease (CCD). On rare occasions, SCN5A mutations are associated with multiple sodium channel biophysical defects insodium channel function and lead to clinical manifestations of both gain (LQT3) as well as loss (BrS, CCD) of sodium channel function8. The first such “overlap” mutation , SCN5A- 1795insD, was described by our group9 in a large family with clinical manifestations of LQT3, BrS and CCD occurring either in isolation or in combinations thereof9, 10. Knock-in mice (Scn5a1798insD/+) carrying the mouse homolog of this mutation recapitulate the diverse 5 clinical phenotype associated with the mutation in patients11. Ventricular cardiomyocytes isolated from Scn5a1798insD/+ mice displayed action potential prolongation as a consequence of an increased tetrodotoxin-sensitive persistent inward Na+ current, explaining the Long QT phenotype of mutation carriers. In addition cardiomyocytes from these mice were also characterized by a slower action potential upstroke velocity caused by a reduction in peak sodium current density, which on the other hand explains the conduction disease and Brugada Syndrome features in mutation carriers. While our initial studies in Scn5a1798insD/+ mice focused on mice of the FVB/N genetic background (FVB/N-Scn5a1798insD/+, hereafter referred to as FVB/N-MUT), we subsequently investigated genetic background effects on disease severity by studying mutant mice of the 129P2 strain (129P2-Scn5a1798insD/+, 129P2- MUT). These studies demonstrated that the mutation was associated with greater severity of conduction and repolarization disease in mice of the 129P2 inbred strain as opposed to the FVB/N strain, indicating an important role for genetic background in modulation of disease severity5.
81 Chapter 5
To identify novel genes and pathways influencing the severity of conduction and repolarization disease associated with sodium channelopathy we here carried out parallel investigations in humans and mice segregating the SCN5A-1795insD mutation. We provide evidence that genetic variation within the RCAN1 gene, encoding a regulator of the calcineurin signaling pathway, modulates the extent of PR-interval prolongation in mutation carriers and that the calcineurin-NFAT pathway modulates atrio-ventricular conduction in the setting of sodium channelopathy.
Materials and Methods SCN5A mutation study populations The discovery study population consisted of a large Dutch kindred of European decent segregating the SCN5A-1795insD mutation associated with manifestations of Long- QT syndrome type 3 (LQT3), Brugada syndrome (BrS) and cardiac conduction disease (CCD) occurring either in isolation or in combinations thereof9. The replication study set consisted of individual probands as well as probands and relatives originating from families segregating mutations in SCN5A (other than SCN5A-1795insD) and presenting clinically with LQT3, BrS/CCD or LQT3+CCD. The replication study set was recruited in the Netherlands, France, Italy and the United Kingdom. All subjects studied were of European descent. Only individuals with available DNA and ECG were included. The study was approved by the institutional review boards of the respective centers and all study participants provided their (written) informed consent.
ECG-phenotyping of the SCN5A-1795insD study population Heart-rate and ECG indices of conduction and repolarization were measured from ECGs acquired in the absence of anti-arrhythmic drugs. All ECGs were digitalized and analyzed using ImageJ (http://rsb.info.nih.gov/ij/). Only sinus rhythm complexes were analyzed. Measurement of all parameters (heart-rate, PR-interval, QRS-duration and QT-interval) was done manually on-screen, in lead II whenever possible36. Parameters were averaged from up to 3 consecutive beats with similar preceding RR-intervals. For QT- and heart-rate corrected QT (QTc), the tangent method with Bazett’s correction was used37.
SNP selection, genotyping Haplotype tagging SNPs (tagSNPs) were selected using Tagger38 based on HapMap39 data release #20 / phase II using the March 2006 NCBI B35 genome assembly and dbSNP build 125 data. The following criteria were applied for tagSNP selection: HapMap CEU population, pairwise only tagging with r2 ≥ 0.8 and a minor allele frequency (MAF) ≥ 10%. To account for genetic variation in regions surrounding each gene, we considered up- and downstream genetic data in the tagging procedure. Included regions were defined by
82 Identification of RCAN1 linkage-disequilibrium (LD), blocks as previously reported40. We included at least 50 kb of both up- and downstream information.
SNP genotyping was performed using a custom assay (GoldenGate) on an Illumina- BeadStation500GX (Illumina Inc., SanDiego, USA). Multiple quality control measures were implemented: sample call rate <0.95, and SNP call rate <0.98. Monomorphic SNPs were removed and genotypes were checked for Mendelian errors using the mistyping method in Mendel 8.0 and and were removed when inconsistent41.A total of 1308 out of 1424 SNPs (92%) passed the quality control criteria and were included in the linkage and association analyses. SNP genotyping in the validation dataset was done using a Taqman assay on a Lightcycler System (Roche).
Linkage and association analysis Phenotypic data for mutation carriers and non-carriers were normally distributed (Shapiro- Wilk-test, W>0.90). Effects of sex, age and mutation-carrier status on ECG characteristics were analyzed using a linear mixed effects model with correction for family relations (lmekin function) implemented in the Kinship package (1.1.0-22)42 in R (http://www.r- project.org/).
Genetic distance maps were calculated using the kosambi mapping function (Mendel) and estimation of (multipoint) identity-by-descent and linkage analyses were carried out using SOLAR. Two linkage models were used; 1) with adjustment for age and sex, and 2) model 1 with additional adjustment for SCN5A mutation carrier status. The latter model was used to identify effects of SNPs independent of the (large) effect of the SCN5A-1795insD mutation. All LOD-scores reported were corrected for inaccuracy caused by possible non- normal trait distribution using the empirical LOD-adjustment in SOLAR. Following Landers 5 & Kruglyak43, who proposed a LOD-score of >1.9 as suggestive linkage and >3.3 as genome wide significant, and correcting for four different phenotypes, we consider LOD-scores >2.4 (p=4.2e-04) and >3.6 (p=2.2e-05) in the present study as thresholds for suggestive and genome-wide significance, respectively.
Family-based association analyses were performed using linear mixed effect models from the Kinship package. Prior to the main analyses, each SNP-phenotype relationship was investigated for dominance deviation using an additive model with a heterozygosity indicator variable. In case of significant (p<0.1) dominance deviation, either a dominant or recessive genetic model was applied in the subsequent analyses (depending on the direction of the dominance deviation). Otherwise an additive genetic model was applied. Similar to the linkage analysis, two models were used in the main association analyses; 1) with adjustment for age and sex, and 2) model 1 with additional adjustment for SCN5A mutation carrier status.
83 Chapter 5
Based on Nyholt’s method44, which is a simple correction for multiple testing using spectral decomposition on the matrices of pairwise LD between SNPs to determine the number of independent SNPs (tests), the number of independent tests in our data set was 1308. Thus the significance threshold for association was calculated by dividing the nominal α (0.05) by the number of independent tests (1308) ´ 4 phenotypes and ´ 2 models, resulting in a Bonferroni corrected significance threshold a* of 4.8e-06.
Within the validation families, association analyses were performed using a linear mixed effects model with adjustment for age, sex, disease phenotype (and carrier status) with the most appropriate genetic model for the SNP to be validated. The significance threshold applied in the validation study was 0.05 divided by the number of SNPs to be validated. (Since only a single SNP was validated, the significance threshold in the validation sample was p < 0.05)
Generation of Scn5a1798insD/+ mice Heterozygous Scn5a-1798insD (Scn5a1798insD/+) mice in the two different genetic backgrounds (129P2-Scn5a1798insD/+ and FVB/N-Scn5a1798insD/+) were generated as described previously5, 11. 129P2-Scn5a1798insD/+ mice were maintained in the 129P2 genetic background by crossing them with wild-type 129P2 mice that were invariably purchased from Harlan. FVB/N- Scn5a1798insD/+ mice were maintained in the FVB/N genetic background by crossing them with wild-type FVB/N mice that were invariably purchased from Charles River Laboratories. Mice were genotyped as described previously11. Scn5a1798insD/+ F2 progeny were generated by crossing FVB/N-Scn5a1798insD/+ and 129P2-Scn5a1798insD/+ mice as described previously19. All experiments were performed on adult (3 to 5 months old, males and females) Scn5a1798insD/+ mice with their wild-type littermates as control, and were in accordance with governmental and institutional guidelines for animal use in research.
Electrocardiographic (ECG) measurements in mice Mice were anesthetized using isoflurane inhalation (0.8-1.0 volume % in oxygen) and surface ECGs were recorded from subcutaneous 23-gauge needle electrodes attached to each limb using the Powerlab acquisition system (ADInstruments). ECG traces were signal averaged and analysed for heart rate (RR-interval), PR-, QRS- and QT-interval duration using the LabChart7Pro software (ADInstruments). QT-intervals were corrected for heart rate using the formula: QTc=QT/(RR/100)1/2 (RR in ms).
Quantitative RT-PCR RNA was isolated from LV samples of F2 mice, and Rcan1 and Kcne1 mRNA expression levels were quantified using the LightCycler system for real-time RT-PCR (Roche Applied Science). Quantitative PCR data was analyzed with the LinRegPCR program. All samples were processed in triplicate and expression levels were normalized to Hprt.
84 Identification of RCAN1
(Persistent) sodium current measurements Cell isolation For the isolation of ventricular myocytes, excised hearts were first perfused in a Langendorff system (37 ºC) with normal Tyrode’s solution for 5 minutes (containing in mmol/l: 140
NaCl, 5.4 KCl, 1.8 CaCl2, 1 MgCl2, 5.5 glucose, 5 HEPES; pH 7.4 (NaOH)). Next, the heart was perfused for 8 minutes with a similar solution in which the calcium concentration was lowered to 1 µM, after which the enzyme Liberase Blendzyme type 4 (Roche; 0.05 mg/ml) and trypsin (Boehringer, 1 µl/ml of 2.5% solution) were added to the low calcium solution for 10 minutes. Digested tissue was gently triturated in the low calcium enzyme solution and isolated cells were washed twice in the low calcium solution and twice in normal calcium Tyrode’s solution, both supplemented with bovine serum albumin (BSA, 50 mg/ml). Quiescent, rod-shaped cells with clear cross-striations and smooth surface were selected for measurements.
Data acquisition Membrane potentials and currents were recorded in the perforated or ruptured whole- cell configuration of the patch-clamp technique using patch pipettes (1-3 MW, borosilicate glass). Signals for sodium current were low-pass filtered with a cut-off frequency of 5 kHz and digitized at 5-10 kHz; action potential measurements were filtered and digitized at 10 and 40 kHz, respectively. Series resistance was compensated by ≥75%. For voltage control, data acquisition, and analysis, custom-made software was used.
Sodium current properties Sodium current properties were determined at room temperature using conventional voltage clamp protocols as indicated in Figure X, with a holding potential of -120 mV and a cycle time of 5 seconds. Current density was calculated by dividing whole-cell 5 current amplitude by cell capacitance (Cm). Data for voltage-dependence of activation and -1 inactivation were fitted with a Boltzmann functiony ( =[1+exp{(V-V1/2)/k}] ), where V1/2 is the half-maximal voltage and k is the slope factor. Data for the recovery from inactivation was fitted by a two-exponential function (y=y0+Af{1-exp[-t/τf]}+As{1-exp[-t/τs]}), where
Af and As are fractions of fast and slow inactivating components andτf and τs are the time constants of the fast and slow inactivating components, respectively. Since the current decay could not be reliably fit with a two-exponential function, the time course of inactivation was instead determined by analyzing the time required for 50% of current decay to occur (t50%). The bath solution contained (in mmol/l): 7.0 NaCl, 133 CsCl, 1.8
CaCl2, 1.2 MgCl2, 11.0 glucose, 5.0 HEPES, and 5 µM nifedipine; pH 7.4 (CsOH). Pipettes were filled with (in mmol/l): 3.0 NaCl, 133 CsCl, 2.0 MgCl2, 2.0 Na2ATP, 2.0 TEACl, 10 EGTA, 5.0 HEPES; pH 7.3 (CsOH).
85 Chapter 5
The persistent sodium current was measured at 37ºC as 30 µM TTX-sensitive current during a descending voltage ramp protocol (Figure X, inset). The bath solution contained
(in mmol/l) 130 NaCl, 10 CsCl, 1.8 CaCl2, 1.2 MgCl2, 11.0 glucose, 5.0 HEPES, and 5 µM nifedipine; pH 7.4 (CsOH). The pipette solution was identical to that used for the sodium current measurements described above.
Intracellular calcium and sodium measurements Ventricular myocytes were attached to a poly-lysine (0.1 g L−1)-treated coverslip on the stage of a microscope (Nikon Diaphot). A perfusable chamber (height 0.4 mm, diameter 10 mm, volume 30 μL), containing two platinum electrodes for field stimulation (8 mm distance, 40 V cm−1 bipolar square pulses of 0.5-ms duration), was pressed onto the cover slip. The microscope stage and perfusion chamber were maintained at 37 °C. The measuring window was adjusted with a rectangular diaphragm to the cellular surface of one quiescent rod-shaped myocyte. For every experimental condition, fluorescence + measurements were taken in 3 to 5 myocytes per heart. Intracellular sodium ([Na ]i) and 2+ calcium ([Ca ]i) concentrations were measured in myocytes loaded with the fluorescent probes sodium-binding benzofuran isophathalate (SBFI) and Indo-1 respectively in HEPES solution without albumin as described previously45. SBFI fluorescence was measured + in dual wavelength emission mode and [Na ]i was calculated using previously obtained values for Rmax, Rmin, β and kd; dual wavelength emission mode provides a more sensitive + 46 and sodium specific measurement of [Na ]i than the dual excitation technique . Because + + [Na ]i does not change on a beat to beat basis, calculated [Na ]i data were averaged over the entire cardiac cycle. Na+/K+-ATPase activity was inhibited with 100 μmol/l ouabain.
Transverse aortic constriction (TAC) Mice were anaesthetized by isoflurane (mean 2.5% in oxygen), intubated with a20G polyethylene catheter, and ventilated (200 µL, 160 strokes/min) with a rodent ventilator (Minivent, Hugo Sachs Electronics, Germany). The thoracic cavity was accessed through a small incision at the left upper sternal border in the second intercostal space. A 7-0 silk suture was passed around the aorta between the right innominate and left common carotid arteries. Constriction of the transverse aorta was performed by tying against a 27 G needle, which was subsequently removed. The same procedure was followed in sham animals, except for the constriction. A gradient of approximately 50mmHg across the aortic valve was confirmed by Doppler echocardiography. In a subset of mice, a radiotelemetry transmitter (Data Sciences International) was surgically inserted into the peritoneal cavity. After 2 days of recovery, ECG signals were measured continuously until mice were sacrificed (2 weeks after TAC).
86 Identification of RCAN1
Electrophysiological assessments in Langendorff-perfused hearts Two weeks after TAC- or sham-surgery, surface ECG analysis was performed and hearts were excised and perfused in a Langendorff set-up. Wenckebach periodicity (WBP) was determined by >2 sec pacing at the left atrium. Starting at 150 ms, the cycle length was reduced in steps of 10 ms until a single stimulation failed to activate the ventricles. For the AV nodal refractory period (AVNRP), the same protocol was used for the ERP. Atrio- ventricular delay (AV-delay) was determined by calculating the difference between the time of LA stimulation and the onset of ventricular activation.
Statistical analysis Data are presented as mean ± SEM, unless otherwise specified. Differences between groups were analyzed by unpaired Student’s t-test, nonparametric test, or ANOVA as appropriate. The level of statistical significance was set to p<0.05, unless otherwise mentioned. The degree of association between PR-interval and LV transcript levels was determined by Pearson’s correlation (r).
Results Phenotypic variability among carriers of the SCN5A-1795insD mutation An extensive genealogical search allowed us to trace the SCN5A-1795insD family back to the eighteenth century, enabling the construction of a highly extended pedigree Figure( 1A). DNA and 12-lead ECGs were available for 215 individuals (100 mutation carriers) from the last 4 generations of the family. As expected, ECG conduction (PR-interval, QRS-duration) and repolarization (QTc-interval) parameters were significantly prolonged in SCN5A-1795insD carriers versus non-carriers (Supplementary Table S1). There was a striking variability in all ECG parameters, both among mutation carriers and non-carriers 5 (Figure 1B). In mutation carriers, while some of the variance in ECG parameters was accounted for by effects of age and sex (Table 1), 53%-89% of the variance in these traits remained unexplained, providing a rationale for searching for genetic modifiers of disease expression in this family.
87 Chapter 5 2 R 15 11 47 36 Carriers only Carriers # 2 R 12 54 45 39 0.92 p-value 3.3e-36 4.4e-12 5.0e-10 β ± SE Carriers vs. Non carriers vs. Carriers 0.25 ± 2.46 57.76 ± 3.77 27.58 ± 3.76 13.72 ± 2.10 0.23 p-value 4.5e-07 9.0e-14 3.82e-05 (*4.48e-04) Age (per year) Age -1795insD carriers and non-carriers -1795insD carriers and interaction if p<0.01). SE: standard error. *Depicts a significant interaction between age between interaction significant a *Depicts error. standard SE: p<0.01). if interaction and # β ± SE SCN5A 0.76 ± 0.10 0.16 ± 0.14 -0.31 ± 0.06 0.24±0.06 (* 0.35±0.10) mutation carriership ( carriership mutation 0.72 0.08 0.052 p-value 9.3e-05 -1795insD mutation carriership on ECG parameters in the large family. in the large parameters ECG on carriership -1795insD mutation SCN5A SCN5A Females vs. males vs. Females β ± SE 1.95 ± 5.35 4.14 ± 2.38 -7.81 ± 4.00 -8.65 ± 2.17 Effects of age, sex and sex of age, Effects
1 | Phenotype le b PR-interval (ms) PR-interval Heart rate (bpm) Heart rate QTc-interval (ms) QTc-interval : Explained variance by sex, age and age sex, by variance Explained : QRS-duration (ms) QRS-duration 2 a T R and carrier ship. and carrier
88 Identification of RCAN1
A.
B. 5
89 Chapter 5
C.
220
200
180 interval
- Non Carriers PR 160 Carriers
140
120 AA AG GG rs2834506
D.
Figure 1 | (A) The pedigree of the large family carrying the SCN5A-1795insD founder mutation. (B) Correlation between ECG parameters in mutation carriers and non carriers. PR-interval vs. QRS-duration and QTc-interval vs. heart rate. (C) Bar graph showing the mean and standard deviation of PR-interval between the different genotype groups for rs2834506 in SCN5A-1795insD carriers and non-carriers. (D) Regional association plot for SNPs at the chromosome 21q22.12 region with PR-interval in the family (Figure generated using SNAP47. Recombination fraction is according to HapMap CEU data. The correlation (r2) of the interrogated SNPs with rs2834506 is represented by red shading, with the strongest red representing the strongest correlation.
90 Identification of RCAN1
SNP rs2834506 on chromosome 21 is associated with PR-interval in the SCN5A- 1795insD family To identify genetic variants modulating disease severity in SCN5A-1795insD mutation carriers, we investigated single nucleotide polymorphisms (SNPs) located in or around 18 candidate genes for effects on heart rate and ECG parameters of conduction and repolarization by linkage and association analyses. Genes investigated were selected on the basis of either being disease-causing genes in cardiac arrhythmia syndromes, functionally important subunits of such genes, or genes significantly associated with the QTc-interval in the general population based on the results of genome-wide association studies (Supplementary Table 2). Haplotype tagging SNPs (tagSNPs) were selected as described previously12 (details also provided in the Methods section). Besides the systematic analysis of each gene, the selected tagSNPs permitted the investigation of flanking regions as tag-SNPs capturing haplotypes at least 50 kb up- and downstream of the candidate genes tested were also included. A total of 1,424 SNPs were identified in this way, of which 1308 passed the quality control criteria (see methods for details) and were used for linkage and association analyses. A complete list of the 1308 SNPs tested is presented in Supplementary Table 3.
As expected, due to the effects of the SCN5A-1795insD mutation, linkage analysis using age and sex as covariates (model 1) revealed high LOD scores for conduction (PR, QRS) and repolarization (QTc) parameters at the region ofSCN5A on chromosome 3 (LOD scores of 12.9, 6.3 and 19.5, respectively). Using a model with additional correction for SCN5A- 1795insD mutation carriership (model 2), uncovered linkage that was not detected in the previous analysis. The chromosome 21 locus in the region of the KCNE1 and KCNE2 candidate genes now showed significant linkage to PR-interval (LOD=3.7), while the chromosome 3 region harboring the genes SCN5A, GPD1L and CAV3 displayed suggestive 5 linkage with PR- interval (LOD=3.1).
In association analysis with correction for age and sex effects, all ECG traits i.e.heart rate, PR-interval, QRS-duration and QTc-interval were, as expected, found to be highly associated with SNPs in and around the SCN5A gene on chromosome 3 (data not shown). After additional correction for SCN5A-1795insD mutation carriership, only a single association exceeded our pre-specified Bonferroni-corrected threshold for statistical significance of 4.8×10-6 (Supplemental Table 3). This SNP, rs2834506, on chromosome 21, was found to be associated with PR-interval (p=9.8×10-8). The best fitting genetic model for the association of this SNP with PR-interval was a dominant genetic model. Adding SNP rs2834506 to the linkage model for PR-interval resulted in a disappearance of the
91 Chapter 5 linkage signal on chromosome 21, indicating that this SNP underlies the linkage signal at this locus in the linkage analysis. SNP rs2834506 accounted for 8% of the variance in PR- interval among mutation carriers, bringing the total explained variance to 55%.
The G-allele of rs2834506 was associated with increased PR-interval (Figure 1C) and although the effect of the G-allele appeared stronger in mutation carriers as compared to non-carriers, no significant interaction was detected when a model including an interaction term between rs2834506 and SCN5A-1795insD carriership was used (pinteraction=0.33; β (±SE)=5.5 ± 5.6 ms). Moreover, no effect of rs2834506 on PR-interval was detected in a large sample of the general population (n=5370, p>0.05), European descent)13 pointing to a possible SCN5A mutation-specific effect of the variant.
Association between rs28934506 and PR-interval in carriers of SCN5A mutations other than SCN5A-1795insD We next explored the effect of rs2834506 in probands and small families harboringSCN5A mutations other thanSCN5A -1795insD. Since the SCN5A-1798insD mutation is associated with disease features of both gain (LQT3) as well as loss of sodium channel function (CCD, BrS) these additional individuals were classified in 3 categories: (i) isolated LQT3, i.e. in the absence of CCD (n=118; 61 mutation carriers), (ii) LQT3 with CCD (n=55 carriers) or (iii) isolated BrS and/or CCD, i.e. in the absence of QTc-prolongation (n=320, 164 carriers). Analyzing the rs2834506 effect in the carriers and non-carriers within these probands and families uncovered a significant SNP*carriership interaction: the effect of rs2834506 was more pronounced in the carriers compared to the non-carriers (b = 10 ± 5 ms, p=0.052). When we combined the data from the SCN5A-1795insD family with the new probands and families, the SNP*carriership interaction effect improved (p=0.045). In this analysis, combining the 1795insD family and the additional sets, carriers with the AG/GG genotype had on average a PR-interval that was 12 ± 3 ms longer than individuals with the AA- genotype, whereas in non-carriers the AG/GG-genotype was associated with an increase of only 5 ± 2 ms.
Considering mutation carriers only from the additional probands and small families (n=280), 123 patients carried at least one G-allele at rs2834506 and these patients showed, on average, 7 ± 4 ms longer PR-interval compared to SCN5A carriers with the AA- genotype (p=0.06, correction for sex, age and disease category). When we considered the two larger phenotypic groups with the distinct/pure phenotypes, that is, isolated LQT3 or isolated CCD/BrS, the effect of the polymorphism appeared stronger in the LQT3 group
92 Identification of RCAN1 suggesting that the interaction may be stemming from the gain-of-function properties of the LQT3-causing SCN5A mutations (Figure 2).
Figure 2 | Forrest plot of the effect (Beta ± SE) in ms of rs2834506 on PR-interval in the the SCN5A- 1795insD family and the validation sets of SCN5A mutation carriers with a LQT3, LQT3+CCD, or CCD/BrS phenotype.
SNP rs2834506 is located within RCAN1, encoding the regulator of calcineurin 1. In our candidate gene approach, rs2834506 was included in the association analysis by virtue of its location upstream of the KCNE1 gene which encodes MinK, a regulatory 14 subunit of the repolarizing potassium current IKs . However, examination of the exact location of rs2834506 showed that it is in fact located within intron 1 of the RCAN1 gene, situated upstream of KCNE1 (Figure 1D). RCAN1 (previously called MCIP and DSCR1) encodes ‘Regulator of Calcineurin 1’ which is highly expressed in heart and regulates calcineurin, a calcium-activated phosphatase that promotes hypertrophic growth of the heart15. Since transgenic mice overexpressing constitutively active calcineurin display premature sudden death and profoundly prolonged PR-intervals16-18, we hypothesized 5 that the effect observed at this locus may in fact be mediated by RCAN1 through the calcineurin pathway. To investigate this possibility, we assessed whether we could detect a correlation between cardiacRcan1 transcript levels and PR-interval in mutant F2 progeny generated by crossing Scn5a1798insD/+ (MUT) mice of the two distinct inbred strains (FVB/N- MUT and 129P2-MUT) displaying different severity of PR-interval prolongation5, 19. Rcan1 mRNA expression levels (corrected for the house-keeping gene Hprt) were significantly correlated with PR-interval (n=56 mice; r=-0.333, p=0.012), whereas no significant correlation was found between Kcne1 mRNA levels and PR-interval (n=56 mice; r=0.113, p=0.405). This data provides support to the idea that RCAN1 may modify the PR-interval in the setting of sodium channelopathy through effects on and calcineurin signaling.
93 Chapter 5
The Scn5a1798insD/+ mutation causes abnormal intracellular sodium and calcium homeostasis The calcineurin/Nfat pathway is activated by increased intracellular levels of calcium 2+ ([Ca ]i). Since the association of theRCAN1 SNP rs2834506 with PR-interval was restricted to SCN5A mutation carriers, we thus hypothesized that the mutation causes abnormal intracellular Ca2+ homeostasis with subsequent calcineurin/Nfat-pathway activation, 2+ thereby providing a target for the regulatory effects of RCAN1. [Ca ]i may be increased + secondary to increased intracellular sodium levels ([Na ]i) as a consequence of the 11 sustained inward sodium current (INa,sus) associated with this mutation . We therefore 2+ + investigated the magnitude of INa,sus, [Ca ]i and [Na ]i in ventricular cardiomyocytes isolated from 3-month old wild-type (WT) and MUT mice of both strains.
As 129P2-MUT mice presented with a more severe phenotype compared to FVB/N-MUT mice, we first explored whether this was also reflected in a greater INa,sus, as a difference in INa,sus between the two MUT mouse strains would allow us to establish whether a dose- dependent relationship existed between the magnitude of INa,sus and the extent of increase + 2+ in [Na ]i and [Ca ]i. Indeed, while 129P2-MUT and FVB/N-MUT mice displayed a similar reduction in peak INa, INa,sus was significantly greater in 129P2-MUT compared to FVB/N-
MUT (Figure 3A-F). Moreover, a gradation was observed in the magnitude of Na,susI among the 4 groups of mice studied, with FVB/N-WT mice being the least affected and 129P2-
MUT mice the most severely affected. Of note, FVB/N-MUT mice displayed an Na,susI that was comparable to 129P2-WT. As expected, the graded severity observed in INa,sus across + 2+ the 4 line of mice, was reflected in the amplitude of steady-state [Na ]i, diastolic [Ca ]i and 2+ Ca transient amplitude with the 129P2-MUT (which has the largest INa,sus) displaying the greatest increase in these parameters (Figure 3G-J). This data demonstrates that the INa,sus associated with this mutation disrupts intracellular Ca2+ homeostasis, thereby setting the stage for activation of the calcineurin/Nfat-signaling pathway and a potential regulatory effect of RCAN1.
94 Identification of RCAN1
5
Figure 3 | (A) Representative example of peak sodium current measurement assessed with conventional voltage clamp (protocol shown as inset); (B) Average current-voltage relationships for sodium current in ventricular myocytes from FVB/N-WT (n=10), FVB/N-MUT (n=11), 129P2-WT (n=14), and 129P2-MUT (n=11) mice; (C) Average peak sodium current (mean±SEM, pA/pF) measured at a holding potential of -40 mV; (D) Representative example of sustained inward sodium (INa,sus) current measurement assessed with a ramp protocol (see top panel); (E) Average current-voltage relationships for INa,sus in ventricular myocytes from FVB/N-WT (n=9), FVB/N-MUT (n=9), 129P2-WT (n=10), and 129P2-MUT (n=10) mice; (F) Average
INa,sus (mean±SEM, pA/pF) measured at a holding potential of -20 mV; (G) Representative examples for intracellular calcium transients in isolated ventricular myocytes; (H) Average intracellular sodium ([Na+] i) concentrations measured at a stimulation frequency of 6Hz in FVB/N-WT (n=13), FVB/N-MUT (n=22), 129P2-WT (n=15), and 129P2-MUT (n=16) isolated ventricular myocytes; (I-J) Average intracellular 2+ diastolic calcium concentrations ([Ca ]i) and calcium transient amplitudes measured at a stimulation frequency of 6 Hz in FVB/N-WT (n=16), FVB/N-MUT (n=26), 129P2-WT (n=16), and 129P2-MUT (n=16) isolated ventricular myocytes.
95 Chapter 5
Figure 4 | (A) Average heart weight to tibia length ratios for WT-sham, WT-TAC, MUT-sham and MUT- TAC mice (untreated or treated with cyclosporine-A, CsA); (B) Survival curve for WT and MUT mice after TAC; (C) Telemetric ECG recordings indicating normal sinus rhythm in WT-TAC and sinus bradycardia and AV-block in MUT-TAC mice; (D) Representative examples of AV-mapping traces in isolated Langendorff- perfused hearts from WT-TAC and MUT-TAC mice, indicating prolonged AV-conduction delay in MUT-TAC; (E) Average values for AV-conduction delay during sinus rhythm in WT-sham, WT-TAC, MUT-sham and MUT-TAC hearts (untreated or treated with cyclosporine-A, CsA); (F) Increased AV-delay in MUT-TAC hearts at shorter coupling intervals; (G) Average values for AV-nodal effective refractory period (ERP) in WT-sham, WT-TAC, MUT-sham and MUT-TAC hearts (untreated or treated with cyclosporine-A, CsA); (H) Average values for Wenckebach periodicity in WT-sham, WT-TAC, MUT-sham and MUT-TAC hearts (untreated or treated with cyclosporine-A, CsA). * indicates p<0.01 versus WT sham, # denotes p<0.01 versus MUT sham
96 Identification of RCAN1
Chronic activation of the calcineurin/Nfat pathway elicits AV-block and sudden death in Scn5a1798insD/+ mice To further investigate the idea that the Scn5a1798insD/+ mutation stimulates the calcineurin/ Nfat pathway and thereby impacts on atrio-ventricular conduction, we subjected FVB/N- MUT and FVB/N-WT mice to Transverse Aortic Constriction (TAC). This surgical intervention leads to chronic pressure overload in the left ventricle, activation of the calcineurin/Nfat pathway, and ultimately development of cardiac hypertrophy. We hypothesized that if the mutation indeed impacts on atrio-ventricular conduction by activation of this signaling pathway, then the setting of increased calcineurin/Nfat signaling during TAC should result in an increased atrio-ventricular slowing in mutant mice as opposed to sham-operated mutant mice. After TAC, WT and MUT mice developed similar extent of cardiac hypertrophy (Figure 4A). However, 8 out of 19 MUT-TAC mice (42%) died suddenly between day 5 and day 14 day post-TAC-surgery, whereas all WT-sham, WT-TAC and MUT-sham mice survived the 2-week TAC period (Figure 4B). Continuous 24-hour telemetry recording revealed that the MUT-TAC mice that died prematurely developed progressively increased PR-intervals and bradycardia, ultimately culminating in (complete) AV-block and death (Figure 4C). In surviving MUT-TAC animals, PR-intervals on surface ECGs were not significantly increased as compared to MUT-sham (data not shown). However, electrophysiological measurements in isolated Langendorff-perfused hearts revealed profound AV-conduction delay in surviving MUT-TAC mice as compared to WT-TAC, WT-sham or MUT-sham during baseline atrial stimulation at 120 ms, and this effect was further exacerbated at shorter coupling-intervals (Figure 4D,E). Similarly, both the AV-nodal refractory period and the Wenckenbach periodicity were significantly prolonged in MUT-TAC hearts compared to the other groups, indicating the development of substantial atrio-ventricular dysfunction secondary to TAC in MUT mice only (Figure 4F,G). Treatment with the calcineurin-inhibitor 5 cyclosporin-A (CsA; 15 mg/kg i.p. twice daily) prevented the development of cardiac hypertrophy in both WT-TAC and MUT-TAC to an equal extent (Figure 4A). However, while CsA had no effect on electrophysiological properties in WT-TAC, WT-sham or MUT-sham hearts, it completely prevented sudden death and normalized atrio-ventricular conduction properties in MUT-TAC mice Fig.( 4D-G). Thus, the secondary effects of the mutation on intracellular Na+ and Ca2+ homeostasis makes Scn5a1798insD/+ hearts more susceptible to the deleterious effects of calcineurin/Nfat-pathway activation on atrio-ventricular conduction and provide a target for the regulatory effects of RCAN1.
Discussion Through linkage and association analysis in an extended family with cardiac sodium channelopathy caused by the SCN5A-1795insD mutation, we identified genetic variation within the RCAN1 gene as a modifier of the PR-interval. Further association analysis in a set of patients harboring other SCN5A mutations, confirmed this association and
97 Chapter 5 established an interaction between the SNP and SCN5A mutation carriership. Studies in mice carrying the homologous mutation (Scn5a1798insD/+) support the concept that intracellular Ca2+ homeostasis is disrupted as a consequence of the increased sustained sodium current characteristic for this mutation, which makes the hearts of mutation carriers more susceptible to the deleterious effects of calcineurin/Nfat-pathway activation on atrio-ventricular conduction and provides a target for the regulatory effects ofRCAN1 .
Like many Mendelian disorders, inherited arrhythmia syndromes typically display reduced penetrance and variable disease expression, which have been attributed to effects of environmental or genetic modifiers. Some important modifiers such as age20, gender21, 22, heart-rate23 and drug treatment10, 24 are already recognized. However, while evidence points to a role for genetic modifiers of disease severity5, 7, 25, these remain largely unknown3. Recent genome-wide association studies in the general population have uncovered a number of SNPs in various genes associated with heart rate, and ECG indices of conduction (PR, QRS) and repolarization (QTc) (reviewed in Kolder et al. 201226). As for most biological traits, the loci identified display small effect sizes and in aggregate explain only a small fraction of the total heritability for a given trait, leaving a large portion of heritability remains unexplained. A complementary approach to such population studies in the identification of genetic variants impacting on these traits comprises family-based studies which although requiring appreciable effort and resources for recruitment of family members, offer distinct advantages27USA. [email protected]
In the present study, the large size of the SCN5A-1795insD pedigree coupled to the variable disease severity and pleiotropic effects of the mutation make it a powerful model for uncovering genetic modifiers. Indeed, our results provide strong evidence in support for a role of genetic variation on chromosome 21q22.12 in modulation of PR-interval. The most significant SNP (rs2834506) within this region displayed a p-value (p=9.8e-08) that not only passed our pre-determined threshold for significance (p<4.8e-06) but was even borderline significant when one considers the commonly used genome-wide significance p-value cut-off of p<5.0e-8 corresponding to Bonferroni adjustment for1 million independent tests28.
98 Identification of RCAN1
A role for the 21q22.12 chomosomal region, harboring KCNE1 and KCNE2, in regulation of atrio-ventricular conduction was at first glance rather unexpected. KCNE1 and KCNE2 encode b-subunits, respectively MinK and MiRP-1, of the major repolarization currents
IKs and IKr. Thus, genetic variation at these genes would be expected to affect cardiac repolarization rather than conduction. Inspection of the association signals within this region showed that the most highly associated SNP, rs2834506, which we selected for genotyping by virtue of the fact that it lies upstream of the KCNE1 gene, actually lies within intron 3 of the RCAN1 gene (previously called MCIP) encoding ‘Regulator of Calcineurin 1’. RCAN1 is highly expressed in heart and regulates calcineurin, a calcium- activated phosphatase that promotes hypertrophic growth of the heart. The availability of transgenic mice carrying the knock-in mutation Scn5a1798insD/+ allowed for further investigation of the relevance of the observed association.Scn5a 1798insD/+ mice recapitulate the diverse clinical phenotype of the patients11 and moreover, Scn5a1798insD/+ mice of two separate genetic backgrounds display varying severity of conduction and repolarization disease5, 19. In mutant F2 progeny generated by crossing Scn5a1798insD/+ mice of these two distinct inbred strains19, we found a significant correlation between cardiac Rcan1 (but not Kcne1) transcript levels and PR-interval, thereby extending the observed clinical association betweenRCAN1 and PR-interval in human SCN5A-1795insD mutation carriers to the cardiac gene expression level.
RCAN1 is highly expressed in heart and regulates calcineurin, a calcium-activated phosphatase that promotes hypertrophic growth of the heart15. RCAN1 directly binds to the catalytic domain of calcineurin and has been suggested to act as a feedback inhibitor of calcineurin, although activation of calcineurin by RCAN1 has also been described29-31. Interestingly, transgenic mice over-expressing constitutively active calcineurin display premature sudden death and profoundly prolonged PR-intervals16-18. Action potential 5 (AP) recordings in neonatal cells from these mouse hearts before the development of hypertrophy showed decreased peak sodium current and decreased sodium channel availability, suggesting that alternations in this current are directly and independently linked to the same calcineurin signaling pathway as myocardial hypertrophy17. Decreased
Scn5a mRNA and Nav1.5 protein levels have also been reported in hearts of calcineurin- overexpressing mice18. In response to sustained elevated intracellular Ca2+, calcineurin becomes activated in the cytoplasm where it dephosphorylates its transcriptional effector, NFAT. This results in translocation of NFAT to the nucleus, interaction with other transcription factors, and induction of hypertrophic gene expression. Here, we have shown that the Scn5a1798insD/+ mutation causes enhanced sustained inward sodium + (INa,sus) current in ventricular cardiomyocytes, with subsequent increased intracellular Na and Ca2+-levels, setting the stage for calcineurin-Nfat pathway activation. Notably, mice + of the 129P2 inbred strain showed more pronounced increase of INa,sus, intracellular Na
99 Chapter 5 and Ca2+, and atrio-ventricular conduction delay secondary to the Scn5a1798insD/+ mutation, indicating a dose-dependent effect. Furthermore, Scn5a1798insD/+ mice developed severe atrio-ventricular conduction abnormalities and sudden death after TAC, in the setting of chronic enhanced calcineurin/Nfat-activation. These TAC-induced effects were prevented by the calcineurin-inhibitor cyclosporin-A, confirming the functional interaction between calcineurin/Nfat pathway activation and atrio-ventricular conduction in the setting of the Scn5a1798insD/+ mutation. Interestingly the only three carriers within this family that have died suddenly although they were implanted with a pacemaker were homozygous for the G-allele of rs2834506.
SCN5A mutations are associated with a wide range of clinical symptoms and syndromes, including long QT syndrome type 3 (LQT3), Brugada syndrome, sinus node dysfunction, (atrio-) ventricular conduction disease, atrial standstill, dilated cardiomyopathy and atrial fibrillation8. From a biophysical point of view, SCN5A mutations causing Brugada syndrome and conduction disease are typically loss-of-function mutations associated with reduced peak sodium current and/or sodium channel availability. In contrast, LQT3 mutations generally cause an increase in non-inactivating, persistent (sustained) inward sodium + (INa,sus), leading to continuous Na -influx during the entire duration of the action potential and subsequent prolongation of action potential and QT-interval. In some instances, as is the case for SCN5A-1795insD, one single mutation can cause both gain- and loss-of- function biophysical alterations in combination with a clinical overlap syndrome of LQT3, Brugada syndrome and/or conduction disease11. Our current findings demonstrate that 1798insD/+ 2+ enhanced INa,sus in Scn5a mice leads to increased intracellular Ca levels, thus setting the stage for the calcineurin/Nfat-pathway and its regulator RCAN1. Since the RCAN1 SNP had only an effect in mutation carriers and displayed no effect in a large sample of the general population, enhanced Na,susI and subsequent calcineurin/Nfat-activation appears essential for the effects of RCAN1 on atrio-ventricular conduction to occur. This is further corroborated by the finding that the association between the RCAN1 polymorphism and PR-interval appeared stronger in carriers of other SCN5A mutations presenting with a LQT3 phenotype, as compared to those presenting with Brugada syndrome and/or conduction disease.
In this study we also report suggestive linkage of chromosome 3 in the region of theCAV3 , GPD1L and SCN5A genes in modulation of heart rate and PR-interval. SCN5A encodes the pore-forming α- subunit of the cardiac sodium channel, while CAV3 and GPD1L encode respectively caveolin-3 and glycerol-3-phosphate dehydrogenase 1- like protein, both interacting with the sodium channel α- subunit32, 33. This effect was detected in the analysis
100 Identification of RCAN1 correcting for carriership of theSCN5A mutation, suggesting that genetic variability within this region, separate from the causal mutation in this family, also impacts on heart rate and atrio-ventricular conduction.
In recently published GWAS meta-analysis studies for QT-interval in the general population34, 35, genetic variation within or upstream of the NOS1AP gene was consistently the most significant association with this trait. In both of these studies rs12143842 emerged as the most strongly associated SNP in this region. Although in our study, this SNP was not genotyped directly, it was captured by SNP rs16847548 with an r2 of 0.82 (HapMap CEU) which was in turn significantly associated with QTc in our data (p=4.0e-04). The SNP most strongly associated with QTc in our data was however rs7539281 (p=4.0e-05).
In conclusion, we have applied a multi-faceted approach in both humans and mice segregating the same mutation in theSCN5A gene and have identifiedRCAN1 as a genetic modifier of phenotype severity in cardiac sodium channelopathy. Moreover, our findings point to abnormal intracellular Na+ and Ca2+ homeostasis in mediating this effect through the activation of calcineurin/Nfat signaling.
Acknowledgements We are indebted to the family for their participation and their cooperation throughout the years. We thank Tineke van der Laan for her genealogical research in the family. We also thank Wilma van der Roest, Murat Kiliçarslan and Maik Grundeken for their help. We are grateful to all the previous participants in studies related to this family as well to all referring physicians. Dr. J.J. Houwing (Leiden University Medical Center) and Dr. E.E.J. Creemers are thanked for helpful discussion. This study was supported by the Netherlands 5 Heart Foundation (NHS 2005T025, NHS 2008B051), the InterUniversity Cardiology Institute of the Netherlands (061.02), and the Division for Earth and Life Sciences (ALW; project 836.09.003) with financial aid from the Netherlands Organization for Scientific Research (NWO).
101 Chapter 5
Reference List 1. Priori,S.G. The fifteen years of discoveries 10. Postema,P.G. et al. Drugs and Brugada that shaped molecular electrophysiology: syndrome patients: review of the literature, time for appraisal. Circ. Res. 107, 451-456 recommendations, and an up-to-date (2010). website (www.brugadadrugs.org). Heart 2. Priori,S.G., Napolitano,C., & Schwartz,P.J. Rhythm. 6, 1335-1341 (2009). Low penetrance in the long-QT syndrome: 11. Remme,C.A. et al. Overlap syndrome of clinical impact. Circulation 99, 529-533 cardiac sodium channel disease in mice (1999). carrying the equivalent mutation of human 3. Scicluna,B.P., Wilde,A.A., & Bezzina,C.R. SCN5A-1795insD. Circulation 114, 2584- The primary arrhythmia syndromes: same 2594 (2006). mutation, different manifestations. Are we 12. Kaab,S. et al. A Large Candidate Gene starting to understand why? J. Cardiovasc. Survey Identifies the KCNE1 D85N Electrophysiol. 19, 445-452 (2008). Polymorphism as a Possible Modulator of 4. Crotti,L. et al. KCNH2-K897T is a genetic Drug-Induced Torsades de Pointes. Circ. modifier of latent congenital long-QT Cardiovasc. Genet.(2011). syndrome. Circulation 112, 1251-1258 13. Chambers,J.C. et al. Genetic variation in (2005). SCN10A influences cardiac conduction. 5. Remme,C.A. et al. Genetically determined Nat. Genet. 42, 149-152 (2010). differences in sodium current characteristics 14. Sanguinetti,M.C. et al. Coassembly of modulate conduction disease severity in K(V)LQT1 and minK (IsK) proteins to form mice with cardiac sodium channelopathy. cardiac I(Ks) potassium channel. Nature Circ. Res. 104, 1283-1292 (2009). 384, 80-83 (1996). 6. Crotti,L.et al. NOS1AP is a genetic modifier 15. Molkentin,J.D. et al. A calcineurin- of the long-QT syndrome. Circulation 120, dependent transcriptional pathway for 1657-1663 (2009). cardiac hypertrophy. Cell 93, 215-228 7. Tomas,M. et al. Polymorphisms in the (1998). NOS1AP gene modulate QT interval 16. Dong,D. et al. Overexpression of calcineurin duration and risk of arrhythmias in the in mouse causes sudden cardiac death long QT syndrome. J. Am. Coll. Cardiol. 55, associated with decreased density of K+ 2745-2752 (2010). channels. Cardiovasc. Res. 57, 320-332 8. Remme,C.A., Wilde,A.A., & Bezzina,C.R. (2003). Cardiac sodium channel overlap 17. Guo,J. et al. Decrease in density of INa is in syndromes: different faces of SCN5A the common final pathway to heart block in mutations. Trends Cardiovasc. Med. 18, murine hearts overexpressing calcineurin. 78-87 (2008). Am. J. Physiol Heart Circ. Physiol 291, 9. Bezzina,C. et al. A single Na(+) channel H2669-H2679 (2006). mutation causing both long-QT and 18. Bierhuizen,M.F. et al. In calcineurin- Brugada syndromes. Circ. Res. 85, 1206- induced cardiac hypertrophy expression 1213 (1999). of Nav1.5, Cx40 and Cx43 is reduced by different mechanisms. J. Mol. Cell Cardiol. 45, 373-384 (2008).
102 Identification of RCAN1
19. Scicluna,B.P. et al. Quantitative trait loci 28. Sabeti,P.C. et al. Genome-wide detection for electrocardiographic parameters and and characterization of positive selection arrhythmia in the mouse. J. Mol. Cell in human populations. Nature 449, 913- Cardiol. 50, 380-389 (2011). 918 (2007). 20. Beaufort-Krol,G.C. et al. Developmental 29. Rothermel,B.A., Vega,R.B., & Williams,R.S. aspects of long QT syndrome type 3 and The role of modulatory calcineurin- Brugada syndrome on the basis of a single interacting proteins in calcineurin signaling. SCN5A mutation in childhood. J. Am. Coll. Trends Cardiovasc. Med. 13, 15-21 (2003). Cardiol. 46, 331-337 (2005). 30. Shin,S.Y., Yang,H.W., Kim,J.R., Do,H.W., 21. Priori,S.G. et al. Risk stratification in the & Cho,K.H. A hidden incoherent switch long-QT syndrome. N. Engl. J. Med. 348, regulates RCAN1 in the calcineurin-NFAT 1866-1874 (2003). signaling network. J. Cell Sci. 124, 82-90 22. Gehi,A.K., Duong,T.D., Metz,L.D., (2011). Gomes,J.A., & Mehta,D. Risk stratification 31. Vega,R.B. et al. Dual roles of modulatory of individuals with the Brugada calcineurin-interacting protein 1 in cardiac electrocardiogram: a meta-analysis. J. hypertrophy. Proc. Natl. Acad. Sci. U. S. A Cardiovasc. Electrophysiol. 17, 577-583 100, 669-674 (2003). (2006). 32. Teng,G.Q. et al. Homozygous missense 23. van Den Berg,M.P. et al. Possible N629D hERG (KCNH2) potassium channel bradycardic mode of death and successful mutation causes developmental defects in pacemaker treatment in a large family the right ventricle and its outflow tract and with features of long QT syndrome type embryonic lethality. Circ. Res. 103, 1483- 3 and Brugada syndrome. J. Cardiovasc. 1491 (2008). Electrophysiol. 12, 630-636 (2001). 33. Vatta,M. et al. Mutant caveolin-3 induces 24. Yang,P. et al. Allelic variants in long-QT persistent late sodium current and is disease genes in patients with drug- associated with long-QT syndrome. associated torsades de pointes. Circulation Circulation 114, 2104-2112 (2006). 5 105, 1943-1948 (2002). 34. Newton-Cheh,C. et al. Common variants 25. Crotti,L. Genetic predisposition to sudden at ten loci influence QT interval duration in cardiac death. Curr. Opin. Cardiol. 26, 46- the QTGEN Study. Nat. Genet. 41, 399-406 50 (2011). (2009). 26. Kolder,I.C., Tanck,M.W., & Bezzina,C.R. 35. Pfeufer,A. et al. Common variants at ten Common genetic variation modulating loci modulate the QT interval duration in cardiac ECG parameters and susceptibility the QTSCD Study. Nat. Genet. 41, 407-414 to sudden cardiac death. J. Mol. Cell (2009). Cardiol. 52, 620-629 (2012). 36. Moennig,G. et al. Clinical value of 27. Laird,N.M. & Lange,C. Family-based designs electrocardiographic parameters in in the age of large-scale gene-association genotyped individuals with familial long QT studies. Nat. Rev. Genet. 7, 385-394 (2006). syndrome. Pacing Clin. Electrophysiol. 24, 406-415 (2001).
103 Chapter 5
37. Postema,P.G., de Jong,J.S., Van der Bilt,I.A., 47. Johnson,A.D. et al. SNAP: a web-based tool & Wilde,A.A. Accurate electrocardiographic for identification and annotation of proxy assessment of the QT interval: teach the SNPs using HapMap. Bioinformatics. 24, tangent. Heart Rhythm. 5, 1015-1018 2938-2939 (2008). (2008). 38. de Bakker,P.I. et al. Efficiency and power in genetic association studies.Nat. Genet. 37, 1217-1223 (2005). 39. A haplotype map of the human genome. Nature 437, 1299-1320 (2005). 40. Gabriel,S.B. et al. The structure of haplotype blocks in the human genome. Science 296, 2225-2229 (2002). 41. Lange,K., Weeks,D., & Boehnke,M. Programs for Pedigree Analysis: MENDEL, FISHER, and dGENE. Genet. Epidemiol. 5, 471-472 (1988). 42. Atkinson,B. & Therneau,T. kinship: mixed- effects Cox models, sparse matrices, and modeling data from large pedigrees. 2009. Ref Type: Computer Program 43. Lander,E. & Kruglyak,L. Genetic dissection of complex traits: guidelines for interpreting and reporting linkage results. Nat. Genet. 11, 241-247 (1995). 44. Nyholt,D.R. A simple correction for multiple testing for single-nucleotide polymorphisms in linkage disequilibrium with each other. Am. J. Hum. Genet. 74, 765-769 (2004). 45. Baartscheer,A. et al. Chronic inhibition of Na+/H+-exchanger attenuates cardiac hypertrophy and prevents cellular remodeling in heart failure. Cardiovasc. Res. 65, 83-92 (2005). 46. Baartscheer,A., Schumacher,C.A., & Fiolet,J.W. Small changes of cytosolic sodium in rat ventricular myocytes measured with SBFI in emission ratio mode. J. Mol. Cell Cardiol. 29, 3375-3383 (1997).
104 Chapter 6
Family-based genome-wide association analysis for the identification of genetic modifiers of heart rate and electrocardiographic indices of conduction and repolarization in a large Dutch family with a muta - tion in SCN5A.
Iris C.R.M. Kolder1,2, Yuka Mizusawa1, Pieter G. Postema1, Nynke Hofman3, Maarten P. Van den Berg4, J. Peter van Tintelen5, Arthur A.M. Wilde1, Michael W.T. Tanck2, Connie R. Bezzina1.
Affiliations: 1Heart Failure Research Center, Department of Clinical and Experimental Cardiology, 2Department of Clinical Epidemiology, Biostatistics and Bioinformatics, and 3Department of Clinical Genetics, Academic Medical Center, Amsterdam, The Netherlands 4Department of Cardiology and 5Department of Clinical Genetics, University Medical Center Groningen, Groningen, The Netherlands Chapter 6
Abstract Background A previous candidate study uncovered a novel genetic modifier in the extended kindred harbouring the SCN5A mutation 1795insD with clinical features of Long QT syndrome, Brugada syndrome and progressive cardiac conduction disease (occurring either in isolation or in combinations thereof). In this study, we continue the search for genetic modifiers that play a role in the phenotypic variability in this family using a genome-wide approach.
Methods and Results Genome-wide association and genome wide interaction analysis in the SCN5A-1795insD family were performed. A total of 120 carriers and 156 of their non-carrier family members were genotyped with the Illumina Human610-Quad assay. One SNP (rs2631864) in the region of GFRA2 passed the genome-wide significance threshold with PR-interval (p-value=3.1 ´ 10-9). In addition, two suggestive associated SNPs were found in LMCD1. This gene is part of the previously found associated calcineurin pathway. There were several other SNPs found to be associated with heart rate and the ECG indices. Multiple interaction effects were detected, but only at a suggestive significance level.
Conclusion We identified another gene in the calcineurin pathway that acts as a modifier ofPR- interval duration in the setting of cardiac sodium channel disease. This gene andthe other possible modifiers we identified may, after validation in future studies, contribute to our increased understanding of the genetic underpinnings of these complex cardiac electrophysiological traits.
Abbreviations and Acronyms ECG = electrocardiogram SNP = single nucleotide polymorphism SCD = sudden cardiac death
106 Family-based genome-wide association
Introduction Our understanding of the genetic basis of the inherited primary arrhythmia disorders has increased drastically over the past 15 years. We have learned that most of these disorders are caused by mutations in a large spectrum of genes that encode ion channel subunits and their modulators. Phenotypic variability in the inherited arrhythmia disorders has been reported in genotype-phenotype studies1. Many of these studies identified variability in clinical manifestation between family members that carry the same gene mutation. This observation points to a role of other genetic variants in modification of the underlying disease mechanism2.
In the last 5 years several large genome-wide association studies (GWAs) have uncovered many SNPs associated with ECG indices3. In a previous candidate gene study in a large Dutch kindred carrying the SCN5A-1795insD mutation with clinical features of Long QT syndrome, Brugada syndrome and progressive cardiac conduction disease (occurring either in isolation or in combinations thereof) we have previously identified genetic variation on chromosome 21 within the RCAN1 gene as a genetic modifier of the PR- interval, a measure of atrio-ventricular conduction [see Chapter 5 of thesis]. Although this finding led to novel mechanistic insight, this study was limited to only candidate genes and as expected a lot of the variability in the severity of the disease features remained unexplained. In the present study we carried out additional genetic studies, namely a genome-wide association analysis, in the same large kindred harboring the SCN5A-1795insD mutation to identify common genetic variation impacting on heart rate and electrocardiographic (ECG) indices of conduction (PR-interval, QRS-duration) and repolarization (QTc-interval).
Methods Study population The study population comprised members of a white Dutch family harbouring the SCN5A-1795insD mutation. This mutation is associated with manifestations of Long QT 6 syndrome (type 3), Brugada syndrome and progressive conduction disease, occurring either in isolation or in combinations thereof. Individuals included in the study had both DNA and ECG data available and provided their (written) informed consent. Carrier status for SCN5A-1795insD was determined by direct sequencing as described previously4. The study was approved by the institutional review board of the clinical center at which the patients were recruited.
107 Chapter 6
Phenotyping Heart-rate and ECG indices of conduction and repolarization were measured from the first available resting ECG in the absence of anti-arrhythmic drugs. All ECGs were digitalized and analyzed using ImageJ (http://rsb.info.nih.gov/ij/). Only sinus rhythm complexes were analyzed. Measurements of all parameters (heart rate, PR-interval, QRS-duration and QT- interval) were done manually on-screen, in lead II whenever possible. Parameters were averaged from up to 3 consecutive beats with similar preceding RR-intervals. For QT and heart-rate corrected QT (QTc), the tangent method with Bazett’s correction was used5.
SNP selection and genotyping SNP genotyping was performed using the Illumina Human610-Quad chip (Illumina Inc., SanDiego, USA). The 610-chip contains 550.000 SNPs and 60.000 copy number variants (CNVs). Multiple quality control measures were implemented. Exclusion criteria included sample call rate <0.95, SNP call rate <0.98 and minor allele frequency (MAF) <0.05. Reported r2 are based on linkage disequilibrium patterns in the general CEU population (www.broadinstitute.org/mpg/snap/).
Statistical analyses The ECG parameters PR-interval, QRS-duration, QTc-interval and heart rate were normally distributed (Shapiro-Wilk-test, W>0.90) and are reported as mean ± standard deviation (SD). Differences in ECG parameters between mutation carriers and non-carriers and effects of sex and age on these ECG-characteristics were analyzed using linear regression (two-sided significance threshold of 0.05). Family-based association analysis was carried out using two different models,
Model 1: corrected for sex, age, SCN5A-1795InsD mutation carriership status and including an interaction term for carriership ´ SNP, and Model 2: as model 1 without the interaction term. An additive genetic model was assumed for both models. All analyses were performed using the linear mixed-effect model function (lmekin) in the Kinship package (1.1.3)6 in R (http://www.r-project.org/), thus correcting for family correlation.
For all GWA analyses, we applied a two-sided significance threshold of 6.3 ´ 10-9 (5×108 / 4 phenotypes *2 models) for genome-wide significance and a arbitrary threshold of 6.3 x 10-6 (5×105 / 4 phenotypes *2 models).
108 Family-based genome-wide association
Results After quality control a total of 276 (120 mutation carriers) family members from the SCN5A-1795insD kindred with ECG data available were included in this study. A total of 503.877 SNPs (92%) passed the quality control criteria (see methods for details) and were used in the association analyses.
Study population and ECG parameters; Influence of age and sex As expected, ECG conduction (PR-interval, QRS-duration) and repolarization (QTc-interval) parameters were significantly prolonged in SCN5A-1795insD carriers versus non-carriers (Table 1). There was a striking variability in all ECG parameters, both among mutation carriers and non-carriers as observed previously in the candidate gene study (Chapter 5).
Table 1 | ECG characteristics of SCN5A-1795insD carriers and non-carriers from the family.
Carriers (n=120) Non-Carriers (n=156) P-value
Male/female (%) 48% 51% 0.62
Age at ECG (years) 25±19 37±21 1.94e-06
Heart-rate (bpm) 75±18 74±15 0.66
PQ interval (ms) 166±28 150±25 2.46e-06
QRS duration (ms) 94±14 86±13 9.86e-06
QTc interval (ms) 453±36 395±22 2.13e-42
Data depicts mean ± standard deviation
Genome-wide (interaction) association study on ECG conduction indices All SNPs with a P-value < 6.3 ´ 10-6 are depicted in Table 2. For the conduction parameter PR- interval, one SNP (rs2631864) on chromosome 8 displayed a p-value of 3.1 ´ 10-9 thereby passing our pre-specified genome-wide significance threshold (figure 1A). This SNPs is intergenic, the closest gene (Figure 1B) being GDNF family receptor alpha 2 (GFRA2). SNP 6 (rs10503700) in that region passed our arbitrary threshold p=4.6 ´ 10-6 (r2<0.5). Other SNPs passing our arbitrary threshold for PR-interval were: 1) two SNPs (rs9820959 and rs17049256; r2=0.68) on chromosome 3 in the gene LIM and cysteine-rich domains 1 (LMCD1), 2) two SNPs (rs10793137 and rs11236611; r2=0.90) on chromosome 11 in the gene SH3 and multiple ankyrin repeat domains 2 (SHANK2) and 3) three intergenic SNPs were found on chromosome 1 (rs9659337), 3 (rs1374879) and 12 (rs750208), respectively. The closest genes to these intergenic SNPs are presented in Table 2.
109 Chapter 6 P-value 4.16e-06 2.18e-06 1.12e-06 3.17e-07 4.56e-06 8.11e-07 4.55e-06 1.62e-06 3.14e-09 P-value 2.89e-06 4.05e-07 8.72e-07 1.39e-06
eta ±SE eta B eta ±SE eta 15.70 ± 2.99 11.48 ± 2.27 20.54 ± 3.35 B 11.02 ± 2.34 12.33 ± 2.63 16.02 ± 3.31 10.43 ± 2.22 14.68 ± 2.99 -10.36 ± 2.08 6.19 ± 1.29 -6.77 ± 1.34 -6.63 ± 1.34 -6.79 ± 1.31 0.18 0.06 0.16 0.09 0.36 0.33 0.32 0.08 0.12 MAF 0.13 0.28 0.28 0.28 MAF T/C T/C T/C C/T C/T T/G A/C G/T G/A G/A G/A G/A A/G Allele Allele Coded / Coded / Non-coded Non-coded Non-coded Non-coded # # # ZNF14 LMCD1 SHANK2 KRT76,KRT3* closest gene(s) closest closest gene(s) closest PDZRN3,CNTN3, APPL2,C12orf75, KRT2,KRT1,KRT77, LZTS1,LOC286114, SIPA1L2,KIAA1383* GFRA2,DOK2,XPO7* MIR4444-1,FAM86DP NUAK1,CKAP4,TCP11L2* with PR-interval, QRS-duration, QTc-interval and heart rate. and heart QTc-interval QRS-duration, with PR-interval, LOC100287814,DISC1, DISC2, LOC100287814,DISC1, -06 SNP SNP rs247772 rs750208 rs1521338 rs9659337 rs1374879 rs9820959 rs2163813 rs2631864 rs17049256 rs10503700 rs10793137 rs11236611 rs17699333
Intronic Intronic Intronic Intronic Intronic Intronic Location Location Location Location Intergenic Intergenic Intergenic Intergenic Intergenic Intergenic Intergenic SNPs found to be associated at a p-value<6.3x10 at be associated to found SNPs
-interval 2 | 1q42.2 3p12.3 3p26.1 8p21.3 11q13.3 12q23.3 R 12q13.13 19p13.11 P le Chromosome Chromosome QRS-duration b a T
110 Family-based genome-wide association P-value 5.34e-06 1.62e-06 5.82e-06 2.54e-06 4.94e-06 2.23e-06 3.89e-07 6.28e-07 1.88e-06
eta ±SE eta B 9.34 ± 2.01 9.32 ± 2.01 6.98 ± 1.45 7.79 ± 1.61 7.75 ± 1.49 8.04 ± 1.57 7.59 ± 1.56 -7.04 ± 1.51 -13.50 ± 2.75 0.11 0.11 0.38 0.15 0.22 0.38 0.38 0.23 0.27 MAF C/A A/C G/T G/A G/A A/G A/G A/G A/G Allele Coded / Non-coded Non-coded # NHSL1 SETD3,CCNK closest gene(s) closest PPP1R1C,PDE1A, LOC283547,SEC23A JAG2,GPR132,CDCA4 FOXA1,SSTR1,CLEC14A, DNAJC10,FRZB,NCKAP1* TSPAN11,LOC100506660* C14orf64,C14orf177,BCL11B, IPO8,CAPRIN2,LOC100287314, SNP rs33163 rs33162 rs2812081 rs1555409 rs4403996 rs4399485 rs4633639 rs10085225 rs10180264 6
Intronic Location Location Intergenic Intergenic Intergenic Intergenic Intergenic Intergenic Intergenic Intergenic 6q23.3 2q32.1 14q21.1 14q32.2 12p11.21 14q32.33 Heart rate Chromosome
111 Chapter 6 P-value 2.09e-06 4.71e-06 4.79e-06 4.95e-06 3.67e-06 6.96e-07 1.47e-06 For SNPs that are are that SNPs For #
eta ±SE eta B 14.71 ± 3.03 12.23 ± 2.59 13.92 ± 2.74 13.53 ± 2.74 -13.42 ± 2.87 -13.57 ± 2.90 -13.41 ± 2.88 0.15 0.43 0.21 0.20 0.30 0.28 0.28 MAF T/C T/C C/T A/C G/A G/A A/G Allele Coded / Non-coded Non-coded # ELMO1 BNIP3L* TPM1,LACTB* closest gene(s) closest ,PTPN1,MIR645* MGC15885,TLN2,MIR190A, RASGRP3,FAM98A,MYADML TMEM189,CEBPB,LOC284751 KCTD9,CDCA2,EBF2,PPP2R2A, SNP rs7340483 rs4871944 rs2392474 rs8040593 rs7169837 rs13043334 rs10495824
Intronic Location Location Intergenic Intergenic Intergenic Intergenic Intergenic Intergenic 2p22.3 8p21.2 7p14.2 15q22.2 20q13.13 QTc-interval Chromosome intragenic, only the gene harbouring the SNP is listed. MAF = minor allele frequency. MAF was based on 1000 Genomes CEU / Hapmap MAF was MAF = minor allele frequency. harbouring the SNP is listed. only the gene intragenic, Multiple SNPs listed at a given chromosomal locus represent multiple independent association signals at the given locus. For SNPs within intergenic regions, multiple all at independent association signals represent a locusSNPsintergenic givenat the chromosomal Multiple given For listed within locus. SNPs listed. are genes 5 nearest the 1Mb, flanking the in present were genes 10 than more *when listed; are 1Mb flanking the within located genes
112 Family-based genome-wide association
6
Figure 1 | Association results for PR-interval. (A) Genome-wide association analysis of SNPs with PR-interval. The red line indicates the pre-set genome-wide significance threshold of p<6.3e-9. The blue line indicates our arbitrary threshold of p<6.3 x 10-6. (B) Locus-specific association map, generated from genotyped SNPs, centered at rs2631864 on chromosome 8. The plot was generated using SNAP14. The correlation (r2) of the interrogated SNPs with rs2631864 is represented by red shading, with the strongest red representing the strongest correlation.
113 Chapter 6
Figure 2 | Genome-wide interaction analysis for PR-interval.The red line indicates the pre-set genome- wide significance threshold of p<6.3e-9. The blue line indicates our arbitrary threshold of p<6.3 x 10-6.
114 Family-based genome-wide association 0.01 non- 0.08 non- P-value carriers P-value carriers 2.01e-05 2.35e-05 1.99e-05 1.97e-04 2.79e-04 1.43e-05 with PR-interval, -6 1.41 1.40 1.45 1.46 1.52 non- 4.94 3.57 3.56 non- eta ±SE eta 6.48 ± 5.82 ± 6.37 ± eta ±SE eta 6.29 ± -6.19 ± -3.52 ± carriers 15.99 ± carriers 18.42 ± B B 0.02 P-value Carriers P-value Carriers 8.88e-04 1.16e-04 2.89e-03 1.69e-03 1.18e-05 4.74e-03 9.01e-03 1.88 1.84 1.86 1.89 2.00 3.48 5.00 4.46 eta ±SE eta 4.59 ± 6.31 ± eta ±SE eta -7.52 ± -5.80 ± -6.47 ± carriers carriers B -16.00 ± -11.85 ± B -15.86 ± P-value P-value P-value P-value 8.77e-07 5.33e-06 1.06e-07 2.48e-06 6.00e-06 3.60e-06 3.93e-07 8.85e-07 interaction interaction 0.40 0.38 0.48 0.48 0.41 MAF 0.35 0.08 0.18 MAF T/C T/C T/C C/T T/G T/C C/T C/T Allele Allele Coded / Coded / Non-coded Non-coded -1795insD atcarriership mutation a threshold p-value of p<6.3x10 Non-coded Non-coded SCN5A LHX2# SHANK2# ALOX15B* HTRA3,ACOX3, Nearest gene(s) Nearest Nearest gene(s) Nearest MIR4325,SPO11* CNTROB, GUCY2D, GUCY2D, CNTROB, KCNAB3, TRAPPC1, KCNAB3, LOC641364, SLC7A11 LOC641364, PCDH18, LOC641365, MeTTL19, GPR78,CPZ* FAM209B,TFAP2C,BMP7, 6 SNP SNP rs720629 rs7696692 rs2631723 rs3780680 rs1327291 rs1887341 rs9808591 rs7501530 Intronic Intronic Intronic Intronic Location Location Location Intergenic Intergenic Intergenic Intergenic SNPs found to have a significant interaction effect with SNPs to found a have significant interaction
3 | 3 -interval 4q28.3 4p16.1 9q33.3 11q13.4 17p13.1 le R 20q13.31 P b Chromosome Chromosome QRS-duration a T QRS-durarion and heart rate. QRS-durarion
115 Chapter 6 0.02 0.01 0.41 0.02 non- P-value carriers 5.91e-03 For SNPs that are are that SNPs For # 1.81 2.48 3.65 1.81 2.46 non- eta ±SE eta -4.34 ± -6.43 ± -1.49 ± -5.62 ± carriers -10.21 ± B P-value Carriers 2.27e-03 3.15e-03 2.67e-06 5.59e-05 2.44e-03 3.04 4.67 2.61 3.34 5.01 eta ±SE eta 9.51 ± 14.13 ± 13.00 ± 14.10 ± 15.58 ± carriers B P-value P-value 2.17e-06 2.24e-06 3.36e-06 4.04e-06 4.17e-06 interaction 0.23 0.13 0.27 0.17 0.06 MAF T/C T/C C/T G/A A/G Allele Coded / Non-coded Non-coded NHSL1# SLC12A3# LINC00301 STAM2,FMNL2, G6PC2,ABCB11, Nearest gene(s) Nearest MeTTL15,KCNA4, DHRS9, LRP2,BBS5* DHRS9, PRPF40A, ARL6IP6,RPRM SNP rs6910844 rs6434394 rs1491818 rs2289119 rs17270248 Intronic Intronic Location Location Intergenic Intergenic Intergenic 16q13 2q23.3 6q23.3 2q31.1 11p14.1 Heart rate Chromosome intragenic, only the gene harbouring the SNP is listed. MAF = minor allele frequency. MAF was based on 1000 Genomes CEU / Hapmap CEU. MAF was MAF = minor allele frequency. harbouring the SNP is listed. only the gene intragenic, Multiple SNPs listed at a given chromosomal locus represent multiple independent association signals at the given locus. For SNPs within intergenic regions, multiple all at independent association signals represent a locusSNPsintergenic givenat the chromosomal Multiple given For listed within locus. SNPs listed. are genes 5 nearest the 1Mb, flanking the in present were genes 10 than more *when listed; are 1Mb flanking the within located genes
116 Family-based genome-wide association
6
Figure 3 | Association results for QRS-duration. (A) Genome-wide association analysis of SNPs with QRS-interval. The red line indicates the pre-set genome-wide significance threshold of p<6.3e-9. The blue line indicates our arbitrary threshold of p<6.3 x 10-6. (B) Locus-specific association map, generated from genotyped SNPs, centered at rs247772 on chromosome 19. The plot was generated using SNAP14. The correlation (r2) of the interrogated SNPs with rs247772 is represented by red shading, with the strongest red representing the strongest correlation.
117 Chapter 6
In the genome-wide interaction analysis for PR-interval (Figure 2), three SNPS on chromosome 11 (rs720629, p=8.8 ´ 10-7), 17 (rs7501530, p=3.6 ´ 10-6) and 20 (rs9808591, p=3.9 ´ 10-7) passed the suggestive significance threshold. These SNPs were associated with a significant decrease in PR-interval in carriers but with a significant increase in non-carriers. SNP (rs720629) lies intronic in SHANK2 and is not in LD with the two SNPS (rs10793137 and rs11236611) mentioned above. The SNPS on chromosome 17 and 20 are intergenic and the closest genes are presented in Table 3.
For QRS-duration (Figure 3), the top three signals (rs247772, rs2163813 and rs17699333, r2=1) lie in the gene zinc finger protein 14 (ZNF14) on chromosome 19 or close to it. Next to these three SNPS, only SNP (rs1521338) on chromosome 12 passed our arbitrary threshold (p=2.9 ´ 10-6). This SNP is intergenic with more than ten genes in the 1MB flanking regions (Table 2).
Figure 4 | Genome-wide interaction analysis for QRS-duration. The red line indicates the pre-set genome-wide significance threshold of p<6.3e-9. The blue line indicates our arbitrary threshold of p<6.3 x 10-6.
118 Family-based genome-wide association
6
Figure 5 | Association results for QTc-interval. (A) Genome-wide association analysis of SNPs with heart rate. The red line indicates the pre-set genome-wide significance threshold of p<6.3e-9. The blue line indicates our arbitrary threshold of p<6.3 x 10-6. (B) Locus-specific association map, generated from genotyped SNPs, centered at rs7340483 on chromosome 2. The plot was generated using SNAP14. The correlation (r2) of the interrogated SNPs with rs7340483 is represented by red shading, with the strongest red representing the strongest correlation.
119 Chapter 6
In the genome-wide interaction analysis for QRS, five SNPs passed the suggestive significance threshold (Figure 4). Three of these SNPs (rs3780680, rs1327291, rs1887341; r2=0.74 – 0.93) were all located on chromosome 9 intronic in the gene LIM homeobox 2 (LHX2). These SNPs were associated with a significant decrease in QRS-duration in carriers but with a significant increase in non-carriers. The other two SNPs, rs7696692 (4q28.3, p=8.8 ´ 10-7) and rs2631723 (4p16.1, p=5.3 ´ 10-6) were independent signals on chromosome 4. These SNPs were associated with an increase in QRS-duration in carriers but decreasing in non-carriers (Table 3).
Figure 6 | Genome-wide interaction analysis for QTc-interval. The red line indicates the pre-set genome-wide significance threshold of p<6.3e-9. The blue line indicates our arbitrary threshold of p<6.3 x 10-6.
120 Family-based genome-wide association 0.22 0.02 0.10 0.11 0.11 0.01 0.07 0.71 0.71 0.71 0.11 0.07 0.09 0.09 0.50 0.07 0.59 0.12 non- P-value carriers with QTc-interval. eta ±SE eta -06 B 1.46 ± 2.71 -3.90 ± 3.15 -6.12 ± 2.61 -5.80 ± 3.50 -5.59 ± 3.45 -4.68 ± 2.91 -7.62 ± 3.02 non-carriers -4.86 ± 2.70 -1.17 ± 3.09 -1.17 ± 3.09 -1.17 ± 3.09 -5.59 ± 3.45 -4.87 ± 2.71 -4.78 ± 2.83 -4.78 ± 2.83 -2.20 ± 3.26 -4.88 ± 2.63 -4.85 ± 3.08 P-value Carriers 4.83e-05 5.86e-04 4.69e-05 5.51e-05 1.53e-04 5.86e-06 3.76e-04 1.45e-04 1.53e-04 1.53e-04 1.53e-04 5.26e-05 5.01e-05 1.83e-04 1.83e-04 8.36e-04 1.18e-03 7.27e-04 4.67 eta ±SE eta carriers -22.26 ± B 24.34 ± 5.76 14.89 ± 4.21 26.44 ± 6.24 25.94 ± 6.19 20.79 ± 5.30 18.78 ± 5.11 16.43 ± 4.17 24.24 ± 6.18 24.24 ± 6.18 24.24 ± 6.18 26.34 ± 6.26 19.16 ± 4.54 20.44 ± 5.28 20.44 ± 5.28 22.19 ± 6.46 15.94 ± 4.79 18.52 ± 5.33 P-value P-value 8.30e-07 4.14e-06 2.47e-06 3.27e-06 6.44e-07 2.94e-06 2.38e-06 2.74e-06 1.61e-06 1.61e-06 1.61e-06 2.26e-06 4.60e-06 5.18e-07 5.18e-07 5.58e-06 6.27e-06 4.41e-06 interaction 0.2 0.24 0.20 0.20 0.20 0.26 0.33 0.25 0.13 0.13 0.13 0.20 0.32 0.21 0.21 0.17 0.47 0.29 MAF T/C T/C T/C C/T C/T C/T A/C A/C G/T G/A G/A G/A A/G A/G A/G A/G A/G A/G Allele Coded / Non-coded Non-coded -1795insD mutation carriership at a p-value threshold of p<6.3x10 threshold at a p-value carriership -1795insD mutation SCN5A # # # # # # WLS BOLL, PDK1 ITGA6 CSMD2 RAPGEF4 RNF144A RAPGEF4, ZAK* Nearest gene(s) Nearest LOC285084,SP9* SESTD1, ZNF385B, ZNF385B, SESTD1, CDCA7, SP3, OLA1, PLCL1,SATB2,FLJ32063 ITGA6, PDK1, LOC91149, PDK1, LOC91149, ITGA6, MIR1258, CWC22, UBE2E3 CWC22, MIR1258, 6 SNP rs836616 rs262265 rs3769321 rs1550545 rs4578181 rs4395226 rs2820500 rs6738196 rs6661190 rs2606063 rs4972603 rs2701268 rs2116047 rs11694597 rs13019331 rs10193942 rs12693008 rs12615678 Intronic Intronic Intronic Intronic Intronic Intronic Intronic Intronic Intronic Intronic Intronic Location Location Intergenic Intergenic Intergenic Intergenic Intergenic Intergenic Intergenic SNPs found to have a significant interaction effect with interaction a significant have to found SNPs
4 | 1p35.1 2p25.2 2q31.1 2q31.3 2q33.1 1p31.3 le b QTc-interval Chromosome a T
121 Chapter 6 0.01 0.02 0.01 0.50 0.01 0.65 0.69 0.69 0.71 non- 0.18 0.02 0.12 P-value carriers 2.80e-03 eta ±SE eta B 3.59 4.70 1.12 ± 2.84 1.12 ± 2.84 1.06 ± 2.85 non-carriers -7.28 ± 2.95 -8.20 ± 3.40 -7.28 ± 2.95 -1.73 ± 2.57 -8.43 ± 3.40 -1.15 ± 2.53 -10.92 ± -11.34 ± -4.81 ± 3.54 -9.77 ± 6.32 P-value Carriers 2.20e-08 9.25e-05 1.40e-03 9.25e-05 4.89e-05 1.39e-03 2.87e-08 2.87e-08 3.67e-08 2.44e-03 4.73e-06 2.03e-04 2.86e-05 eta ±SE eta carriers 10.40 B 45.42 ± 23.77 ± 5.86 23.77 ± 5.86 21.00 ± 4.97 25.95 ± 7.92 25.29 ± 7.72 -25.58 ± 4.24 -25.67 ± 4.30 -25.67 ± 4.30 -25.70 ± 4.34 25.78 ± 8.31 26.67 ± 5.54 37.96 ± 9.88 P-value P-value 1.84e-06 1.63e-06 3.71e-06 1.63e-06 2.96e-06 4.72e-06 2.71e-06 2.71e-06 3.21e-06 1.83e-06 1.22e-06 1.77e-07 2.95e-06 interaction 0.16 0.21 0.17 0.46 0.23 0.40 0.37 0.37 0.37 MAF 0.18 0.18 0.06 0.03 C/T C/T C/T A/C G/T G/A G/A G/A C/T C/T A/G A/G A/G Allele Coded / Non-coded Non-coded * # # # # FHIT PCDH7 CSMD3 ICOS,PARD3B HMMR,MAT2B LOC100507117 Nearest gene(s) Nearest FAM49B, ASAP1, ASAP1, FAM49B, LYRM2/ANKRD6 CCNG1,NUDCD2, RAPH1,CD28,CTLA4, LOC728724, GSDMC, GSDMC, LOC728724, TRIP13, LOC100506688, TRIP13, LOC100506688, NKD2, SLC12A7, MIR4635 NKD2, SLC12A7, SNP rs158842 rs166078 rs158845 rs158851 rs4975526 rs2164384 rs6717395 rs2594264 rs1522059 rs4425724 rs1179905 rs1522060 rs1179907 Intronic Intronic Intronic Intronic Intronic Location Location Intergenic Intergenic Intergenic Intergenic Intergenic Intergenic Intergenic Intergenic 5q34 6q15 4p15.1 2q33.2 3p14.2 8q23.3 5p15.33 8q24.21 Chromosome
122 Family-based genome-wide association 03 0.28 0.04 0.21 0.84 0.58 0.11 0.30 0.62 0.17 0.20 0.96 0.75 0.09 non- 9.50e- P-value carriers 4.83 eta ±SE eta -12.69 ± B 3.51 ± 3.39 1.25 ± 2.49 0.91 ± 2.89 -4.49 ± 4.12 -2.27 ± 2.46 -3.22 ± 2.59 -0.58 ± 2.84 -1.90 ± 3.44 -4.17 ± 2.59 non-carriers -5.79 ± 4.17 -3.77 ± 2.95 -0.19 ± 4.20 -6.45 ± 3.74 For SNPs that are intragenic, intragenic, are that SNPs For # P-value Carriers 9.92e-06 1.28e-04 1.30e-04 1.26e-07 1.58e-05 4.67e-06 5.74e-06 7.91e-06 1.72e-03 3.53e-05 3.83e-05 5.55e-05 6.00e-07 5.15e-05 eta ±SE eta carriers B 35.69 ± 7.71 18.20 ± 4.59 16.83 ± 4.25 27.49 ± 4.86 29.15 ± 6.45 21.32 ± 4.43 34.11 ± 7.91 22.23 ± 5.18 39.62 ± 9.45 26.19 ± 4.94 24.54 ± 5.83 33.19 ± 10.33 -29.72 ± 6.23 -21.51 ± 4.59 P-value P-value 8.01e-08 5.50e-06 3.38e-06 3.92e-08 6.68e-07 1.25e-07 1.10e-06 5.02e-06 1.82e-06 2.02e-07 5.23e-07 3.01e-06 1.71e-06 1.77e-06 interaction 0.09 0.38 0.49 0.44 0.20 0.43 0.13 0.27 0.10 0.09 0.32 0.08 0.42 0.22 MAF T/C T/C T/C T/C C/T T/G T/G C/A C/A G/T G/A A/G A/G A/G Allele Coded / Non-coded Non-coded # # # # # # # FLRT2 POP4 MYOF MCTP2 ARNT2 ZNF347 PPP1R32 FAM19A5 ARHGAP22 Nearest gene(s) Nearest 10,CDRT15,HS3ST3B1 HS3ST3A1,CDRT15P1,COX 6 SNP rs787650 rs3019200 rs3851552 rs1756575 rs1894543 rs5768683 rs7914646 rs1860124 rs1806437 rs4238521 rs10411749 rs11635007 rs10423215 rs12218054 synon coding- Intronic Intronic Intronic Intronic Intronic Intronic Intronic Intronic Intronic Intronic Intronic Location Location Intergenic Intergenic 19q12 17p12 11q12.2 14q31.3 15q26.2 15q25.1 10q11.22 22q13.32 19q13.42 10q23.33 Chromosome only the gene harbouring the SNP is listed. MAF = minor allele frequency. MAF was based on 1000 Genomes CEU / Hapmap MAF was MAF = minor allele frequency. harbouring the SNP is listed. only the gene Multiple SNPs listed at a given chromosomal locus represent multiple independent association signals at the given locus. For SNPs within intergenic regions, all genes all regions, intergenic within SNPs For locus. given at the signals association independent multiple represent locus chromosomal given a at listed SNPs Multiple listed. are genes 5 nearest the 1Mb, flanking the in present were genes 10 than more *when listed; are 1Mb flanking the within located
123 Chapter 6
Genome-wide (interaction) association study on repolarization (QTc-interval) Six out of seven SNPs that passed the arbitrary threshold with QTc were intergenic (Figure5). Two SNPs (rs7340483 and rs10495824, r2=0.95) on chromosome 2, two SNPs (rs8040593 and rs7169837; r2=0.96) on chromosome 15, one (rs4871944) on Chromosome 8 and one (rs13043334) on chromosome 20. The genes closest to these SNPs are listed in Table 2. The intronic SNP (rs2392474) was found on chromosome 7p14.2 in the gene engulfment and cell motility protein 1 (ELMO1)
A total of 45 SNPs divided over 24 chromosomal regions passed the suggestive threshold in the genome-wide interaction analysis with QTc (Figure 6). The lowest p-value was for rs3851552 (p=3.92 ´ 10-8), which lies on chromosome 10 intronic in the gene Rho GTPase activating protein 22 (ARHGAP22). Three other SNPs (rs7914646, rs1806437 and rs12218054; r2<0.59) in ARHGAP22 were observed. For the complete list of all the signals see Table 4.
Genome-wide (interaction) association study on heart rate There were nine suggestive associated SNPs with heart rate (Figure 7). On chromosome 14, there are five intergenic SNPS: one (rs2812081) at 14q21, one (rs1555409) at 14q32.2 and three (rs4403996, s4399485 and rs4633639; r2<0.51) at 14q32.33. Genes in the area of theses intergenic SNPs are depicted in table 2. Two SNPs (rs33163 and rs33162; r2=0.83) were found on chromosome 12, one SNP (rs10180264) on chromosome 2, and one SNP (rs10085225) was intronic in the gene NHS-like 1 (NHSL1) on chromosome 6.
The genome-wide interaction analysis revealed five SNPs passing the suggestive threshold (Figure 8). Two independent intergenic signals (rs6434394 and rs17270248; r2<0.5) were found on chromosome 2 and one intergenic SNP (rs1491818) was found on chromosome 11. The genes within the 1MB flanking regions of these SNPs are shown in table 3. The other two SNPs were intronic, i.e. one on chromosome 6 (rs6910844) and one on chromosome 16 (rs2289119). SNP rs6910844 lies in the gene NHSL1, but is not in LD with the other SNP found in this gene mentioned above. SNP rs2289119 lies in the gene solute carrier family 12 (sodium/chloride transporters), member 3 (SLC12A3). All these SNPs seem to increase heart rate in the carriers and decrease heart rate in the non-carriers.
Discussion We here conducted an unbiased genome-wide association study to uncover genetic modifiers of disease severity in a family with cardiac sodium channelopathy. Specifically we investigated the role of common genetic variants spread throughout the genome in modulation of heart rate and ECG indices of conduction and repolarization, important determinants of disease in this family. Furthermore, by using a statistical model accounting
124 Family-based genome-wide association
6
Figure 7 | Association results for heart rate. (A) Genome-wide association analysis of SNPs with heart rate. The red line indicates the pre-set genome-wide significance threshold of p<6.3e-9. The blue line indicates our arbitrary threshold of p<6.3 x 10-6. (B) Locus-specific association map, generated from genotyped SNPs, centered at rs4403996 on chromosome 14. The plot was generated using SNAP14. The correlation (r2) of the interrogated SNPs with rs4403996 is represented by red shading, with the strongest red representing the strongest correlation.
125 Chapter 6
Figure 8 | Genome-wide interaction analysis for heart rate.The red line indicates the pre-set genome- wide significance threshold of p<6.3e-9. The blue line indicates our arbitrary threshold of p<6.3 x 10-6.
for possible SNP x mutation interaction we searched for genetic variants that impacted on the ECG indices differently in the SCN5A mutation carriers and non-carriers in the family.
Our analysis uncovered several associations with high statistical significance. We chose to declare genome-wide statistical significance at a stringent Bonferroni-corrected p-value threshold, correcting for the testing of one million SNPs7. Such correction for one million SNPs is very often used in calculating genome-wide Bonferroni-corrected significance thresholds in association studies on unrelated individuals of European descent. One could however argue that this threshold may be too conservative in a family-based study as ours, as among related individuals linkage-disequilibrium (LD) extends over a greater distance as compared to unrelated individuals, meaning that SNPs that are not in LD among unrelated individuals may be in high LD among related individuals in a family. Nevertheless, statistical evidence for association, even at stringent p-value thresholds may still be due to chance8. Thus, to exclude the possibility that the signals we detected are due to chance, it is therefore vital to replicate our findings in other patients with sodium channelopathy and/or in samples of the general population in future studies.
126 Family-based genome-wide association
Only one SNP passed our pre-specified genome-wide significance threshold. This was rs2631864 on chromosome 8p21.3 which was associated with the PR-interval with a p-value of 3.14x10-09. This SNP is located in an intergenic region and the closest characterized gene is GFRA2 encoding GDNF (glial cell line-derived neurotrophic factor) family receptor alpha 2. Of note, mice that are knock-out for this gene [Gfra2(-/-)] showed reduced cholinergic innervation by 40% in the ventricles and by 60% in the ventricular conduction system9. This would support a role for this gene at this locus in mediation of the effect observed on PR-interval.
Although not reaching genome-wide statistical significance another interesting locus associating with PR-interval was on chromosome 3p26.1 at rs9820959 (p=3.17x10-07) and rs17049256 (p=1.62x10-06). Both SNPs are located within the LMCD1 gene encoding LIM and cysteine-rich domains protein 1, a player in calcineurin/Nfat signaling10. This finding is of particular interest as in the same family we previously identified genetic variants in RCAN1 - a player in the same pathway – as modulators of PR-interval. Further studies into the possible role of LMCD1 in mediating the effect on the PR-interval at this locus are therefore highly warranted and are expected to shed more light on the relevance of the calcineurin/Nfat pathway in modulation of atrio-ventricular conduction in the setting of cardiac sodium channelopathy. Along this line, it will be pertinent to determine in future studies whether the effect of theLMCD1 locus is restricted to carriers of an SCN5A mutation as demonstrated for the RCAN1 locus.
By identifying chromosomal loci previously unlinked to the traits of interest, our findings potentially provide us with several new avenues for research into mechanisms involved in cardiac electrical function and susceptibility to arrhythmia. However, GWAS point us to the chromosomal regions involved, but tell us nothing about the causal gene or mechanism. Linking the genetic variant or haplotype to a specific gene and ultimately to an electrophysiological mechanism remains a major task11, 12. One major obstacle along this path is the fact that our understanding of the function of the non-protein coding regions of the genome is still quite rudimentary, while most of the common DNA variants modulating complex traits have been mapped there. A systematic strategy aimed at 6 understanding a particular GWAS locus will necessitate the generation of a comprehensive catalog of all the genetic variation within the region (which can be obtained by targeted sequencing) followed by annotation of the variable regulatory elements (e.g. enhancers, promoters, insulators, silencers). Overlaying GWAS data with other genomic data, for instance data on the location of transcription factor and enhancer binding sites (obtained by Chip-Seq)13 may be an efficient approach in prioritizing candidate regulatory sites and
127 Chapter 6 likely functional variants for downstream reporter assaysin vitro and/or in vivo. Expression QTL data, yet unavailable for human heart, is also expected to help in bridging the gap between functional variant and causal gene. Finally, once a transcript is associated with a risk allele, its involvement in the trait will ultimately need to be proven in functional follow-up studies relevant to the phenotype.
In conclusion we searched for genetic modifiers of heart rate and ECG indices of conduction and repolarization in a large family with theSCN5A -1795insD mutation. Several potentially interesting loci were identified, some of which may exert an effect in a mutation-specific manner. After careful validation in additional families with SCN5A mutations, the results herein may form the basis for future functional studies. This work clearly demonstrates the utility of the hypothesis-free genome-wide approach in identifying novel loci.
128 Family-based genome-wide association
Reference List 1. Scicluna,B.P., Wilde,A.A., & Bezzina,C.R. 10. Bian,Z.Y. et al. LIM and cysteine-rich The primary arrhythmia syndromes: same domains 1 regulates cardiac hypertrophy mutation, different manifestations. Are we by targeting calcineurin/nuclear factor of starting to understand why? J. Cardiovasc. activated T cells signaling.Hypertension 55, Electrophysiol. 19, 445-452 (2008). 257-263 (2010). 2. Cerrone,M., Napolitano,C., & Priori,S.G. 11. Arking,D.E. & Chakravarti,A. Understanding Genetics of ion-channel disorders. Curr. cardiovascular disease through the lens of Opin. Cardiol.(2012). genome-wide association studies. Trends 3. Kolder,I.C., Tanck,M.W., & Bezzina,C.R. Genet. 25, 387-394 (2009). Common genetic variation modulating 12. Freedman,M.L. et al. Principles for the cardiac ECG parameters and susceptibility post-GWAS functional characterization of to sudden cardiac death. J. Mol. Cell cancer risk loci. Nat. Genet. 43, 513-518 Cardiol. 52, 620-629 (2012). (2011). 4. Bezzina,C. et al. A single Na(+) channel 13. May,D. et al. Large-scale discovery of mutation causing both long-QT and enhancers from human heart tissue. Nat. Brugada syndromes. Circ. Res. 85, 1206- Genet.(2011). 1213 (1999). 14. Johnson,A.D. et al. SNAP: a web-based tool 5. Postema,P.G., de Jong,J.S., Van der Bilt,I.A., for identification and annotation of proxy & Wilde,A.A. Accurate electrocardiographic SNPs using HapMap. Bioinformatics. 24, assessment of the QT interval: teach the 2938-2939 (2008). tangent. Heart Rhythm. 5, 1015-1018 (2008). 6. Atkinson,B. & Therneau,T. kinship: mixed- effects Cox models, sparse matrices, and modeling data from large pedigrees. 2009. Ref Type: Computer Program 7. Hoggart,C.J., Clark,T.G., De,I.M., Whittaker,J.C., & Balding,D.J. Genome- wide significance for dense SNP and resequencing data. Genet. Epidemiol. 32, 179-185 (2008). 8. Gordon,D. & Finch,S.J. Factors affecting 6 statistical power in the detection of genetic association. J. Clin. Invest 115, 1408-1418 (2005). 9. Hiltunen,J.O., Laurikainen,A., Airaksinen,M.S., & Saarma,M. GDNF family receptors in the embryonic and postnatal rat heart and reduced cholinergic innervation in mice hearts lacking ret or GFRalpha2. Dev. Dyn. 219, 28-39 (2000).
129
Chapter 7
General Discussion And Future Perspective Chapter 7
General discussion In this thesis, we have focused on identifying genetic modifiers of disease in families with Mendelian cardiac disorders associated with a high risk of sudden cardiac death (SCD). Our aim was to identify genetic modifiers that could explain, at least in part, the phenomena of reduced penetrance and variable disease expression observed in these families, features commonly encountered in Mendelian disorders in general. Part of the work focused on the more-prevalent Mendelian cardiac disorders, namely hypertrophic cardiomyopathy (HCM) and Long QT Syndrome Type 2 (LQT2). Work also focused on a large family presenting with the rare phenotype of “overlap” cardiac sodium channel disease (Long QT Syndrome, Brugada Syndrome, Conduction Disease). In the primary electrical disorders we searched for genetic modifiers of heart rate and electrocardiographic (ECG) indices of conduction and repolarization as these constitute primary determinants of the disease. In HCM, we searched for genetic modifiers of echocardiographic parameters reflecting the extent of cardiac hypertrophy. Our studies included a linkage analysis and/ or association analysis approach and investigated candidate genetic variants and genes as well as an unbiased approach interrogating genetic variants genome-wide. In some of our studies we exploited the occurrence of founder mutations, allowing for the detection of genetic modifier effects in the setting of one or a limited number of primary gene defects, avoiding the heterogeneity in disease severity stemming from the different mutations encountered in patient sets with different primary gene defects. Besides confirming or refuting previously published associations at candidate loci, our studies uncovered multiple new loci modulating the traits of interest, providing a strong basis for future genetic validation studies and targeted functional studies in genes and pathways previously unlinked to cardiac electrical function.
In our studies on HCM, we sought to reproduce earlier findings from two studies1, 2 that reported a pro-left ventricle hypertrophy effect of five single nucleotide polymorphisms (SNPs) within genes active in the renin-angiotensin-aldosterone system (RAAS). For this, we used a set of 389 individuals with HCM carrying one of three functionally-equivalent truncating founder mutations in the MYBPC3 gene that are prevalent in the Netherlands. Contrary to the previous studies, our findings do not provide support for a marked effect of these genetic variants on the phenotypic expression of left ventricular hypertrophy. However, our study was limited to five candidate polymorphisms and the role of other genetic variants in explaining the phenotypic variability among HCM patients, certainly merits investigation perhaps in future genome-wide studies.
In a study on LQT2 probands and families harbouring mutations in the KCNH2 gene (438 patients), we investigated the possible modulatory effect of common genetic variants residing within or around 18 candidate genes on the extent of QTc-interval prolongation
132 General Discussion and Future Perspective in this disorder. For the first time, in this study we accounted for the varying effect of the primary gene defect (KCNH2 mutation in this case) on the extent of QTc-interval prolongation. In doing so, we confirmed and extended previous observations thatthe type and location of the KCNH2 mutation has a marked impact on disease severity3, providing a very strong argument for correcting for these effects in association studies searching for genetic modifiers. Our analysis confirmed previous observations that SNPs in the NOS1AP gene, known to impact on the QTc-interval in the general population4, 5, modulate the severity of QTc-interval prolongation in LQT2. Importantly, this study provided the first evidence that specific genetic variation inKCNH2 (the gene which also harbours the primary gene defect in these patients), which does not impact on the QTc- interval in the general population, is associated with the QTc-interval in patients with LQT2. This raises the intriguing possibility that this specific variant exerts its effect only in the setting of a KCNH2 mutation, where repolarization is already compromised, thereby providing a setting permissive to the identification of variants that are silent in the general population. The robustness of this association however must in the future be investigated in additional LQT2 probands and families. Studies in Long QT Syndrome patients with other gene defects will also be required to shed light as to whether this genetic variant is specific to KCNH2 mutation carriers as opposed to carriers of other Long QT Syndrome- causing gene defects such as mutations in KCNQ1 and SCN5A.
In this thesis, extensive work was carried out as to genetic modifiers in a large Dutch pedigree segregating the SCN5A-1795insD mutation, in which we searched for modifiers by linkage and association analysis in candidate genes as well as genome-wide. This family which presented clinically with both gain- as well as loss-of-function features of sodium channel disease, provided us with the unique opportunity to search for genetic modifiers of both the cardiac conduction as well as the repolarization process. The most important finding of the candidate gene study was the novel finding that genetic variation at the RCAN1 locus is associated with the PR-interval. RCAN1 encodes regulator of calcineurin 1 and acts in calcineurin/Nfat signaling, a pathway already implicated in modulation of atrio-ventricular conducton6, 7. The identification RCAN1of as a possible modulator of the PR-interval may be considered as a serendipitous finding as this region of chromosome 21 was included in our analysis only by virtue of the fact that it lies upstream of the KCNE1 gene, which is in turn a strong candidate for the modulation of the QTc-interval but has no likely role in PR-interval modulation. Through extensive functional studies in mice carrying the homologous mutation (Scn5a1798insD/+) we demonstrate that abnormal 7 intracellular calcium homeostasis as a consequence of the SCN5A-1795insD mutation provides a setting for activation of the calcineurin/Nfat pathway, subsequently impacting on the PR-interval. We argue that the activation of the calcineurin/Nfat pathway in this way provides a target for the action of the RCAN1 gene product and the related genetic
133 Chapter 7 variation in modulation of the PR-interval. This would imply that the effect of the RCAN1 variant would only be evident in SCN5A mutation carriers. In line with this, no effect of this SNP was detected in the general population. Furthermore we demonstrated a statistically significant interaction effect with SCN5A mutation-carriership when the study set was enlarged with additional SCN5A mutation carriers. To our knowledge, this is the first time that a SNP could be shown to interact with the primary genetic defect and thereby modulate the phenotype of a Mendelian rhythm disorder.
Further evidence implicating the calcineurin/Nfat pathway in modulation of thePR- interval was provided in our genome-wide association study in the family. In this analysis, genetic variation within theLMCD1 gene encoding LIM and cysteine-rich domains protein 1, a player in calcineurin/Nfat signaling8 was also found to modulate the PR-interval at high statistical significance. It will be very interesting to determine in future studies whether the effect of this genetic variant is restricted to carriers of anSCN5A mutation. In addition, the genome-wide association study uncovered multiple other potentially interesting loci modulating heart rate and ECG indices of conduction and repolarization, some of which with an SCN5A mutation-specific effect. These novel loci will benefit from replication studies in additional families with SCN5A mutations. This work clearly demonstrates the utility of the hypothesis-free genome-wide approach in identifying novel genes that were previously unlinked to cardiac electrical function and with the potential of opening new avenues for research.
Perspectives for the future This thesis identified common genetic variants impacting on intermediate traits relevant to the disease. The identification of genes previously unlinked to the cardiac traits of interest provides novel targets for future research into cardiac electrical function at baseline and in disease and for development of new therapies. Importantly, the identification of such variants may aid, in the future, after careful evaluation, in risk stratification in patients, contributing to a “personalized medicine” approach to patient care. In this respect, it is important to realize that the common genetic variants dealt with in this thesis are generally associated with modest effect sizes and may therefore have a more-relevant role in the determination of risk when considered in aggregate. Furthermore, future studies in the patients sets studied in this thesis should also encompass other strategies, including the analysis of rare genetic variants which may be associated with larger effects9. The rapid increase in the number of genomes and exomes being sequenced now enables the identification of such rare variants and these may in the future be tested in patient sets using specialized chips (e.g. exome chip). Furthermore, one could also sequence
134 General Discussion and Future Perspective
Figure 1 | An updated schematic representation of a cardiomyocyte depicting genes encoding channel subunits and interacting proteins involved in the primary electrical disorders or in cardiac electrical function. the exome and genome of the patients directly, with the added advantage that this has the ability of identifying private variants or variants that may be specific to the Dutch population.
Genetic variation may impact on a given trait and disease process either by impacting on protein function or by altering gene expression. In line with findings in genome- wide association studies in general10 our genome-wide approach identified a number of associations in intergenic regions or gene-poor chromosomal areas. Such genetic variants are likely to exert their modulatory function by impacting on gene regulation and thereby gene expression. Expression quantitative trait locus (eQTL) analysis, not yet available in cardiac tissue, would aid in identifying any cis- or trans-eQTL effects of these genetic variants, providing a genetic mechanism for the effect observed and candidate 7 gene identification for functional studies11. Combining the data generated here with other .emerging resources such as Chip-seq in cardiac tssue12 will also aid this process
135 Chapter 7
Undoubtedly the expansion of bio-banks and related databases of patients with Mendelian cardiac disorders should be pursued as this will provide increased statistical power for variant and gene discovery in addition to the resources necessary for validation of findings in discovery sets. To this aim, effort in collecting and characterizing large sets harbouring founder mutations will be especially useful.
In the case of the SCN5A-1795insD mutation in particular, where knock-in mice of two distinct inbred strains (FVB/NJ and 129P2) carrying the homologous mutation exist13, 14, future work should consider overlapping data generated from the gene-finding efforts of this thesis with those generated from studies in the segregating population of mice arising from the FVB/N-Scn5a1798insD/+ x 129P2-Scn5a1798insD/+ cross14. Such an effort provides unique possibilities for gene prioritization especially due to the possibility of a systems genetic approach in mice, for instance incorporating cardiac gene expression analysis, that may not be possible in patient studies.
136 General Discussion and Future Perspective
Reference List 1. Ortlepp,J.R. et al. Genetic polymorphisms 9. Asimit,J. & Zeggini,E. Rare variant in the renin-angiotensin-aldosterone association analysis methods for complex system associated with expression of left traits. Annu. Rev. Genet. 44, 293-308 ventricular hypertrophy in hypertrophic (2010). cardiomyopathy: a study of five 10. Hindorff,L.A. et al. Potential etiologic and polymorphic genes in a family with a functional implications of genome-wide disease causing mutation in the myosin association loci for human diseases and binding protein C gene. Heart 87, 270-275 traits. Proc. Natl. Acad. Sci. U. S. A 106, (2002). 9362-9367 (2009). 2. Perkins,M.J. et al. Gene-specific modifying 11. Cookson,W., Liang,L., Abecasis,G., effects of pro-LVH polymorphisms involving Moffatt,M., & Lathrop,M. Mapping the renin-angiotensin-aldosterone system complex disease traits with global gene among 389 unrelated patients with expression. Nat. Rev. Genet. 10, 184-194 hypertrophic cardiomyopathy. Eur. Heart J. (2009). 26, 2457-2462 (2005). 12. Blow,M.J. et al. ChIP-Seq identification of 3. Shimizu,W. et al. Genotype-phenotype weakly conserved heart enhancers. Nat. aspects of type 2 long QT syndrome. J. Am. Genet. 42, 806-810 (2010). Coll. Cardiol. 54, 2052-2062 (2009). 13. Remme,C.A. et al. Genetically determined 4. Newton-Cheh,C. et al. Common variants differences in sodium current characteristics at ten loci influence QT interval duration in modulate conduction disease severity in the QTGEN Study. Nat. Genet. 41, 399-406 mice with cardiac sodium channelopathy. (2009). Circ. Res. 104, 1283-1292 (2009). 5. Pfeufer,A. et al. Common variants at ten 14. Scicluna,B.P. et al. Quantitative trait loci loci modulate the QT interval duration in for electrocardiographic parameters and the QTSCD Study. Nat. Genet. 41, 407-414 arrhythmia in the mouse. J. Mol. Cell (2009). Cardiol. 50, 380-389 (2011). 6. Bierhuizen,M.F. et al. In calcineurin- induced cardiac hypertrophy expression of Nav1.5, Cx40 and Cx43 is reduced by different mechanisms. J. Mol. Cell Cardiol. 45, 373-384 (2008).
7. Dong,D. et al. Overexpression of calcineurin in mouse causes sudden cardiac death associated with decreased density of K+ channels. Cardiovasc. Res. 57, 320-332 (2003). 8. Bian,Z.Y. et al. LIM and cysteine-rich domains 1 regulates cardiac hypertrophy 7 by targeting calcineurin/nuclear factor of activated T cells signaling.Hypertension 55, 257-263 (2010).
137
Summary Summary
Summary Sudden cardiac death (SCD) is one of the most prevalent causes of death in Western societies. It underlies 20% of total mortality, and 50% of cardiovascular mortality. In young individuals (below 40 years of age) SCD often occurs in the setting of disorders displaying Mendelian inheritance, with the cardiomyopathies (chapter 3) and primary electrical disorders (chapters 4 - 6) being the most prevalent. Here, the inheritance of very rare genetic variants with large effects potentially increases risk for SCD substantially. The primary electrical disorders have been linked primarily to mutations in genes encoding ion channel subunits or their interacting proteins(Figure 1 chapter 1). On the other hand, the cardiomyopathies are caused by mutations affecting genes coding for the contractile apparatus and structural components of the cardiomyocyte such as the sarcomere and desmosomes
Genotype-phenotype studies in these disorders have clearly established that they are not spared from the phenomena of reduced penetrance and variable expression typical of Mendelian diseases. For instance, in the primary arrhythmia syndromes, extensive variability in clinical manifestations is often observed among family members carrying an identical ion channel gene mutation, with some individuals exhibiting overt abnormalities on the electrocardiogram (ECG) and suffering potentially fatal arrhythmias, whereas others do not display any ECG changes and do not develop rhythm disturbances throughout life. Probands and families with these Mendelian disorders, harboring known disease-causing mutations, likely provide a permissive, genetically sensitized setting for the identification of novel genes and pathways modulating cardiac (electrical) function.
In this thesis we employ the phenotypic variability evidenced among probands and their relatives with Mendelian cardiac disorders to identify genetic modifiers of disease expression. We focused on two distinct groups of disorders associated with increased risk of SCD, namely the primary electrical disorders (Long QT Syndrome, Brugada Syndrome, Conduction Disease) and hypertrophic cardiomyopathy (HCM). The aim of this thesis was to identify such genetic modifiers using both linkage and (family based) association analyses. A candidate SNP / gene approach as well as a genome-wide unbiased approach were used in the study of common genetic variants as possible modifiers of disease severity.
In chapter 2, we reviewed the available literature on the genetic and allelic architecture of SCD. In this review we focused on the common genetic variation that has been recently identified through genome-wide association studies to modulate risk ofSCD and to modulate heart rate and ECG indices of conduction (PR-interval, QRS-duration)
140 Summary and repolarization (QTc-interval) as intermediate phenotypes of SCD. Several studies reported that a family history of SCD increases an individual’s risk for SCD giving evidence for a heritable component. In the general population however, the genetic and allelic architecture remains largely unknown.
In patients with hypertrophic cardiomyopathy (HCM) the occurrence of phenotypic variability even in the presence of an identical pathogenic mutations suggests a role for genetic modifiers (chapter 3). The renin-angiotensin-aldosterone system (RAAS) plays a regulatory role in cardiac function, blood pressure and electrolyte homeostasis making it an interesting candidate system that could modify phenotypic expression in HCM patients. Five Single nucleotide polymorphisms (SNPs) in this system were previously suggested to modify the extent of hypertrophy in HCM patients. In order to investigate the effect of these SNPs, we selected a large cohort of carriers of one of 3 functionally-equivalent truncating mutations in the MYBPC3 gene. We used family based association to analyze the effect of these five RAAS polymorphisms (ACE, rs4646994; AGTR1, rs5186; CMA, rs1800875; AGT, rs699; CYP11B2, rs1799998) on interventricular septum (IVS) thickness and the Wigle score. We detected two modest associations. Carriers of the CC genotype in the AGT gene had less pronounced IVS thickness compared to CT and TT genotype carriers. The DD polymorphism in the ACE gene was associated with a high Wigle score (p=0.01). In our large study population of HCM patients with functionally-equivalent mutations in the MYBPC3 gene we did not find major effects of genetic variation within genes of the RAAS system on phenotypic expression of HCM or the previously described association between the pro-LVH score and IVS thickness /Wigle score.
In chapter 4 we studied a large set of individuals (probands and, their family-members where available) carrying a mutation in the KCNH2 gene and presenting clinically with Long QT syndrome type 2 (LQT2). LQT2 is a cardiac repolarization disorder that is caused by mutations in theKCNH2 gene which encodes kv11.1 (HERG) underlying the Ikr repolarizing K+ current. Although mutation type and location are known to impact on the severity of clinical manifestations, the occurrence of phenotypic variability among patients harboring the same mutation points to the presence of additional modulatory genetic factors. Here we comprehensively investigated the effect of 1201 haplotype-tagging SNPs in and around 18 candidate genes on the QTc-interval in 438 patients with LQT2. We performed family based association analysis, taking the effect of KCNH2 mutation type and location into account in our analysis for modifiers of QTc-interval. Two SNPs passed the Bonferroni- corrected significance threshold for association (p<4.16×10-5). rs16847548 located immediately 5’upstream of the NOS1AP gene, and rs956642, located in the vicinity of KCNH2, were associated with the QTc-interval). Of these, rs956642 was also found to be associated with cardiac events (p=0.02). Two other SNPs in NOS1AP, rs10494366 and
141 Summary rs12567211, identified previously by GWA studies, were also significantly associated with QTc in the LQT2 patients studied (p-values, 9.96×10-3 and 2.24×10-4, respectively). We provide the first evidence that common genetic variants at the KCNH2 locus modulate severity of clinical manifestations in the Long QT Syndrome type 2. Furthermore, we extend previous observations that common genetic variants in NOS1AP modulate the extent of QTc-prolongation in this disorder.
In the last two chapters we studied a large Dutch kindred with the SCN5A mutation 1795insD. An extensive genealogical search allowed us to trace this family back to the eighteenth century, enabling the construction of a highly extended pedigree. Individuals in this kindred present with manifestations of Long QT syndrome, Brugada syndrome and progressive conduction disease occurring either in isolation or in combinations thereof.
In chapter 5, we performed linkage and association analysis with heart rate and ECG indices of conduction and repolarization using 1308 haplotype-tagging SNPs inand around 18 candidate genes in 215 family members (100 carriers) of the 1795insD mutation. Both significant linkage (LOD=3.7) and association (p=9.8e-08) with PR-interval was found at the region of chromosome 21 harboring the KCNE1 and KCNE2 candidate genes. The SNP displaying the most significant association within this region (rs2834506, p=9.8e-08), was observed within intron 3 of the RCAN1 (Regulator of Calcineurin 1) gene. This association was subsequently validated in an independent set of patients harboring different mutations in SCN5A, also indicating a significant stronger effect of rs2834506 on PR-interval in the SCN5A mutation carriers compared to non-carriers. The KCNE1 and KCNE2 genes were considered unlikely modulators of PR-interval, so we sought additional evidence for a role of RCAN1 in mediating the observed effect. InScn5a 1798insD/+ F2 progeny of FVB/N and 129P2 mice displaying variable conduction disease severity, a significant correlation was found between ventricular Rcan1 mRNA transcript levels and PR-interval (n=56 mice; r=-0.333, p=0.012). Because RCAN1 is a regulator of the pro-hypertrophic calcineurin/Nfat-pathway, we hypothesized that the Scn5a1798insD/+ mutation disrupts intracellular Ca2+-homeostasis, thereby setting the stage for calcineurin-activation which subsequently impacts on PR-interval. In cardiomyocytes of Scn5a1798insD/+ mice elevated intracellular Na+ and diastolic Ca2+ levels were observed. Additionally, chronic activation of the calcineurin/Nfat pathway through application of Transverse Aortic Constriction (TAC) elicited extreme AV-dysfunction, AV-block and sudden death in Scn5a1798insD/+ mice, which was prevented by treatment with the Nfat-inhibitor cyclosporine-A. This evidence all pointed us to the calcineurin/Nfat pathway as a possible modifier of the PR-interval in the setting of sodium channelopathy, making it a very interesting pathways for further future studies.
142 Summary
Additionally in this family we performed a genome-wide association study (GWAs)(chapter 6). DNA and ECG data were available for 276 family members (120 carriers) of the SCN5A- 1795insD family. We performed family based genome wide association analysis. One SNP (rs2631864) in the region of GFRA2 passed the genome-wide significance threshold with PR-interval (p-value=3.1 ´ 10-9). This SNP is located in an intergenic region and the closest characterized gene is GFRA2 encoding GDNF (glial cell line-derived neurotrophic factor) family receptor alpha 2. Of note, mice that are knock-out for this gene [Gfra2(-/-)] showed reduced cholinergic innervation by 40% in the ventricles and by 60% in the ventricular conduction system. This would support a role for this gene at this locus in mediation of the effect observed on PR-interval. In addition, two suggestive associated SNPs were found in LMCD1. Both SNPs are located within the LMCD1 gene encoding LIM and cysteine-rich domains protein 1. This gene, like RCAN1 plays a role in the calcineurin signaling pathway, providing further evidence in support of this pathway in modulation of cardiac conduction. Further studies into the possible role of LMCD1 in mediating the effect on the PR-interval at this locus are therefore highly warranted and are expected to shed more light on the relevance of the calcineurin/Nfat pathway in modulation of atrio-ventricular conduction in the setting of cardiac sodium channelopathy. Along this line, it will be pertinent to determine in future studies whether the effect of theLMCD1 locus is restricted to carriers of an SCN5A mutation as demonstrated for theRCAN1 locus. Several other SNPs found to be associated with heart rate and the ECG indices and multiple interaction effects were also detected, but all at a suggestive significance level. Future mechanistic work on the candidate genes found in this study will likely provide further insight into the molecular underpinnings of cardiac electrophysiological function.
In summary, in this thesis, we have focused on identifying genetic modifiers of disease in families with Mendelian cardiac disorders associated with a high risk of SCD. Our aim was to identify genetic modifiers that could explain, at least in part, the phenomena of reduced penetrance and variable disease expression observed in these families, features commonly encountered in. Mendelian disorders in general. We uncovered several interesting new candidate genes and pathways associated with heart rate and ECG indices of conduction (PR-interval, QRS-duration) and repolarization (QTc-interval) that form prime candidates for functional studies in the future.
143
Samenvatting Samenvatting
Samenvatting Plotse hartdood (SCD) is een van de meest voorkomende doodsoorzaken in de westerse samenlevingen. Het is verantwoordelijk voor 20% van de totale mortaliteit, en 50% van de cardiovasculaire mortaliteit. Bij jonge mensen (onder de 40 jaar) komt SCD vaak voor in de context van Mendeliaanse overerving; de cardiomyopathieën (hoofdstuk 3) en primaire elektrische stoornissen (hoofdstuk 4 - 6) zijn de meest voorkomende. Hier word het risico op SCD aanzienlijk verhoogd door het erven van zeldzame genetische varianten met een groot effect. De primaire elektrische stoornissen zijn in de eerste plaats geassocieerd met mutaties in genen die coderen voor onderdelen die samen de ionenkanalen vormen of die kunnen binden aan deze kanalen (figuur ,1 hoofdstuk 1). De cardiomyopathieën echter zijn meestal geassocieerd met mutaties in genen die coderen voor de contractiele apparatuur en onderdelen van de cardiomyocyten, zoals het sarcomeer en desmosomen.
Genotype-fenotype studies in deze stoornissen hebben duidelijk aangetoond dat er vaak sprake is van verminderde penetrantie en variabele expressie van het fenotype. Bijvoorbeeld, in de primaire aritmie syndromen, is uitgebreide variatie in klinische verschijnselen waargenomen bij familieleden met een identieke ionkanaal genmutatie. Waar een aantal mensen afwijkingen vertonen op het elektrocardiogram (ECG) en potentieel fatale hartritmestoornissen kunnen ontwikkelen, hebben anderen geen abnormaal ECG en ontwikkelen ook geen ritmestoornissen gedurende hun leven. Probands en hun familieleden met deze Mendeliaanse aandoeningen, hebben bekende gen variaties, die zorgen voor een genetisch gevoelige systeem die daardoor kunnen helpen in de identificatie van nieuwe genen en systemen die de hart (elektrische) functie moduleren.
In het onderzoek beschreven in dit proefschrift maken we gebruik van de fenotypische variabiliteit onder probands en hun familieleden met Mendeliaanse hartafwijkingen om genetische factoren te identificeren die de expressie van de ziekte moduleren. We concentreerden ons op twee verschillende groepen van aandoeningen geassocieerd met een verhoogd risico op SCD, namelijk de primaire elektrische stoornissen (lange QT syndroom, Brugada Syndroom, geleiding ziekte) en hypertrofische cardiomyopathie (HCM). Het doel van dit proefschrift was om dergelijke genetische modifiers met behulp van zowel linkage en (familie) associatie analyses te identificeren. Een kandidaat SNP/ gen aanpak en een genoom wijde benadering werden gebruikt in de studie van veel voorkomende genetische varianten die mogelijk de ernst van de ziekte modificeren.
Hoofdstuk 2 is een review gericht op de veel voorkomende genetische variatie die onlangs geïdentificeerd zijn door middel van genoom wijde associatie studies voor het moduleren van het risico op SCD en die hartslag en ECG-indices van geleiding (PR-interval, QRS-
146 Samenvatting duur) en repolarisatie (QTc-interval) moduleren als intermediaire fenotypes van SCD. Verschillende studies laten zien dat een familiegeschiedenis van SCD het risico van een individu op SCD vergroot; dit is bewijs voor het bestaan van een erfelijke component in deze ziekte. In de algemene bevolking echter, blijft de genetische en allelische architectuur grotendeels onbekend.
Patiënten met hypertrofische cardiomyopathie (HCM), met een identiek pathogene mutatie hebben vaak verschillende fenotypes, wat een rol voor genetische modifiers suggereert (hoofdstuk 3). Het renine-angiotensine-aldosteron systeem (RAAS) speelt een regulerende rol in hartfunctie, bloeddruk en de elektrolytenhuishouding, waardoor het een interessant kandidaat-systeem is dat de fenotypische expressie in HCM patiënten zou kunnen veranderen. Vijf Singel nucleotide polymorphismes (SNPs) in dit systeem zijn eerder geassocieerd met variabele hypertrofie in HCM patiënten. Om het effect van deze SNPs te onderzoeken, selecteerden we een grote groep dragers van 3 functioneel equivalente truncatie mutaties in het MYBPC3 gen. We gebruikten familie associatie analyse om het effect van de vijf RAAS polymorfismen (ACE, rs4646994, AGTR1, rs5186, CMA, rs1800875, AGT, rs699, CYP11B2, rs1799998) te associëren met interventriculaire septum (IVS) dikte en de Wigle score. We hebben twee bescheiden associaties geïdentificeerd. Dragers van het CC genotype in het AGT-gen hadden minder uitgesproken IVS dikte ten opzichte van CT-en TT-genotype dragers. Het DD polymorfisme in het ACE-gen werd geassocieerd met een hoge Wigle score (p = 0,01). In onze grote populatie studie van HCM patiënten met een functioneel-equivalente mutatie in het gen MYBPC3 vonden we geen grote effecten van genetische variatie binnen genen van het RAAS-systeem op de fenotypische expressie van HCM of de eerder beschreven associatie tussen de pro-LVH score en IVS dikte/Wigle score.
In hoofdstuk 4 bestudeerden we een groot aantal mensen (probands en hun familieleden indien beschikbaar) met het lange QT syndroom en een mutatie in het genKCNH2 dragen en presenteren met klinisch lange QT-syndroom type 2 (LQT2). LQT2 is een cardiale repolarisatie aandoening die wordt veroorzaakt door mutaties in het KCNH2 gen dat + Kv11.1 (HERG) codeert en ten grondslag liggen aan de Ikr repolarisatie K stroom. Hoewel bekend is dat het type mutatie en locatie van de mutatie invloed hebben op de ernst van de klinische verschijnselen, wijst het ontstaan van fenotypische variabiliteit bij patiënten met dezelfde mutatie op de aanwezigheid van extra modulerende genetische factoren. Hier hebben we onderzoek gedaan naar het effect van 1201 haplotype-tagging SNPs in en rond de 18 kandidaat-genen, op QTc-interval duur in 438 patiënten met LQT2. In de familie associatie analyse werd rekening gehouden met heteffect vanKCNH2 mutatie type. Twee SNPs passeerden de Bonferroni-gecorrigeerde drempel voor associatie (p <4,16 ´ 10-5): rs16847548, direct 5’ upstream van het NOS1AP gen, en rs956642, gelegen in de nabijheid
147 Samenvatting van KCNH2, werden in verband gebracht met het QTc-interval. Hiervan werd rs956642 ook geassocieerd met cardiale events (p = 0,02). Twee andere SNPs in NOS1AP, rs10494366 en rs12567211, die in eerdere GWA studies geïdentificeerd waren, waren ook significant geassocieerd met QTc in de LQT2 patiënten (p-waarden, 9,96 × 10-3 en 2,24 × 10-4, resp.). Dit is de eerste studie die aantoont dat de veel voorkomende genetische varianten in KCNH2 de ernst van de klinische manifestaties in LQT2 kan moduleren. Verder bevestigen we eerder gevonden observaties met veel voorkomende genetische varianten inNOS1AP die de QTc- moduleren in deze aandoening
In de laatste twee hoofdstukken bestudeerden we een grote Nederlandse familie met de SCN5A mutatie 1795insD. Door middel van uitgebreid genealogische onderzoek konden we deze familie terug voeren tot de achttiende eeuw, waardoor er een zeer uitgebreid stamboom ontstond. Personen in deze familie hebben het lange QT syndroom, Brugada- syndroom en progressieve geleiding ziekte, of een combinatie daarvan.
In hoofdstuk 5 hebben we linkage en associatie analyse uitgevoerd voor hartslag en ECG- indices van geleiding en repolarisatie door middel van 1308 haplotype-tagging SNPs in en rond 18 kandidaat-genen in 215 leden (100 dragers) van de 1795insD mutatie-familie. Zowel significante linkage (LOD = 3.7) en associatie (p = 9.8e-08) met PR-interval werd gevonden in de regio van chromosoom 21 waar de kandidaat genen KCNE1 en KCNE2 bevinden. De SNP met de meest significante associatie in deze regio (rs2834506, p = 9.8e-08), bevind zich in intron 3 van het RCAN1 (Regulator van calcineurin 1) gen. Deze associatie werd vervolgens gevalideerd in een onafhankelijke set van patiënten met verschillende mutaties in SCN5A: het effect van rs2834506 op PR-interval in de SCN5A mutatiedragers is groter dan in de patienten zonder mutatie. De genen KCNE1 en KCNE2 waren niet de meest voor de hand liggende modulatoren van PR-interval, dus we zochten aanvullend bewijs voor een rol van RCAN1 in het moduleren van het waargenomen effect. Hiervoor werd gebruik gemaakt van een genetisch gemodificeerd muismodel met de scn5a-1798insD mutatie (equivalent aan de 1795insD mutatie in de mens). In Scn5a1798insD/+ F2 nakomelingen van FVB/N en 129P2 muizen die verschillende mate van geleidings problemen laten zien, werd een significante correlatie gevonden tussen de ventriculaire Rcan1 mRNA transcript niveaus en PR-interval (n = 56 muizen; r = -0,333, p = 0,012). Omdat RCAN1 een regulator is van het pro-hypertrofische calcineurin/ NFAT-systeem, veronderstelden we dat de Scn5a1798insD/+ mutatie de intracellulaire Ca2+- homeostase verstoort, waardoor calcineurin geactiveerd kan worden, die vervolgens van invloed is op PR-interval. In hartspiercellen van Scn5a1798insD/+ muizen werden verhoogde intracellulaire Na+ en diastolische Ca2+ niveaus waargenomen. Bovendien veroorzaakte chronische activering van de calcineurin/NFAT systeem door het toepassing van dwars aorta vernauwing (TAC) extreme AV-disfunctie, AV-blok en het plots overlijden van
148 Samenvatting
Scn5a1798insD/+ muizen. Dit laatste kon worden voorkomen door behandeling met de NFAT- inhibitor cyclosporine-A . Dit bewijs wijst allemaal naar de calcineurin/NFAT route als een mogelijke modifier van PR-interval in de context van een natriumkanaal ziekte, waardoor het een zeer interessant systeem is voor toekomstige studies.
Verder hebben we in de SCN5A-1795insD familie een genoom wijde associatie studie (GWAS) gedaan (hoofdstuk 6). DNA en ECG gegevens waren beschikbaar voor 276 familieleden (120 dragers) van deze familie. We hebben een familie based associatie analyse uitgevoerd. Een SNP (rs2631864) in de regio van GFRA2 passeerde de genoom wijde significatie drempel voor PR-interval (p-waarde = 3,1 ´ 10-9). Deze SNP is gelegen tussen genen in, het dichtstbijzijnde gen is GFRA2 dat codeert voor GDNF (gliale cellijn afgeleide neurotrofe factor) familie receptor alpha 2. Interessant is dat knock-out muizen voor dit gen [Gfra2 (- / -)] 40% verminderde cholinerge innervatie in de ventrikels vertoont en 60% vermindering in het ventriculaire geleidingssysteem. Dit is extra bewijs dat dit gen een rol zou kunnen spelen in het waargenomen effect op PR-interval. Daarnaast vonden we twee suggestieve geassocieerde SNPs in LMCD1. Beide SNPs zijn gelegen in het LMCD1 gen dat codeert voor LIM en cysteïne-rijke domeinen eiwit 1. Dit gen, speelt, net als RCAN1, een rol in het calcineurin/NFAT systeem wat het eerder gevonden bewijs ondersteunt in de rol van dit systeem in de modulatie van de geleiding in het hart. Verdere studies naar de mogelijke rol van LMCD1 in het moduleren van PR-interval zijn dan ook zeer gerechtvaardigd en zullen naar verwachting meer licht werpen op de relevantie van de calcineurin/NFAT systeem in de modulatie van de atrio-ventriculaire geleiding in een achtergrond van hart-natrium kanaal ziekten. Het is relevant om te bepalen in toekomstige studies of het effect van de LMCD1 locus is beperkt tot dragers van een mutatie in SCN5A zoals is gebleken voor de RCAN1 locus. Verschillende andere SNPs waren ook geassocieerd met hartslag en de ECG-indices en meerdere interactie-effecten werden ook gevonden, maar allemaal met suggestieve associatie. Toekomstig moleculair werk op de kandidaat-genen gevonden in deze studie zal meer inzicht kunnen geven in de moleculaire onderbouwing van de cardiale elektrofysiologische functie.
Samenvattend, in het onderzoek beschreven in dit proefschrift hebben we ons gericht op het identificeren van genetische modifiers in families met Mendeliaanse cardiale aandoeningen geassocieerd met een hoog risico op SCD. Ons doel was om genetische modifiers te identificeren die, de verschijnselen van verminderde penetrantie en variabele expressie van ziekte die waargenomen wordt binnen families (ten dele) zouden kunnen verklaren. We ontdekten een aantal interessante nieuwe kandidaat-genen en systemen die geassocieerd zijn met hartslag en ECG-indices van de geleiding (PR-interval, QRS-duur) en repolarisatie (QT-interval) die interessante kandidaten zijn voor functionele studies in de toekomst.
149
Dankwoord Dankwoord
Dankwoord Een proefschrift maak je nooit alleen dus ik wil graag iedereen bedanken die eenrol gespeeld heeft in het tot stand komen van mijn proefschrift. Sommigen wil ik hier graag in het bijzonder noemen.
Mijn promotores, Prof. Dr. Arthur Wilde & Prof. Dr. Koos Zwinderman en mijn co- promotores, Dr. Michael Tanck en Prof. Dr. Connie Bezzina. Arthur, bedankt voor al je inzet en je hulp, de wetenschappelijke input en discussies. Koos, ik heb het enorm gewaardeerd dat ondanks dat je niet mijn dagelijkse begeleider was, je deur toch altijd voor mij open stond. Met veel humor stond je open voor de acties (“cadeaus”) vanuit 207 en ging je zelfs mee naar de epsteinbar om een biertje te doen. Michael, bedankt voor alle hulp en input die je me de afgelopen jaren gegeven hebt, ik heb veel van je geleerd. De “Koffie!” pauzes waren altijd gezellig; soms bespraken we wat werk maar vaak kwamen ook de mooie verhalen over de kermis en niet te vergeten de Carnaval in Lievelde ter sprake :-). Wij representeerden het AMC als enige AMC’ers in de boerenkelder en de grote markt in Leuven. Ik weet zeker dat we in de toekomst nog weleens samen zullen werken! Connie, thank you for all your help over the last years. As it was very stressful at times, you were never worried and “no…don’t worry” has become sort of your catch-phrase. In the end you were right, because here we are. :-) I’m sure with your new appointment as a professor you’ll get very far!
Prof. Dr. F. Baas, Prof.Dr. R.N.W. Hauer, Prof. Dr. J.J.P. Kastelein, Prof. Dr. H. Meijers- Heijboer en Dr. M.M.A.M. Mannens: bedankt voor jullie bereidheid om zitting te nemen in de commissie. Dr. J.J. Schott, thank you for your willingness to take part in this committee.
Mijn paranimfjes Roos Kolder en Tamara Koopmann, Heel erg bedankt voor al jullie hulp en inzet ik kan me geen beter paar bedenken die de 29ste achter mij staat. Al hoewel de geheime FB berichtjes me toch lichtelijk bang maken voor het avondprogramma ;-)
Tamara, je bent mijn grote steun en toeverlaat geweest of het nu ging om labwerk of mijn eerste praatje bij de AHA. Team KoKo heeft mooie avonturen gehad Nashville, AHA ,Hockey pucks, Roze olifanten en niet te vergeten de Schietvereniging te Almelo ;-) Ik weet zeker dat met jouw doorzettingsvermogen en humor je van je nieuwe avontuur in Toronto een groot succes zal gaan maken en ik kom je zeker opzoeken!
Roos, ben erg blij geweest met onze spontane lunches, lekker bijkletsen en af en toe even lekker ventileren! Ik ben heel trots op jou als zus; je bent een doorzetter, een Kolder en ik weet zeker dat dit een heel mooi jaar gaat worden! Ik ga je missen, maar ik weet dat je me snel komt opzoeken hier in Engeland voor een Australië 2.0, zonder de knuffelbeestjes en mudcake, maar in ieder geval met een paar topshops :-P.
152 Dankwoord
Teodora, “sharing is caring”: gezien je dit de afgelopen jaren een paar keer per dag zei, leek het me wel toepasselijk om het aan m’n stellingen toe te voegen, gezien we ons hele promotie traject en nu ook onze promotie dag en feest mogen delen! Ook al zat je niet officieel in 207, je had er wel je eigen stoel en je was altijd van de partij voor random dingen. Een hoogtepunt was denk ik wel ons Gøtenborg avontuur: “Last women standing”. Het is dan ook zeer toepasselijk dat jij degene was die het bewijs vond dat wij Kolders ons genetisch aangepast hebben aan onze omgeving ;-). Ik weet zeker dat wij de 29ste die titel weer kunnen claimen. De Patho gaat zeker blij met je zijn!!
207, Voor de meeste mensen bekend als de leukste kamer op de KEBB en waarschijnlijk het hele AMC. Ik heb hier mijn hele promotietijd doorgebracht en dat was een avontuur! De laatste (en waarschijnlijk meest beruchte) formatie van 207 bestond uit De guru (Raha), Enrico (Erik), Pinkman (Michel), T en ik. Kerstbomen, Snipen, Jersey shore, Eppen (inc/exl wodkoe), lekker dwalen en vooral veel randomness ;-).
Erik en Michel, de een kan niet genoemd worden zonder de andere! Vanaf het begin dat Michel zijn intrek nam in 207 heb ik jullie Bromance zien ontwikkelen. Jullie hebben mijn tijd in 207 echt heel gezellie gemaakt. Heel veel succes met jullie onderzoek, lekker blijven eppen en niet vergeten af en toe lekker te dwalen, maar niet naar Amstelveen, liever naar Cambridge ofzo ;-). T en ik zorgen dat de Cøla speciaal koud staat de 29ste!
Raha, thanks for the support and insights. I liked our talks a lot! I wish you all the best with finalizing your thesis and of course with “baby GWAs”. I’m sure you’ll come up with an awesome name.
Joost, meneer laten we 1 biertje doen…….. Werden er toch altijd een paar meer ;-). Volgend jaar met zijn 3en naar NKOTBSB, vind het super jammer dat ik het gemist heb. Zo gauw ik weet wanneer de Bird heropend wordt, laat ik het je weten! Na al die verhalen ben ik zoooo benieuwd!!!!
Suthesh, Sje mijn bijna buurman, co-host van de wetenschap quiz. We hebben een leuke tijd gehad. Niet alleen met het Aprove bestuur, het geweldige Nerdenbal, maar ook de “gewone” feestjes! Met als hoogtepunt natuurlijk de Illumina VIP party op Hawaii ;-). Gelukkig kom je vaak in de UK, dus dat komt helemaal goed!
Al mijn collega’s bij de KEBB: Miranda, het liefst hadden we natuurlijk een triple promotie gehad, maar ja, 11-11-11 was niet echt haalbaar. Sterkte met de laatste loodjes; ik zal voor je duimen. Barcelona is een super stad kom vast wel een keer die richting op. Sandra, bedankt voor de gezellige lunches. Ik mis je mooie taart creaties.Hans, de mooie verhalen en de gezellig lunches waren altijd top. Inge, koffie?....Biertje? Dankzij onze drukke
153 Dankwoord leventjes zijn we er niet meer aan toe gekomen, dus ik zeg: Biertje de 29ste!! Jammbe, we gaan de 29ste zeker een “cøla” drinken, met of zonder speciaal. Umesh, iMesh put your iPhone down. No more pics!!! Kimiko, ook al was je veblijf in 207 maar van korte duur, ik vond het super. De kamer was nooit meer zo roze en zen ;-). Veel succes in de opleiding. Marjolein, heerlijke nuchterheid! Hopelijk is je boekje ook bijna af, maar gelukkig ben je al lekker in opleiding. Succes daarmee! Annetski, je gaat T en mij net voor. Hoe blij zijn we dat het af is, yeah! Parvin, geweldig hoe snel je het Nederlands oppikt. Mij ben je in ieder geval al voorbij gegroeid. Heel veel succes met je promotie! Anne-floor en Esther ,heel erg leuk dat jullie de aio meetings voorzitten. Succes met iedereen gemotiveerd houden (bier helpt!) Barbera, erg leuk dat je de eerste bekende was die ik in Cambridge zag. Zat er ten slotte al twee weken :-). Het was een uitdaging om een bar te vinden; Watson en Crick’s Eagle sloot al om 11 uur! Maar gelukkig is het ons gelukt! Rosa, Eleanor, Fleur en Shayan , heel veel succes met jullie promoties! Angela, zeker na jullie KEBB uitje was je vaak in 207: oplader lenen en de nieuwe “gevonden” items bewonderen. Altijd gezellig en wij gaan zeker de dansvloer onveilig maken de 29ste ;-). Aldo, heb jou volgens mij nog nooit zonder glimlach gezien :-). Lekker sociaal en altijd gezellig een praatje maken! Hoop dat we in de toekomst nog eens zullen samenwerken. Barbara, Joke, Joost, Mariska, Susan, Rob, Hans, Silvia, Antoine & Tessa, jullie ook bedankt voor alles. Het secretariaat niet te vergeten natuurlijk: Gré, Petra en Hannie. Gré bedankt voor al je hulp met de laatste loodjes, maar vooral bedankt voor alle gezelligheid die jij aan de KEBB en 207 bracht! De kerstboom, het leuke KEBB uitje en natuurlijk de lekkere borrels!
De Bezzina group: Carol ann, hoe blij zijn we dat RCAN1 nu toch eindelijk de deur uit is! Margriet, onze beruchte ep fotograaf ;-). Bedankt voor je werk aan LQT2s en je geweldige skill om overal en dan ook overal Passoã vandaan te toveren! Wat elke feestje natuurlijk nog leuker maakt ;-). Leander, samples en nog meer samples pipetteren voor de GWAs…. Wat waren het er toch veel, maar met jouw hulp en humor kan ik nu eindelijk zeggen: Frack, I’m finished man! The french connection ;-),Mathilde and Julien. Brendon, I’m sure you can appreciate my book cover ;-), cheers! Roos, denk dat we beiden Stockholm nooit meer zullen vergeten. Die lift, de posterkoker :-).Yuka, thanks for the wonderful Sushi and all the ECG’s you read for my studies! Elisabeth, Maaike, Annalisa, Britta, Anne, Rianne, Rob, Nina, jullie ook bedankt voor alles.
Mijn andere collega’s bij de Cardiologie,Klinische genetica en het hartfaalcentrum en co- auteurs: Dr. Phil, always up for some dancing and laughing. Luckily I select the music on the 29th :-). Cheers! Ingrid, ook voor jou de laatste loodjes , sterkte! Als je nog wat ontspanning nodig hebt geloof ik dat we met de andere dames nog een 80s party zouden doen :-). Minder glitter en meer haarlak!Christian, listen C….Het zit er voor mij op, Jij bent gelukkig ook bijna klaar en lekker in opleiding! Heel veel succes met alles. Ciao. Esther,
154 Dankwoord nu zijn we allemaal in de positie voor KoKoES. Yeah! Pieter, bedankt voor al je werk aan de 1795insD familie! Fijn dat de paper nu eindelijk de deur uit is. Nynke, bedankt dat ik de woensdagen bij jou mocht zitten om alle stambomen en de database te maken. Het was een geworstel met blauwe mappen en antieke computer systemen, maar het is ons gelukt! Alex,Yigal, Amin, Imke, Michelle en Dennis jullie ook bedankt voor alles.
Mijn mede Aprove commissie leden: Suthesh, Matthijs, Fleur, Anniek, Danny, Olaf, Hamid en Marije, bedankt voor alle leuke vergaderingen en natuurlijk het bestuursuitje!
My new team the Barret Group: Jeff, Kate, Luke, Yang and James, thanks for the warm welcome!
Mijn lieve vrienden, De Chica’s: a.k.a. Lotte, Ewa, Hanna en Marjolein. Bedankt voor de gezellige jaarclub etentjes/weekendjes en al jullie steun en afleiding <3. Sandra en Vinny, de geweldige VIS avondjes/uitjes. Ik denk dat we bijna alle achtbanen in Europa gehad hebben ;-). We missen alleen de Engelse nog en daar kunnen we nu zeker aan gaan werken gezien jij, San, nu in Londen woont. Roadtrip….!! Manuel, super bedankt voor de mooie kaft van mijn boekje! Bedankt (Thanks) Anna, Dave, Karin, Jacob, Judith, Alexander, Kirsten, Helen, Jetske, Anne, Maartje, Jelena, Aldred, Anja voor alle interesse en afleiding! Iedereen bij Blackbear bedankt voor de heerlijke uitlaadklep voor al mijn frustratie.
Wietse, je bent een top broer. We hebben af en toe lekker gestuiterd op feestjes en ik weet zeker dat we dat in de toekomst nog vele malen gaan doen. In ieder geval de 29ste, want als dj Wietse draait dan komt het helemaal goed!!
Oudertjes: Querido papá y lieve mama, bedankt voor al jullie steun en liefde. Het waren pittige tijden de afgelopen 4,5 jaar, maar jullie hebben altijd in me geloofd!
-X- Iris
155