Linkage Analysis and Candidate Gene Resequencing

The search for genetic variants underlying AF has evolved with improvements in genetic techniques and with the recognition of a widespread heritable basis for AF. Early investigations used classical linkage analysis to identify monogenic disease susceptibility loci in families. Linkage analysis relies on the use of genetic markers at known locations of the genome that can help determine which loci transmit along with disease.

The first genetic locus for AF was described in 1997 by Brugada et al, who identified an AF susceptibility region on chromosome 10 in a large family with autosomal dominant AF (Table 49-1).⁸ Since this initial report, linkage analysis has repeatedly been used to identify genetic mutations underlying AF in a number of large families with AF.

Chen et al narrowed an AF susceptibility locus to a region on chromosome 11p15 in a family with autosomal dominant AF spanning four generations.⁹ Investigators extended their analysis by sequencing KCNQ1, a candidate gene in the region that encodes a potassium channel α-subunit. In the first transmembrane segment of the channel, they discovered a highly conserved serine residue that was mutated to a glycine in all affected family members. Subsequent characterization revealed that the mutant protein increases the repolarizing I_Ks current density when expressed with the KCNE1-subunit. This gain-of-function mutation is expected to result in shortened atrial refractory periods and thereby promote reentry, a well-founded mechanism underlying AF.⁵¹ Our group used a similar approach to map a mutation in the third transmembrane domain of KCNQ1.¹⁰ The S209P variant demonstrated a similar gain-of-function effect again implicating this as a disease susceptibility gene for AF.

Linkage analysis also has identified mutations in SCN5A that segregate with AF, conduction abnormalities, cardiomyopathy, and possibly early-onset ischemic stroke.24–26 SCN5A encodes a sodium channel α-subunit responsible for the depolarizing I_Na current. The D1275N mutation that results in an aspartic acid for asparagine substitution was independently identified by different investigators,^24–26 and associates with atrial standstill when co-segregating with connexin-40 (GJA5) mutations.⁵²

In yet another large multigenerational family with prevalent AF, Hodgson-Zingman et al mapped a frameshift mutation to NPPA, which encodes atrial natriuretic peptide.⁴⁷ The two–base-pair deletion eliminates a stop codon, resulting in an additional 12 amino acids at the carboxy terminus of the mature 28-residue long atrial natriuretic peptide (ANP) protein. Ex vivo rabbit hearts bathed in the mutant peptide demonstrated significantly shortened atrial effective refractory periods as compared with those bathed with a wild-type atrial natriuretic peptide, again consistent with reentry as a pathogenic model for AF.

In aggregate, linkage analysis has implicated potassium (KCNQ19,10) and sodium (SCN5A^24–26) channel mutations, a nuclear envelope protein (NUP155^40,41), NPPA,⁴⁷ and loci on chromosomes 6⁴⁹ and 10.^8,50 Newer genetic techniques such as exome and whole genome sequencing have emerged in recent years that make mapping of these families even more efficient. However, the large, multigenerational families needed for linkage analysis remain rare. More often AF is observed in smaller families and thus is it is often hard to establish the causality of apparently disease-causing mutations with such limited genetic information.

Candidate Gene Association Studies

In contrast to linkage mapping, in which inferences about recombination are made on the basis of segregation of markers in a pedigree, investigators have also selected and screened candidate genes for mutations in cohorts of patients with and without AF. As an extrapolation of initial studies on KCNQ1, numerous additional candidate gene association studies have focused on cardiac ion channels (see Table 49-1).

In one report, genetic variants in KCNQ1 were identified that predisposed to AF in a stretch-sensitive fashion,¹² illustrating the potential for a concealed predisposition for AF to be elicited by an acquired exposure (e.g., valvular disease, heart failure). In another report implicating KCNQ1, a mutation was identified in a patient with long QT syndrome.¹³ Diverging effects on I_Ks were observed depending on whether the mutant protein was expressed with KCNE1 or KCNE2 β-subunits, each of which are differentially expressed in the atria and ventricles, leading the authors to speculate that the mutation was the cause of both long QT syndrome and AF. Gain-of-function mutations in both KCNQ1¹¹ and KCNH2²⁰ have been discovered in AF in the context of short QT syndrome.

In contrast to the enhanced atrial repolarization mechanism invoked by most discovered potassium channel mutations, a nucleotide substitution resulting in a premature stop codon in KCNA5 has been described that manifested a loss-of-function of the Kv1.5 channel protein.²³ The mutation effectively abolished I_Kur and prolonged the action potential duration, increasing susceptibility to early afterdepolarization-induced atrial tachycardia and fibrillation. This mechanism has been described as “atrial torsades,” and observations from this report underscore the heterogeneity of mechanisms that can lead to AF.

Candidate gene association studies in patients with AF also have identified variants in genes encoding sodium channel subunits. Both loss-of-function²⁷ and gain-of-function^28,29 mutations in SCN5A have been identified in patients with lone AF. Mutations in sodium channel β-subunits, SCN1B and SCN2B, have been described that decrease sodium current amplitude and alter channel gating kinetics when coexpressed with SCN5A.³⁰ These mutations are speculated to predispose to AF through shortening of the atrial action potential duration or by conduction slowing, both of which may facilitate reentry.

Mutations in GJA5, which encodes the gap junction connexin-40, have been described in 4 of 15 patients with AF screened in one candidate gene association study.³² The GJA5 mutations in three of the four patients were present in cardiac tissue but not in lymphocyte specimens, suggesting that these mutations were acquired or somatic rather than inherited or germline mutations. Other candidate gene association studies have linked common genetic variation in GJA5 to AF.^31,33

Non-ion channel variants also have been discovered in candidate gene association studies. In a recent large-scale cardiovascular candidate gene association study, an intronic variant in the IL6R receptor was associated with AF.⁴⁸ Mutations in LMNA, which encodes the nuclear envelope proteins lamin A and lamin C, have been identified in patients with cardiomyopathy and AF.^36,37 Mutations in these proteins underlie a diverse spectrum of disorders that include Emery-Dreifuss syndrome, Charcot-Marie-Tooth disease, and premature aging syndromes. Nevertheless, LMNA mutations appear to be rare causes of AF.³⁸

Recently, loss-of-function variants in ANK2 have been identified in kindreds with early-onset AF, in which family members had frequently progressed to permanent AF.³⁴ ANK2 encodes ankyrin-B, which was previously implicated in the long QT syndrome in a family in which a substantial proportion of members was affected by AF.^35,53 Loss-of-function mutations in ANK2 appear to decrease the expression and trafficking of Ca_V 1.3 channels, resulting in decreased I_Ca,L current.³⁴ Heterozygous ANK2 knockout mice exhibit a lack of discrete P waves and increased susceptibility to pacing-induced atrial arrhythmias, including AF.

Similar to linkage mapping, candidate gene association studies have provided insight into the pathogenic mechanisms of AF. These studies demonstrate that mutations in ion channels and genes hypothesized to be involved in AF are rare.54–56 Furthermore, like linkage analysis, candidate gene association studies have generally identified rare mutations underlying monogenic forms of AF that are private to individual families. The validity and generalizability of more common variants identified in some candidate gene association studies have been called into question owing to the lack of replication of many associations.⁵⁷

Nevertheless, the potential importance of such rare variants was underscored in a recent analysis, in which sequencing for potassium channel mutations in a sample of 80 probands with AF and 240 controls demonstrated an excess number of nonsynonymous variants among those with AF.⁵⁸ Despite their rarity, modeling of the effects of multiple potassium channel mutations suggested that the combination of such variants could result in substantial changes in action potential duration and dispersion of repolarization, potentially establishing a substrate for AF.

Genome-Wide Association Studies

Over the past decade, systematic efforts to examine human genetic variation such as the international HapMap project⁵⁹ have resulted in appreciation of genetic diversity in different ancestral groups. The HapMap project revealed the presence of about 10 million common genetic variants occurring with a frequency of 5% or greater in the general population, most of which were single-nucleotide polymorphisms (SNPs). Such efforts provided a reference against which observed genetic variation could be compared in cohorts of individuals assembled for investigation. High-density chips were developed that allowed the simultaneous genotyping of hundreds of thousands of SNPs across the genome, enabling efficient genetic profiling of individuals.

Genome-wide associations emerged in which genotypes at each of hundreds of thousands to millions of known SNP positions were tested for association with human traits using standard epidemiologic techniques. Unlike linkage analysis, genome-wide association testing does not require multigenerational cohorts for assessment of variant transmission. In contrast to candidate gene association studies, the presence of genome-wide genotyping allowed investigators to efficiently test variants in genes or genetic regions that were not previously suspected of being involved in AF pathogenesis.

Unique study design and interpretation considerations are applicable to genome-wide association studies. First, owing to the massive number of tests performed in each genome-wide association study, stringent significance thresholds (e.g., P < 5 × 10⁻⁸