Improved understanding of the pathophysiology of atrial fibrillation through the lens of discrete pathological pathways

Atrial fibrillation (AF) is a common disorder with a complex and incompletely understood pathophysiology. Genetic approaches to understanding the pathophysiology of AF have led to the identification of several biological pathways important in the pathogenesis of the arrhythmia. These include pathways important for cardiac development, generation and propagation of atrial electrical impulses, and atrial remodeling and fibrosis. While common and rare genetic variants in these pathways are associated with increased susceptibility to AF, they differ substantially among patients with lone versus typical AF. Furthermore, how these pathways converge to a final common clinical phenotype of AF is unclear and might also vary among different patient populations. Here, we review the contemporary knowledge of AF pathogenesis and discuss how derangement in cardiac development, ion channel dysfunction, and promotion of atrial fibrosis may contribute to this common and important clinical disorder.


INTRODUCTION
Atrial fibrillation (AF), the most common sustained heart rhythm disorder in adults, is becoming increasingly prevalent. 1,2 It is projected that 12-16 million people in the US will have AF by 2050. 3,4 The lifetime risk of developing AF for Americans reaching 80 years of age is approximately one in four. 5 The risk of stroke in patients with AF increases five-fold with resultant adverse functional outcome and , 2-fold increase in mortality. 6 More recently AF has been associated with increased risk for incident myocardial infarction. 7 Because of the global health burden of AF, identification of those at increased risk may improve diagnoses and management. Traditional risk factors for AF include male gender, advanced age, Caucasian race, hypertension, diabetes mellitus type II, valvular heart disease, and other forms of underlying heart disease. Recently, some newer risk factors for AF have been identified including the metabolic syndrome, obstructive sleep apnea, and obesity. 8,9,10,11 In addition, multiple studies support the contribution of genetic factors to AF susceptibility. Lone AF, which is AF in the absence of underlying hypertension, valvular, ischemic or underlying structural heart disease, is present in 10 -30% of all AF patients. 12 Given its clinical and genetic heterogeneity, 13 AF is likely to represent a final common phenotype with the convergence of multiple pathways.
Some of the dominant theories on AF pathogenesis include electrical and structural remodeling and inflammation. In addition, common and rare genetic variants modulating cardiac development, altering ion channel function and promoting atrial fibrosis are considered to play an important role in both acquired and lone AF. 14 An improved understanding of the pathophysiology of AF is critical as it is a complex and highly heterogeneous disease. While our current armamentarium of treatments for this disorder has recently expanded, it is still limited by a lack of efficacy, concerns about adverse effects, and, by a rudimentary process for selecting specific strategies for individual patients. The overarching goal of studying AF should be to improve treatment. Improving our understanding of AF by examining different pathophysiological pathways that contribute to the final phenotype holds the promise of a mechanism-based, individualized approach to treatment. Here, we review the substrate, development, and progression of AF, based on different pathophysiological pathways, explore how these pathways converge to promote AF, and highlight how these data can be used to optimize treatment strategies.

CARDIAC DEVELOPMENTAL PATHWAYS
Normal cardiac development during embryogenesis plays an essential role in the development of AF later in life. Recent positional cloning, candidate gene, and genome-wide association studies have identified common and rare variants in genes important in cardiac development. Here, we will discuss these in some detail.

Genome-wide association studies
In 2007, a genome-wide association study (GWAS) identified and replicated two common AF susceptibility SNPs on chromosome 4q25 in Caucasians of European descent. 15 While these SNPs are intragenic, the closest gene is the paired-like homeodomain transcription factor 2 (PITX2) that is critical not only for cardiac formation but also for pulmonary vein development. PITX2 is involved in left-right signaling during embryogenesis; loss of function of PITX2 results in severe cardiac malformations including right cardiac isomerism, double-outlet right ventricle, malformation of the great vessels and paired sino-atrial node formation. 16 In electrically-stimulated adult mice with structurally normal hearts, PITX2 haploinsufficiency has demonstrated predisposition to AF. 17 In atrial biopsy studies, PITX2 expression is severely diminished in AF patients compared to those with no history of AF. 18 PITX2c isoform deficiency in mice can result in right atrial isomerization, loss of suppression of a sinus node development default pathway in the left atrium and loss of the pulmonary myocardium, predisposing to AF. 17,19,20 A recent meta-analysis of AF GWAS identified a novel locus on chromosome 1q24 that associated with AF. 21 The closest gene to this locus, PRRX1, encodes paired-related homeobox gene 1, a transcription factor that is important for normal cardiac development. Defects in PRRX1 and PRRX2 cause abnormal development of the great vessels and pulmonary vasculature. 22,23 Since the pulmonary vein myocardial sleeve is a target during AF ablation, 17,24 modulation of AF pulmonary myocardial development may increase AF risk.

Positional cloning and linkage analysis
In 2004 an autosomal recessive inheritance of AF was reported in a large family from Uruguay. 25 The disease was severe with AF occurring at the fetal stage or during infancy with associated sudden cardiac death and ventricular arrhythmias. Linkage analysis mapped the locus to chromosome 5p13(arAF1) 25 and a homozygous mutation (R391H) was identified subsequently in a nucleoporin gene (NUP155). 26 NUP155 is a major component of the nuclear pore complex (NPC) within the nuclear envelope that facilitates the transport of DNA and mRNA from the nucleus to the cytoplasm. 27 Heterozygous deletions in NUP155 in mice leads to spontaneous AF. 26 In vitro analysis showed that this mutation impaired nuclear permeability to NUP155 gene products. 26 Recently, NUP155 was shown to interact with histone deacetylase 4 (HDAC4) and the possibility that the phenotype may also be related to altered transcriptional activity should be considered. 28 However, the mechanistic link between the NUP155 mutation and AF remains unclear. One postulated mechanism relates to a reduction in nucleocytoplasmic transport due to NUP155 deficiency altering the expression of atrial genes and likely influencing maturation of cardiac ion channel proteins, which may modulate action potential duration (APD) and cause AF. 29 Further studies are required in order to define the underlying mechanisms by which NUP155 mutations cause AF.
Mayo clinic investigators identified a novel mutation in the natriuretic peptide precursor-A (NPPA) gene on chromosome 1p36-p35, which encodes atrial natriuretic peptide (ANP), in a large kindred with early-onset AF and atrial conduction disease in an autosomal dominant pattern of inheritance. 30 Furthermore a Chinese study has reported an association between lone AF and a common NPPA variant. 31 These findings however could not be replicated in Americans of European descent. 32 ANP plays a central physiological role in regulating vascular tone and blood volume and induces diuresis, natriuresis, and vasodilation by activating the intracellular second messenger cGMP. 33 In addition cardiac ion channels (sodium, potassium and calcium) are regulated by ANP through cGMP signaling. 34,35,36 ANP can promote AF either by shortening of the atrial action potential duration (APD) and effective refractory period (ERP), 37 or by autonomically-mediated shortening of the atrial monophasic action potential (MAP) duration and the ERP in dogs. 38 Therefore, it is hypothesized that NPPA mutation can lead to AF. However, the precise mechanism(s) by which NPPA mutations cause AF has not yet been fully delineated and studies in transgenic mouse models will provide further insights.

ION CHANNEL MODULATION PATHWAYS
Cardiac ion channels play an important role in cardiac function. Cardiac disease can alter ion channel trafficking, with adverse consequence to both the electrical and mechanical function of the heart. 39 Studies have identified various mutations in cardiac ion channel genes which may increase susceptibility to AF.

Linkage and candidate-gene studies
In 2003 a mutation in KCNQ1 was identified in a Chinese kindred and linked to familial AF. 40 KCNQ1 encodes for cardiac ion channel subunits involved in conduction of the delayed rectifier potassium current (I ks ) responsible for the terminal phase of the action potential plateau. 41,42 Gain-of-function mutation in KCNQ1 likely results in prolongation of atrial APD and resultant refractoriness causing increased susceptibility to AF. 40 Similar effect is postulated with mutations in KCNE2 which encodes for subunits of I ks conducting ion channel and KCNJ2 which encodes the strong inward rectifier Kþ channel protein Kir2.1. 43,44 A gain-of-function mutation (E299V) in KCNJ2 is responsible for increased susceptibility to atrial reentry and AF. 44 A KCNE5 mutation has been linked with non-familial or acquired forms of AF. KCNE5 is expressed in both atria and ventricles. 45,46 Its gene product expressed in heterologous systems causes suppression of I ks and is hypothesized to compete with the KCNE1 b subunit for the KCNQ1 a subunit resulting in its down regulation 47 ; thus making KCNE5 a possible AF candidate gene. Ravn et al. in 2008 reported AF association with L65F which involves substitution of phenylalanine for leucine at position 65 in the KCNE5. 48 KCNA5 encodes for Kv1.5, a voltage gated potassium channel conducting ultra-rapid delayed rectifier potassium current I kur , which is an atrial-specific repolarizing current that may play a role in AF pathogenesis. 29 Loss-of-function mutations in KCNA5 have been identified in AF families, 49,50 which prolong the APD and trigger premature after depolarization in human atria, predisposing to stress-provoked triggered activity and AF. 29 Our group discovered a rare KCNA5 variant in an AF family with AF occurring at a young age. This variant is predicted to disturb a tyrosine kinase proline-rich motif important in regulating I Kur . This mutation causes the channel to be resistant to kinase with reduction in the wild type current. Thereby, these changes might cause AP shortening for the development of AF. 51 Another potassium channel gene important for atrial and ventricular repolarization is KCNH2 which encodes for the rapid delayed rectifier current I kr . Mutations in the KCNH2 are associated with a higher incidence of AF. 52 Rare non-synonymous variants in the b-subunits of cardiac sodium channels (SCN1B and SCN2B) have been found in AF patients. 53 These mutations occur in the extracellular domain, which regulate the sodium channel gating and cell surface expression. 54 Reduced sodium current resulting from mutations in SCN1B and SCN2B can generate AF -prone substrate through multiple mechanisms such as shortening of refractoriness and slowing of conduction. 53 Furthermore, loss-of-function mutations in the b-subunits can cause a Brugada-type AF. 53 Cardiac L-type calcium channel mutations involving the a1and b 2b-subunits have been linked to the Brugada syndrome and other inherited cardiac channelopathies. 55 CACNA1C gene on chromosome 12p13.3 forms the a1c subunit of the L-type calcium channel (Cav1.2) which is the binding site for all currently available calcium channel blockers 56,57 whereas CACNB2 gene encodes the b-subunit Cav b2. Abnormal intracellular calcium handling may play a crucial role in AF initiation and continuation. 58,59 A higher rate of spontaneous sarcoplasmic reticulum calcium release is found in atrial myocytes of patients with chronic or paroxysmal AF, which is considered pro-arrhythmogenic. 60,61 About 20% to 30% cases of Brugada syndrome and short-QT syndrome are known to have associated AF. 62,63 Genome-wide association studies Multiple AF related loci affecting ion channels have been identified in recent GWASs. Some of the important associations include PITX2, PRRX1, SCN5A, SCN10A, KCNN3 and CAV-1. 21 Both PITX2 and PRRX1 play a key role in pulmonary myocardial development. 19,23 Although the precise mechanism(s) by which these AF loci increase AF susceptibility is unclear, an up-regulation of KCNQ1 results from PITX2 suppression in transgenic mice, in addition to differential expression and distribution of the inward-rectifier potassium I k1 current. 18,64,65,102 GWAS in individuals of European descent have identified a genomic region related to AF on chromosome 1q21 (KCNN3). 66 KCNN3 gene encodes for small conductance voltage-independent calcium-activated potassium channels that are expressed in various excitable tissues including the brain 67 and the heart. 68,69 KCNN3 channels are distributed in the atria and the ventricles and have a prominent role during the late phase of cardiac action potential. 68,69 Other members of the family expressed predominantly in the atria are KCNN1 and KCNN2. Li et al. observed APD prolongation and an increased number of early after depolarization in the atrial myocytes of KCNN2 2/2 knockout mice. Atrial arrhythmias, mainly pacing-induced AF, were more frequent in both heterozygous and homozygous KCNN2 mice. 70 Variation in either KCNN2 or KCNN3 may modulate ion-channel function, as in vitro both subunits exhibit co-assembly in channel complexes. 71 Caveolin-1 gene (CAV1) on chromosome 7q31 encodes a signal transduction cell-membrane protein expressed selectively in the atria. 72 The CAV1 protein by co-localizing, negatively regulates the activity of KCNH2 protein, a potassium channel involved in cardiac repolarization that has been linked to AF. 52 Dilated cardiomyopathy has been reported in CAV1 knockout mice. 73 Mutations in HCN4, on chromosome 15q24, have shown association with dysfunction of the sinus node. HCN4 encodes the potassium-sodium hyperpolarization-activated cyclic nucleotide-gated channel which constitutes the predominant sino-atrial node pacemaker channel. 74,75 SCN5A and SCN10A cardiac sodium channel related genes are associated with AF. The voltage-gated sodium channel Nav1.5, involved in conduction of the inward sodium current I Na responsible for initiating cardiac AP, is encoded for by SCN5A. 76,77 Screening for SCN5A variants in a large AF cohort found mutations in 5.9% of those with AF. 78 AF has been associated with loss-of-function mutations in SCN5A, in addition to dilated cardiomyopathy, sinus node dysfunction, and/or conduction disorder. 76 The H558R polymorphism in SCN5A has also been linked to incidence of AF. 79 SCN10A encodes the Nav1.8 channel and has been associated with PR and QRS duration variability, atrioventricular and intraventricular conduction velocity. 80,81 The Nav1.8 channel is known to be expressed in the dorsal root ganglia 82,83 and the retina. 79 The functional characterization of Nav1.8 current in heterologous expression experiments indicate that the current activates much more slowly than does Nav1.5 current, and this difference may in turn contribute to variable cell-to-cell charge transfer that underlies fast conduction in heart. 84

INFLAMMATION AND ATRIAL FIBROSIS PATHWAYS
In 1997, Frustaci et al. reported myocarditis like histological changes in the atria of lone AF patients, supporting involvement of inflammation in the pathogenesis of AF. 85 The high incidence of AF after cardiac surgery and the role of anti-inflammatory drugs in preventing AF support the association of inflammation with AF. 86,87 Higher baseline levels of C-reactive protein (CRP), an acute-phase protein, have been linked to future development of AF, and patients with AF have been shown to have elevated levels of CRP. 88,89 It has also been postulated that an elevated CRP level may be an outcome of the underlying cardiac pathology rather than a marker of AF itself. 90 In conditions such as oxidative stress and ischemia, which promote apoptosis and energy depletion, CRP may initiate AF by disrupting the normal structure of cell membrane. 91 Atrial fibrosis may play an important role in both initiation and maintenance of AF. 92,93 Induced atrial fibrosis in animal studies has shown a greater tendency for developing AF. 94,95,96 Atrial tissue specimens from patients with AF have been shown to have increased fibrosis. 97,98 Fibrosis may induce AF by premature or burst atrial pacing that in a normal heart will not lead to AF. 95,99 Severity of left atrial (LA) fibrosis correlates with favorable treatment outcomes in AF patients. 100,101 There is substantial evidence suggesting the potential of multiple common AF-susceptibility variants to modulate atrial fibrosis. In addition, all these variants are likely to mediate their effect not only by regulating atrial conduction slowing, but also by modulating electrical remodeling that promote AF, such as shortening of the ERP.
Genome-wide association studies PITX2 encodes a transcription factor critical for left-right patterning in the atria, suppression of sino-atrial node formation on the left, and development of the pulmonary myocardium during embryogenesis. Given its critical role in cardiac development, its dysfunction may be predicted to cause atrial structural remodeling. In addition, it has been shown that PITX2 is expressed in adult human hearts, especially in the left atrium, suggesting that this transcription factor plays a physiological role beyond cardiac development. 102 Chinchilla et al. found that PITX2c 2/2 mutant mice had enlarged cardiac chambers, increased fibrosis seen on histological staining of the ventricles, and increased expression of collagen precursor genes in the atria. 18 In a separate study, PITX2c þ/2 mice were found to have structurally normal hearts, however differential expression of genes involved in Wnt signaling (a key fibrosis signaling pathway) was demonstrated in microarray analysis. 102 Another locus associated with AF in a recent meta-analysis of AF GWAS was at 7q31 in CAV1. This gene encodes a cellular membrane protein caveolin-1, which is expressed in the atria. CAV1 is associated with myocardial fibrosis, and its variants have been associated with pulmonary hypertension. 103 CAV1 knockout mice display dilated cardiomyopathy. 73 Caveolin-1 associated lipid rafts internalize TGF-b receptors resulting in inhibition of TGF-b signaling. 104 While the role of several other variants associated with AF in GWAS is unknown, a growing body of evidence implicates atrial fibrosis as a plausible link between these genes and AF. The transcription factor zinc finger homeobox 3 (encoded by ZFHX3) is a tumor-suppressor gene that by associating with runt-related transcription factor 3 (RUNX3) translocates in response to an important fibrosis mediator, TGF-b. 105,106 SYNPO2L knock-down in zebra fish resulted in abnormal skeletal and cardiac muscle development and function. The gene encodes an actin-binding protein which is a component of the cytoskeleton and is expressed in the heart. 107 Nesprin-2, encoded by SYNE2, in combination with nesprin-1, creates a network that links the actin cytoskeleton and nuclear membrane structures to the nucleoskeleton in muscle. Alpha-catenin's interaction with emerin and nesprin-2 regulates the Wnt signaling-dependent transcription, which is a fibrosis related pathway in the heart, lung and kidney. 108,109 The homeodomain transcription factor encoded by PRRX1 is expressed in the developing heart, especially the connective tissue. It is involved in normal pulmonary development and is linked to fibrosis in the lungs and liver. 110,111 A higher expression of C9ORF3 is shown in TGF-b stimulated normal and scleroderma fibroblasts. 112 Linkage analyses A possible pathophysiological mechanism of NPPA variants in AF may involve exposure to higher levels of mutant ANP causing structural changes promoting atrial fibrosis. 30 In fact, a missense mutation in NPPA was recently identified in four families in Northeast Italy with the clinical phenotype of atrial myopathy, massive biatrial enlargement, and atrial standstill. 113

INTERTWINING OF PATHWAYS
Emerging evidence suggests that development of AF might be predicated on the interaction between clinical and genetic factors. In this proposed 'two hit' or 'multiple hit' paradigm [ Figure 1], deleterious changes in confluent pathways might predispose patients to develop this complex clinical disorder. We previously studied 11 AF families in which we found 42 mutations in AF candidate genes. When affected individuals were also genotyped for 4q25 SNPs, a striking interaction between candidate gene mutations and 4q25 variants emerged. Zero out of 13 patients who had the wild-type 4q25 allele developed AF prior to the age of 50, whereas 21 of 29 who had 4q25 variant alleles developed AF before age 50 (P ¼ 1.12 £ 10 25 ). 114 A frameshift mutation in NPPA was previously linked to familial AF. Functional characterization revealed that the mutant ANP was associated with shortened APD and ERP, suggesting that ion channel modulation was the pathophysiological link with AF. In addition, several studies have implicated NPPA mutations in cardiac structural changes. NPPA knock-out mice were found to develop left ventricular hypertrophy (LVH), even when fed a normal diet. 115 A separate mouse study found that NPPA knock-out mice developed LVH and had a several-fold increase in heart interstitial collagen volume compared with wild-type littermates. 116 In humans, as discussed above, a missense mutation in NPPA was linked to a familial syndrome of atrial standstill and extreme biatrial enlargement, a striking example of an atrial myopathy with relative sparing of the ventricles. 113 Thus, genetic variants in NPPA may affect all three pathways that can potentially lead to the clinical phenotype of AF: ion channel modulation as evidenced by changes in APD and ERP, cardiac development as NPPA is part of the signaling pathway for PITX2 and atrial fibrosis [ Figure 2a].
Variants in CAV1 were associated with AF in GWAS. The gene encodes a membrane protein thought to play a role in cardiac structural changes and fibrosis. A frameshift mutation in CAV1 was found in affected members of a large family with primary pulmonary hypertension. 103 Furthermore, CAV1 knockout mice display dilated cardiomyopathy. 73 However, CAV1 is also thought to play a role in cardiac electrophysiology. CAV1 SNPs have been associated with longer PR interval (a known AF risk factor) in GWAS. Interestingly, CAV1 was shown to co-localize with KCNH2, the channel encoded by HERG (also called KCNH2) which is responsible for the I kr current. A CAV1 knock-down model resulted in decreased I kr current, whereas overexpression of CAV1 resulted in increased current. 117 KCNH2 is itself a gene associated with AF in GWAS. 21 Therefore, CAV1 seems to play a dual role in the pathophysiology of AF, by modulating cardiac fibrosis and ion channel function.
PITX2, a homeobox transcription factor, has been associated with AF 15 and has a complex role in cardiac development most importantly left-right signaling during embryogenesis. 16 Experiments in mice models with PITX2c deletions have revealed atrial enlargement, increased atrial expression of collagen precursor genes and increased fibrosis. 18,102 Electrophysiological findings in similar mice models with PITX2c deletions included complete AV block, AF, a more depolarized atrial resting membrane potential, and a lower amplitude atrial action potential. In addition decreased atrial expression of SCN5A, SCN1B, KCNJ2, KCNJ4, and KCNJ12, was observed as compared to control mice. 102 Thus, PITX2 seemingly plays a vital role in AF pathophysiology with contributions to cardiac development, electrical remodeling, and atrial fibrosis [ Figure 2b].

IMPLICATIONS FOR TREATMENT OF AF
Recent advances in catheter ablation, hybrid surgical and catheter ablation, and new oral anticoagulants have improved treatment options for patients with AF. However, there is much room for improvement especially in selecting individualized treatment strategies for patients that will maximize efficacy and minimize risks. For example, an evidence-based strategy to select patients for treatment with rate control therapy, AADs, and invasive treatment does not exist and is sorely needed. Improved understanding of the genetic mechanisms that underlie AF in individual patients should lead to personalized strategies to maximize efficacy and minimize adverse events. The I kr blockers sotalol and dofetilide maintain sinus rhythm by prolonging repolarization and increasing refractoriness, thus abolishing the excitable gap necessary for arrhythmia propagation. These agents may be particularly effective in maintaining sinus rhythm in AF patients who harbor potassium channel gene variants. Conversely, sodium channel-blockers such as flecainide and sotalol, which abolish reentry by slowing conduction, might cause ventricular pro-arrhythmias in AF patients with loss-of-function sodium channel variants. These important considerations have not been tested in human pharmacogenomic studies.
Many patients with AF are treated with AV nodal blockers to prevent tachycardia and its associated complications. Mounting recent evidence from pharmacogenomic studies suggests that these agents, particularly beta-blockers, are less effective in patients with variants in the beta-adrenergic signaling pathway. Several studies have shown that common SNPs in the beta-adrenergic receptor genes ADRB1 and ADRA2C modulate the response to beta-blocker therapy in HTN 118,119,120 and HF. 120,121,122 We tested whether the polymorphisms in the ADRB1 gene are associated with response to rate control therapy in 543 patients in the Vanderbilt AF registry. The Arg389Gly variant showed association with adequate rate control (adjusted OR 1.42) as defined by the AFFIRM criteria. 123 Several recent studies highlight the challenges of pharmacological and catheter-based rhythm control strategies in AF patients with 4q25 (PITX2) variants. Our group showed that AF patients with the wild-type allele at the 4q25 SNP rs10033464 were significantly more likely to respond to Class I and III AADs as compared with carriers of the variant allele (OR 4.7, P ¼ 0.001). 123 Husser et al. demonstrated that carriage of 4q25 variants was associated with post-catheter ablation recurrence of AF. 124 In the Vanderbilt AF registry, 4q25 variants were associated with unfavorable outcomes after catheter ablation, with shorter time to recurrence of atrial arrhythmias (survival time ratio 0.76) that was especially evident for patients with left atrial diameter . 5 cm. 125 A similar outcome was shown after patients underwent DC-cardioversion. 126 Collectively, these findings highlight the importance of the 4q25 variants in creating an atrial pro-arrhythmic substrate due to deleterious electrical and structural remodeling, with subsequently difficult-to-treat sub-type of AF.

CONCLUSIONS
Data from linkage analyses in AF families, candidate gene and GWA approaches and animal models, have provided important insights into the underlying pathophysiology of this complex and heterogeneous disorder. Despite these advances however, AF persists as a common and the most challenging arrhythmia to treat in clinical practice. Here, we proposed viewing AF through the lens of biological pathways important in the pathogenesis of the arrhythmia. This perspective will not only improve our understanding of disease onset and progression, but inform us on how to best individualize therapy to maximize efficacy and minimize adverse effects.

COMPETING INTEREST
None of the authors have any competing interests to declare.