Introduction
Much progress has been made in identifying genetic loci linked to hereditary arrhythmia syndromes over the past decade and a half. Linkage analyses for Mendelian diseases have been powerful in the discovery phases. Considerable challenges remain however, for the clinician faced with individual patients and families when the clinical symptoms are atypical or intermediate and when novel mutations or polymorphisms are reported in the course of genetic testing. To unambiguously define the deleterious nature of any given mutation, additional functional analyses are required. Such studies should not only detect the functional consequence of mutations but also the degree of severity and mechanisms that bring about the deleterious behavior. These principles apply not only to cardiac arrhythmia syndromes but also to any hereditary genetic disease. In practice, this is not always feasible or possible with current technology. This is particularly problematic when standard genetically manipulable animals (mouse) differ considerably from human, as they do in cardiac electrophysiology. An additional obstacle occurs when the target organ is not amenable to biopsy without considerable risk (e.g. heart, brain, etc.). For evaluation of genetic mutations in cardiac arrhythmia syndromes, heterologous expression of affected genes has helped tremendously.
Hereditary arrhythmia syndromes include: the long QT syndrome, the Brugada syndrome, catecholaminergic polymorphic ventricular tachycardia, the short QT syndrome, and arrhythmogenic right ventricular dysplasia. We will restrict our discussion to the long QT syndrome; however, the basic principles of verifying functional consequences of mutations also applies to the other syndromes. In this topic we will review the progress in characterizing arrhythmia-linked genetic mutations. Several areas of recent technical advancement have been achieved which we will discuss in detail. We will also highlight how biophysical, biochemical and cell-biological studies may be used to help inform clinicians in managing the more subtle and varied aspects of patients with specific mutations. Lastly, we will discuss how such studies may eventually point to therapeutic modalities that will lead to gene-specific, or personalized medicine.
Overview of the long QT syndrome
Congenital long QT syndrome (LQTS) was first described by Jervell and Lange-Nielsen in 1957, who presented a family in which four of six children were born deaf, had episodes of syncope, prolonged QT interval and early sudden death (Jervell & Lange-Nielsen, 1957). In 1963 and 1964, independent reports of a similar constellation of findings in patients, but without hearing loss, were made (Romano et al., 1963; Ward, 1964). Subsequently, these were classified as autosomal-recessive (Jervell-Lange-Nielsen syndrome, with hearing loss) and autosomal dominant (Romano-Ward syndrome) forms of LQTS. The incidence of hereditary LQTS has been estimated to be as high as 1 in 2500 (Crotti et al., 2008). Notably, disease severity varies widely — from patients who are mostly asymptomatic, to ones who suffer multiple episodes of syncope and/or sudden cardiac death at a young age. The common pathophysiological feature of LQTS is delayed repolarization, manifest on electrocardiogram (ECG) as a prolonged QT interval corresponding to a prolonged action potential duration (APD). Delayed repolarization occurs either due to an excess of sodium (Na+) or calcium (Ca2+) influx, or to deficient potassium (K+) efflux. This disruption in the normal ionic currents across the cell membrane undermines the highly regulated electrical activity in the heart required for normal, rhythmic beating, and leaves patients at risk for potentially lethal arrhythmias. Abnormal currents can result from congenital mutations in the ion channels, or from pharmacological agents and acquired disease that can alter cardiac ion channel function. When a ventricular myocyte action potential is prolonged, abnormal depolarizations may develop, known as early afterdepolarizations (EAD) that occur during the plateau or repolarization phases of the action potential (i.e. a type of depolarization that occurs before an action potential has completed repolarization). An EAD can then trigger an action potential that is self-perpetuating, leading to a particularly deadly type of arrhythmia known as polymorphic ventricular tachycardia or "torsade de pointes" which may degenerate into ventricular fibrillation. Furthermore, intracardiac imbalances of ion currents may lead to disperson of refractoriness that may play a role in susceptibility to micro-reentry. Symptoms include syncope (fainting), palpitations and sudden cardiac death.
Linkage studies
The hereditary long QT syndrome (LQTS) is now recognized as a genetically heterogeneous disorder with at least 13 different proposed loci (Table 1). Most of the loci contain genes of cardiac ion channels, accessory subunits, or channel-associated scaffolding proteins. The approaches taken by researchers in the 1990s to initially characterize hereditary LQTS relied on classical genetics with pedigree analysis of large families using microsatellite markers and logarithm of odds (LOD) score calculation. LOD scores indicate the likelihood of linkage of two loci by comparing the calculated recombination frequency against chance. A positive LOD score signifies linkage, whereas a negative score signifies the absence of linkage. The major goal of the early studies was to connect symptomatic LQTS patients with a common genetic feature. Originally, LQTS was thought to be a single-gene disorder linked to chromosome 11 (Keating et al., 1991a; Keating et al., 1991b). Subsequent refinement revealed that heterogeneity and multiple loci were involved (Worley et al., 1992; Benhorin et al., 1993). In 1994, analysis of multiple LQTS families using LOD scores showed that some had linkage to chromosome 7, others linked to chromosome 3, and both excluded chromosome 11 linkage (Jiang et al., 1994). Other of the families in the study did not show linkage to any of the three known loci, suggesting the existence of additional loci. Once the first three LQTS loci were identified, several groups worked to identify the genes responsible for the phenotypes. In 1995, Wang et al. used linkage analysis to show that locus LQT3 contained SCN5A, a Na+ channel that was previously cloned and characterized in 1992 (Gellens et al., 1992; Wang et al., 1995a; Wang et al., 1995b).
|
Locus |
Gene |
Protein Function |
Chromosome |
Other Diseases |
|
LQT1 |
KCNQ1 |
KvLQTl K+ channel a subunit |
11p15.5 |
Short QT Syndrome (SQTS1) Familial Atrial Fibrillation (FAF) |
|
LQT2 |
KCNH2 |
HERG K+ channel a subunit |
7q35-q36 |
SQT1 |
|
LQT3 |
SCN5A |
Na+ Channel a subunit |
3p21 |
Brugada Syndrome (BrS1) Conduction & Sinus node disease |
|
LQT4 |
ANK2 |
Ankyrin B adaptor protein |
4q25-q27 |
LQTS |
|
LQT5 |
KCNE1 |
minK P subunit |
21q22.1-2 |
Atrial Fibrillation, Deafness |
|
LQT6 |
KCNE2 |
MiRP1 P subunit |
21q22.1-22.2 |
Hypothyroidism, Periodic paralysis |
|
LQT7 |
KCNJ2 |
Kir2.1 K+ Channel |
17q23.1-q24.2 |
Andersen’s Syndrome, myotonia |
|
LQT8 |
CACNAlc |
Ca2+ channel a subunit |
12p13.3 |
Timothy Syndrome, BrS3 |
|
LQT9 |
CAV3 |
Caveolin 3 membrane scaffold |
3p25 |
Cardiomyopathy |
|
LQT10 |
SCN4B |
Na+ Channel P subunit |
11q23 |
Conduction Disease |
|
LQT11 |
AKAP9 |
Yotiao PKA scaffold |
7q21-q22 |
LQTS |
|
LQT12 |
SNTA1 |
Syntrophin a 1 scaffold protein |
20q11.2 |
LQTS |
|
LQT13 |
KCNJ5 |
Kir3.4 K+ Channel |
11q24 |
Neonatal hyper-insulinemia |
Table 1. The Hereditary Long QT Syndrome Loci
In 1995 Curran et al. analyzed LQTS families using markers linked to locus LQT2 on chromosome 7q35-36 (Curran et al., 1995). Physical mapping using yeast artificial chromosomes (YACs) and fluorescent in situ hybridization (FISH) indicated that a candidate gene with homology to potassium ion channels (K+ channels) resided in that position. This gene had been previously identified as the human ether-a-go-go related gene (HERG or KCNH2) (Warmke & Ganetzky, 1994). Patient sample analysis for mutations in HERG with single-strand conformation polymorphisms (SSCP) detected the presence of genetic variants and functional expression of the cDNA in Xenopus oocytes showed that HERG encoded a channel that carried the rapidly activating delayed rectifier K+ current (IKr) and confirmed the deleterious nature of the mutations. Further positional cloning showed that LQT1 on chromosome 11 encoded KvLQT1/KCNQ1, another K+ channel (Wang et al., 1996). Concurrently, Schott et al. used similar linkage techniques to map the LQT4 locus to chromosome 4q25-27 (Schott et al., 1995). The gene responsible for LQT4 was identified in 2003 by the Mohler group as ankyrin-B, a scaffolding protein which when mutated causes aberrant targeting of essential cardiac channel proteins (Mohler et al., 2004).
Identification of other LQTS loci was done through a variety of techniques ranging from classical genetics to modern genomic methods (Chevillard et al., 1993; Duggal et al., 1998; Abbott et al., 1999; Fodstad et al., 2004; Vatta et al., 2006; Ueda et al., 2008). The loci include other channel proteins such as the Kir2.1 channel encoded by KCNJ2 and the voltage-gated Cav1.2 calcium channel encoded by CACNA1c, K+ channel accessory subunits (KCNE1 and KCNE2), as well as scaffolding proteins such as AKAP9 and syntrophin. While the genes are numerous and diverse, the overall themes of cardiac ion channel function/dysfunction and alterations in regulation unify the genetic causes of LQTS.
Heterologous expression of arrhythmia-linked genes
The ideal system for studying behavior of cardiac ion channels would be isolated cardiac myocytes that survive in culture for a long time period. Such primary cells however, entail significant risk to patients and are extremely difficult to maintain long term in culture. The next option is to express the channel proteins in a cell type that can be maintained and manipulated as necessary.
Xenopus oocytes
Xenopus laevis oocytes are an established system for studying ion channels using electrophysiological techniques. The procedure consists of creating cRNAs of the gene of interest followed by injection into oocytes, which contain all the necessary cellular machinery for protein expression (Gurdon et al., 1971; Barnard et al., 1982). Two-electrode voltage clamp is a relatively easy method to use with oocytes given their large size and provides a rapid way to functionally characterize many of the genes involved in LQTS, and many of the first studies utilized this method. Xenopus oocytes however, contain an endogenous K+ channel similar to KCNQ1, thus confounding some of the early studies on KCNE1 and KCNQ1. Moreover, oocytes are maintained at 16-19°C, a temperature that may permit mutant proteins to properly fold and traffic to the cell surface thereby masking a misfolding phenotype that would normally occur at human physiological temperatures. Such an occurrence was noted in the initial analysis of the cystic fibrosis transmembrane conductance regulator protein (CFTR)(Cheng et al., 1990; Denning et al., 1992).
Mammalian cultured cell systems
Another approach is to use immortalized mammalian cell lines such as human embryonic kidney (HEK 293), Chinese hamster ovary (CHO), or COS-7 cells. Unlike primary cell lines, immortalized cell lines can be propagated many times and maintain baseline characteristics. The cells are incubated at 37°C and contain all the necessary components for protein transcription, translation, trafficking and degradation. They are more amenable than oocytes for immunoblotting, immuno-precipitation, high-resolution immuno-fluorescence, trafficking assays, cell-surface expression assays, and patch clamp electrophysiology. The cells may also have endogenous K+ current; however, the magnitude is small and does not usually interfere with measurements of over-expressed channels currents. Mammalian cell lines more closely mimic native systems than oocytes and are useful for analyzing biological consequences of LQTS mutations. A caveat to this system is that the LQT-linked channels may exist in macromolecular complexes in vivo. Such complexes may comprise accessory subunits and regulatory proteins, which may not be recapitulated by heterologous expression system.
Purified proteins for biochemistry and structural analysis
Functional expression in oocytes and cells allows the study of many aspects of mutations but the fundamental mechanism of mutational effects ultimately relies on structural analysis. The primary challenge is finding conditions under which a large quantity of protein can be expressed and purified. This process can be relatively straightforward for soluble, cytosolic proteins, but is more difficult for membrane proteins such as ion channels. Because of the large amount of protein needed for purification, transfection of mammalian cells, or even the use of stably transfected mammalian cells lines, may not be feasible. Alternative systems of expression have been developed for bacteria, yeast, and insect cells, but determining the best host for producing a particular protein is usually an empirical process.
One of the most commonly used expression systems is the bacteria Escherichia coli. There are several technical and economic advantages: the ease of introducing DNA via transformation with a plasmid expression vector, rapid growth, and simple growth media. Problems do exist, though, in expressing mammalian membrane proteins in bacteria. These include alternative translation, posttranslational modification and trafficking mechanisms. Certain limitations can be overcome by changing growth conditions, co-expressing necessary chaperones, or creating fusions with prokaryotic partners such as maltose-binding protein (MBP) or glutathione S-transferase (GST) to improve their solubility and stability. Even with these modifications, it can still be difficult to express the full length of a protein (for example, the full length of KCNQ1 is nearly 700 amino acids, and Nav1.5 is around 2000 amino acids). An alternative expression host is the yeast Pichia pastoris, which has a eukaryotic protein synthesis pathway and is capable of post-translational modifications, though it is not entirely equivalent to a mammalian system. The first (and thus far, only) mammalian K+ channel to be crystallized was expressed in P. pastoris, whereas several previously crystallized bacterial channels were expressed in E. coli (Doyle et al., 1998; Jiang et al., 2002; Jiang et al., 2003; Long et al., 2005). A higher eukaryotic system that may be used is insect cells, with baculovirus as the vector for protein expression. Insect cells are even better equipped with the machinery needed for proper protein folding and for post-translational modifications. While they provide high expression levels and can be grown to high density, a disadvantage is that the growth media is more expensive than for bacteria or yeast.
If a system for high-yield expression and purification of a protein can be achieved, the protein can then be used in a multitude of biochemical and structural experiments. The highest resolution is crystal structure; however, this is a difficult and time consuming task. The difficulty of this task is evident in the small fraction of membrane proteins that have been crystallized, compared to soluble proteins. An alternative that has been used successfully is solution nuclear magnetic resonance (NMR) structure. Besides the obvious advantage of not needing crystals, NMR may yield structure that is closer to native form, since formation of crystals may impose non-native constraints on the protein.
Animal models of inherited human arrhythmias
Ideally, it is desirable to create an animal model of a disease — acquired or hereditary — in order to study pathophysiological mechanisms, and to design and test therapeutic options. To accomplish this it is important that the model recapitulate the human condition as closely as possible. For hereditary diseases it is necessary that the animal be genetically manipulable and that homologues of the genes of interest exist and be expressed in the same tissues as humans. Here we will discuss animal models that have been proposed and used for LQTS.
Rodent
Mice and rats are valuable systems for modeling a variety of human diseases, especially in terms of organ system pathophysiology and immune diseases. Since they can be easily genetically manipulated, they are good surrogates and provide clues to the hierarchy of genetic pathways and regulation that occur in the healthy and disease states. Once the genetic loci were identified, investigators created knock-out (null) and knock-in mice to model the LQTS phenotype. The knock-out is done by creating an exogenous construct based on the sequence of the mouse gene, but where the relevant allele has been inactivated or nullified by inserting a stop codon or deletion/insertion to inhibit expression of the native protein. This construct is then injected into mouse embryos where homologous recombination occurs and the endogenous mouse gene is replaced by the null construct, which contains a marker so that recombinant mice may be distinguished. The recombinant mouse must then be bred to create heterozygous and homozygous null mice in subsequent generations. A knock-in mouse is created by a similar method, where the construct is a human gene (or mouse ortholog) that contains a known functional mutation. The engineered heterozygous mice will express one copy of endogenous mouse gene and one of the transgenic mutated gene.
There are at least 40 mouse models of LQTS genes. Two mouse models were created that disrupted exons 1 and 2 in KCNQ1 (Lee et al., 2000; Casimiro et al., 2001). Interestingly, the mouse with mutation in exon 1 did not show any ECG abnormalities. However, this mouse did have auditory-vestibular aspects of the Jervell and Lange-Nielsen syndrome. The mouse with KCNQ1 exon 2 disruption showed abnormal T-wave morphology on in vivo ECGs and inner ear abnormalities. In a third study, mice were created that expressed a dominant negative isoform of KCNQ1; these mice had QT prolongation on ECG as well as torsade de pointes arrhythmias (Demolombe et al., 2001). For KCNE1 null mice, the models exhibit deafness, but no baseline QT prolongation (Schulze-Bahr et al., 1997). One KCNE1 null mouse showed abnormal rate adaptation, which is similar to the phenotype seen in humans with KCNQ1/KCNE1 mutations upon exercise challenge (Charpentier et al., 1998; Warth & Barhanin, 2002). Mice express the HERG ortholog Merg1 in the heart. The Merg1 homozygous knock-out mouse is embryonic lethal as it dies early in development (London, 1998). A mouse model that expresses the dominant-negative HERG-G628S mutation showed a normal ECG phenotype (Babij et al., 1998).
While these studies yielded valuable information about pathogenesis of LQTS, they also highlighted how mice have limited value in studying inherited cardiac arrhythmias resulting from mutations in delayed rectifier K+ current channels. Mice have a baseline heart rate of ~600 beats per minute. As such, they have a short action potential and repolarization phase that is largely dependent on the transient outward K+ current (Ito) and have little to no IKs or IKr (Nerbonne, 2004; Milan & MacRae, 2005). So while they are genetically tractable, they may not be electrophysiologically similar enough to humans to provide a good model system. In contrast to the limitations of modeling human repolarization in the mouse, more success has been achieved for the depolarizing currents, which are more akin to those in the human. A LQTS mouse model generated by knock-in of an LQT3 mutation (KPQ deletion in SCN5A) (Nuyens et al., 2001). The transgenic mice had prolonged APD and polymorphic ventricular tachycardia.
Early studies of guinea pig ventricular myocytes revealed that two components making up the repolarization current IKr and IKs (Sanguinetti, 1990). This work was the original characterization of two repolarizing K+ currents and forms the basis for many of the subsequent studies. Considering that isolated guinea pig ventricular myocytes was the in vitro system that launched a whole field of study, some groups have used an interesting approach by injecting adenoviral vectors containing wild-type or mutant KCNE1 or HERG into guinea pig myocardium (Hoppe et al., 2001). This group found that myocytes expressing the HERG G628S mutant, IKr was reduced, but action potential duration was not shortened however, beat-to-beat variability increased as did EADs. They also expressed the KCNE1-D76N mutant which suppressed IKs, significantly slowing repolarization, leading to frequent EADs and QT prolongation on ECG.
Rabbit
Given the limitations of rodent models, larger animals with cardiac electrophysiology more similar to humans might be considered. These included study of dogs, ferrets and rabbits. The Koren group has developed transgenic rabbits expressing human LQT mutations (Brunner et al., 2008). To create the transgenic rabbits the investigators injected embryos with a cDNA construct that contained either mutant HERG or KCNQ1 under a cardiac specific promoter, so that the transgene will only be active in the heart. These animals have enabled the investigators to gain significant insights by ECG analysis in awake freely moving animals, optical mapping of repolarization waves using voltage-sensitive dyes, and at the cellular level by recording from isolated rabbit myocytes. To date, this may be the most accurate model system that exists for hereditary LQTS.
Zebrafish
The newest model system to be explored is the zebrafish, Danio rerio. These are genetically tractable animals that expresses an endogenous ortholog of HERG (zERG) (Langheinrich et al., 2003). zERG is expressed specifically in both heart chambers of zebrafish embryos, is similarly composed of six transmembrane domains, and displays a particularly high degree of amino acid conservation in the S6 helix and pore domain. One specific mutant that was characterized named breakdance displayed prolonged ventricular APD, spontaneous EADs, and 2:1 atrio-ventricular block in the embryonic stages of development. The group of Scholz et al. expressed cloned zERG in Xenopus oocytes and showed current characteristics similar to the human channel however the details of its kinetics and gating were distinctly different (Scholz et al., 2009). Arnaout et al. recently performed a forward genetic screen and identified two zebrafish HERG mutants s213 and s290. They showed that homozygous animals had virtually no ventricular contraction and impaired calcium handling in the ventricles. Heterozygous animals showed increased APD and prolonged QT-interval on ECG (Arnaout et al., 2007). These studies show that given the conserved channel function, zebrafish does represent a valuable genetic model system to investigate HERG channel mutations.
Primary isolated myocytes
To find a more native system to study ion channels, researchers have sought methods to isolate and maintain primary cardiac myocytes. Primary isolated myocytes are best suited for short-term culture (approximately four days) and electrophysiological or immunofluorescence experiments that require only a low yield of viable cells (10s compared to 10,000s needed for biochemistry experiments) (Nuss & Marban, 1994). Some of the technical challenges involved include obtaining fresh healthy heart samples, appropriate and not over-digestion of the tissue by enzymes, purification of myocytes from fibroblasts and matrix, calcium tolerance of the freshly isolated myocytes, and finding the correct conditions for culture. Most adult myocytes have been isolated from mouse, rat, guinea pig, and rabbit since the animals are readily available and economical. Fresh human heart samples for cardiomyocyte isolation are difficult to obtain routinely for ethical reasons. Rat neonatal cardiomyocytes have provided a fairly easy-to-obtain and widely applicable system in recent years (Chlopcikova et al., 2001). Since the rat neonatal cardiomyocytes may only transiently express the relevant channel, another approach is to use adenoviral or lentiviral vectors containing the cDNA of interest to infect the cells and allow adequate expression for study in a more native system. Comparing the behavior of wild type HERG and KCNQ1 channels with previously characterized deleterious mutants in rat neonatal myocytes has confirmed initial phenotypic characterization (Li et al., 2001; Lin et al., 2010). These groups found that the wild-type and mutant channel behaved generally the same as in cultured cells with some slight differences. Additionally some groups used the neonatal cardiomyocyte system to understand localization and interaction of the HERG, KCNQ1 and P accessory subunits (Rasmussen et al., 2004; Wu et al., 2006).
