Phenotypic heterogeneity is the result of variations originating from genetic and environmental factors, as well as stochastic biomolecular events (1
). Before the sequencing of the human genome, it was evident that mutations in genes could be related to diseases (3
). Since the sequencing of the human genome and advances in high-throughput techniques, it is now clear that mutated gene products associated with disease phenotypes interact with other proteins to alter regulatory network behavior (5
). Although compensatory mechanisms often allow these networks to remain robust to changes in a single component, mutations or gene variants, such as single-nucleotide polymorphisms (SNPs) or copy number variants that sufficiently alter the function of cellular components beyond a threshold, result in disease (8
). Silent variations and mutations that do not lead to phenotypic changes can become unmasked through interactions of the organism with the environment (11
). Networks that identify relationships among gene products that are responsible for phenotypic behavior can provide insight into the interaction between genes and the environment.
The construction of disease-centered networks of cellular interactions may also help explain the origins of variable responses to therapeutic or adverse effects of drugs. Drugs can be considered “environmental signals” because their targets often serve to link signaling networks to cellular machines and are responsible for the phenotypic changes (12
). If the adverse response to a drug produces a phenotype similar to that of an inherited disease, it is plausible that this drug acts on the same molecular pathways that are altered in the inherited disease. This line of reasoning leads to the hypothesis that identification of networks related to a clearly observable phenotype could be useful for understanding drug responses.
Long QT syndrome (LQTS) is a congenital or drug-induced change in electrical activity of the heart that can lead to fatal arrhythmias. LQTS is defined by a specific change (lengthening of the QT interval) in the electrocardiogram (ECG), and is thus a readily observable phenotype. Therefore, we analyzed the relationship between mutations in genes that lead to congenital LQTS and drugs that induce LQTS as an adverse event to test the hypothesis that network analysis could be useful for understanding drug responses.
The ECG represents an integrated organismal measure of the electrical conduction system of the heart, and the different parts of an ECG pattern are labeled with individual letters (). In a healthy heart, depolarization of cardiac atria generates the P wave of the ECG. This is followed by the Q, R, and S peaks representing the depolarization of the cardiac ventricles. The T wave represents the repolarization of the ventricles. The interval between the start of the Q peak and the end of the T wave is the QT interval. Changes in the QT interval are risk indicators for arrhythmias (), which can be fatal. For example, torsades de pointes (TdP) is a potentially fatal arrhythmia associated with LQTS (14
Fig. 1 Multiscale relationships between mutated genes and QT interval. LQTS can be described across several organizational scales. (A) Organ level: LQTS symptoms result from disruptions of blood flow as a result of electrical disturbances of the heart that are (more ...)
Malfunction of specific ion channels in cardiac myocytes is often the cause of cardiac failure. At the cellular level, several major currents contribute to cardiac action potential () (14
), and genetic mutations that alter the functional properties of the ion channels that produce these currents can alter the duration of the action potential and thus change the QT interval (). Mutations in eight genes that encode ion channels and four genes encoding proteins known to interact with these channels have been associated with familial forms of LQTS (). However, identical mutations can affect different members of a family to varying extents ranging from individuals with no observable disease to those that experience sudden cardiac death. This variability is indicative of complex interactions between mutated genes and other cellular components (15
). This complex network could account for some of the variability in penetrance of LQTS mutations, as well as differences in susceptibility to acquired LQTS, which can be induced by certain drugs or metabolic disturbances (16
Drugs used to treat noncardiovascular diseases can have cardiovascular risks as side effects (17
). Many seemingly unrelated drugs, often used for noncardiac indications, cause QT interval prolongation and TdP as an adverse event (18
). This rare, but dangerous, side effect occurs frequently enough that, in clinical practice, drugs with the risk of inducing acquired LQTS are avoided in patients with the congenital syndrome (18
) and has also resulted in removal of drugs from the market. For example, the gastrointestinal drug cisapride was withdrawn from human usage because it could cause QT prolongation, leading to fatal arrhythmias (19
), and the allergy medication terfenadine was withdrawn because of its propensity to cause LQTS (20
). Although many drugs that cause LQTS interact with the product of the KCNH2
gene, the HERG potassium channel that regulates myocyte action potential, it would be useful to know whether other targets of drugs that cause QT prolongation are related to genes involved in LQTS. Because the degree of blockade of the HERG ion channel is not directly related to the risk of TdP and is not the only risk factor for TdP (21
), it is likely that other genes associated with LQTS are targets of drugs that cause acquired LQTS.
LQTS can be considered a channelopathy related to altered cell signaling. An underlying defect in ion channel function decreases cardiac repolarization reserve, and then additional signals, such as QT-prolonging medications or adrenergic activation due to stress, can precipitate dangerous arrhythmias. The importance of cell signaling pathways is evident because the main treatment for patients with congenital LQTS is β-adrenergic blocking drugs (14
We integrated protein-protein interactions with disease-associated genes to identify a signaling network that regulates the ion channels involved in LQTS. We show that such a network, based on known disease genes and protein-protein interactions, can be used in combination with clinical data sets, such as the U.S. Food and Drug Administration’s (FDA’s) Adverse Event Reporting System (AERS), to understand previously undetected relationships between drugs and adverse events.