Ion channel inhibition by targeted recruitment of NEDD4-2 with divalent nanobodies.

Targeted recruitment of E3 ubiquitin ligases to degrade traditionally undruggable proteins is a disruptive paradigm for developing new therapeutics. Two salient limitations are that <2% of the ~600 E3 ligases in the human genome have been exploited to produce proteolysis targeting chimeras (PROTACs), and the efficacy of the approach has not been demonstrated for a vital class of complex multi-subunit membrane proteins- ion channels. NEDD4-1 and NEDD4-2 are physiological regulators of myriad ion channels, and belong to the 28-member HECT (homologous to E6AP C-terminus) family of E3 ligases with widespread roles in cell/developmental biology and diverse diseases including various cancers, immunological and neurological disorders, and chronic pain. The potential efficacy of HECT E3 ligases for targeted protein degradation is unexplored, constrained by a lack of appropriate binders, and uncertain due to their complex regulation by layered intra-molecular and posttranslational mechanisms. Here, we identified a nanobody that binds with high affinity and specificity to a unique site on the N-lobe of the NEDD4-2 HECT domain at a location physically separate from sites critical for catalysis- the E2 binding site, the catalytic cysteine, and the ubiquitin exosite- as revealed by a 3.1 Å cryo-electron microscopy reconstruction. Recruiting endogenous NEDD4-2 to diverse ion channel proteins (KCNQ1, ENaC, and CaV2.2) using a divalent (DiVa) nanobody format strongly reduced their functional expression with minimal off-target effects as assessed by global proteomics, compared to simple NEDD4-2 overexpression. The results establish utility of a HECT E3 ligase for targeted protein downregulation, validate a class of complex multi-subunit membrane proteins as susceptible to this modality, and introduce endogenous E3 ligase recruitment with DiVa nanobodies as a general method to generate novel genetically-encoded ion channel inhibitors.

Divergent regulation of KCNQ1/E1 by targeted recruitment of protein kinase A to distinct sites on the channel complex.

The slow delayed rectifier potassium current, IKs, conducted through pore-forming Q1 and auxiliary E1 ion channel complexes is important for human cardiac action potential repolarization. During exercise or fright, IKs is up-regulated by protein kinase A (PKA)-mediated Q1 phosphorylation to maintain heart rhythm and optimum cardiac performance. Sympathetic up-regulation of IKs requires recruitment of PKA holoenzyme (two regulatory - RI or RII - and two catalytic Cα subunits) to Q1 C-terminus by an A kinase anchoring protein (AKAP9). Mutations in Q1 or AKAP9 that abolish their functional interaction result in long QT syndrome type 1 and 11, respectively, which increases the risk of sudden cardiac death during exercise. Here, we investigated the utility of a targeted protein phosphorylation (TPP) approach to reconstitute PKA regulation of IKs in the absence of AKAP9. Targeted recruitment of endogenous Cα to E1-YFP using a GFP/YFP nanobody (nano) fused to RIIα enabled acute cAMP-mediated enhancement of IKs, reconstituting physiological regulation of the channel complex. By contrast, nano-mediated tethering of RIIα or Cα to Q1-YFP constitutively inhibited IKs by retaining the channel intracellularly in the endoplasmic reticulum and Golgi. Proteomic analysis revealed that distinct phosphorylation sites are modified by Cα targeted to Q1-YFP compared to free Cα. Thus, functional outcomes of synthetically recruited PKA on IKs regulation is critically dependent on the site of recruitment within the channel complex. The results reveal insights into divergent regulation of IKs by phosphorylation across different spatial and time scales, and suggest a TPP approach to develop new drugs to prevent exercise-induced sudden cardiac death.

Pharmacological rescue of specific long QT variants of KCNQ1/KCNE1 channels.

The congenital Long QT Syndrome (LQTS) is an inherited disorder in which cardiac ventricular repolarization is delayed and predisposes patients to cardiac arrhythmias and sudden cardiac death. LQT1 and LQT5 are LQTS variants caused by mutations in KCNQ1 or KCNE1 genes respectively. KCNQ1 and KCNE1 co-assemble to form critical IKS potassium channels. Beta-blockers are the standard of care for the treatment of LQT1, however, doing so based on mechanisms other than correcting the loss-of-function of K+ channels. ML277 and R-L3 are compounds that enhance IKS channels and slow channel deactivation in a manner that is dependent on the stoichiometry of KCNE1 subunits in the assembled channels. In this paper, we used expression of IKS channels in Chinese hamster ovary (CHO) cells and Xenopus oocytes to study the potential of these two drugs (ML277 and R-L3) for the rescue of LQT1 and LQT5 mutant channels. We focused on the LQT1 mutation KCNQ1-S546L, and two LQT5 mutations, KCNE1-L51H and KCNE1-G52R. We found ML277 and R-L3 potentiated homozygote LQTS mutations in the IKS complexes-KCNE1-G52R and KCNE1-L51H and in heterogeneous IKS channel complexes which mimic heterogeneous expression of mutations in patients. ML277 and R-L3 increased the mutant IKS current amplitude and slowed current deactivation, but not in wild type (WT) IKS. We obtained similar results in the LQT1 mutant (KCNQ1 S546L/KCNE1) with ML277 and R-L3. ML277 and R-L3 had a similar effect on the LQT1 and LQT5 mutants, however, ML277 was more effective than R-L3 in this modulation. Importantly we found that not all LQT5 mutants expressed with KCNQ1 resulted in channels that are potentiated by these drugs as the KCNE1 mutant D76N inhibited drug action when expressed with KCNQ1. Thus, our work shows that by directly studying the treatment of LQT1 and LQT5 mutations with ML277 and R-L3, we will understand the potential utility of these activators as options in specific LQTS therapeutics.

Potassium Channels as Therapeutic Targets in Pulmonary Arterial Hypertension.

Pulmonary arterial hypertension (PAH) is a devastating disease with high morbidity and mortality. Deleterious remodeling in the pulmonary arterial system leads to irreversible arterial constriction and elevated pulmonary arterial pressures, right heart failure, and eventually death. The difficulty in treating PAH stems in part from the complex nature of disease pathogenesis, with several signaling compounds known to be involved (e.g., endothelin-1, prostacyclins) which are indeed targets of PAH therapy. Over the last decade, potassium channelopathies were established as novel causes of PAH. More specifically, loss-of-function mutations in the KCNK3 gene that encodes the two-pore-domain potassium channel KCNK3 (or TASK-1) and loss-of-function mutations in the ABCC8 gene that encodes a key subunit, SUR1, of the ATP-sensitive potassium channel (KATP) were established as the first two potassium channelopathies in human cohorts with pulmonary arterial hypertension. Moreover, voltage-gated potassium channels (Kv) represent a third family of potassium channels with genetic changes observed in association with PAH. While other ion channel genes have since been reported in association with PAH, this review focuses on KCNK3, KATP, and Kv potassium channels as promising therapeutic targets in PAH, with recent experimental pharmacologic discoveries significantly advancing the field.

Two small-molecule activators share similar effector sites in the KCNQ1 channel pore but have distinct effects on voltage sensor movements.

ML277 and R-L3 are two small-molecule activators of KCNQ1, the pore-forming subunit of the slowly activating potassium channel IKs. KCNQ1 loss-of-function mutations prolong cardiac action potential duration and are associated with long QT syndrome, which predispose patients to lethal ventricular arrhythmia. ML277 and R-L3 enhance KCNQ1 current amplitude and slow deactivation. However, the presence of KCNE1, an auxiliary subunit of IKs channels, renders the channel insensitive to both activators. We found that ML277 effects are dependent on several residues in the KCNQ1 pore domain. Some of these residues are also necessary for R-L3 effects. These residues form a putative hydrophobic pocket located between two adjacent KCNQ1 subunits, where KCNE1 subunits are thought to dwell, thus providing an explanation for how KCNE1 renders the IKs channel insensitive to these activators. Our experiments showed that the effect of R-L3 on voltage sensor movement during channel deactivation was much more prominent than that of ML277. Simulations using a KCNQ1 kinetic model showed that the effects of ML277 and R-L3 could be reproduced through two different effects on channel gating: ML277 enhances KCNQ1 channel function through a pore-dependent and voltage sensor-independent mechanism, while R-L3 affects both channel pore and voltage sensor.

Exploring mutation specific beta blocker pharmacology of the pathogenic late sodium channel current from patient-specific pluripotent stem cell myocytes derived from long QT syndrome mutation carriers.

The congenital long QT syndrome (LQTS), one of the most common cardiac channelopathies, is characterized by delayed ventricular repolarization underlying prolongation of the QT interval of the surface electrocardiogram. LQTS is caused by mutations in genes coding for cardiac ion channels or ion channel-associated proteins. The major therapeutic approach to LQTS management is beta blocker therapy which has been shown to be effective in treatment of LQTS variants caused by mutations in K+ channels. However, this approach has been questioned in the treatment of patients identified as LQTS variant 3(LQT3) patients who carry mutations in SCN5A, the gene coding for the principal cardiac Na+ channel. LQT3 mutations are gain of function mutations that disrupt spontaneous Na+ channel inactivation and promote persistent or late Na+ channel current (INaL) that delays repolarization and underlies QT prolongation. Clinical investigation of patients with the two most common LQT3 mutations, the ΔKPQ and the E1784K mutations, found beta blocker treatment a useful therapeutic approach for managing arrhythmias in this patient population. However, there is little experimental data that reveals the mechanisms underlying these antiarrhythmic actions. Here, we have investigated the effects of the beta blocker propranolol on INaL expressed by ΔKPQ and E1784K channels in induced pluripotent stem cells derived from patients carrying these mutations. Our results indicate that propranolol preferentially inhibits INaL expressed by these channels suggesting that the protective effects of propranolol in treating LQT3 patients is due in part to modulation of INaL.

Human iPSC-derived cardiomyocytes and pyridyl-phenyl mexiletine analogs.

In the United States, approximately one million individuals are hospitalized every year for arrhythmias, making arrhythmias one of the top causes of healthcare expenditures. Mexiletine is currently used as an antiarrhythmic drug but has limitations. The purpose of this work was to use normal and Long QT syndrome Type 3 (LQTS3) patient-derived human induced pluripotent stem cell (iPSC)-derived cardiomyocytes to identify an analog of mexiletine with superior drug-like properties. Compared to racemic mexiletine, medicinal chemistry optimization of substituted racemic pyridyl phenyl mexiletine analogs resulted in a more potent sodium channel inhibitor with greater selectivity for the sodium over the potassium channel and for late over peak sodium current.

Antiarrhythmic Hit to Lead Refinement in a Dish Using Patient-Derived iPSC Cardiomyocytes.

Ventricular cardiac arrhythmia (VA) arises in acquired or congenital heart disease. Long QT syndrome type-3 (LQT3) is a congenital form of VA caused by cardiac sodium channel (INaL) SCN5A mutations that prolongs cardiac action potential (AP) and enhances INaL current. Mexiletine inhibits INaL and shortens the QT interval in LQT3 patients. Above therapeutic doses, mexiletine prolongs the cardiac AP. We explored structure-activity relationships (SAR) for AP shortening and prolongation using dynamic medicinal chemistry and AP kinetics in human-induced pluripotent stem cell-derived cardiomyocytes (hiPSC-CMs). Using patient-derived LQT3 and healthy hiPSC-CMs, we resolved distinct SAR for AP shortening and prolongation effects in mexiletine analogues and synthesized new analogues with enhanced potency and selectivity for INaL. This resulted in compounds with decreased AP prolongation effects, increased metabolic stability, increased INaL selectivity, and decreased avidity for the potassium channel. This study highlights using hiPSC-CMs to guide medicinal chemistry and "drug development in a dish".

Reengineering an Antiarrhythmic Drug Using Patient hiPSC Cardiomyocytes to Improve Therapeutic Potential and Reduce Toxicity.

Modeling cardiac disorders with human induced pluripotent stem cell (hiPSC)-derived cardiomyocytes is a new paradigm for preclinical testing of candidate therapeutics. However, disease-relevant physiological assays can be complex, and the use of hiPSC-cardiomyocyte models of congenital disease phenotypes for guiding large-scale screening and medicinal chemistry have not been shown. We report chemical refinement of the antiarrhythmic drug mexiletine via high-throughput screening of hiPSC-CMs derived from patients with the cardiac rhythm disorder long QT syndrome 3 (LQT3) carrying SCN5A sodium channel variants. Using iterative cycles of medicinal chemistry synthesis and testing, we identified drug analogs with increased potency and selectivity for inhibiting late sodium current across a panel of 7 LQT3 sodium channel variants and suppressing arrhythmic activity across multiple genetic and pharmacological hiPSC-CM models of LQT3 with diverse backgrounds. These mexiletine analogs can be exploited as mechanistic probes and for clinical development.

Detection of Nav1.5 Conformational Change in Mammalian Cells Using the Noncanonical Amino Acid ANAP.

Nav1.5 inactivation is necessary for healthy conduction of the cardiac action potential. Genetic mutations of Nav1.5 perturb inactivation and cause potentially fatal arrhythmias associated with long QT syndrome type 3. The exact structural dynamics of the inactivation complex is unknown. To sense inactivation gate conformational change in live mammalian cells, we incorporated the solvatochromic fluorescent noncanonical amino acid 3-((6-acetylnaphthalen-2-yl)amino)-2-aminopropanoic acid (ANAP) into single sites in the Nav1.5 inactivation gate. ANAP was incorporated in full-length and C-terminally truncated Nav1.5 channels using mammalian cell synthetase-tRNA technology. ANAP-incorporated channels were expressed in mammalian cells, and they exhibited pathophysiological function. A spectral imaging potassium depolarization assay was designed to detect ANAP emission shifts associated with Nav1.5 conformational change. Site-specific intracellular ANAP incorporation affords live-cell imaging and detection of Nav1.5 inactivation gate conformational change in mammalian cells.

Regulation of IKs Potassium Current by Isoproterenol in Adult Cardiomyocytes Requires Type 9 Adenylyl Cyclase.

The subunits KCNQ1 and KCNE1 generate the slowly activating, delayed rectifier potassium current, IKs, that responds to sympathetic stimulation and is critical for human cardiac repolarization. The A-kinase anchoring protein Yotiao facilitates macromolecular complex formation between IKs and protein kinase A (PKA) to regulate phosphorylation of KCNQ1 and IKs currents following beta-adrenergic stimulation. We have previously shown that adenylyl cyclase Type 9 (AC9) is associated with a KCNQ1-Yotiao-PKA complex and facilitates isoproterenol-stimulated phosphorylation of KCNQ1 in an immortalized cell line. However, requirement for AC9 in sympathetic control of IKs in the heart was unknown. Using a transgenic mouse strain expressing the KCNQ1-KCNE1 subunits of IKs, we show that AC9 is the only adenylyl cyclase (AC) isoform associated with the KCNQ1-KCNE1-Yotiao complex in the heart. Deletion of AC9 resulted in the loss of isoproterenol-stimulated KCNQ1 phosphorylation in vivo, even though AC9 represents less than 3% of total cardiac AC activity. Importantly, a significant reduction of isoproterenol-stimulated IKs currents was also observed in adult cardiomyocytes from IKs-expressing AC9KO mice. AC9 and Yotiao co-localize with N-cadherin, a marker of intercalated disks and cell-cell junctions, in neonatal and adult cardiomyocytes, respectively. In conclusion, AC9 is necessary for sympathetic regulation of PKA phosphorylation of KCNQ1 in vivo and for functional regulation of IKs in adult cardiomyocytes.

Loss-of-Function ABCC8 Mutations in Pulmonary Arterial Hypertension.

BACKGROUND: In pulmonary arterial hypertension (PAH), pathological changes in pulmonary arterioles progressively raise pulmonary artery pressure and increase pulmonary vascular resistance, leading to right heart failure and high mortality rates. Recently, the first potassium channelopathy in PAH, because of mutations in KCNK3, was identified as a genetic cause and pharmacological target. METHODS: Exome sequencing was performed to identify novel genes in a cohort of 99 pediatric and 134 adult-onset group I PAH patients. Novel rare variants in the gene identified were independently identified in a cohort of 680 adult-onset patients. Variants were expressed in COS cells and function assessed by patch-clamp and rubidium flux analysis. RESULTS: We identified a de novo novel heterozygous predicted deleterious missense variant c.G2873A (p.R958H) in ABCC8 in a child with idiopathic PAH. We then evaluated all individuals in the original and a second cohort for rare or novel variants in ABCC8 and identified 11 additional heterozygous predicted damaging ABCC8 variants. ABCC8 encodes SUR1 (sulfonylurea receptor 1)-a regulatory subunit of the ATP-sensitive potassium channel. We observed loss of ATP-sensitive potassium channel function for all ABCC8 variants evaluated and pharmacological rescue of all channel currents in vitro by the SUR1 activator, diazoxide. CONCLUSIONS: Novel and rare missense variants in ABCC8 are associated with PAH. Identified ABCC8 mutations decreased ATP-sensitive potassium channel function, which was pharmacologically recovered.

The Impact of Heterozygous KCNK3 Mutations Associated With Pulmonary Arterial Hypertension on Channel Function and Pharmacological Recovery.

BACKGROUND: Heterozygous loss of function mutations in the KCNK3 gene cause hereditary pulmonary arterial hypertension (PAH). KCNK3 encodes an acid-sensitive potassium channel, which contributes to the resting potential of human pulmonary artery smooth muscle cells. KCNK3 is widely expressed in the body, and dimerizes with other KCNK3 subunits, or the closely related, acid-sensitive KCNK9 channel. METHODS AND RESULTS: We engineered homomeric and heterodimeric mutant and nonmutant KCNK3 channels associated with PAH. Using whole-cell patch-clamp electrophysiology in human pulmonary artery smooth muscle and COS7 cell lines, we determined that homomeric and heterodimeric mutant channels in heterozygous KCNK3 conditions lead to mutation-specific severity of channel dysfunction. Both wildtype and mutant KCNK3 channels were activated by ONO-RS-082 (10 µmol/L), causing cell hyperpolarization. We observed robust gene expression of KCNK3 in healthy and familial PAH patient lungs, but no quantifiable expression of KCNK9, and demonstrated in functional studies that KCNK9 minimizes the impact of select KCNK3 mutations when the 2 channel subunits co-assemble. CONCLUSIONS: Heterozygous KCNK3 mutations in PAH lead to variable loss of channel function via distinct mechanisms. Homomeric and heterodimeric mutant KCNK3 channels represent novel therapeutic substrates in PAH. Pharmacological and pH-dependent activation of wildtype and mutant KCNK3 channels in pulmonary artery smooth muscle cells leads to membrane hyperpolarization. Co-assembly of KCNK3 with KCNK9 subunits may provide protection against KCNK3 loss of function in tissues where both KCNK9 and KCNK3 are expressed, contributing to the lung-specific phenotype observed clinically in patients with PAH because of KCNK3 mutations.

KCNE1 and KCNE3 modulate KCNQ1 channels by affecting different gating transitions.

KCNE ß-subunits assemble with and modulate the properties of voltage-gated K+ channels. In the heart, KCNE1 associates with the α-subunit KCNQ1 to generate the slowly activating, voltage-dependent potassium current (IKs) in the heart that controls the repolarization phase of cardiac action potentials. By contrast, in epithelial cells from the colon, stomach, and kidney, KCNE3 coassembles with KCNQ1 to form K+ channels that are voltage-independent K+ channels in the physiological voltage range and important for controlling water and salt secretion and absorption. How KCNE1 and KCNE3 subunits modify KCNQ1 channel gating so differently is largely unknown. Here, we use voltage clamp fluorometry to determine how KCNE1 and KCNE3 affect the voltage sensor and the gate of KCNQ1. By separating S4 movement and gate opening by mutations or phosphatidylinositol 4,5-bisphosphate depletion, we show that KCNE1 affects both the S4 movement and the gate, whereas KCNE3 affects the S4 movement and only affects the gate in KCNQ1 if an intact S4-to-gate coupling is present. Further, we show that a triple mutation in the middle of the transmembrane (TM) segment of KCNE3 introduces KCNE1-like effects on the second S4 movement and the gate. In addition, we show that differences in two residues at the external end of the KCNE TM segments underlie differences in the effects of the different KCNEs on the first S4 movement and the voltage sensor-to-gate coupling.

Gating mechanisms underlying deactivation slowing by two KCNQ1 atrial fibrillation mutations.

KCNQ1 is a voltage-gated potassium channel that is modulated by the beta-subunit KCNE1 to generate IKs, the slow delayed rectifier current, which plays a critical role in repolarizing the cardiac action potential. Two KCNQ1 gain-of-function mutations that cause a genetic form of atrial fibrillation, S140G and V141M, drastically slow IKs deactivation. However, the underlying gating alterations remain unknown. Voltage clamp fluorometry (VCF) allows simultaneous measurement of voltage sensor movement and current through the channel pore. Here, we use VCF and kinetic modeling to determine the effects of mutations on channel voltage-dependent gating. We show that in the absence of KCNE1, S140G, but not V141M, directly slows voltage sensor movement, which indirectly slows current deactivation. In the presence of KCNE1, both S140G and V141M slow pore closing and alter voltage sensor-pore coupling, thereby slowing current deactivation. Our results suggest that KCNE1 can mediate changes in pore movement and voltage sensor-pore coupling to slow IKs deactivation and provide a key step toward developing mechanism-based therapies.

Coupling Data Mining and Laboratory Experiments to Discover Drug Interactions Causing QT Prolongation.

BACKGROUND: QT interval-prolonging drug-drug interactions (QT-DDIs) may increase the risk of life-threatening arrhythmia. Despite guidelines for testing from regulatory agencies, these interactions are usually discovered after drugs are marketed and may go undiscovered for years. OBJECTIVES: Using a combination of adverse event reports, electronic health records (EHR), and laboratory experiments, the goal of this study was to develop a data-driven pipeline for discovering QT-DDIs. METHODS: 1.8 million adverse event reports were mined for signals indicating a QT-DDI. Using 1.6 million electrocardiogram results from 380,000 patients in our institutional EHR, these putative interactions were either refuted or corroborated. In the laboratory, we used patch-clamp electrophysiology to measure the human ether-à-go-go-related gene (hERG) channel block (the primary mechanism by which drugs prolong the QT interval) to evaluate our top candidate. RESULTS: Both direct and indirect signals in the adverse event reports provided evidence that the combination of ceftriaxone (a cephalosporin antibiotic) and lansoprazole (a proton-pump inhibitor) will prolong the QT interval. In the EHR, we found that patients taking both ceftriaxone and lansoprazole had significantly longer QTc intervals (up to 12 ms in white men) and were 1.4 times more likely to have a QTc interval above 500 ms. In the laboratory, we found that, in combination and at clinically relevant concentrations, these drugs blocked the hERG channel. As a negative control, we evaluated the combination of lansoprazole and cefuroxime (another cephalosporin), which lacked evidence of an interaction in the adverse event reports. We found no significant effect of this pair in either the EHR or in the electrophysiology experiments. Class effect analyses suggested this interaction was specific to lansoprazole combined with ceftriaxone but not with other cephalosporins. CONCLUSIONS: Coupling data mining and laboratory experiments is an efficient method for identifying QT-DDIs. Combination therapy of ceftriaxone and lansoprazole is associated with increased risk of acquired long QT syndrome.

Cardiac Delayed Rectifier Potassium Channels in Health and Disease.

Cardiac delayed rectifier potassium channels conduct outward potassium currents during the plateau phase of action potentials and play pivotal roles in cardiac repolarization. These include IKs, IKr and the atrial specific IKur channels. In this article, we will review their molecular identities and biophysical properties. Mutations in the genes encoding delayed rectifiers lead to loss- or gain-of-function phenotypes, disrupt normal cardiac repolarization and result in various cardiac rhythm disorders, including congenital Long QT Syndrome, Short QT Syndrome and familial atrial fibrillation. We will also discuss the prospect of using delayed rectifier channels as therapeutic targets to manage cardiac arrhythmia.

An Integrative Data Science Pipeline to Identify Novel Drug Interactions that Prolong the QT Interval.

INTRODUCTION: Drug-induced prolongation of the QT interval on the electrocardiogram (long QT syndrome, LQTS) can lead to a potentially fatal ventricular arrhythmia known as torsades de pointes (TdP). Over 40 drugs with both cardiac and non-cardiac indications are associated with increased risk of TdP, but drug-drug interactions contributing to LQTS (QT-DDIs) remain poorly characterized. Traditional methods for mining observational healthcare data are poorly equipped to detect QT-DDI signals due to low reporting numbers and lack of direct evidence for LQTS. OBJECTIVE: We hypothesized that LQTS could be identified latently using an adverse event (AE) fingerprint of more commonly reported AEs. We aimed to generate an integrated data science pipeline that addresses current limitations by identifying latent signals for QT-DDIs in the US FDA's Adverse Event Reporting System (FAERS) and retrospectively validating these predictions using electrocardiogram data in electronic health records (EHRs). METHODS: We trained a model to identify an AE fingerprint for risk of TdP for single drugs and applied this model to drug pair data to predict novel DDIs. In the EHR at Columbia University Medical Center, we compared the QTc intervals of patients prescribed the flagged drug pairs with patients prescribed either drug individually. RESULTS: We created an AE fingerprint consisting of 13 latently detected side effects. This model significantly outperformed a direct evidence control model in the detection of established interactions (p = 1.62E-3) and significantly enriched for validated QT-DDIs in the EHR (p = 0.01). Of 889 pairs flagged in FAERS, eight novel QT-DDIs were significantly associated with prolonged QTc intervals in the EHR and were not due to co-prescribed medications. CONCLUSIONS: Latent signal detection in FAERS validated using the EHR presents an automated and data-driven approach for systematically identifying novel QT-DDIs. The high-confidence hypotheses flagged using this method warrant further investigation.