Dysfunctional Sodium Channels and Arrhythmogenesis: Insights Into the Molecular Regulation of Cardiac Sodium Channels Using Transgenic Mice

Dysfunctional Sodium Channels and Arrhythmogenesis: Insights into the Molecular Regulation of Cardiac Sodium Channels Using Transgenic Mice

Jeffrey Michael Abrams

Submitted in partial fulfillment of the requirements for the degree of Doctor of Philosophy under the Executive Committee of the Graduate School of Arts and Sciences

COLUMBIA UNIVERSITY

2017

ABSTRACT

Dysfunctional Sodium Channels and Arrhythmogenesis: Insights into the Molecular

Regulation of Cardiac Sodium Channels Using Transgenic Mice

Jeffrey M. Abrams

Proper functioning of the voltage gated sodium channel, NaV1.5, is essential for maintenance of normal cardiac electrophysiological properties. Changes to the biophysical properties of sodium channels can take many forms and can affect the peak component of current carried during phase zero of the action potential; the “persistent” or

“late” current component conducted during the repolarizing phases of the action potential; the availability of the channel as seen by changes in window current; and the kinetics of channel transitions between closed, opened and inactivated states.

Mutations in NaV1.5 that alter these parameters of channel function are linked to a number of cardiac diseases including arrhythmias such as atrial fibrillation. In addition, mutations in many of the auxiliary proteins that form part of the sodium channel macromolecular complex have likewise been associated with diseases of the heart.

Mutations in regions of the sodium channel responsible for interactions with these auxiliary proteins have also been linked to various dysfunctional cardiac states. Indeed, a large number of disease causing mutations are localized to the C-terminal domain of

NaV1.5, a hotspot for interacting proteins.

Using a transgenic mouse model, we show that expression of a mutant sodium channel with gain-of-function properties conferring increased persistent current, is

sufficient to cause both structural and electrophysiological abnormalities in the heart driving the development of spontaneous and prolonged episodes of atrial fibrillation. The sustained and spontaneous atrial arrhythmias, an unusual if not unique phenotype in mice, enabled explorations of mechanisms of atrial fibrillation using in vivo (telemetry), ex vivo

(optical voltage mapping), and in vitro (cellular electrophysiology) techniques.

Since persistent sodium current was the driver of the structural and electrophysiological abnormalities leading to atrial fibrillation, we subsequently pursued studies exploring the mechanisms of persistent sodium current. Prior work of heterologously expressed sodium channels identified calmodulin as a regulator of persistent current. Mutation of the calmodulin binding site in the C-terminus of the cardiac sodium channel caused increased persistent current when the channel was expressed heterologously. The role of calmodulin in the regulation of the sodium channel in cardiomyocytes has not been definitively determined. We created transgenic mice expressing human sodium channels harboring a mutation of the calmodulin binding site.

Using whole cell patch clamping, we found, in contrast to previously reported findings, that ablation of the calmodulin binding site did not induce increased persistent sodium current. Instead, loss of calmodulin binding stabilized the inactivated state by shifting the

V50 for steady-state inactivation in the hyperpolarizing direction.

Furthermore, loss of calmodulin binding sped up the transition to the inactivated state demonstrated by a significantly shortened tau of inactivation. In contrast to studies performed in heterologous expression systems, our findings thus suggest that in heart cells, calmodulin binding increases availability, similar to its role in regulating NaV1.4 channels.

The studies were then expanded to explore the role of other interacting proteins, fibroblast growth factor (FGF) homologous factors (FHF), in the presence and absence of calmodulin binding. Using whole cell patch clamping, we found that a mutation

(H1849R) of the sodium channel causing decreased FHF binding affinity leads to a rightward shift in steady-state inactivation and a slowed rate of inactivation of INa. A third mutant channel, with concurrent decreased FHF and calmodulin binding affinity similarly results in a rightward shift in steady-state inactivation suggesting a dominant effect of the

H1849R mutation. Persistent current was not elevated in either of these mutant channels.

Importantly, the methodology that we report enables us and other groups to carry out studies of human sodium channels in the native environment of NaV1.5. Our investigation into calmodulin’s role, which yielded conclusions distinct from prior findings in heterologous expression systems, demonstrates the value of this approach.

Table of Contents LIST OF CHARTS, GRAPHS, ILLUSTRATIONS………………………………….iv ACKNOWLEDGMENTS…………………………………………...………………….vi DEDICATION…………………………………………………………………………viii INTRODUCTION……………………………………………………………………….1 Action Potential…………………………………………………………………………...1 Sodium Channel Early Discoveries……………………………………………………….3 Voltage Dependent Activation……………………….…………………………….….. 4 Selectivity…………………….…………………………………………………………5 Inactivation………………………………………………………………………..……5 NavAb Structure…………………………………………………………………………..6 Persistent Current……………………………………………………………………...…..7 Auxiliary Proteins……………………………………………………………………..…12 Calmodulin…………………………………………………………………………….13 Fibroblast Growth Factor Homologous Factors………...……………………………14 Sodium Channel in Disease……………………………………………………………...17 Gain-of-Function and Loss-of-Function Effects………………………………………17 Atrial Fibrillation……………………………………………………………………...…18 Cardiac Conduction………………………………………………………………...…18 Morbidities and Mortalities in Patients with Atrial Fibrillation…………...………....19 Arrhythmia Basics………………………………………………………………….….19 Early-Afterdepolarization……………………………………………………………..20 Delayed-Afterdepolarization……………………………………………………….….21 Automaticity……………………….…………………………………………………..22 Reentrant Activity……………….…………………………………………………..…22 Triggered Activity in Veins………………………………………….……………...…23 CaMKII and Atrial Fibrillation………………………………...………………..……23 Genetics of Atrial Fibrillation…………………………...……………………………23

Nav1.5 in Atrial Fibrillation...………………………...…………………………….…25 Remodeling…………………………...……………………………………………….26 Treatments for Atrial Fibrillation………………...…………….………………...……27

i Small Animal Models of Atrial Fibrillation……………………….…………………..28 Large Animal Models of Atrial Fibrillation…………………………………….……..29 Dissertation Outline………………………………………………………………...……30

Materials and Methods………………………………………………………………....32 Mouse Models……………………………………………………………………………32 Mouse In-Vivo Characterization………………………………………………………....33 Biochemistry/Molecular Biology………………………………………………………...34 Cellular Electrophysiology………………………………………………………………36 Epicardial Optical Voltage Mapping…………………………………………………….39 Histology……………………………………………………………………………..…..39 Transmission Electron Microscopy………….…………………………………………..40 Statistical Analyses..……………………………………………………………..………40

Chapter 1: Exploration of atrial fibrillation molecular mechanisms through use of transgenic mouse model………………………………………………………………..41 Introduction………………………………………………………………….………….42 Experimental Design and Approach………………………………………………….….43 Results………………………………………………………………………………....…44 Discussion………………………………………………………………………………..59 Summary………………………………………………………………………………....64

Chapter 2: Calmodulin’s effect on NaV1.5 within cardiomyocytes………………… 65 Introduction…………………………………………………………………….……….66 Experimental Design and Approach………………………………………………..…....67 Results………………………………………………………………………………...….68 Discussion……………………………………………………………………………..... 80 Summary……………………………………………………………………………...... 85

Chapter 3: FHF’s effect on NaV1.5 within cardiomyocytes…..………………………... 86 Introduction……………………………………………………………………….….... 87

ii Experimental Design and Approach………………………………………….………….88 Results………………………………………………………………………….….……. 88 Discussion………………………………………………………………………………..98 Summary…………………………………………………………………………....…..101

Conclusions……………………………..………………………....……………...……103

Bibliography……………………………..……………………….………….……...…106

Appendix: Supplementary Tables……………………………..…………..…...…….118

iii

List of Charts, Graphs, Illustrations

Introduction Figure 1. Ventricular action potential 2

Figure 2. NaV1.5 membrane topology 4

Figure 3. NaVAb structure 7 Figure 4. Persistent current 9 Figure 5. Triggered activity and reentrant waves 21

Chapter 1 Figure 1. Cardiac-specific, FLAG-tagged F1759A 45

NaV1.5-expressing transgenic mice

Figure 2. Lidocaine resistance in F1759A-NaV1.5 47 + Figure 3. F1759A-NaV1.5 increases persistent Na 48 current in atria and ventricles Figure 4. Persistent current is due to F1759A transgenic channels 49 Figure 5. Functional and Structural changes in atria and ventricle 51 Figure 6. Histological Changes in Atria and Ventricle 52 Figure 7. Prolonged QT interval and spontaneous 54 atrial fibrillation in F1759A-dTG mice Figure 8. Optically recorded atrial action potentials 56 and voltage maps of AF Figure 9. Inhibition of Na+-Ca2+ exchanger (NCX) attenuates 58 atrial and ventricular arrhythmogenesis in F1759A-dTG mice Figure 10. Proposed mechanism of arrhythmogenesis 60

Chapter 2 Figure 1. Cardiac-specific, FLAG-tagged 68

C374Y NaV1.5-expressing TG mice

iv Figure 2. TTX-sensitive current in pWT and IQ/AA TG mice 70 Figure 3. ECG analysis of TG mice 71 Figure 4. Persistent current is not increased in IQ/AA mice 74 Figure 5. Schematic of conditions in which steady-state 75 activation/inactivation (A/I) parameters were measured Figure 6. Voltage dependence of activation in IQ/AA mice 77 Figure 7. IQ/AA results in loss-of-function 79

Figure 8. Structure of NaV1.5 C-terminal (CT) domain 83

Chapter 3

Figure 1. Cardiac-specific, FLAG-tagged C374Y NaV1.5-expressin 89 TG mice and TTX-sensitive current Figure 2. Electrophysiological parameters in H1849R 90 and H1849R-IQ/AA mice Figure 3. Persistent current is not increased in H1849R mice 91 Figure 4. Voltage dependence of activation in H1849R mice 92 Figure 5. H1849R results in gain-of-function 93 Figure 6. Persistent current is not increased 95 in H1849R –IQ/AA mice Figure 7. Voltage dependence of activation 96 in H1849R-IQ/AA mice Figure 8. Increased availability in H1849R-IQ/AA 98

Appendix

Table 1. Summary of work exploring Calmodulin regulation of NaV1.5 118

Table 2. Summary of work exploring FHF regulation of NaV1.5 119

v Acknowledgments

This thesis has been made possible only through countless contributions and assistance from others.

One of the great benefits of having conducted my graduate studies at Columbia has been the opportunity to interact with and learn from the many experts in biophysics, cardiac electrophysiology and ion channel research. In particular, I have greatly benefited from significant input and guidance from the members of my thesis committee, Drs.

Henry Colecraft, Robert “Rocky” Kass and Penelope Boyden. Being afforded the chance to present my research to them, and receive their feedback was something that I greatly cherished. Our meetings were always valuable, each one leaving me with questions to ponder, interesting directions to consider and experiments to attempt. Thanks are also owed to Dr. Masayuki Yazawa for serving on my defense committee.

I have similarly benefited from- and am similarly grateful for- having had the chance to work with the members of the Marx lab. From my first day in lab, I was made to feel welcome. Each member has contributed to my education offering technical advice on experiments, theoretical advice on the underlying science, not to mention personal support. Drs. Lin Yang and Guoxia Liu provided significant assistance in new molecular biology and biochemistry techniques. Dr. Sergey Zakharov offered guidance in learning electrophysiological techniques. Dr. Alexander Katchman helped with electrophysiological experiments. Dr. Elaine Wan was of significant help in carrying out animal phenotypic studies. Drs. John Morrow and Leroy Joseph offered experimental advice as well as general guidance. I would also like to thank Dr. Richard Weinberg, with

vi whom I had the pleasure of working side-by-side in my first year in lab. Working closely with Dr. Weinberg was an incredible learning experience.

I have also been supported by a number of people outside of lab itself. Zaia Sivo has made navigating through the Integrated Program seamless. Jeffrey Brandt has similarly been extremely helpful during my time at Columbia. Dr. Jamie Rubin expertly advised me in applying for various grants. And each of the MD/PhD directors, Drs.

Steven Reiner, Ronald Liem, Michael Shelanski and Patrice Spitalnik has likewise offered help whenever needed and support at all times.

My parents have been extremely supportive of me both during the PhD years, and more importantly, in the formative years in which I became interested in science. The true beginning of my career as a scientist was discussing math, chemistry, and physics problems with my dad as a kid. Thank you mom and dad for everything! My fiancé Ilana

Yoffe has made my years pursuing a PhD far more enjoyable and meaningful than they otherwise would have been.

Finally, I am immensely appreciative of having been able to conduct my studies under the guidance of Dr. Steven Marx. Dr. Marx always had an open door. Between coming up with an initial proposal during the start of my tenure in lab and writing this thesis towards its end, Dr. Marx has offered assistance with pleasure. To the extent that

I’ve progressed as a scientist during my years as a graduate student, I owe significant thanks to Dr. Marx.

vii

Dedication

To my grandpa Nathaniel, granny Freeda, grandma Faye and grandpa Rubin.

viii Introduction

NaV1.5: An Exploration of Protein Function in Health and Disease

The heart is a four-chambered organ responsible for delivering blood throughout the organs and tissues of the body. To carry out this task, the heart functions as a pump.

In a rhythmic manner, the right and left ventricular chambers of the heart eject blood into the pulmonary artery and aorta respectively. The heart’s ability to repeatedly contract and thus to generate pressures that propel blood forward is ultimately derived from the unique features of the individual cells that together make up the heart, cardiomyocytes.

Cardiac contraction is the result of coordinated and sequential contractions - or shortenings - of individual, connected cardiac cells, via the process known as excitation- contraction coupling. Numerous scientists have contributed to the discovery of this cellular phenomenon, but it was Luigi Galvani, in 1791, who first demonstrated that muscle twitching – or contraction – could be generated via electrical sparks [4]. This finding thus linked electrical signaling with muscular contractions. Since then, much has been learned regarding the precise mechanisms by which individual cells convert electrical signals into mechanical contraction. Numerous steps are required for the critical, life enabling process of cardiac contraction. Amongst the most essential is a

microscopic conformational change in the voltage gated cardiac sodium channel, NaV1.5.

Action Potential

An action potential is an electrical signal generated within cells. In the 1950s,

Alan Hodgkin and Andrew Huxley investigated action potentials in a squid giant axon in a series of experiments, which greatly enhanced our knowledge and understanding of the mechanisms responsible for the membrane potential changes characteristic of an action

1 potential [5]. The action potential within a ventricular cardiomyocyte shares many features with the neuronal action potential studied by Hodgkin and Huxley though it differs in important respects (Figure 1). In the ventricular cardiac cell, the initiating event in the action potential is inward sodium movement through voltage gated sodium channels, depolarizing the cell. Within 2 milliseconds [6], sodium channels inactivate and voltage gated potassium channels open, leading to the initial repolarization characteristic of phase 1 of the cardiac action potential. During phase 2, calcium entry through voltage activated calcium channels balances potassium efflux, resulting in a plateau in the membrane potential. As calcium channels inactivate, potassium efflux is unopposed resulting in the repolarization seen during phase 3 of the action potential. In phase 4, the cell resides at its resting membrane potential [7]. Importantly, all of the ionic movements that together create the cardiac action potential are the result of specific ions moving down their electrochemical gradients through ion channels that open and close in a highly-orchestrated manner.

Figure 1. Ventricular action potential. Sequential openings of ion channels and subsequent movement of ions down their electrochemical gradients results in the characteristic ventricular action potential waveform. The ventricular action potential is divided into four phases based on the specific channels that are predominantly active. Figure adapted from Luo & Rudy, 1994 [2]

2 Sodium Channel Early Discoveries

Through the combined efforts of electrophysiology, genetics, molecular biology and structural biology, impressive insights into the sodium channel protein have been attained. Direct evidence that the sodium current detected by Hodgkin and Huxley was the result of ion movement through a specific molecule first arose from studies demonstrating that specific drugs- tetrodotoxin and saxitoxin – were able to block sodium current with a dose response curve indicative of a single toxin molecule interacting with a single receptor [5]. In the 1980s, experiments using various sodium channel-binding toxins identified the proteins comprising the sodium channel. Catterall and Beneski demonstrated that the sodium channel consisted of a 260 kDa alpha subunit along with associated 30-40 kDa beta subunits [8] (Figure 2). Subsequent work demonstrated that single channel currents from a planar bilayer with co-expressed alpha and beta subunits were comparable to those seen from sodium channels in intact cells, further suggesting that the alpha and beta subunit comprised the sodium channel [9]. It has since been discovered that many of the defining features of sodium channels, such as toxin binding sites, voltage sensitivity, and gating properties result solely from properties of the alpha subunit though auxiliary proteins such as the beta subunit impact various aspects of sodium channel function [5].

Ten sodium channel alpha subunits are expressed in mammals. The tissue expression pattern of these channels differs, as do properties such as binding affinity with drugs and toxins and the kinetics of inactivation [10] . The predominant cardiac isoform,

NaV1.5, encoded by the SCN5A gene, has been mapped to human chromosome 3 [11].

3 The rapid progress in molecular biology during the 1980s led to the discovery of the cDNA sequence of the sodium channel alpha subunit [12]. Utilizing the primary sequence, Guy and Seetharamulu employed molecular modeling techniques to make predictions regarding the two-dimensional structure of the sodium channel [13].

Impressively, many of these predictions have been confirmed, such as that the channel consists of four homologous domains, each linked by large intracellular loops and that the

N- and C- termini of the sodium channel alpha subunits reside on the cytoplasmic side of the membrane (Figure 2) [14]. These studies also correctly suggested that each homologous domain consists of 6 transmembrane, alpha-helical segments (Figure 2) [13].

It was also believed that a large reentrant loop, which actually entered into the membrane region, connected segments S5 and S6, forming the outer pore of the channel [13].

Figure 2. Nav1.5 membrane topology. The alpha subunit of NaV1.5 interacts with one to two beta subunits in the membrane. The NaV1.5 alpha subunit consists of four homologous domains, each containing 6 transmembrane segments. The positively charged S4 segments in each domain function as voltage sensors. S5 and S6 together with the intervening P-loop form the pore of the channel. The III-IV loop plays a critical role in inactivation. Numerous proteins interact with the C-terminal domain, including FHF and calmodulin.

Voltage Dependent Activation

Sodium channels are capable of transitioning between three conformations or states [1]. With membrane depolarization, sodium channels switch from their closed 4 state, into an open state in which sodium ions can diffuse down their electrochemical gradient. Much research has been focused on determining the mechanisms linking membrane potential changes to channel opening. The current consensus is that membrane depolarization causes voltage sensors, comprised of the four S4 segments, to move through the membrane towards the extracellular side [15]. This results in conformational changes within the channel that culminate in S6 segments moving to open the pore [14].

Evidence favoring this hypothesis came from the insight that S4 transmembrane segments contain positively charged residues capable of dynamically responding to changes in membrane potential. Subsequent functional studies provided support for the role of the S4 segment as the voltage-sensing and activating component of sodium channels [16].

Selectivity

A key property of ion channels is their selectivity. For voltage-gated sodium channels, the S5 and S6 segments along with the connecting reentrant loop form the ion selectivity filter (Figure 2). Key evidence that the S5 and S6 segments formed the pore came from studies of chimeric potassium channels in which the unique ion selectivity of particular potassium channels was shown to result from a 21 amino acid stretch linking the S5-S6 segments, analogous to the loop present in sodium channels [17].

Inactivation

Inactivation represents a critical process for voltage gated sodium channels. Initial evidence that an intracellular gate was responsible for inactivation came from studies showing that a protease was able to prevent inactivation when applied intracellularly but not extracellularly [18]. These studies supported a model in which the inactivation gate

5 could enter and occlude the pore only after channel opening thus suggesting that inactivation could not occur from the closed state [18]. In papers published in the late

1980s, Vassilev et al offered evidence that the intracellular III-IV linker acted as the inactivation gate [19]. Utilizing site directed antibodies targeting different intracellular domains of the sodium channel, only antibodies against the III-IV linker led to delayed inactivation [19]. Further elucidation of the mechanisms governing inactivation came from work in the early 1990’s in which it was shown that three specific hydrophobic residues in the III-IV linker - Ile-1488, Phe-1489 and Met-1490 - were necessary for inactivation (Figure 2) [20]. Subsequent work by Eaholtz et al. (1994) showed that an artificial peptide, acetyl-KIFMK-amide, containing the IFM motif, “rescued” the loss of inactivation seen in wild type sodium channels exposed to an inactivation-impairing scorpion toxin [21]. These studies thus demonstrated that this hydrophobic trio of amino acids was necessary for the process of fast inactivation [21].

NavAb Structure

A great deal of information regarding sodium channel structure has come from high resolution images of the crystal structure of NaVAb, a voltage gated sodium channel found in the species Arcobacter butzleri. Unlike NaV1.5, this channel consists of four distinct but identical domains, and thus lacks the large intracellular loops connecting domains in the cardiac sodium channel [22]. While this sodium channel diverges from

NaV1.5 and other eukaryotic sodium channels in certain respects, many similarities exist.

Utilizing X-ray crystallography, a 2.7-angstrom image of the channel in the “pre-open” state was obtained. This structure offered insights into sodium selectivity. The selectivity

6 filter was shown to be 4.6 angstroms wide with four glutamate residues lining the a filter [22]. The model suggested sodium ions could conduct through the selectivity filter in a partially hydrated state. Full rehydration of sodium was shown to be possible upon diffusing through the b selectivity filter, after which hydrated sodium could enter the central cavity of the channel before moving into the cytoplasm through an opened activation gate [22]. An c additional interesting finding from the structure was that voltage sensing motifs of a given domain are linked with pore forming Figure 3. NavAb Structure. (a) Top view motifs of an adjacent domain, an arrangement of 3-D crystal structure of bacterial sodium channel, NaVAb (2.7-Angstroms). Pore forming domains surround the central pore thought to foster concerted gating of the four of the channel. (b) Structural components of NaVAb with S1-S6 segments and subunits comprising NaVAb [23]. intervening P loop of a single subunit highlighted. (c) Schematic of the pore architecture. Figure from [1]

Persistent Current

In a cell with sodium channels “peak current” represents the maximum inward sodium current in response to a depolarizing stimuli. Peak current is reached very shortly after a depolarizing stimulus due to the sodium channel’s rapid activation process.

Persistent or late current represents sodium current that persists throughout the duration

7 of the cardiac action potential (Figure 4) [24]. In 1989, Kiyosue and Arita measured persistent current in guinea pig ventricular myocytes, showing that it was sensitive to 60 microM tetrodotoxin and had a magnitude of 12 to 50 pA at -40 mV [25]. They also used cell-attached patch to measure single channels, and showed that two varieties of late current were present, background type and burst type. In the background type, there were individual short lasting openings. In the burst type late current, there were longer lasting openings with frequent quick closures [25]. In a subsequent study of rat and rabbit cardiomyocytes treated with lysophosphatidylcholine – in order to model cardiac ischemic injury- burst type openings in sodium channels were similarly detected [26]. In

1992, Saint et al. reported on the presence of persistent current, measured at positive potentials outside the range of sodium channel activation, in rat ventricular myocytes

[27]. The presence of persistent current at positive potentials is clinically significant, as the plateau phase of the ventricular action potential occurs at positive potentials [28]. A

1988 study identified a slowly inactivating sodium current in canine ventricular myocytes

[29]. This current was sensitive to both lidocaine and tetrodotoxin and its block shortened ventricular action potential [29]. In 1998, Maltsev et al. demonstrated the presence of persistent current in cardiomyocytes isolated from the left ventricle of healthy individuals and those in heart failure using whole cell patch clamp. They reported that the persistent current amplitude was independent of peak sodium current and that it had maximal density at -30 mV. This represented the first demonstration of persistent current in human cardiomyocytes [30].

8 In a 2006 paper, Maltsev and Undrovinas characterized three components of sodium currents: transient mode (reflecting peak current), late scattered openings, and burst mode [31]. They also characterized the timing of each mode, with the first 40 ms of sodium current decay comprised of all three types, the intermediate range (40 to 300 ms) consisting of both late scattered openings and burst mode, and the slow mode (300 ms and beyond) solely consisting of late scattered openings [31]. They showed that transient currents were roughly 1000-fold larger than persistent current (50 nA vs 50 pA), but that the persistent current duration was significantly longer (2 s vs 2 ms) [31]. Because it is conducted for the duration of the action potential, the intracellular sodium accumulation due to “late” sodium current approximates that from “peak” sodium current [28].

Figure 4. Persistent current. (a) a Incomplete inactivation of sodium channel results in persistent current. (b) Sodium currents rapidly reach a peak value and subsequently inactivate within 2 milliseconds. A small non- inactivating component of current persists throughout the action potential. Alterations to the biophysical properties of Nav.1.5 – Persistent Current due to genetic, post-translational or b other modifications – can result in pathological increases in persistent current. Persistent current is sensitive to TTX. (b) adapted from [3]

An interesting feature of persistent current is its frequency dependence. A 2002 report on the long QT 3 (LQT3) mutation Y1759C showed that the magnitude of persistent current measured during voltage pulses was inversely related to the interval between pulses [32]. A study of cells from the canine heart similarly demonstrated an inverse relationship between pulse frequency and persistent current [33]. These findings 9 suggest a basis for the inverse relationship between heart rate and QT interval in LQT3 patients [34].

Persistent current in humans can be elevated due to genetic or acquired abnormalities. In 1995, Bennett et al. reported on an SCN5A mutation in which lysine

1505, proline 1506 and glutamine 1507 were deleted (delta KPQ) [35]. This mutation was identified in patients with hereditary long QT syndrome. When expressed in Xenopus laevis oocytes, the mutant channel did not completely inactivate during a 200 ms depolarization. In single channel analysis, the delta KPQ sodium channel experienced burst mode re-openings 20 ms after the depolarizing pulse [35]. Rivolta and Kass reported on an LQT3 mutation with a histidine substituted for tyrosine at amino acid

1795 in SCN5A (Y1795H) [36]. In whole cell patch clamp experiments of this mutant sodium channel, persistent current was elevated [36]. Veldkamp et al. similarly reported on a sodium channel mutation at amino acid 1795, 1795insD, in patients with a prolonged

QT interval [37]. Persistent current was elevated in 1795insD mutant channels expressed in HEK cells [37].

Mutations in proteins that interact with the sodium channel can also lead to increased persistent current. For example, mutations in caveolin 3 have been identified in patients with long QT syndrome. In HEK293 cells with stable expression of NaV1.5, the mutant CAV3 led to greater late current than did wild type CAV3 [38]. A subsequent 2007 study identified three CAV3 mutations in a cohort of patients who died from sudden infant death syndrome SIDS, which also led to increased persistent current when expressed in HEK cells with stable expression of NaV1.5 [39]. Mutations in alpha1- syntrophin (SNTA1) have also been linked to increased persistent current. One mutation

10 in SNTA1 was identified in a group of LQT patients who did not have mutations in any of the known ion channel candidates [40]. SNTA1 interacts with NaV1.5 and links neuronal nitric oxide synthase (nNOS) to PMCA4B, an inhibitor of nNOS. The identified mutant did not link nNOS with PMCA4b however, and led to increased nitrosylation of NaV1.5

[40]. When studied in HEK cells, the mutant SNTA1 led to increased persistent current compared to wild-type SNTA1, an effect that was reduced by NOS blockers [40]. In a follow-up study by Cheng et al., additional SNTA1 mutations were identified in SIDS patients, three of which led to increases in persistent current [41].

A number of studies have investigated the effects of post-translational modifications on persistent current. Many have focused on the role of CaMKII. Studying the effects of both acute and chronic over expression of CaMKII-deltac, Wagner et al showed that in both rabbit and mouse ventricular myocytes, late current was increased

[42]. They also showed that intracellular sodium accumulation increased with CaMKII- deltac overexpression. QT interval was prolonged in vivo and ventricular tachycardia

(VT) was induced more easily in mice with CaMKII-deltac over-expression as well [42].

The CaMKII inhibitor KN93 was able to prevent the increased persistent current and intracellular sodium in the case of acute CaMKII-deltac over-expression [42].

PI3-kinase signaling has similarly been shown to regulate persistent current. In patients with diabetes, QT interval is more frequently prolonged and PI3K signaling is perturbed [43]. Based on these findings, Lu et al. explored the role of PI3K signaling on cardiac myocytes derived from type 1 or 2 diabetes mouse models. In these myocytes,

APD was prolonged [43]. Expression of PI3K, which is down-regulated in diabetes, or its downstream mediator, PIP3 restored normal APD. Insulin similarly shortened the APD,

11 an effect not seen in cardiomyocytes treated with a PI3K inhibitor. Furthermore, myocytes from the diabetic mice had increased persistent current, which was returned to normal with both PI3K and PIP3 [43]. Treatment of the myocytes with the late current blocker mexiletine also reduced the APD prolongation [43]. Further evidence of the role of PI3K came from studies of the potassium channel blocker, dofetilide. Mouse cardiomyocytes exposed to dofetilide experienced greater than tenfold increases in persistent sodium current [44]. However, these increases were not seen when pipette solution included PI3P. In human induced pluripotent stem cell-derived cardiomyocytes as well as Chinese Hamster Ovary cells with expression of NaV1.5, dofetilide similarly increased persistent current, and PI3P blocked this increase [44].

Auxiliary Proteins

The NaV1.5 alpha subunit does not exist within cardiac membranes in isolation.

Rather, the channel is one constituent within a macromolecular complex, in which the primary pore-forming alpha subunit interacts with a variety of auxiliary proteins. In addition to the well-characterized beta subunits, many other proteins interact with

NaV1.5, impacting the channel in myriad ways. Broadly speaking, the actions of interacting auxiliary proteins can be fitted into three classes: trafficking and sub-cellular localization, post-translational modification, and alteration of biophysical properties [45].

Much has been learned regarding the NaV1.5 interactome, though for many of the individual proteins, further work is necessary to more clearly define the functional role.

Many of the auxiliary proteins bind to the large intracellular domains of the sodium channel, including the I-II, II-III, and III-IV linkers, as well as the C-terminal domain [45]. The C-terminal domain in particular, is a relative hotspot with numerous

12 interacting proteins, including syntrophins, dystrophin, protein tyrosine phosphatase, and synapse-associated protein 97 [45]. Nedd4-2, a ubiquitin-protein ligase, has been shown to bind and ubiquitinate NaV1.5 [46]. Allouis et al showed that 14-3-3n modulates biophysical properties of NaV1.5 in COS cells, and interacts with the I-II linker [47].

CaMKII(Delta)C, a serine/threonine kinase, similarly interacts with the I-II linker where it has been shown to modulate biophysical properties via phosphorylation as discussed in the persistent current section [42]. Prominently explored auxiliary proteins that interact with the II-III linker include Ankyrin G, which mediates anchoring and trafficking of

NaV1.5 to the membrane [48]. A variety of studies have similarly suggested that MOG1 interacts with NaV1.5. Experiments in HEK cells offered evidence that MOG1 assists in membrane trafficking, as both surface expression, and sodium current were increased when MOG1 was co-transfected with NaV1.5 [49].

Calmodulin

Calmodulin, which is a ubiquitous, 17-kDa protein, possesses four EF-hand motifs enabling it to function as a calcium sensor [50]. Calmodulin interacts with and modulates a variety of ion channels, such as calcium channels, sodium channels and ryanodine receptors. Calmodulin confers calcium-dependent inactivation for L-type calcium channels [51]. Calmodulin similarly interacts with voltage-gated sodium channels including NaV1.5, where it binds to the channel’s C-terminal domain [52]. A number of studies have indicated that calmodulin specifically interacts with the “IQ” motif located within the C-terminal domains of channels [53]. Mutations within or close to the IQ motif of NaV1.5 have been linked to cardiac arrhythmias, including LQTS [54].

However, despite the abundance of evidence suggesting calmodulin interacts with

13 NaV1.5, no clear consensus has emerged regarding the role of this interaction. The scientific literature is replete with conflicting reports of calmodulin’s effects on NaV1.5 function [45]. In 2001, Balser’s group reported that calmodulin binds to the “IQ” domain of NaV1.5, with the interaction “enhancing slow inactivation” [52]. The calmodulin interaction with the CTD of NaV1.5 was demonstrated using a yeast two-hybrid assay. In patch clamp experiments, Balser’s group showed that administration of a peptide acting as a calmodulin binding antagonist, right-shifted (depolarizing direction) NaV steady-state inactivation by 6 mV, only in the presence of 1 microM free calcium [52]. In a 2004 study, Kim et al. showed that in HEK cells, an IQ/AA substitution in NaV1.5 biochemically compromised calmodulin’s interaction with the channel, and electrophysiologically led to increased persistent current when expressed in HEK cells, with similar effects across differing calcium concentrations [53]. More recently and in contrast to the Balser work, the Yue group showed that calcium does not regulate NaV1.5, neither in heterologously expressed NaV1.5 nor in the native environment [55]. They also showed that calcium regulated NaV1.4 channels, and further showed that a chimeric

NaV1.5 channel with a swapped NaV1.4 CTD was in fact calcium regulated [55]. Other studies from the Yue group suggested that apocalmodulin’s binding to NaV1.4 increases channel open probability [56]. Despite the extensive research into calmodulin’s effects on the cardiac sodium channel, a definitive understanding of its role has not yet been attained, especially in its native environment, the cardiomyocyte.

Fibroblast growth factor homologous factors

Fibroblast growth factor homologous factors are FGF proteins that remain intracellular due to the absence of a signal secretion sequence [23]. A number of FHF

14 proteins bind to the NaV1.5 C-terminal domain [45]. Many studies have indicated that

FHF’s can modulate gating kinetics of NaV1.5 [45]. The first report of FHF’s interaction with NaV1.5 came from the Waxman group [57]. They showed that FHF1B, a member of the FGF family, binds the C-terminal domain of NaV1.5 and causes a hyperpolarizing shift in the voltage dependence of inactivation [57]. Furthermore, they reported on an

LQT3 mutation, D1790G, which led to loss of FHF1B interaction with NaV1.5, suggesting that loss of FHF interaction had a gain-of-function effect upon the channel

[57]. Mohammadi’s group showed that wild-type FHF2A and FHF2B caused a depolarizing shift in voltage dependence of inactivation but that mutant versions of these

FHF’s with reduced NaV binding affinity, caused no such shifts in V1/2 [58]. This finding was in contrast to the Waxman report in which FHF1B expression caused a hyperpolarizing shift in voltage dependence of inactivation, perhaps because Waxman’s studies were on NaV1.5 expressed in HEK cells whereas Mohammadi’s were on NaV1.2-

NaV1.4 and NaV1.7, the endogenous sodium channels expressed in a neuroblastoma cell line [55][56][59].

In a 2011 study, Wang et al. showed that FGF13 was the predominant FHF cardiac isoform in mouse ventricular myocytes [60]. They also demonstrated that FGF13 binds to NaV1.5 and co-localizes to it in the membrane of cardiomyocytes [60]. Knocking down FGF13 with shRNA led to a nearly 9 mV hyperpolarizing shift in steady state inactivation [60]. A more recent study showed that FGF13 increased window current by shifting steady state availability in a depolarizing direction, sped up transitions to the slow-inactivated state and most strikingly slowed recovery from inactivation [61]. In

2015, Mohler and colleagues described a family harboring an amino acid substitution in

15 SCN5A with cardiac phenotypes including Long QT syndrome, atrial fibrillation (AF),

PVC’s, as well as syncope [62]. They showed that H1849R resulted in decreased binding affinity of FGF with the NaV1.5 CTD [62]. This group also reported a rightward shift in the steady-state inactivation and prolongation of the time constant of inactivation of

NaV1.5 in HEK cell experiments, and a rightward shift in the steady-state inactivation in transfected cultured neonatal myocytes [62]. These gain of function findings in this sodium channel with decreased FGF binding affinity, are not consistent with earlier described examples in which FGF expression caused a rightward shift in steady-state inactivation [61]. One of our goals was to explore the effect of the H1849R mutation in adult cardiomyocytes.

Interesting structural work on the interaction of FGF, calmodulin and the NaV1.5

C-terminal domain has been carried out. In 2012, Pitt’s laboratory group reported on a ternary complex of the NaV-CTD in complex with FGF and apocalmodulin at a resolution of 2.2 angstroms [63]. The ternary complex exhibited 1:1:1 stoichiometry and its molecular weight was comparable to the sum of the three constituents. The NaV-CTD was shown to consist of a globular domain containing alpha helices 1-5, and a protruding segment (alpha helix 6) of 29 amino acids. FGF13’s binding surface was localized to the alpha 4 helix within the globular domain [63]. Two amino acids, His1849 and Asp1852 on this alpha 4 helix formed part of a critical interaction with FGF. Calmodulin’s interaction with the NaV-CTD predominantly involved the extended alpha 6 helix [63].

The C-lobe of calmodulin was shown to interact with Ile1908 and Gln1909, critical amino acids within the IQ motif [63].

16 In summary, a large volume of data from electrophysiological experiments on sodium channels to patient data, suggest that FGF interacts with and modulates sodium channel function, though the precise manner of this modulation is, as with calmodulin, not entirely settled.

Sodium Channel in Disease

Gain-of-function and loss-of-function effects

The function of sodium channels can be modified in a number of ways. At the genetic level, mutations in the SCN5A gene frequently result in sodium channels with altered biophysical properties. The altered properties tend to be classified as having either gain- or loss-of-function effects [64]. Gain-of-function mutations tend to cause increased persistent current, or increased window current. On the other hand, loss-of-function mutations often result in decreased peak current due to impaired membrane trafficking or reduced window current [64].

Regulation of sodium channels is also possible at the different stages leading from gene to channel expression at the membrane [64]. Therefore, abnormalities in transcription, translation or post-translational modifications can likewise alter sodium channel expression and function. Diseases such as LQT3, Brugada syndrome, cardiac conduction disease, sick sinus syndrome, dilated cardiomyopathy and AF often result from these dysfunctional channels [64].

In LQTS, the ventricular action potential duration is prolonged resulting in a lengthened QT interval on surface ECG’s. The LQTS can result in cardiac arrest and sudden death [65]. LQT3 syndrome is caused by gain-of-function mutations in SCN5A, often leading to increased persistent sodium current [65].

17 Loss of function mutations in SCN5A resulting in decreased peak sodium current have been linked to sick sinus syndrome (SSS) [65]. In SSS, patients experience sinus bradycardia as well as sinus arrest and atrial standstill, as a result of sinoatrial node dysfunction [23].

Brugada syndrome is another cardiac disease linked to loss-of-function mutations in SCN5A [23]. Patients with Brugada syndrome have increased risk of experiencing lethal ventricular arrhythmias, AF as well as structural heart disease. SCN5A mutations associated with Brugada syndrome can lead to production of truncated channels, channels with decreased conductance, or impaired channel trafficking to the membrane. SCN5A mutations associated with Brugada syndrome have also been shown to cause biophysical defects in the channel including faster inactivation and increased slow inactivation, changes which lower channel availability [23].

Mutations in SCN5A have also been linked to dilated cardiomyopathy [64]. In one family harboring a D1275N SCN5A amino acid mutation, individuals with this particular aspartic acid to asparagine substitution presented with ventricular dilatation [66].

Atrial Fibrillation

Cardiac Conduction

A human heart typically contracts between 60 -100 times every minute. These contractions progress through the heart in a highly synchronized manner, initiating in a region appropriately called the “pacemaker,” and electrically spreading through the atrial chambers to the ventricles. An ECG recording can capture this typical pattern with the classic P wave representing atrial contraction, the PR interval representing the spread of excitation from atria to ventricles and the QRS complex demonstrating ventricular

18 activation. In normal sinus rhythm, these recordings demonstrate a regular gap between subsequent P-waves and a consistent pattern in which P waves precede QRS complexes.

Electrical abnormalities within the heart however, can alter this state leading to abnormal spontaneous firing in regions outside the sinoatrial node. In AF, there is rapid, non-coordinated, heterogeneous atrial activation [67]. ECG recordings of AF show an absence of regular P waves (indicative of a loss of coordinated atrial contraction) with irregular QRS complexes indicative of an irregular ventricular contraction [67].

Morbidities and Mortality in Patients with Atrial Fibrillation

AF is the most commonly observed arrhythmia in the USA. About 6 million

Americans currently have AF and that number is expected to rise in the next 15 years due to aging of the population [68]. Based on the Framingham Heart Study, in men and women over 40 years, the lifetime risk for developing AF was 26 and 23 percent respectively [69]. For those with lone AF (no HF or MI), the lifetime risk was estimated as roughly 16 percent [69]. AF increases risk of stroke as well as mortality with data from

2009 implicating AF as the primary cause of over 15,000 deaths in the US. In addition to increased mortality, AF is associated with increased risk of stroke [69]. Amongst those

80 to 89 years old, nearly 24 percent of strokes are attributable to AF. In addition, AF increases the risk of dementia 2-fold. Due to the health burden it imposes, AF is very costly, with estimates putting the annual cost in the US at more than 25 billion dollars

[69].

Arrhythmia Basics

Atrial arrhythmias emerge as the result of triggered activity in the myocardium

(due to early or delayed afterdepolarizations), enhanced automaticity or re-entrant

19 activity (Figure 5) [70]. Early afterdepolarizations are seen in the context of action potential duration prolongation with delayed repolarization. Delayed afterdepolarizations are seen in the context of improper calcium handling. Reentrant activity is usually seen in atria that have undergone structural or electrical remodeling [70].

Early afterdepolarizations

Early afterdepolarizations (EADs), frequently a trigger of arrhythmias, are depolarizing stimuli that initiate during phase 2 or 3 of the cardiac action potential, leading to premature action potentials. Early afterdepolarizations are often caused by changes in membrane currents that increase the action potential duration [70] such as decreases in outward repolarizing potassium currents [71]. In the human heart, these can be the result of abnormally lowered currents through the slow delayed-rectifier channel

(IKs), the rapid delayed rectifier (IKr) or the ultrarapid delayed rectifier (IKur) [7, 70].

Similarly, increases in depolarizing currents can also lead to EADs. Increased persistent current through NaV1.5 or increased calcium currents through the L-type calcium channel often serve as the source of these depolarizing currents [70, 72]. In a normal action potential, calcium channels inactivate allowing outward potassium currents to repolarize the cardiac cell back to its resting membrane potential. However, with disturbances that prolong action potentials, there is increased potential for calcium channels to reactivate

[70, 73]. Reactivated calcium channels can then conduct sufficient inward current to initiate a premature action potential, referred to as an EAD [70, 73].

20 Delayed afterdepolarizations

Calcium-handling abnormalities can lead a to triggered activity through delayed- afterdepolarizations (DADs) [70, 74]. With

DADs, calcium is released from the b sarcoplasmic reticulum spontaneously, rather than in response to inward calcium currents.

These spontaneous SR Ca2+ release events result in increased depolarizing currents via the c sodium-calcium exchanger, which removes the excess calcium by bringing in three sodium ions

[70, 74]. This thus leads to net inward current (1 Figure 5. Triggered activity (EAD’s, DAD’s) and reentrant waves are Ca2+ removed, 3 Na+ brought in, net inward believed to initiate and drive cardiac arrhythmias. (a) EAD’s result from prolonged action potentials enabling current of +1 per exchange). If the depolarizing reactivation of L-type calcium channels. (b) DAD’s result from calcium leak from currents carried by the sodium-calcium the sarcoplasmic reticulum. (c) Reentrant waves emerge in atrial tissue with a exchanger reach threshold, action potentials are favorable substrate. triggered [70, 74]. As in the case of EAD’s, specific molecular phenomena favor DAD’s.

SR calcium leak can result from CaMKII phosphorylation of RyR2, from decreases in the

RyR2 stabilizing proteins FKB12.6 and junctophilin-2, and from phospholamban hyperphosphorylation, which leads to subsequent increased SR calcium load [70] [75]

[76].

21 Automaticity

When threshold is spontaneously reached in atrial cells prior to arrival of the next sinus beat from the SA node, automatic focal ectopic activity results [70].

Reentrant Activity

In addition to triggered activity, reentrant circuits can also act as AF drivers that maintain the arrhythmia [77]. Reentrant activity can be seen as both single and multiple circuits within the atria [77]. The two predominant models of reentry are the leading circle and spiral wave models [77, 78]. In the leading circle model of reentry, the wavefront and wavetail meet. Because depolarization is continuously spreading towards the center of the wave’s path of curvature, tissue inside of the leading circle is refractory according to this model [79]. Reentrant waves are often modeled as circular waves with wavelengths equal to the product of conduction velocity and refractory period [77, 78].

Reentrant activity is promoted by changes that enable reentrant waves to form within the relatively small size of the atria [80]. One such change is increased atrial size, as seen in atrial dilation [80]. Reduced wavelengths due to either decreased conduction velocity or refractory period also favor reentrant activity [80]. Rotors have also been explored as drivers of AF. In a rotor, there is no gap between the wavefront and wavetail. The tissue medial to the rotor’s path is also not refractory [79]. The point of convergence between the wavefront and wavetail is referred to as the singularity [79]. The two-dimensional representation of a rotor is referred to as a spiral wave. Two-dimensional reentry has been experimentally demonstrated in left ventricular epicardial muscle isolated from sheep

[81].

22 Triggered Activity in Veins

A seminal paper in the AF field was Haisaguerre’s 1998 investigation into the initiation of AF [82]. Studying patients with AF, multielectrode catheters were used to map the origin of the arrhythmia. In the patients where one point of initiation could be localized, 65 percent were localized to the pulmonary veins [82] suggesting that pulmonary veins represented a common source of AF initiation. Radiofrequency ablation was successful in preventing AF recurrence amongst this group [82].

CaMKII and Atrial Fibrillation

A number of animal studies have suggested that increased CaMKII activity could play a role in AF. Chelu et al. showed that via rapid atrial pacing, mice with a gain of function mutation in RyR2 demonstrated increased CaMKII phosphorylation of RyR2

[83]. The rapid atrial pacing led to an increased incidence of AF in these mice [83].

Inhibition of CaMKII phosphorylation of RyR2, either pharmacologically or genetically, prevented the pacing induced AF in these mice [83].

Genetics of Atrial Fibrillation

AF is known to be heritable. Many studies utilizing differing methods for investigating genetic risk have shown that the disease has a genetic component. The first report of familial AF in the medical literature was a 1943 New England Journal of

Medicine case report on two sets of brothers with AF [84]. More recently, a study using participants in the Framingham Heart Study, as well as their family members, demonstrated that there was a 1.4-fold increase in AF risk in those with familial AF [85].

A 1997 New England Journal of Medicine study by Ramon Brugada looking at a Spanish family with an autosomal dominant pattern of AF inheritance mapped the locus using

23 genetic-linkage analysis [86]. A 2003 Science paper detailed the first known genetic cause of familial AF, a gain-of-function mutation in KCNQ1, which codes the pore- forming subunit of the IKs channel [87]. Subsequent studies have identified further gain- of-function mutations in KCNQ1 amongst patients with AF [88]. Mutations in other potassium channel genes including KCNA5, KCND3 and KCNJ2 have also been identified amongst AF patients [88]. Most of these mutations have similarly been gain-of- function in nature, with resultant decreases in the atrial action potential duration, a change believed to promote reentrant waves via decreased refractory period [88].

However, loss-of-function potassium channel mutations have also been identified, which cause prolonged action potential duration, and susceptibility to EADs [89].

AF has also been linked to mutations in the voltage gated sodium channel.

Mutations in both the alpha subunit, as well as the beta subunits have been found in AF patients [88]. As with the potassium channel mutations, included amongst the identified sodium channel mutations are both loss-of-function and gain-of-function variants [88].

Mutations in non-ion channel genes, such as GJA5, NPPA, NKX2.5 and GATA4 have also been associated with AF. GJA5, which encodes the gap junction protein, connexin40 is of particular interest as loss-of-function mutations in this channel could reduce atrial conduction velocity, a condition that promotes re-entrant activity [88].

GWAS studies, in which the presence of single nucleotide polymorphisms (SNP) in individuals with and without a given disease is compared in order to locate potential disease causing loci, have been used to study the genetics of AF. A variety of potentially disease causing loci have been identified using GWAS studies. Amongst these, the greatest relative risk (1.64 fold) is seen in those with a particular SNP in chromosome

24 4q25 [88]. PITX2C maps to this genomic region. A study of 210 patients with idiopathic

AF showed that two of these patients had heterozygous mutations in PITX2C that were not found in healthy controls [90]. PITX2C regulates a number of cardiac embryonically expressed genes and functional studies of the mutant PITX2C proteins showed that they resulted in lower transcriptional activity [90].

Nav1.5 in Atrial Fibrillation

A number of studies have suggested that dysfunctional NaV1.5 can contribute to

AF pathogenesis. Amongst a cohort of 375 patients with AF, SCN5A variants were identified amongst 6 percent of patients [91]. Further evidence of a role for dysfunctional

NaV1.5 in AF came from a report of a three-generation kindred in which members with the Y1795C SCN5A mutation all had long QT intervals and all adults had paroxysmal AF

[92]. This report thus linked a gain of function SCN5A mutation with AF [92]. Another

2008 study looked at incidence of early onset (under 50 years of age) AF in patients with various LQTS [93]. In the LQT3 cohort, 1 out of the 59 patients had early onset AF due to an E1784K mutation, which in prior studies was shown to cause a gain-of-function effect on the sodium channel [93].

Further insights into the contributions of NaV1.5 on AF came from work by

Sossalla et al [94]. In atrial myocytes from patients in AF, there was 26 percent more persistent current than in atrial myocytes from those in sinus rhythm [94]. Furthermore, ranolazine, an inhibitor of persistent current, significantly decreased persistent current from AF derived myocytes. Studies on atrial trabeculae showed that ranolazine decreased proarrhythmic activity induced by exposure to high calcium concentrations or isoprenaline [94]. This study thus implicated increased persistent current as a component

25 of and potential contributor to AF pathophysiology [94].

A number of pharmacological studies also offered evidence of persistent current contributing to AF. In the MERLIN trial, 6,560 patients hospitalized with acute coronary syndrome were assigned to receive either ranolazine or placebo and were monitored for arrhythmia for 7 days [95]. Those receiving ranolazine had AF less often than those receiving placebo [95]. More recently, results from the Harmony trial supported the potential efficacy of ranolazine in treating AF [96]. In this trial, patients receiving ranolazine along with the multi-channel blocker dronedarone had a markedly lower AF burden. Those receiving neither ranolazine nor dronedarone or just one of the two did not have a reduction in AF burden [96].

Remodeling

Important in the pathogenesis of AF is remodeling, the structural and function changes occurring within atrial tissue to increase susceptibility to AF. AF is believed to progress from a sporadic, self-terminating form, termed paroxysmal, into persistent and subsequently permanent forms [77, 97, 98]. These transitions occur in part due to remodeling electrically through changes in ion channel and transporter expression and function, as well as structurally through increased fibrosis and atrial size [77, 99].

Electrical remodeling tends to result due to the increased calcium loading during tachycardia. These resultant changes can take the form of reduced L-type calcium channel expression or increased IK1 and IKAch currents [77, 100] [101]. These changes shorten action potential duration [100], a desirable result when the goal is to counteract the increased calcium loads. However these shorter action potential durations are favorable for re-entry, as they decrease the reentrant wavelengths [77, 102, 103].

26 Remodeling can also increase susceptibility to delayed afterdepolarizations by increasing calcium leak during diastole [104]. Fibrosis can be pro-arrhythmic because it tends to slow conduction velocity as well as disrupt electrical propagation through cardiac tissue

[77, 99, 105]. Remodeling of the autonomic nervous system can also contribute to AF

[106, 107]. Sympathetic activity can lead to hyperphosphorylation of RyR2, with subsequent increases in diastolic leak, thus promoting DAD activity [77, 108]. Increased vagal activity on the other hand, leads to increased IKAch currents, which shortens action potential duration, a condition favorable to reentrant pathways [77, 109].

Treatments for Atrial Fibrillation

Both pharmaceutical and surgical approaches are utilized in the treatment of AF.

The goals of pharmaceutical treatment are rhythm control, which entails restoration of normal sinus rhythm or rate control to reduce the rapid heart rates seen in AF. In a seminal study comparing the two approaches, only 23 percent of patients in the rhythm control arm were converted to sinus rhythm. Symptom improvement was seen in 60.8 and 55.1 percent of patients in the rate and rhythm control groups respectively [110].

Surgically, ablation of the source of ectopic activity can be effective. The “Updated

Worldwide Survey on the Methods, Efficacy, and Safety of Catheter Ablation for Human

Atrial Fibrillation” study reported successful outcomes without need for anti-arrhythmic drugs in 74.9 percent of patients with paroxysmal AF, 64.8 percent with persistent AF and 63.1 percent with permanent AF. In this study, 4.5 percent of patients experienced major complications [111]. As a result of the high rates of AF refractory to treatment seen across both pharmacological and surgical approaches, development of improved AF therapies is a major goal of electrophysiology research.

27 Small Animal Models of Atrial Fibrillation

Various small animal models have been developed and used in order to study AF.

However, many of these animal models have aspects, which limit their usefulness as models of human AF. In many cases, the AF recorded in animal models is seen only with burst pacing or premature extrastimuli. In these cases, it is not spontaneously present.

Furthermore, the duration of AF tends to be far shorter than that seen in humans. In these studies, AF is defined by a duration of at least 1 second [112]. These models can thus offer information on structural cardiac changes, which enable AF formation and propagation, but are less informative with respect to the triggers leading to AF initiation.

Here, I will describe a few of these models. In 2007, Adam et al reported a mouse model with spontaneous AF in mice with cardiac specific overexpression of RacET, which codes for the Rac1 GTPase protein [113]. By 16 months, 75 percent of these mice showed spontaneous AF. Additionally, these mice had increased fibrosis in the left atria, as well as cardiac hypertrophy and decreased fractional shortening [113]. A 2012 model reported by Choi et al, described a mouse with increased cardiac specific expression of

TGF-beta1 [114]. These mice had increased atrial fibrosis, shortened action potential durations with increased calcium transients, and AF that emerged following burst pacing

[114]. They also showed that the right atria were both more fibrotic and more frequently the source of triggered activity than the left atria, thus linking fibrosis severity with AF occurrence [114]. Furthermore, they reported that recurrence of AF could be prevented by administration of ryanodine at concentrations that block SR calcium release [114]. In

2015, Glukhov et al published data of a calsequestrin 2 knockout mouse model [115].

These mice showed increased fibrosis in the pacemaker complex, abnormal calcium

28 release, and reentrant AF in mice induced by burst pacing. The average AF duration following rapid pacing was 31 seconds [115]. Based on the D1275N SCN5A mutation identified in a family with dilated cardiomyopathy discussed above, a mouse with

D1275N SCN5A expression was developed and described by Watanabe et al [116]. These mice exhibited decreased conduction velocity, dilated cardiomyopathy, as well as ventricular tachycardia and AF in ECG’s recorded under light anesthesia [116]. Notably, fibrosis was absent. Cardiomyocytes from mice with the D1275N sodium channel showed reduced peak sodium currents, increased persistent sodium currents, as well as slowed time constant of inactivation [116]. In addition, action potential duration was prolonged in these mice due to the increased persistent sodium current [116].

Large Animal Models of Atrial Fibrillation

Larger animal models of AF have also been described. Carneiro et al reported on a subset of pigs in which AF was induced through administration of acetylcholine followed by epinephrine [117]. In 6 pigs tested, all developed AF using this protocol with an average burden of 12 minutes [117]. Application of GS-967, a selective blocker of the persistent component of sodium current, blocked AF induction in 5 of the 6 pigs tested.

This report thus suggested the feasibility of late current blockers as an AF therapeutic option [117]. In a dog model described by Nishida et al, in vivo optical mapping was performed on 40 dogs, 8 days after MI was induced [118]. Spontaneous atrial ectopy was increased in these animals, as was triggered activity. Following beta-adrenergic stimulation, spontaneous calcium sparks were seen more frequently, and NCX currents were increased [118]. Increased heterogeneity in border zone conduction was demonstrated. With burst pacing, long lasting AF was induced with an average duration

29 of 1146 seconds in the post MI dogs compared to 30 seconds in the control group [118].

While important insights have come from these and other AF animal models, further advances in understanding can be made from models with alternative molecular mechanisms and with phenotypic characteristics closer to those seen in human AF.

Dissertation Outline

There are two primary components to my dissertation. In my first aim, I investigate the effects of a gain of function mutation in the voltage-gated sodium channel

NaV1.5 on the electrophysiological and structural properties of the heart using a transgenic mouse model. We hypothesize that the F1759A mutant sodium channel, which confers increased persistent sodium current, is sufficient to cause electrophysiological and structural abnormalities in the heart. A combination of cellular electrophysiology, whole heart ex-vivo electrophysiology, and in-vivo phenotyping are utilized to address this question.

In my second aim, I focus on the effects of two auxiliary proteins-calmodulin and fibroblast homologous growth factors (FGF)- on the function of voltage gated sodium channels expressed in cardiomyocytes. This cardiac specific expression is of central importance as many ion channel findings from experiments in HEK cells have failed to be replicated when repeated in the channel’s native environment. In order to explore the roles of these proteins on channel function, whole cell patch clamp recording on isolated cardiomyocytes is utilized.

In the first part of this aim, we specifically explore the effects of calmodulin. We hypothesize that as in HEK cells, loss of calmodulin binding to voltage-gated sodium

30 channels in cardiac cells will lead to increased “late” or “persistent” current. To do this, we create mice with transgenic expression of IQ1908AA NaV1.5, a sodium channel harboring two mutated residues in the known calmodulin binding site. Our primary focus was determining whether loss of calmodulin binding increased the “late” or “persistent” component of sodium current as compared to our transgenic control.

In the second part of this aim, we explore the effects of FHF’s. We hypothesize that decreased FHF binding affinity to NaV1.5 in cardiomyocytes will cause a depolarizing shift in steady-state inactivation, based on prior studies exploring the role of

FHF’s. We further hypothesize that dual decreased FHF and calmodulin binding will lead to increased persistent current. As detailed in chapter 2, our investigation of calmodulin revealed that this protein does not protect against late current in cardiac cells. We hypothesize that FHF, which is missing in HEK cells, but is naturally present in cardiac cells, protects against late current in the cardiomyocyte. Two additional transgenic experimental lines were generated for this hypothesis, H1849R Nav1.5, a channel with a key FHF binding site mutated and H1849R / IQ1908-1909AA-Nav1.5, a channel with the key calmodulin and FHF binding sites mutated.

31 Materials and Methods

Mouse Models

All transgenic mice were generated through the CUMC Shared Research

Facilities Transgenic Mouse core. Plasmids used for pronuclear DNA microinjection were created in the laboratory.

F1759A

The F1759A line was generated by fusing human heart sodium channel α-subunit cDNA (hH1) ([119, 120]) to clone 26 vector containing the modified murine α-MHC, tetracycline-inducible promoter (“responder line”) [121, 122]. SCN5A was engineered to be lidocaine-resistant by introducing an Ala-substitution at position F1759, and a 3X

FLAG epitope was ligated in-frame to the N-terminus. The F1759A transgenic (TG) positive mice in a B6CBA/F2 hybrid background were bred with cardiac-specific rtTA mice in a FVB/N background (obtained via MMRRC) [123] to generate F1759A-dTG mice and single TG (rtTA-positive or F1759A-positive) littermate controls.

C374Y / C374Y-IQ1908-1909AA / C3374Y- H1849R / C374Y-H1852R/IQ1908-

1909AA

The four C374Y lines were generated by fusing human heart sodium channel α- subunit cDNA (hH1) [119] [120] to clone 26 vector containing the modified murine α-

MHC, tetracycline-inducible promoter (“responder line”) ([121, 122] ). SCN5A was engineered to be tetrodotoxin-sensitive by introducing a Tyr-substitution at position C374

32 ([124]), and a 3X FLAG epitope was ligated in-frame to the N-terminus. In three lines, additional amino-acid substitutions were made: Ala/Ala at position IQ1908-1909 to decrease calmodulin binding affinity; Arg at position H1852 to decrease FHF binding affinity; and both Ala/Ala at position IQ1908-1909 and Arg at position H1852 to decrease both calmodulin and FHF binding affinities respectively. The C374Y TG positive mice (from each of the four lines) in a B6CBA/F2 hybrid background were bred with cardiac-specific reverse tet-transactivator (rtTA) mice in a FVB/N background

(obtained via MMRRC) [123] to generate C374Y, C374Y –IQ1908-1909AA, C374Y-

H1852R, and C374Y-H1852R-IQ1908-1909AA - dTG mice and single TG (rtTA- positive or the respective C374Y-positive) littermate controls.

Mouse In-Vivo Characterization

Telemetry and ECG Analysis

For all telemetry experiments presented here, Dr. Elaine Wan performed implantations. Joseph Bayne, Elaine Wan and Jeffrey Abrams carried out analysis.

Telemetry devices (Data Sciences International, St. Paul, MN) model (ETA-F10) were implanted in 4-11 month old mice. Recordings were started 1-week after implantation.

PR, RR, QRS and QT intervals were measured manually using Ponemah 3 software.

Reviewers were blinded and independently assessed the burden of atrial and ventricular arrhythmias. SEA-0400 (Chemscene Chemicals), dissolved in DMSO, was administered via IP injection.

Anesthetized Four-lead ECG

33 Subcutaneous four-lead electrocardiograms of isoflurane-anesthetized mice were performed using EMKA ECG and recorded using Iox. EMKA ECGAuto, post-processing software was used for analysis. AF was defined as absence of P waves and irregular R-R intervals for >1 sec. Where noted, QTc was calculated as QT//(√RR).

Echocardiography

All echocardiography experiments presented here were conducted in tandem with

Drs. Richard Weinberg and Elaine Wan. Drs. Richard Weinberg and Elaine Wan, as well as Jeffrey Abrams independently performed analysis. Transthoracic echocardiography was performed on anesthetized mice using a VisualSonics Vevo 2100 high-resolution imaging system with a 30-mHz imaging transducer through the CUMC Shared Research

Facilities Small Animal Imaging Center. Left atrial diameter (LAD) was measured at end-systole in parasternal long axis view, and left ventricular diastolic dimension

(LVEDD) was measured at end-diastole in parasternal short axis views and modified 4 chamber views. Left ventricular ejection fraction (LVEF) was measured in parasternal, modified 4 chamber and short axis views using Simpson’s biplane formula.

Biochemistry/Molecular Biology

Western Blots

Cardiomyocytes were homogenized in a 1% Triton X-100 buffer containing (in mM): 50 Tris-HCl (pH7.4) 150 NaCl, 10 EDTA, 10 EGTA and protease inhibitors. The lysates were incubated on ice for 30 min, centrifuged at 14K rpm at 4o C for 10 min and supernatants collected. Proteins were size-fractionated on SDS-PAGE, transferred to nitrocellulose membranes and probed with anti-FLAG (HRP conjugated) (Sigma, catalog

34 A8592), anti-NaV1.5 (Alomone, catalog ASC-005) and anti-tubulin (Santa Cruz, catalog sc-12462-R) antibodies. Anti-Flag antibody was diluted 1000-fold in 5% milk in TBST, and incubated with the membrane for one hour or overnight at 4o C. For FLAG immunoblotting, secondary antibody was not necessary as the primary antibody was HRP conjugated. Six 5-minute washes in TBST were subsequently performed.

Anti-Nav1.5 antibody was also diluted 1000-fold in 5% milk in TBST and incubated with membrane for one hour or overnight at 4o C. Three 5-minute washes with

TBST were performed following incubation with primary antibody. Following washes, the membrane was incubated for one hour at room temperature with anti-goat HRP conjugated secondary antibody diluted 4000-fold in 5% milk in TBST. Anti-tubulin antibody detection followed the same steps as anti-Nav1.5 antibody, with the exception that primary antibody was diluted 500-fold in 5% milk in TBST. Prior to development, the membrane was washed with TBST for six 5-minute washes.

For all western blotting detection, Pierce ECL Western Blotting Substrate Kit was used. Membranes were incubated with kit reagents diluted 5-fold (200.0 µL of each reagent added to 800.0 µL DI H2O) for 1 minute. Detection and quantification was performed with a CCD camera (Carestream Imaging) and ImageQuant software respectively. Membranes were imaged for 20 seconds, 1 minute and 5 minutes.

Immunofluorescence

For immunofluorescence, cover slips were incubated with laminin diluted in

PBS fifty-fold for one hour at 37°C. After one hour, laminin solution was removed.

Isolated cardiomyocytes were subsequently placed on each chamber of the slide.

Cardiomyocytes were incubated on the slide for one hour at room temperature. After this

35 they were washed two times with PBS. Following these washes, cardiomyocytes were fixed for 15 min in 4% paraformaldehyde. Cells were then twice washed with 100 mM glycine in PBS. Cells were permeabilized with 0.5% Triton X-100 in PBS for 5 minutes at room temperature. Cells were then incubated with 10% v/v goat serum in PBS for one hour for blocking prior to antibody incubation. Indirect immunofluorescence was performed using a 1:200 rabbit anti-FLAG antibody (Sigma) incubated with cells overnight at 4°C. and 1:160 FITC-labeled goat-anti-rabbit antibody (Sigma) incubated with cells for 1 hour at room temperature. Images were acquired using a confocal microscope.

Quantitative real-time polymerase chain reaction (qRT-PCR)

PCR was performed using Applied Biosystems StepOne Plus PT-PCR system and inventoried following inventoried primers:mouse actin B: Mm01205647_g1; mouse

NaV1.5: Assay name: Mm01342518_m1; human (TG) NaV1.5 Assay name:

Hs00165693_m1(Applied Biosystems). A custom set of primers was created by Applied

Biosystems to detect the combined expression of mouse Scn5a and human SCN5A. PCR was performed, in duplicate, for 40 cycles with automated detection of crossing threshold. Results were presented as fold difference for each gene against actin using 2-

ΔΔCT, and then normalized to NTG mice.

Cellular electrophysiology

Whole Cell Patch Clamp

Both Alexander Katchman and Jeffrey Abrams performed the patch clamp experiments. Cardiomyocytes were isolated as described previously[125] from mice at

36 least 8 weeks of age. Prior to euthanasia, surface ECG was obtained. For F1759A-dTG mice, all were shown to be in AF prior to isolation. Experiments were performed at room temperature. Membrane currents from non-contracting rod-shaped cells with clear striations were measured by the whole-cell patch-clamp method using a MultiClamp

700B amplifier (Axon Instruments). The pipette resistance was 0.4-1.0 MΩ in order to minimize voltage clamp error. The liquid junction potential was corrected and series resistance was compensated up to 60%. The leak current was subtracted using a P/4 protocol.

For studies of F1759A-Nav1.5, the intracellular (pipette) solution contained (in mM): 5 NaCl, 15 CsCl, 115 CsF, 10 HEPES and 10 BAPTA, pH 7.4, titrated with CsOH.

For persistent Na+ current determinations, the bath solution contained (in mM): 100

NaCl, 45 TEA-Cl, 10 HEPES, 1 MgCl2, 1 CaCl2, 5 glucose, pH 7.4, titrated with CsOH.

The bath solution was then changed to reduce Na+ concentration to minimize Na+ current; in mM: 5mM NaCl, 140 TEA-Cl, 10 HEPES, 1 MgCl2, 1 CaCl2, 5 glucose, pH 7.4, titrated with CsOH. Lidocaine (3 mM) was superfused to determine the lidocaine- resistant current. Persistent current was evaluated by 190-ms depolarization from -100 mV to -30 mV in the absence and presence of 20 µM tetrodotoxin (TTX, Tocris) [126].

Digital subtraction of the time-averaged responses in the absence and presence of TTX yielded a small TTX-sensitive persistent Na+ current. The mean value of the last 10 ms of the 190-ms pulse was normalized to the peak Na+ current recorded using 5 mM Na+ in both intracellular and extracellular solutions.

For studies of the C374Y-Nav1.5 lines of mice, measurements were made as with

F1759A-Nav1.5 with minor exceptions. Intracellular (pipette) solution was changed from

37 5 to 3 mM NaCl. For peak current measurements, the bath solution was similarly changed from 5 to 3 mM NaCl. Persistent current was assessed using 50 µM ranolazine rather than 20 nM tetrodotoxin. Tetrodotoxin (20 nM) was superfused to determine the tetrodotoxin-sensitive current.

In order to ensure that cardiomyocytes were adequately voltage-clamped, two approaches were taken. For all I-V recordings, we looked for symmetrical current-voltage curves, indicative of a well-clamped cardiomyocyte. Additionally, we used large electrodes, with sizes ranging from 300 kilo-Ohms to 1 Mega-ohm, which also led to improved voltage clamping.

In many of our measurements, long recordings were utilized. In some cases, measurements lasted for many minutes with measurements in the presence of drugs- lidocaine, ranolazine, or TTX- occurring at later time points. To ensure that changes in measured parameters reflected the effects of these drugs, rather than changes in stability of the current, we often repeated the measurements in the absence of drugs after long washouts, to ensure that no changes related to altered stability had occurred.

Ca2+ transients

Myocytes were loaded with Fura-2/AM and intracellular Ca2+ transients were recorded with a dual-excitation fluorescence photomultiplier system (Ionoptix). In brief, cells were bathed in extracellular solution containing (in mM): 135 NaCl, 5.4 KCl, 1.8

CaCl2, 1 MgCl2, 10 glucose, 10 HEPES, pH 7.4. Myocytes were stimulated to contract at

1-Hz. For Ca2+ transient peak measurements, autofluorescence and background emissions were first subtracted and the peak was measured as the difference from the baseline using

IonWizard software (IonOptix) [127].

38 Epicardial optical voltage mapping of Langendorff-perfused hearts:

For all optical mapping experiments, Drs. Elaine Wan and Sergey Zakharov were instrumental in conducting experiments and analysis. High-resolution optical mapping experiments were performed on 4-12 month old F1759A-dTG and littermate control mice. Hearts were isolated and perfused by the Langendorff method with temperature controlled (37o C) oxygenated Krebs Henseleit buffer (pH 7.4). One AgCl wire was attached to the metal aortic cannula and another AgCl wire was positioned near the surface of the heart to record a pseudo-ECG. Cardiac contraction was inhibited with 5

µM blebbistatin. The right and left atria were splayed using Tungsten pins. The heart was then stained with the voltage sensitive dye Di4ANEPPS (8 µL of 2 mM stock solution dissolved in DMSO) and illuminated via epifluorescence. Fluorescence was acquired through a 715 nm long pass filter using a CMOS camera (MICAM Ultima, SciMedia) with 1 cm X 1 cm field of view. One to two-second recordings were captured at 1000 Hz.

All optical signals were processed with custom MATLAB software downloaded from I.

Efimov’s laboratory website (Washington University, St Louis, MO) and as previously described [128, 129].

Histology

Excised hearts were placed in 10% formalin and cut in coronal sections. Every other slice (10 µm) was stained with hematoxylin-eosin and Masson’s-Trichrome staining for fibrosis. Each slide was reviewed and photographed under light microscopy

20-400X. Fibrosis was evaluated as a ratio of total blue pixels:total myocardial area in

Masson’s trichrome stained slices, using cellSens imaging software (Olympus). Two

39 pathologists, who were blinded to the identity of the genotypes, independently reviewed the quantitative analyses.

Transmission electron microscopy

The hearts were fixed with 2.5% glutaraldehyde in 0.1 M Sorenson’s buffer (pH

7.2), and then post-fixed with 1% OsO4 in Sorenson’s buffer for 1-hour. After dehydration, the hearts were embedded in Lx-112 (Ladd Research Industries, Inc.). Thin sections (60 nm) were cut on a PT-XL ultramicrotome. The sections were stained with uranyl acetate and lead citrate and examined under a JEOL JEM-1200 EXII electron microscope. Images were captured with an ORCA-HR digital camera (Hamamatsu) and recorded with an AMT Image Capture Engine.

Statistical analyses: Data was analyzed using Origin, Microsoft excel, Prism and

Python. All statistical analyses were performed using built in functions of Prism 6.0 and

Microsoft Excel. Results are presented as mean ± SEM. For comparisons between two groups, unpaired two-tailed Student’s t test was used. For comparisons between more than two groups, Anova and multiple comparison tests were performed as described.

Differences were considered statistically significant at P < 0.05.

40 Chapter 1: Exploration of atrial fibrillation molecular mechanisms through use of transgenic mouse model

41 Introduction

Atrial fibrillation (AF), estimated to affect about 6 percent of Americans who are

65 years of age and older, doubles the risk of death, and accounts for 15-20% percent of all strokes [70]. It is the most frequent sustained arrhythmia observed in clinical practice.

Yet current treatments for AF are limited by the relatively low efficacy of pharmaceuticals and radiofrequency ablation/surgery, high rates of recurrence, and poor safety of certain anti-arrhythmic drugs [70]. As a result, improved therapeutic strategies and options for AF remain a major goal of cardiac electrophysiology research efforts.

A better understanding of the molecular mechanisms leading to this arrhythmia is one tool that can contribute to this effort. Mouse models that recapitulate features of a human disease represent one avenue towards obtaining molecular mechanistic insights.

Unfortunately, studies of AF have been hindered by the lack of a mouse model that accurately recapitulates the spontaneous initiation and prolonged periods of AF observed in humans. Most, if not all studies use non-physiological methods including high- frequency burst pacing to induce very short episodes in mice, typically lasting several seconds. This is in contrast to AF as seen in humans, which can last for hours, days or longer.

This project initially began with the intention of developing a novel methodology to study the effects of human Na+ channel mutations in the context of a cardiomyocyte

(see Chapter 2). Our strategy was to express mutant human NaV1.5 channels using a doxycycline-inducible system, thereby controlling the amount of expression, and selectively distinguish between endogenous channels and mutant channels by introducing lidocaine resistance via a mutation (F1759A) in the local anesthetic binding site.

42 Although an F1759A-NaV1.5 TG mouse was created as our control, the F1759A-NaV1.5

TG line, when crossed with the αMHC-rtTA driver line, was found to have increased persistent current due to the F1759A mutation. The increased persistent current was sufficient to cause both structural remodeling of the atria, and electrophysiological dysfunctions including spontaneous and prolonged episodes of AF, and became the focus of my first project in the laboratory, the details of which are presented below.

Experimental Design and Approach

We generated TG mice with doxycycline-inducible and titratable, cardiac-specific expression of FLAG-epitope-tagged human NaV1.5 with a mutation (F1759A) in the local anesthetic binding site [130], which causes increased window current and persistent

Na+ current. We found that two founder lines had doxycycline-independent low expression of the mutant Na+ channels. Mice derived from these “leaky” expression founders served as the experimental group of this study. For cellular electrophysiological characterization, whole cell patch clamp of isolated atrial and ventricular cardiomyocytes was performed. For characterization of the phenotype of the TG mice, a variety of parameters were monitored including atrial diameter and ventricular ejection fraction via transthoracic echocardiography and presence and duration of AF and premature ventricular contractions via limb-lead ECG and telemetry. Histological changes within the atria including levels of fibrosis and atrial area, were assessed via H&E and Masson’s trichrome staining. Our control group consisted of single-TG and non-TG (NTG) littermate mice.

43 Results

Expression of F1759A-NaV1.5 in atria and ventricles

In generating TG mice with mutant sodium channels, we utilized a Tet-On strategy.

In this system, expression of our “gene of interest,” F1759A- human NaV1.5 is driven by co-expression of the “driver” gene reverse tet-transactivator (see methods section for details). The driver protein generally requires the presence of doxycycline in order to bind to the promoter of the “gene of interest” and thereby turn on its transcription.

Notably however, “leaky” or doxycycline-independent transcription of the “gene of interest” does occur [123].

We initially created four lines of F1759A-NaV1.5 TG mice. After these mice were crossed with α-MHC-rtTA mice (driver gene) [123], double TG (dTG) mice with doxycycline-regulated cardiac-specific expression of FLAG-F1759A-NaV1.5 were created (Figure 1a). Of the four lines of FLAG-F1759A-NaV1.5 that were initially established, two distinct founder lines were of particular interest to us because offspring with both the F1759A and rtTA transgenes developed a cardiomyopathy and spontaneous prolonged periods of AF in the absence of doxycycline treatment.

One concern with TG mice is that the emergent phenotype results not from properties of the transgene itself, but rather from the genomic location in which it inserts. If the phenotype was secondary to the genomic location in which either of the two transgenes inserted, then the phenotype would be seen in at least one line of single-TG mice and would have been independent of actual expression of the transgene. However, littermate mice with single transgenes, either rtTA-positive or F1759A-positive, were indistinguishable from each other and from NTG mice in terms of the lack of a cardiac

44 phenotype and F1759A transgene expression. Thus, random insertion of the NaV1.5 gene could not be the cause of the phenotypes, including AF, because the F1759A-NaV1.5 mice without rtTA failed to demonstrate any phenotype. As a result, rtTA-positive,

F1759A-positive, and NTG mice were used interchangeably as the littermate control group.

Figure 1. Cardiac-specific, FLAG-tagged F1759A NaV1.5-expressing transgenic mice. (a) Schematic of the binary transgene system. αMHC-rtTA is the standard cardiac-specific, reverse tetracycline-controlled transactivator system. The αMHCMOD construct is a modified αMHC promoter containing the tet-operon for regulated expression of FLAG-tagged, lidocaine-resistant NaV1.5. (b) Quantification of mouse Scn5a and human SCN5A transcripts in ventricle and atrium of F1759A-dTG mice using qRT-PCR. Results are presented as fold difference for each gene against actin by use of 2-ΔΔCT method, followed by normalization to expression of either mouse Scn5a or mouse Scn5a and human SCN5A determined in NTG mice. N= 3 NTG and N=4 F1759A-dTG mice in each group. (c) Immunostaining of F1759A-dTG mice cardiomyocytes with or without anti-FLAG antibody and FITC-conjugated secondary antibody. Images were obtained with confocal at X40 magnification. (d-e) Anti-FLAG antibody, anti-NaV1.5 antibody and anti-tubulin antibody immunoblots of cardiac homogenates showing FLAG-epitope-tagged NaV1.5 expression in ventricle (d) and atrium (e) of single and double TG (dTG) mice in the absence of doxycycline-impregnated food.

45 Expression of the FLAG-F1759A NaV1.5 channels in the atria and ventricles was dependent upon cardiac-specific co-expression of rtTA and did not require doxycycline administration (Figure 1b-e). Doxycycline, if administered, did increase expression, but this was unnecessary for our studies, as a unique cardiac phenotype of a cardiomyopathy and spontaneous prolonged episodes of AF was apparent in the dTG mice in the absence of doxycycline, but never in the single TG or NTG mice. Despite an approximate two- fold increase in the amount of total (TG human + native) NaV1.5 transcript in the dTG compared to NTG or single TG littermate mice (Figure 1b) assessed with quantitative real-time PCR, the combined expression of native and TG NaV1.5 protein was nearly equivalent, assessed semi-quantitatively with an anti-NaV1.5 antibody that recognizes both mouse and human NaV1.5 (Figure 1d-e), and quantitatively using cellular electrophysiology (Figure 3c). It is possible that the discrepancy represents post- transcriptional regulation, or alternatively, that our real-time PCR experiments falsely reported total Na+ channel expression due to the presence of contaminant genomic DNA.

Real-time PCR probes are designed to span exons so that genomic DNA contaminants are not amplified. As our transgene lacked introns however, genomic DNA contaminants would in fact increase the signal in our dTG mice, but not in single TG rtTA or NTG mice.

In assessing mean peak NaV1.5 currents, INa in isolated atrial and ventricular cells, we utilized 5 mM Na+ rather than physiologic Na+ in the extracellular solution to mitigate voltage clamp errors. The F1759A substitution markedly diminishes use-dependent lidocaine block[131], enabling the differentiation of the electrophysiological profiles of endogenous and mutant-TG Na+ channels in cardiomyocytes. We selected 3 mM as the

46 optimal concentration of lidocaine based upon 96% ±

0.5% block in wild-type NaV1.5 and 65.2% ± 1.6% block in heterologously expressed F1759A NaV1.5 (Figure 2).

Measurement of the peak current before and after application of lidocaine, thus enabled us to determine the relative levels of transgenic and endogenous current (See methods). Consistent with the western blotting data, mean peak, INa in isolated atrial and ventricular cells was not significantly different in F1759A-dTG mice compared to Figure 2. Lidocaine control littermates (Figure 3a-b insets and Figure 3c). The resistance in F1759A- NaV1.5. 3 mM lidocaine relatively low protein expression of FLAG-F1759A- blocked 96% ± 0.5% of INa in wild-type cardiomyocytes as well as NaV1.5 in both atrial and ventricular cardiomyocytes in in heterologously expressed WT-NaV1.5 but the absence of doxycycline is likely due to a low basal only 65.2% ± 1.6% of heterologously expressed F1759A-NaV1.5. level of rtTA protein binding to the Tet operator sequences in the absence of doxycycline, thereby driving transcription in some cells

[123]. It is likely that the presence of TG NaV1.5 resulted in downregulation of endogenous NaV1.5 in dTG animals, thus leading to near normal total NaV1.5. Real-time

+ PCR for endogenous Na channel showed a down regulation of endogenous NaV1.5 in dTG compared to control mice (Figure 1b).

F1759A Mutation Causes Persistent Na+ Current

Greater than 5% of peak Na+ current remained after exposure to 3 mM lidocaine in more than 60% of atrial and ventricular cardiomyocytes isolated from F1759A-dTG

47 mice (Figure 3a-b, d). This amount of residual lidocaine-resistant sodium current was not observed in NTG or TG mice that were either rtTA-positive or F1759A-positive.

The F1759A mutation prevented complete inactivation of NaV1.5, thereby increasing persistent Na+ current in ventricular (Figure 3a, e) and atrial (Figure 3b, e) cardiomyocytes isolated from F1759A-dTG mice, as was previously reported when this mutation was studied in heterologously expressing HEK cells [132]. The V50 for activation of NaV1.5 in F1759A-dTG ventricular cardiomyocytes was shifted in a hyperpolarizing direction by 3.3 mV (P <0.05, t-test; N= 23 control; N= 13 dTG) and inactivation was shifted in the depolarizing direction by 1.9 mV (P =0.07, t-test) leading to enhanced window current.

+ Figure 3. F1759A-NaV1.5 increases persistent Na current in atria and ventricles. (a-b) Exemplar + whole cell Na current (INa) traces of ventricular (A) and atrial (B) cardiomyocytes isolated from control (CONT- single TG and NTG) and F1759A-dTG mice. Persistent INa was evaluated with a 190-ms depolarization from a holding potential of -110 mV to -30mV in the absence (black) and presence (green) of 20 µM TTX, and 5 mM Na+ was used in the intracellular solution and 100 mM Na+ was used in the extracellular solution. Inset: For the assessment of peak INa and the fraction of lidocaine-resistant current, whole cell current traces were recorded with 5 mM Na+ in both extracellular and intracellular solutions, in the absence (black) and presence (red) of 3 mM lidocaine. (c) Bar graph of peak INa density recorded with 5 + mM external Na . Data are mean + SEM. P= NS, t-test. (d) Graph of fraction of peak INa resistant to 3 mM lidocaine. *** P <0.001; **** P < 0.0001, t-test. (e) Graph of persistent INa. Red dashed line is maximal persistent INa in cardiomyocytes isolated from single and NTG (CONT) mice. Mean + SEM. ** P <0.01; *** P <0.001, t-test.

To rule out post-translational modifications as a cause of the persistent Na+ current, we showed that the persistent current was resistant to ranolazine. We also observed that after normalization to the amount of peak Na+ current, persistent current correlated with the amount of lidocaine-resistant current in atrial and ventricular cells (Figure 4). These

+ results implied that the persistent Na current was solely due to the TG F1759A-NaV1.5 channels - which are resistant to ranolazine and lidocaine, unlike endogenous NaV1.5 channels - and not due to post-translational modifications of endogenous NaV1.5 channels.

Figure 4. Persistent current is due to F1759A transgenic channels (a) Exemplar whole cell Na+ current traces of ventricular cardiomyocyte isolated from F1759A-dTG mice. Persistent Na+ current was evaluated with a 190-ms depolarization from a holding potential of -110 mV to -30 mV in the absence (black) and presence (red) of 50 microM ranolazine. The intracellular solution contained 5 mM Na+ and the extracellular solution contained 100 mM Na+. Persistent Na+ current is resistant to 50 microM ranolazine in the F1759A-dTG mice. (b-c) The ratio of persistent Na+ current to peak Na+ current correlated with the ratio of lidocaine-resistant peak to total peak Na+ current in atrial and ventricular cardiomyocytes isolated from F1759A dTG mice. Y-axis is ratio of persistent Na+ current determined using 100 mM Na+ in the extracellular solution to peak Na+ current determined using 5 mM Na+ in the extracellular solution. See methods. The vertical red dotted line corresponds to 5% residual peak Na+ current after exposure to 3 mM lidocaine, representing the upper limit of lidocaine-resistant current observed in control (NTG and single TG) mice. Ventricle: Pearson r = 0.70, P<0.0001; N = 31. Atrium: Pearson r = 0.82, P<0.0001; N = 30.

Increased Na+ influx is sufficient to cause atrial and ventricular cardiomyopathy

+ + Elevated [Na ]i, caused by increased persistent Na current, is a hallmark of both animal models and human heart failure[30, 133, 134]. In humans, some gain-of-function

SCN5A mutations are associated with familial dilated cardiomyopathy. It is not known, however, whether increased persistent Na+ current is sufficient to cause a cardiomyopathy. We found that the atria and ventricles of F1759A-dTG mice progressively enlarged over several months post-birth (Figures 5 & 6). We serially performed transthoracic echocardiograms of F1759A-dTG and littermate control mice. At

20-30 days of age, the left ventricular (LV) ejection fraction was reduced non- significantly by 14% compared to controls. Over the ensuing ~2 months, the LV ejection fraction modestly decreased, with a 27% and 35% relative reduction at 41-50 days and

51-80 days respectively (Figure 5b). The reduction in LV ejection fraction is likely multifactorial, due to both intrinsic contractility defects and secondarily due to a tachycardia-induced cardiomyopathy. Furthermore, there are inherent difficulties in assessing LV function in the setting of both AF and ventricular arrhythmias. Atrial remodeling was detectable by echocardiography at very early stages post-birth. The left atrial size was increased by 48% compared to controls at 20-30 days post-birth, which persisted through at least 3 months of age (Figure 5c). Histopathological studies at 3-4 months of age demonstrated atrial hypertrophy, an increase of right and left atrial areas by 52% and 54% respectively (Figures 5a, 6b), increased fibrosis (Figure 6c) and myocyte disarray (Figure 6b-c) in the F1759A-dTG mice.

Figure 5. Functional and Structural changes in atria and ventricle (a) Photographs of littermate control (CONT) and F1759A-dTG hearts at 4 weeks and 3.5 months post-birth. (b-c) Graphs of ejection fraction and left atrial (LA) diameter (in mm) of littermate control and F1759A-dTG mice, assessed by echocardiography at ages shown. Mean + SEM. N > 5 mice per age group per genotype.

Transmission electron microscopy (TEM) images from mice at 6- and 12-weeks-old also demonstrated atrial and ventricular structural remodeling in the F1759A-dTG mice

(Figure 6d). At 6-weeks of age, mild disarray of the myofibrils was noted with increased collagen deposition in the atria, which progressed along with glycogen deposition at 12- weeks of age. These changes mimic the findings observed in large animal models of AF and patients with AF. In the ventricle at both 6-and 12-weeks of age, the diameter of T- tubules (Figure 6d, red arrows) was increased by 2.5-fold which has been observed in other animal models of heart failure [135]. By 12-weeks of age, atrial and ventricular cardiomyocytes from the F1759A-dTG mice demonstrated mitochondrial injury, with circular and swollen mitochondria and ruptured outer membranes (Figure 6d). These findings establish that increased persistent Na+ current is sufficient to cause progressive structural changes in the atria and ventricles in vivo, including mitochondrial injury,

+ 2+ potentially because elevation of [Na ]i accelerates mitochondrial Ca efflux and promotes reactive oxygen species formation and oxidative stress [136, 137].

Figure 6 Histological changes in atria and ventricle. (a) H&E-stained cross-section of ventricle. Scale bars are 1 mm. (b) H&E-stain of right and left atria showing bi-atrial enlargement in F1759A-dTG mice. Bar graph of left atrial and right atrial area of littermate control and F1759A-dTG mice at 3-4 months of age. Mean + SEM; * P <0.05, t-test. N= 5 for each group. (c) Masson’s trichrome stain of atria of littermate control and F1759A-dTG mice. Scale bars are 50 µm. Bar graph quantifying atrial fibrosis. Mean + SEM; * P <0.05, t-test. N >5 for each group. (d) Representative two-dimensional TEM images (n=2 for dTG and littermate control) showing left to right, gross morphology of atrial and ventricular samples at 6 weeks (first row) and 12 weeks (second and third rows) in littermate control mice compared to dTG mice. At 6 weeks, dTG mice show signs of myofibril disarray with loss of congruous parallel myofibrils in the atrium. By 12-weeks of age, atrial and ventricular cardiomyocytes from the F1759A-dTG mice demonstrated mitochondrial injury, with circular and swollen mitochondria and ruptured outer membranes In the ventricle, the red arrows point to T-tubule cross-sections, which are larger in the dTG mice compared to littermate control. Scale bars are 500 nm.

52 Spontaneous atrial fibrillation in F1759A-dTG mice

Extensive electrocardiographic analyses were performed to explore both the frequency and mechanisms of atrial and ventricular arrhythmias (Figure 7). QT intervals

(measured as the time from start of a QRS complex to end of a T wave) were prolonged in F1759A-dTG mice (Figure 7a-d) both when recorded in mice under isoflurane anesthesia with limb-lead ECG’s as well as in free-roaming mice with implantable telemeters, consistent with the increased persistent Na+ current leading to lengthening of the action potential duration (APD). The QT interval, in contrast, was normal in both single TG littermate genotypes (rtTA or F1759A) and NTG mice.

In addition to QT prolongation, a markedly increased frequency of premature ventricular complexes (PVCs) and non-sustained polymorphic ventricular tachycardia

(VT) was observed in F1759A-dTG mice by surface limb-lead ECG and in non- anesthetized F1759A-dTG mice using telemetry (Figure 7g, upper tracing). Ventricular arrhythmias were never observed in the single TG or NTG mice.

In the two independent lines of F1759A-dTG mice, spontaneous AF, which was identified in anesthetized mice during a 3-minute recording period of a surface limb-lead

ECG (Figure 7e), was found as early as 5-6 weeks of age and by 10-weeks of age in greater than 80% of F1759A-dTG mice (Figure 7f). AF was present in many mice for the entire 3-minute recording period of a surface limb-lead ECG (Figure 7e, upper tracing) but in others, the AF was paroxysmal (Figure 7e, lower tracing). AF was never observed in NTG and single TG mice at any time-point (Figure 7f).

Figure 7. Prolonged QT interval and spontaneous atrial fibrillation in F1759A-dTG mice. (a) Representative limb-lead surface electrocardiograms of isoflurane-anesthetized littermate control and F1759A-dTG mice. The QT interval is marked by brackets. (b) Bar graph of R-R, PR and QT intervals from isoflurane-anesthetized littermate control mouse and F1759A-dTG mouse. Mean + SEM. ****P <0.0001, t-test. (c) Representative telemetry recordings of non-anesthetized littermate control and F1759A- dTG mice. (d) Graph comparing R-R, PR and QT intervals of non-anesthetized littermate control and F1759A-dTG mice from telemetric ECG recordings. **** P <0.0001, t-test. (e) Representative surface electrograms of isoflurane-anesthetized F1759A-dTG mice showing AF. In the lower trace, paroxysmal AF was recorded, with P waves at the right side of the recording. (f) Bar graph showing incidence of AF at range of days post-birth. Surface electrograms of isoflurane-anesthetized mice were recorded for 3 min. AF was considered present if the duration of AF persisted for >10 ventricular complexes. Control (single TG + NTG): N= 27, 21-40 days; N= 8, 41-60 days; N= 5, 61-80 days. F1759A-dTG: N= 26, 21-40 days; N= 23, 41-60 days; N= 17, 61-80 days; N= 18, 81-100; N= 9, >100 days. Each age group consisted of a distinct set of mice. (g) Representative telemetry recordings of non-anesthetized F1759A-dTG mice showing ventricular tachycardia (VT, upper tracing) and AF (lower tracing). (h) Graph showing percent of AF during 20-hour recording in control and F1759A-dTG mice. Control, N=3 mice; F1759A-dTG, N= 8 of which 5 were male and 3 were female.

Since slowing of the heart rate under anesthesia could provoke atrial arrhythmias, especially due to the inverse relationship of heart rate and persistent Na+ current, we implanted subcutaneous ECG telemeters. Spontaneous AF in non-anesthetized mice

(Figure 7g, lower tracing) was detected in all of the 11 F1759A-dTG mice with implantable ECG telemeters, indicating a high penetrance. The average AF burden was

37 + 10% during a 20-hour period (Figure 7h) and the longest continuous episode of AF was 1 hour, 52 minutes. Spontaneous AF was never observed in NTG or single TG littermates implanted with ECG telemeters (N=10).

54 These findings are in marked contrast to the vast majority of previously reported mouse models of AF, including the delta-KPQ Na+ channel knock-in mice, in which atrial arrhythmias could only be elicited by very aggressive pacing and was not sustained for longer than several seconds [72, 112, 138]. Our results imply that increased persistent

Na+ current is sufficient to alter the substrate and serve as a triggering factor for the initiation and perpetuation of AF.

Furthermore, epicardial surface optical voltage mapping of the anterior surface of explanted Langendorff-perfused hearts of 12 F1759A-dTG mice and 10 single TG and

NTG littermate control hearts was performed. In 9 of the 12 F1759A-dTG mice, spontaneous AF was present for the duration of ex-vivo experiments. The other 3

F1759A-dTG mice were in paroxysmal AF. Amongst those 3 F1759A-dTG mice, spontaneous AF was rarely interrupted by brief bursts of sinus rhythm, enabling us to compare the APD of F1759A-dTG mice to littermate controls. The APD80 of the

F1759A-dTG mice was prolonged 4-fold in the right atrium and 2.5-fold in the left atrium (Figure 8a-b), consistent with a gain-of-function Na+ channel phenotype.

Increased heterogeneity of the APD was apparent in both the right and left atria (Figure

8c-d), likely caused by the variable expression of the F1759A Na+ channels in atrial cardiomyocytes, as observed in the cellular electrophysiological studies.

Figure 8. Optically recorded atrial action potentials and voltage maps of AF. (a-b) Optically-recorded action potentials from right atrium (RA) and left atrium (LA) of control and F1759A-dTG mice in sinus rhythm. Bar graphs showing right and left atrial APD80 of control mice (N=10 mice) and F1759A-dTG mice (N=3 mice). Mean + SEM, **** P < 0.0001, t-test. The mean atrial rate in sinus rhythm: NTG mice - 338 + 9.5 beats per minute; F1759A-dTG mice - 314 + 9.8 beats per minute. (c-d) Representative APD regional maps and all point histograms of control and F1759A-dTG mice showing increased APD80 dispersion in the RA and LA of F1759A-dTG mice. Color legend of APD80 (ms) is shown to the right of each optical map.

56 NCX inhibition reduces frequency of atrial and ventricular arrhythmias

The relatively low efficacy of pharmaceuticals and radiofrequency ablation/surgery, and high rates of recurrence have plagued clinical treatment of AF. The F1759A-dTG mice have the potential to offer a means to test novel therapeutic approaches, either using pharmacological tools or genetically altered mice. The mechanism for the increased persistent Na+ current in AF patients [94] is not known. It may be due to increased

NaV1.1 expression or due to increased CaMKII phosphorylation or oxidative stress, both of which are known to increase persistent Na+ current [42, 139-141]. A large clinical trial

(MERLIN-TIMI) showed ranolazine, an inhibitor of persistent current, may reduce the frequency of paroxysmal AF in patients with acute coronary syndrome [95]. F1759A-

Na+ channels are resistant to therapeutic blood levels of ranolazine (Figure 4).

Prior studies have demonstrated that a relatively specific inhibitor of the Na+-Ca2+ exchanger (NCX), SEA-0400, suppressed dofetilide-induced torsade de pointes (TdP) in anesthetized dogs [142] and in Langendorff-perfused rabbit hearts with drug and

2+ hypokalemia-induced models of LQT2 and LQT3 [143], possibly by reducing [Ca ]i.

The efficacy of NCX inhibitors in AF, however, is unknown. We determined the fraction of time in AF vs. sinus rhythm during a 20-hour period before and after a single 0.4 mg/kg IP injection of SEA-0400, by manually scanning the entire ECG record. We found that SEA-0400 markedly reduced the fraction of AF over the 20-hour period in 4 of 5 mice (Figure 9B).

The mean reduction in AF burden during the 20-hour recording period for the four responding mice was 69% and for all five mice was 59% (Figure 9a). The single IP injection of SEA-0400 also markedly reduced the frequency of premature ventricular

57 complexes by 76% (Figure 9b), without affecting the QT interval (Figure 9c). Thus, a single injection of SEA0400 markedly reduced AF and ventricular ectopy in vivo, without lengthening the QT interval, which frequently limits administration of many anti- arrhythmic drugs including ranolazine.

Figure 9. Inhibition of Na+-Ca2+ exchanger (NCX) attenuates atrial and ventricular arrhythmogenesis in F1759A-dTG mice. (a) Graph (left) summarizing fraction of AF during 20-hour period before (black bars) and after (blue bars) single IP injection of SEA-0400 for each mouse. Percent change is shown above each data pair. Graph (right) showing fraction AF/hour pre- and post-SEA injection for the five F1759A-dTG mice. Mean + SEM. * P< 0.05, t-test. (b) Quantification of number of PVC normalized to pre-SEA. Mean + SEM. * P< 0.05, t-test. (c) Bar graph showing QT interval before and after SEA-0400.

58 Discussion

Gain-of-function Na+ channel leads to AF

Our data show that gain-of-function Na+ channel perturbations are sufficient to cause structural and functional changes within the atria and ventricles of mice. This is consistent with emerging observations that persistent Na+ current plays an important role in progression of heart failure [42]. Both gain-of-function and loss-of-function Na+ channel dysfunctions are associated with familial dilated cardiomyopathy, and persistent

Na+ current is associated with acquired diastolic and systolic dysfunction and electrical instability. For instance, an R222Q SCN5A gain-of-function variant, which left-shifted the steady state parameters of activation and inactivation, caused reversible ventricular ectopy and a dilated cardiomyopathy, which were substantially reversed with amiodarone or flecainide [144]. The F1759A-dTG mice mimic many of these structural and functional abnormalities including cardiac enlargement and ventricular dysfunction, and our studies reveal, at least in part, the downstream mechanisms by which gain-of-function

SCN5A mutants cause a dilated cardiomyopathy and substrates for both atrial and ventricular arrhythmogenesis. Inhibiting reverse mode Na+-Ca2+ exchange acutely reduced the burden of both spontaneous atrial and ventricular arrhythmias, identifying a potentially new pharmacological approach for treatment of AF in humans.

It is likely that the primary effects of incomplete NaV1.5 inactivation on cardiomyocyte electrophysiology, namely prolongation and dispersion of the APD, and the secondary downstream effects on chamber enlargement, fibrosis and mitochondrial injury/reactive oxygen species (ROS) synergistically cause the unique phenotype of spontaneous and prolonged episodes of AF in mice, mimicking human disease. In the

59 vast majority of previously reported mouse models of AF, atrial arrhythmias could only be elicited by very aggressive burst pacing, suggesting that whereas there may be a substrate for atrial arrhythmia, this can be well tolerated and undetected in the absence of a triggering factor [112]. AF is defined in these studies by a duration of at least 1 sec and most previously reported mouse models demonstrated these relatively short episodes of

AF [112]. Conversely, the presence of a triggering factor but absence of structural changes also precludes the perpetuation of sustained atrial arrhythmia. Uniquely the

F1759A-dTG mice show spontaneous and relatively prolonged episodes of AF in vivo and ex vivo, implying that both the substrate and triggering factors induced by the dysfunctional Na+ channels are intrinsic to the heart. Moreover, during epicardial surface optical voltage mapping experiments of isolated hearts from these mice, rotors, waves and wavelets, were frequently observed, similar to the findings in patients with AF and the classic large animal models of AF [81, 145, 146].

Figure 10. Schematic depicting proposed mechanisms of arrhythmogenesis in F1759A-dTG mice.

60 The presence of spontaneous and prolonged episodes of atrial arrhythmias in this mouse model differs from prior studies using other mouse models of defective Na+ channel inactivation. For instance, the delta-KPQ heterozygous knock-in mice were not reported to have spontaneous or prolonged episodes of AF [147, 148] and the frequency of extra-stimuli induced atrial arrhythmias was not different compared to control mice, although a very aggressive protocol of repetitive intermittent high rate burst stimulation resulting in short-long-short sequences provoked atrial tachyarrhythmias in ~2/3 of delta-

KPQ mice [149]. Left atrial diameter was mildly increased by 10% in delta-KPQ mice >

5 months old. In other studies, TG mice lines expressing the WT NaV1.5 or the LQT3- mutant N1325S did not develop atrial arrhythmias [120, 150]. In the delta-KPQ mouse, the mutant sodium channel was knocked in. Therefore, it is likely that there was less heterogeneity in persistent current. In addition, the amount of persistent current resulting from this mutation could differ from the amount resulting from the F1759A mutation, another potential factor accounting for the absence of spontaneous AF in this mouse model.

Patients with lone AF and permanent AF have increased persistent Na+ current, and patients with LQT3 syndrome have an increased incidence of AF[91, 93, 94], demonstrating the relevance of this animal model to humans. Although mice have a different cardiac ion channel profile compared to humans, with a markedly higher basal heart rate, the importance of NaV1.5 and its complete inactivation is conserved between species. The F1759A-dTG mice demonstrate the structural and electrophysiological changes that are known to be associated with AF in humans, and are thus a unique model to explore the molecular mechanisms responsible for the most common arrhythmia in

61 + humans. Taken together, the enhanced Na influx, via incomplete NaV1.5 inactivation, and APD prolongation with dispersion, the resultant downstream maladaptive effects including increased intracellular [Ca2+] via increased Ca2+ entry and reverse mode Na+-

Ca2+ exchange, and the structural changes within the atrium are likely responsible and sufficient for uniquely triggering and perpetuating AF (Figure 10).

A number of questions do remain regarding the nature of AF in these mice, which future explorations hope to answer. In these mice, there was heterogeneity in the expression of the transgene as assessed by patch clamp experiments. This heterogeneity was similarly observed in optical mapping experiments in which the action potential durations of the left and right atria were shown to vary in the AF animals more than in controls. Whether the same phenotype would be seen with a more consistent level of transgene expression remains to be seen, but could be answered with a knock-in model of this mouse, which would likely have less cell-to-cell variability in level of persistent current.

Interestingly, some experiments have suggested that the efficacy of SEA0400 in reducing arrhythmia burden in animal models is due to its normalizing APD heterogeneity [143]. This would lend some support to the thesis that heterogeneity is a critical component of the AF in our mouse model. SEA0400 could alternatively function by reducing triggered activity. Triggered activity caused by DAD’s results from increased

SR calcium load leading to diastolic calcium release. During diastole, NCX extrudes this

Ca2+, exchanging three Na+ ions for each removed Ca2+ ion. If this depolarizing current reaches threshold, an action potential is triggered. Inhibition of NCX could suppress this

DAD activity. A recent paper showed that ORM-10962, a selective inhibitor of NCX,

62 decreased the amplitude of DAD’s induced by digoxin in dog Purkinje fibers. This NCX inhibitor also decreased the incidence of ventricular extrasystoles in anesthetized guinea pigs [151]. Another study looking at ouabain induced arrhythmias- of particular note as this models the increased Na+ seen in our mouse model – also showed that KB-R7943 and 3',4'-dichlorobenzamil hydrochloride, two inhibitors of NCX, were able to inhibit these arrhythmias in isolated guinea-pig atria. They also showed that KB-R7943 inhibits ouabain-induced arrhythmias in anaesthetized guinea pigs [152]. On the other hand, it is also possible that its efficacy was due to reducing APD heterogeneity. Milberg et al. reported on the effects of SEA0400 in suppressing TdP in both an intact rabbit heart model of LQT2/LQT3 and in a computational model of a rabbit cardiomyocyte [153].

They showed that part of SEA0400’s efficacy was due to its reducing dispersion of repolarization [153].

The specific mechanism leading to increased cardiac fibrosis is also not entirely clear.

Expression of the transgenic sodium channel requires rtTA as assessed by western blot

(Figure 1 d-e). The rtTA gene is under the control of the cardiac specific alpha-MHC promoter and is as a result, not expressed in fibroblasts. Therefore, it is unlikely that the increased fibrosis is the result of sodium channel transgene expression in fibroblasts.

Cardiac electrophysiological properties of mice differ in important ways from humans. In a mouse ventricle, the APD is on the order of 50-milliseconds, roughly five times shorter than the 250-millisecond action potential duration seen in the human ventricle [154]. There are also differences in the ion channels contributing to the action potential morphology. Nonetheless, our model has features that make it particularly well suited to studying AF. Our AF mouse model phenocopied gain-of-function human

63 SCN5A mutations, some of which have been implicated in dilated cardiomyopathy and hypertrophy, and arrhythmias such as long QT syndrome, TdP and AF. Furthermore,

SCN5A is conserved between human and mouse heart.

It is unlikely that a single mechanism can explain the nature of AF as it appears in all patients. Indeed, the mechanisms of AF are highly complex and multifactorial, as evidenced by the variety of genetic changes found in patients with heritable forms of the disease. The mouse model described above does not mimic all mechanisms of AF, nor can it identify a cure for all AF patients. Nonetheless, it does recapitulate key forms of the disease, which should enable it to provide further mechanistic insights that cannot be obtained from other animal models or from humans, given the inaccessibility of cardiac tissue from AF patients.

Summary

This chapter explores the effects of a gain-of-function Na+ channel mutation on cardiac function. We show that heterogeneous expression of this channel within a mouse heart leads to a number of cardiac abnormalities including AF, ventricular arrhythmia, increased atrial area and reduced ejection fraction. The AF is of particular interest given its spontaneous initiation and its relatively long duration. We also show that SEA0400, an inhibitor of the NCX exchanger, reduces the burden of both ventricular and atrial arrhythmias in these mice.

64 Chapter 2: Calmodulin’s effect on Nav1.5 within cardiomyocytes

65 Introduction

Calmodulin is a “ubiquitous Ca2+ sensing protein” [50]. Calmodulin is known to interact with and modulate a number of voltage gated ion channels. Calmodulin interacts with voltage gated Na+ channels via the conserved “IQ” motif of the C-terminal domain

[52]. Mutations in this site diminish calmodulin binding [53]. However, elucidation of the functional effects of calmodulin’s binding has been clouded by inconsistent reports within the literature [155].

Mutations in calmodulin have been linked to severe forms of the long-QT syndrome

[156] though the extent to which that is the result of lost interaction with NaV1.5 is unclear given its important role in calcium dependent inactivation of Cav1.2 amongst other ion channels. Mutations in the IQ motif have been linked to LQT3 [157]. Reports in the literature have been highly varied with some suggesting calmodulin protects against persistent current, others that it shifts steady-state inactivation in a hyperpolarizing direction, and others still that it confers no calcium dependent regulation on NaV1.5 [55].

Many of the prior studies have been carried out in non-cardiac cells, such as the tsA201 cell line or the HEK293 cell line. In order to explore the role of calmodulin in NaV1.5, we sought to study its effects within its native environment, cardiomyocytes. The aim of this chapter is thus to determine the effects of calmodulin binding to NaV1.5 within cardiomyocytes, with specific emphasis on persistent current, steady state activation, steady state inactivation and time constant of inactivation. Based upon prior studies in heterologous expression systems, we hypothesized that loss of calmodulin binding will lead to increased persistent current.

66 Experimental Design and Approach

We generated “pseudo-wildtype” (pWT) transgenic (TG) mice with doxycycline- inducible and titratable, cardiac-specific expression of FLAG-epitope-tagged human

NaV1.5 with a mutation C374Y (Figure 1a), which confers increased sensitivity to tetrodotoxin (TTX) [124]. We simultaneously created mice with expression of FLAG- epitope-tagged human NaV1.5 with the C374Y mutation and a mutation of the calmodulin binding site, IQ1908-1909AA (IQ/AA) (Figure 1a) (Kim et al, 2004). Three distinct founder lines of pWT and two distinct founder lines of IQ/AA mice were initially established. Utilization of lidocaine resistance (F1759A) to distinguish TG from endogenous Na+ channels was not tenable, as we had previously shown that the F1759A mutation itself resulted in significantly increased persistent current as well as shifts in steady-state activation and inactivation (see Chapter 1). After crossing each with α-MHC reverse tet-transactivator protein (rtTA) mice ([123]), pWT double TG (dTG) and IQ/AA dTG mice were generated (Figure 1). Comparison of INa in pWT and IQ/AA derived cardiomyocytes enables exploration of the role of calmodulin binding to NaV1.5 in regulating persistent current, steady state activation and steady-state inactivation as well as time constant of inactivation. For this characterization, whole cell patch clamp of isolated ventricular cardiomyocytes was utilized. Steady-state activation, inactivation and rate of inactivation were measured with 3 mM Na+ with and without perfusion of 20 nM

TTX in pWT and IQ/AA TG lines. This approach enabled us to compare across TG lines

(pWT vs IQ/AA with no TTX). It also enabled us to compare within given lines, using the naturally present endogenous channels as a second control (no TTX vs 20 nM TTX).

Additionally, limb-lead ECG was performed to explore presence of arrhythmia and

67 changes in whole heart electrophysiological properties, in particular QT interval, which offers a whole-heart readout of changes in persistent Na+ current.

Results

Expression of C374Y-NaV1.5 in ventricles

Expression of TG NaV1.5 channels in ventricles of pWT and IQ/AA mice was assessed with an anti-FLAG antibody specific for TG channel (Figure 1b). Additionally, immunocytochemistry using an anti-FLAG antibody demonstrated presence of TG channel expressed at the membranes of ventricular cardiomyocytes from both lines of mice. Expression of channels was titratable with doxycycline. For most experiments, between 1 and 5 days of doxycycline was utilized. In one of two IQ/AA lines, doxycycline was not required for expression, although expression of rtTA was required.

Figure 1. Cardiac-specific, FLAG- a TTX C374Y tagged C374Y NaV1.5-expressing TG mice. (a) Membrane topology of NaV1.5. C374Y confers significantly increased TTX sensitivity [103]. (b) Anti-FLAG CaM IQ1908-1909AA antibody immunoblots of cardiac homogenates showing FLAG-epitope- b c tagged NaV1.5 expression in the ventricles of IQ/AA and pWT mice. (c) NTG IQ/AA pWT + pWT - + IQ/AA - Immunostaining of pWT (left) and IQ/AA kDa (right) cardiomyocytes with or without 250 anti-FLAG antibody and FITC-conjugated secondary antibody. Images were 130 obtained with confocal at 20 X magnification.

Endogenous cardiac NaV1.5 has significantly increased TTX resistance compared to the TG channels with the engineered C374Y mutation. We found that 20 nM TTX which offers nearly complete block of C374Y-Nav1.5 [124] had almost no effect on

68 endogenous currents (Figure 2a-b).

To establish the presence of functional channels at the membrane, we measured peak currents in isolated ventricular cells in the absence and presence of 20 nM TTX. Peak

Na+ current measurements were made with 3 mM Na+ in the extracellular solution in order to reduce voltage clamp errors. Current inhibited by the application of 20 nM TTX was predominantly from TG channels. In NTG cardiomyocytes, the average peak current density was 20.09 + 1.78 pA/pF (mean + SEM) with an average current of 19.67 + 1.74 pA/pF remaining in the presence of TTX (average fraction TTX sensitive current =

2.09%) (Figure 2b, 2e). In pWT cardiomyocytes, the average peak current density was

19.75 + 1.92 pA/pF, with a current of 9.24 + 1.20 pA/pF remaining in the presence of

TTX, demonstrating a mean TG current of 10.51 + 1.48 pA/pF (mean fraction TTX sensitive current = 53.2%) (Figure 2c, 2e). In IQ/AA mice, the average peak current density was 30.41 + 2.28 pA/pF, mean endogenous current was 12.34 + 1.55 pA/pF, and mean TTX-sensitive current was 18.07 + 2.14 pA/pF (average fraction TTX sensitive current = 59.4 %) (Figure 2d, 2e). In both the pWT and IQ/AA transgenic lines, expression levels of the transgene were near 50%. This level of expression is similar to what would be seen in heterozygous mice with wild type and mutant alleles. As a result, the transgenic lines functioned as models of heterozygous mice.

Figure 2. TTX-sensitive current in pWT and IQ/AA TG mice. (a) Exemplar whole cell Na+ current trace of ventricular cardiomyocyte from TG mouse in the absence (black) and presence (blue) of 20 nM TTX. (b-d) Graphs of mean Na+ current in the absence (TG + endogenous currents) and presence (endogenous currents) of 20 nM TTX. Cardiomyocytes were isolated from NTG (b), pWT (c) and IQ/AA (d) mice. Peak current measured with 3 mM Na+ in both extracellular and intracellular solutions. Data are mean + SEM. (e) Average percent transgenic current in NTG, pWT and IQ/AA transgenic lines.

QT Interval is not increased in IQ/AA mice

Electrocardiographic analyses were performed on all analyzed mice. Given our hypothesis that the IQ/AA NaV1.5 channels would display increased persistent current, we were particularly interested in the QT interval, which is prolonged in the setting of increased persistent current as observed in the F1759A-NaV1.5 expressing mice (see

Chapter 1, Figure 3a-b, 7a-b). Sinus rhythm was observed in all pWT mice (17/17) tested.

Compared to NTG, neither QT interval nor RR interval were prolonged in pWT animals.

The IQ/AA mice frequently demonstrated sinus rhythm, but in several mice, sinus rhythm

70 was intermittently interrupted by runs of junctional rhythm. As with pWT animals, neither QT interval nor RR intervals were significantly changed compared to NTG mice.

In contrast, the QT interval in F1759A-NaV1.5 mice was substantially longer. In both pWT and IQ/AA mice, the PR intervals were shorter than in NTG mice. This finding is consistent with previously created TG mouse models with wild-type NaV1.5 overexpression [120] and likely reflects TG channel expression in pWT and IQ/AA animals

(Figure 2c-e). The absence of QT prolongation in the IQ/AA mice compared to both pWT and NTG suggests that the cardiac action potential duration (APD) was not increased.

b RR Interval 0.20

) 0.15

s Figure 3. ECG analysis of TG mice. ( 0.10 (a) Representative limb-lead surface me i

T electrocardiograms of isoflurane- 0.05 anesthetized NTG, pWT, IQ/AA and 0.00 F1759A TG mice. (b) Bar graph of RR, PR and QT intervals from isoflurane-

NTG pWT IQ/AA anesthetized NTG (white), pWT F1759A (black), IQ/AA (red) and F1759A (grey) mice. Mean + SEM. ** P<0.01; PR Interval **** P<0.0001, ordinary one-way 0.04 Anova, Multiple Comparisons: Tukey’s

) 0.03 multiple comparisons test (all s

( ** ** 0.02 comparisons vs NTG). N=5 NTG mice; me i

T N=17 pWT mice; N=13 IQ/AA mice; 0.01 N=5 F1759A mice 0.00

NTG pWT IQ/AA F1759A

QT Interval 0.10 **** 0.08 )

s ( 0.06 me i 0.04 T 0.02

0.00

NTG pWT IQ/AA F1759A

71 Persistent current is not increased in IQ/AA mice

While the lack of QT prolongation in IQ/AA mice suggested that persistent current was not increased (Figure 3b), patch clamp electrophysiological measurements are necessary for definitive determination. For wild-type NaV1.5, the amount of persistent current tends to correlate with peak current. As a result, in order to accurately assess changes in persistent current due to inherent properties of a channel – rather than amount of channel expressed – it is essential to normalize the persistent current to peak current.

This is of particular importance in an inducible TG model where levels of channel expression and thus of peak current can vary. Because of the large Na+ currents in cardiomyocytes, which makes an accurate measurement of peak current difficult due to voltage clamp errors, we employed a protocol in which both persistent current and peak current were measured using 100 mM Na+ or 3 mM Na+ in the extracellular solution respectively. For our measurement of persistent current, we look at the difference in currents during the final 10 ms of a 190-ms depolarization with 100 mM Na+ in the extracellular solution, before and after application of 50 µM ranolazine, a blocker of persistent Na+ current. Normalization of this difference to the peak current measured with

+ 3 mM Na in the extracellular solution yields a late current ratio, RLate, enabling comparison of persistent currents. In the IQ/AA mice, the persistent current ratio was not increased compared to the persistent current ratio in NTG or pWT mice (Figure 4). This finding contrasts with reports from studies of IQ/AA mutant channels expressed in HEK cells where the IQ1908-1909AA mutation resulted in a nearly 7-fold increase in persistent current [53], demonstrating the importance of studying these channels in native cardiomyocytes. In certain recordings, persistent current was slightly higher after

72 ranolazine perfusion due to random variations in measurements. To account for this noise in the data, we analyzed persistent current in three ways: leaving these values in our analysis unchanged, zeroing these values, or removing them from our dataset. The results were consistent with all three approaches and each is presented (Figure 4). It is worth noting that because we are not blocking endogenous current when measuring persistent current, our measurements of persistent current conducted by TG channels is partially diluted. It is possible that this caused us to underestimate the true amount of persistent current conducted by TG channels. However, we detected no correlation between fraction

TTX-sensitive current and persistent current, and the TTX-sensitive current in IQ/AA mice was nearly 60% of the total current. As a result, it is unlikely that the presence of endogenous channels prevented detection of persistent current in the IQ/AA mice.

73 a b Ranolazine 0 Late Current Ratio -2 Control 0.10 (all values included)

100 0 -4 0 -100 -200 nA -200 0.05 , nA , -300 -400 e pA

Na -400 , Na Na , nA Lat

I I - I -500 I

-6 I -600 Control Ranolazine

-600 R Na

I -700 -800 -800 0.00 -900 -1000 -8 -1000 4040 60 80 100100 120 140 160160 180 Time,Time, ms ms -10 -0.05 0 20 40 60 80 100 120 140 160 180 Time,Time, m sms NTG pWT IQ/AA c d Late Current Ratio Late Current Ratio 0.10 0.10 (Negative values removed) (Negative values zero’d) 0.08 0.08

0.06 e 0.06 Late Lat R 0.04 R 0.04

0.02 0.02

0.00 0.00

NTG pWT NTG pWT IQ/AA IQ/AA Figure 4. Persistent current is not increased in IQ/AA mice. (a) Exemplar whole cell Na+ current traces of ventricular cardiomyocyte isolated from IQ/AA mice. Persistent Na+ current was evaluated with a 190- ms depolarization from a holding potential of -110 mV to -30 mV in the absence (black) and presence (red) of 50 µM ranolazine. The intracellular solution contained 3 mM Na+ and the extracellular solution contained 100 mM Na+. (b-d) The ratio of persistent Na+ current to peak Na+ current in ventricular cardiomyocytes isolated from NTG, pWT, and IQ/AA mice. Y-axis is ratio of persistent Na+ current determined using 100 mM Na+ in the extracellular solution to peak Na+ current determined using 3 mM Na+ in the extracellular solution. Data shown with negative values present (b), removed (c) or set to zero (d). Mean + SEM. N=17 NTG ventricular cardiomyocytes; N=8 pWT ventricular cardiomyocytes; N=19 IQ/AA ventricular cardiomyocytes. Dunnett’s multiple comparisons test

Steady State Activation

In addition to persistent current, we also sought to determine whether calmodulin affected properties of INa including voltage-dependence of activation, steady-state inactivation and rate of inactivation. Whole cell patch clamping of cardiomyocytes was similarly utilized to assess these parameters. Steady-state activation was measured with 3 74 mM Na+ in cardiomyocytes from pWT and IQ/AA transgenic lines. Measurements were made both in the absence and presence of TTX. Recordings in the absence of TTX provided a measure of both endogenous and transgenic channels. Recordings with 20 nM

TTX on the other hand, provided a measure only of endogenous channels (Figure 5).

Figure 5. Schematic of conditions in which steady-state activation/inactivation (A/I) parameters were measured. Application of 20 nM TTX blocks transgenic channels. Comparison of A/I properties between pWT and IQ/AA cardiomyocytes in the absence of TTX enables comparison of transgenic mutant IQ/AA to transgenic control pWT channels. Comparison of A/I parameters in the absence and presence of TTX enables comparison of the transgenic channels to endogenous channels in a given line. Comparison of A/I parameters between pWT and IQ/AA cardiomyocytes in the presence of TTX enables comparison of endogenous channels in pWT and IQ/AA lines.

We observed that the voltage-dependence of activation was shifted in a hyperpolarizing direction non-significantly in IQ/AA cardiomyocytes compared to pWT cardiomyocytes (Figure 6a-c).

We next compared steady-state activation in the presence and absence of TTX enabling comparison of TG to endogenous channels within each mouse line. In cardiomyocytes isolated from pWT mice, steady-state activation was non-significantly

75 shifted 1 mV in the presence of TTX suggesting that the pWT TG channels did not significantly differ from mouse endogenous channels in that line of mice.

In cardiomyocytes isolated from the IQ/AA mice, steady-state activation was significantly shifted by 5 mV in a hyperpolarizing direction in the presence of TTX.

Steady-state activation of endogenous channels within the IQ/AA cardiomyocytes were also significantly left-shifted compared to endogenous channels within the pWT cardiomyocytes (Figure 6a-b). This finding interestingly implies that the endogenous channels within the IQ/AA mice were significantly shifted in a hyperpolarizing direction

(see Discussion).

76 Figure 6. Voltage dependence of activation in IQ/AA mice. (a-b). Boltzmann fits of the voltage- dependent activation from cardiomyocytes isolated from pWT (a) and IQ/AA (b) in the absence (black) and presence (red) of 20 nM TTX. (C). Vmid values from pWT and IQ/AA cardiomyocytes in absence (black) and presence (red) of 20 nM TTX. Boltzmann fits of the voltage-dependent activation profile showed non- significantly left-shifted Vmid in IQ/AA mice (-62.93 vs -65.91 for pWT and IQ/AA respectively). Mean + SEM. * P<0.05; ** P<0.01; **** P<0.0001, ordinary one-way Anova, Multiple Comparisons: Tukey’s multiple comparison test. pWT vs pWT-TTX:not significant (NS). IQ/AA vs IQ/AA-TTX: NS. pWT vs IQ/AA: NS. N=16 pWT cardiomyocytes with and without TTX; N=12 IQ/AA cardiomyocytes with and without TTX.

Decreased availability in IQ/AA mice

Similar analyses were performed for steady-state inactivation as well as for time

constant of inactivation. In cardiomyocytes from pWT mice, steady-state inactivation did

77 not differ with or without TTX perfusion. Thus, the transgenic pWT channels had comparable availability to endogenous channels in the pWT mice. In contrast, expression of the IQ/AA transgenic channels resulted in a hyperpolarizing shift in steady-state inactivation (Figure 7). Thus, availability of NaV1.5 was decreased in the IQ/AA mice.

The significant left-shift of steady-state inactivation seen in the IQ/AA mice was similarly seen when we assessed endogenous channels from IQ/AA cardiomyocytes by perfusion of 20 nM TTX. This result thus suggests that the IQ/AA TG channels in some way impact upon endogenous channels (Figure 7 c-d). Notably, given the unexpected nature of this finding, similar results were found in both lines of IQ/AA mice.

Amongst pWT cardiomyocytes, tau of inactivation was not significantly different whether or not TTX was in the extracellular solution. Thus, the inactivation kinetics of the pWT transgenic channels did not differ from endogenous channels in pWT mice.

NaV1.5 channels in IQ/AA cardiomyocytes transitioned to the inactivated state more quickly than did those in pWT cardiomyocytes suggesting that the mutant IQ/AA channels accelerated the rate of inactivation of NaV1.5.

In IQ/AA cardiomyocytes, block of transgenic channels via perfusion of TTX led to a normalization of inactivation kinetics. Thus, the accelerated kinetics of inactivation in

IQ/AA cardiomyocytes was predominantly driven by the TG channels.

Both the left shift in steady-state inactivation and the shortening of tau suggest that loss of calmodulin binding confers loss-of-function properties, with decreased channel availability and more rapid inactivation. These findings suggest that apocalmodulin binding increases channel availability and slows the transition to the inactivated state.

Figure 7. IQ/AA results in loss-of-function. (a-b) Boltzmann fits of the steady-state inactivation from cardiomyocytes isolated from pWT (a) and IQ/AA (b) in the absence (black) and presence (red) of 20 nM TTX. ( c) Vmid values from pWT and IQ/AA cardiomyocytes in absence (black) and presence (red) of 20 nM TTX. Boltzmann fits of the voltage-dependent inactivation profile showed significantly shifted Vmid in pWT vs IQ/AA (-80.55 vs -87.65 for pWT and IQ/AA respectively. **** P<0.0001) pWT in 20 nM TTX vs IQ/AA in 20 nM TTX (-83.59 vs – 90.02 in pWT 20 nM TTX and IQ/AA 20 nM TTX respectively. *** p<0.001). pWT in 20 nM TTX vs IQ/AA (-83.59 vs -87.65 in pWT 20 nM TTX and IQ/AA respectively. * p<0.05). pWT vs IQ/AA in 20 nM TTX (-80.55 vs -90.02 in pWT and IQ/AA in 20 nM TTX respectively. ****p<0.0001) pWT vs pWT in 20 nM TTX: NS. IQ/AA vs IQ/AA in 20 nM TTX: NS. N=15 pWT cardiomyocytes with and without TTX. N=9 IQ/AA cardiomyocytes with and without TTX. Tukey’s multiple comparisons testd) Tau values from pWT and IQ/AA cardiomyocytes in absence (black) and presence (red) of 20 nM TTX. One-exponential fits of the decay phases for the pWT (left) and IQ/AA (right). Tau of inactivation shorter in IQ/AA. Tau significantly changed in: pWT vs IQ/AA (1.983 vs 1.392 ms in pWT and IQ/AA respectively. **** p<0.0001). pWT in 20 nM TTX vs IQ/AA in 20 nM TTX (2.160 vs 1.728 ms in pWT 20 nM TTX and IQ/AA 20 nM TTX respectively. ** p<0.01). pWT in 20 nM TTX vs IQ/AA (2.160 vs 1.392 ms in pWT and IQ/AA respectively. **** p<0.0001). N=15 pWT cardiomyocytes with and without TTX. N=9 IQ/AA cardiomyocytes with and without TTX. Tukey’s multiple comparison test.

79 Discussion

Here we report on the effects of the IQ1908-1909AA mutation within the “IQ motif” of the NaV1.5 C-terminal domain. We show that in ventricular cardiomyocytes, AA substitution of the IQ motif does not lead to increases in persistent current, or QT prolongation on surface limb-lead ECG. We show that steady-state activation is left- shifted non-significantly compared to the TG control. Interestingly, we show that the inactivated state is stabilized, as steady-state inactivation is significantly left-shifted.

Furthermore, the rate of inactivation is significantly accelerated. These findings suggest that binding of apocalmodulin results in a gain-of-function phenotype with increased channel availability and slower kinetics of inactivation, a finding in contrast to studies in which the channels were heterologously expressed in non-muscle cells and without many of the known macromolecular complex components [53]. Importantly, all of our studies are carried out with low calcium concentrations. Determining whether our results would differ in the presence of calcium is a goal of future investigations.

To our significant surprise, when assessing the endogenous channels expressed within IQ/AA cardiomyocytes (through perfusion of 20 nM TTX), we detected similar changes in steady-state inactivation. This finding offers evidence that the properties of the endogenous channels are themselves altered by the presence of the IQ/AA mutant channels, an unexpected finding, but consistent with emerging work demonstrating that

NaV1.5 channels can couple via their C-termini, with calmodulin located at the interface of the proposed dimer [158].

Importantly, we created a system to study the channels in their native state using an approach that enables inducible, cardiac specific TG expression. Our goal was to

80 create a system in which each myocyte could act as its own control, by comparing mutant and endogenous channels. By using an inducible system, we were able to control the expression of the TG channels, as one of our goals was to reduce the probability of channel mislocalization or dilution of endogenous modulators of the channels. A number of investigators have previously explored the role of calmodulin regulation on voltage gated sodium channels however most have restricted their investigations to artificial systems such as the HEK293 and tsA-201 cell types (supplementary Table 1) [52][53].

These findings are surprising for a number of reasons. While the literature on calmodulin regulation of NaV1.5 has not been entirely clear or consistent (supplementary

Table 1) [52][53], a number of reports suggest that loss of calmodulin binding results in various gain-of-function properties when NaV1.5 is heterologously expressed and studied.

Balser et al. showed that calmodulin binding had no functional effect upon steady-state inactivation in the absence of calcium in contrast to our findings. With calcium present, they did see a shift in steady-state inactivation – however block of calmodulin binding led to a depolarizing rather than hyperpolarizing shift as we observed. These studies were all carried out in tsA-201 cells, potentially explaining the discrepancy. Furthermore, they prevented the calmodulin interaction by applying a peptide inhibitor or by mutating residues 1908 and 1921 to E and R. Pitt and Kass also explored calmodulin’s role, and did so using the IQ/AA mutation that we employed. In HEK cells, they showed that persistent current increased significantly- roughly 7-fold – with the IQ/AA mutation and low calcium concentrations. This differed from our findings of unchanged persistent current with the IQ/AA mutation. Here, the biggest differentiating factor was that our investigation occurred in cardiomyocytes whereas Pitt and Kass’ work was in HEK293

81 cells.

Winkel et al. characterized the effects of a Q1909R mutation found in a sudden infant death syndrome case [159]. Functionally, this mutation was shown to increase the persistent current by 2-fold, slow inactivation kinetics, and shift steady-state activation leftwards with no effect on steady-state inactivation. The functional effects were explored in HEK293 cells. Interestingly the Amzel lab also explored this Q1909R mutation and saw right-shifted voltage dependence of activation and steady-state inactivation, unlike

Winkel et al. They also showed that the IQ/AA mutation resulted in a right shift in steady-state inactivation in contrast to our findings [158].

Our findings of a left-shift in steady-state inactivation and accelerated rate of inactivation with the IQ/AA mutant channels are thus in contrast to much of the reported literature. The biggest differentiating factor remains that our work has been carried out in ventricular cardiomyocytes. We have also seen similar changes in two lines of TG mice.

Nonetheless, further exploration remains needed to resolve the effects of calmodulin. One approach is to create a knock-in mouse with an IQ/AA mutation in NaV1.5 using

CRISPR/CAS9.

Another particularly surprising finding in our investigation was the altered properties of endogenous INa assessed in IQ/AA cardiomyocytes with 20 nM TTX.

Endogenous INa in IQ/AA cardiomyocytes displayed a significant left-shift in voltage- dependence of activation and steady-state inactivation.

A few possibilities could account for the changes in properties of endogenous channels. One possibility is that the change was due to some intrinsic property of the mouse line. For example, the genomic location of the transgene insertion could in some

82 way affect properties of endogenous channels. However, we observed similar phenomena in both of our IQ/AA lines. It is also possible that changes in the whole heart properties, such as increased incidence of arrhythmia or structural changes such as hypertrophy, lead to post-translational alterations of all Na+ channels.

While extensive structural analyses of the ventricles have not been performed, we did not find significant differences between IQ/AA and pWT mice in terms of ECG findings. A third possibility is that differences in peak current density and the resultant alterations in the stoichiometry of interactions with other macromolecular partners is contributing to the changes. Average peak current density was higher in the IQ/AA cardiomyocytes, raising the Figure 8. Structure of NaV1.5 C-terminal possibility that interactions with other auxiliary (CT) domain. (a-b) Front and side views of the CTNav1.5-CTNav1.5 dimer. One CTNav1.5 in green, the other in orange. proteins was diminished. However, we did not (C). Attachment of each CTNav1.5 of dimer with S6 of domain IV (rectangles). detect correlations between steady-state Calmodulin in purple. Adapted from Gabelli et al., 2014. [136] inactivation and peak current density in either the pWT or IQ/AA cardiomyocytes. Had the increased peak current in IQ/AA cardiomyocytes been responsible for the left shift in steady-state inactivation in these cells, then we would have expected to see a negative correlation between peak current density and steady-state inactivation.

83 A fourth possibility and one that warrants further investigation is that the TG channels are in some way impacting upon the endogenous channels. While speculative, this nonetheless raises the question of how the IQ/AA mutant channels could impact upon endogenous channel function (as will be seen in Chapter 3, TG channels did not appreciably affect endogenous channels in our other models). With respect to this question, interesting work comes from the Tomaselli lab. In 2014, they reported on the structure of the NaV1.5 C-terminal domain with calmodulin. Here, it was shown that sodium channels dimerize via the C-terminal domain. Interestingly, the dimerization involves helix alpha-VI –which contains the “IQ motif” of one CTD binding to the EF- hand like motif of the interacting CTD. In this dimer of NaV1.5-CTD’s, there are also two calmodulins present. It was proposed that this dimer represented the “non-inactivated” state of the channel, and that dimerization increased stability of the “non-inactivated conformation of the channel” [158]. In studying mutations that disrupted this dimer, it was shown that steady-state inactivation was significantly shifted in a hyperpolarizing direction. Thus, one possibility is that our expression of a significant fraction of IQ/AA channels compromised the ability of endogenous channels to dimerize or prevented the functional effects of dimerization. By either preventing formation of a dimer or allosterically preventing the function of dimerization, the IQ/AA mutants perhaps resulted in endogenous channels that were stabilized in the inactivated state, a finding consistent with our measured changes in endogenous currents. Further efforts are necessary in order to adequately address this hypothesis.

One potential concern raised by the potential coupling of channels is that the absence of persistent current could be the result of endogenous channels acting upon the

84 transgenic channels to suppress persistent current. However, in the F1759A mouse model, the increased persistent current caused by the F1759A channels was present in cardiac myocytes even in cells with greater than 50% endogenous expression.

Furthermore, within the IQ/AA cardiomyocytes, persistent current was not increased even in cells with close to 100 % transgenic channel expression. If endogenous channels were responsible for preventing elevated persistent current in the IQ/AA cardiomyocytes, then we would have expected the cells with nearly 100% IQ/AA channels to display increased persistent current. In order to address this question, as well as to further explore the effects of coupling, future work in HEK cells- with expression of pWT channels,

IQ/AA channels, and co-expressed pWT and IQ/AA channels – can be utilized.

Summary

Within cardiomyocytes, loss of calmodulin binding to NaV1.5 via IQ/AA mutation caused a hyperpolarizing shift in steady state inactivation and accelerated the rate of inactivation. The IQ/AA mutation did not result in increased persistent current. Contrary to many reports suggesting loss of calmodulin binding results in gain-of-function effects, we see that the IQ/AA mutation instead results in stabilization of the inactivated state of the channel. Within cardiomyocytes from IQ/AA mice, we also observe that endogenous currents display stabilization of the inactivated state suggesting functional dimerization.

85 Chapter 3: FHF’s Effect on Nav1.5 in Cardiomyocytes

86 Introduction

Fibroblast growth factor homologous factors (FHF’s) are members of the FGF family that remain intracellular and play important roles in regulating ion channels [160].

Like calmodulin, FHF’s interact with NaV1.5 via its C-terminal domain.

Given our goal of studying mechanisms of persistent current and in exploring regulators of NaV1.5, we had a natural interest in exploring the role of FHF’s. Numerous studies have implicated FHF’s as modulators of voltage gated sodium channels including

NaV1.5. However, as with calmodulin, there is disagreement regarding the functional effects of FHF’s binding to NaV1.5 (Supplementary Table 2).

Recently, Musa et al. reported on a particularly interesting mutation in the

NaV1.5-CTD, resulting in decreased FHF binding affinity and a phenotype of LQT3, AF and PVC’s in a five-generation family [62]. This mutation, H1849R, formed the basis of our studies.

An additional factor prompting our studies of FHF’s interaction with NaV1.5 came from our studies of calmodulin’s interactions with NaV1.5. As reported in Chapter

2, loss of calmodulin binding via IQ/AA NaV1.5 did not result in increased persistent current, a result in contrast to findings in HEK cells [53]. A natural question was whether some unaccounted-for factor, present in cardiomyocytes but missing in HEK293 cells could explain the divergence of these results. One possibility that emerged were FHF’s.

FHF’s interaction with the NaV1.5-C-terminal domain (CTD) is in close proximity to the calmodulin binding site, as demonstrated in crystal structures of the ternary NaV-CTD-

Calmodulin-FHF complex (Figure 1a). FHF’s are not expressed in HEK cells. Thus, we had interest in assessing whether the H1849R mutation, in tandem with an IQ1908-

87 1909AA mutation, might result in persistent current comparable to that seen in HEK cells with expression of the IQ/AA mutant NaV1.5 [53].

Experimental Design and Approach

We generated two transgenic (TG) mice, one with a mutation in NaV1.5 conferring decreased FHF binding affinity (H1849R) and one with H1849R as well as

IQ/AA mutations (H1849R-IQ/AA). In both cases, the sodium channels were made TTX- sensitive so that we could both assess the level of TG expression and distinguish between

TG and mutant channels in the same cardiomyocyte. After crossing each with α-MHC reverse tet-transactivator protein (rtTA) mice ([123]), H1849R-rtTA double TG (dTG) and H1849R-IQ/AA-rtTA dTG mice were generated. As in Chapter 2, whole cell patch clamp of isolated ventricular cardiomyocytes was utilized in order to study Na+ currents.

Limb-lead ECG’s were acquired to determine the presence of arrhythmia and changes in whole heart electrophysiological parameters. The pWT described in Chapter 2 served as the TG control and all data of pWT and IQ/AA presented here are from the experiments described in Chapter 2.

Results

Expression of transgene in ventricles

Expression of TG channels in the ventricles of H1849R and H1849R-IQ/AA was assessed via western blot analysis using anti-FLAG antibody (Figure 1B). Expression of channels was titratable with doxycycline. Most experiments utilized between 1 and 3 days of doxycycline administration, which achieved > 50% expression of transgenic channels.

88 In order to distinguish TG and endogenous currents, 20 nM TTX was utilized

(See Chapter 2, also [124]). Peak Na+ currents were measured using whole cell patch clamp with 3 mM Na+ in the extracellular solution. In cardiomyocytes from H1849R mice, peak Na+ current density was 39.17 + 7.52 pA/pF, current in the presence of 20 nM

TTX was 14.35 + 3.79 pA/pF and average percent transgenic current was thus 61.1 + 8.1

%) (Figure 1c). In H1849R-IQ/AA mice, peak Na+ current density was 25.76 + 2.81 pA/pF, with 11.49 + 1.91 pA/pF remaining in the presence of 20 nM TTX. Percent transgenic current was 52.27 + 4.4 % (Figure 1c).

Figure 1. Cardiac-specific, FLAG-tagged C374Y NaV1.5-expressing TG mice and TTX-sensitive current. (a) Membrane topology of NaV1.5. C374Y confers significantly increased TTX sensitivity ([103]) (B) Anti-FLAG antibody immunoblots of cardiac homogenates showing FLAG-epitope-tagged NaV1.5 expression in the ventricles of H1849R and H1849R-IQ/AA mice. (C) Effect of 20 nM TTX in cardiomyocytes isolated form H1849R (left) and H1849R-IQ/AA (middle) mice. Peak current measured with 3 mM Na+ in both extracellular and intracellular solutions in the absence (black) and presence (red) of 20 nM TTX. Percent transgenic current in pWT, H1849R and H1849R-IQ/AA mice (right). Data are mean +SEM.

89 Whole Heart Electrophysiological Properties

Surface limb-lead ECG’s were acquired from mice under isoflurane anesthesia. All

H1849R mice were in sinus rhythm (14/14). Furthermore, the QT interval and the RR interval were normal (Figure 2). In the H1849R-IQ/AA mice, several arrhythmias were noted including occasional PVC’s, short runs of AF, and sinus block (Figure 2a). In sinus rhythm, the QT interval and the RR interval were normal in H1849R-IQ/AA mice (Figure

2b). In both H1849R and H1849R-IQ/AA mice, the PR intervals were shortened compared to NTG mice, likely due to NaV1.5 over-expression (see Chapter 2, also see

[120]). The normal QT interval in both H1849R and H1849R-IQ/AA mice suggested that action potential duration was not significantly prolonged and is in contrast to the QT prolongation seen in humans with the H1849R mutation.

a H1849R H1849R-IQ/AA H1849R-IQ/AA H1849R-IQ/AA

b 0.20 RR Interval 0.04 PR Interval 0.06 QT Interval 0.15 0.03

) ) 0.04 s s * ** ** ) ( (

0.10 0.02 s ( me me i i T T

me 0.02 0.05 0.01 i T

0.00 0.00 0.00

NTG pWT NTG pWT NTG pWT H1849R H1849R H1849R

H1849R-IQ/AA H1849R-IQ/AA H1849R-IQ/AA

Figure 2. Electrophysiological parameters in H1849R and H1849R-IQ/AA mice. (a) Representative limb-lead surface electrocardiograms of isoflurane-anesthetized H1849R and H1849R-IQ/AA mice. H1849R-IQ/AA mice occasionally display arrhythmias. (b) Bar graph of RR, PR and QT intervals from isoflurane-anesthetized NTG (white), pWT (black), H1849R (red) and H1849R-IQ/AA (grey) mice. Mean + SEM; * P<0.05; ** P<0.01; Tukey’s multiple comparisons (significance reflects comparisons vs NTG). N=5 NTG mice; N=17 pWT mice; N=14 H1849R mice; N=11 H1849R-IQ/AA mice (not shown: N=13 IQ/AA mice; N=5 F1759A mice) 90

Assessing Presence of Persistent Current in H1849R Mice

A primary goal was to determine whether the H1849R mutation resulted in increased persistent current. To compare persistent currents across TG lines, we normalized the persistent current to peak current as described in Chapter 2. In cardiomyocytes from

H1849R mice, persistent current was not increased (Figure 3a-c). As in Chapter 2, analysis was performed with negative values unchanged, zeroed or removed from analysis and the results were essentially the same with each approach (Figure 3a-c). In combination with the lack of QT prolongation in H1849R mice, these data suggest that the H1849R mutation does not result in increased persistent current.

a b c Late Current Ratio Late Current Ratio Late Current Ratio (all values included) (Negative values removed) (Negative values zero’d) 0.10 0.10 0.10 0.08 0.08 e e 0.05 e 0.06 0.06 Lat Lat Lat 0.04 R R 0.04 R 0.00 0.02 0.02 0.00 -0.05 0.00 -0.02

pWT pWT pWT H1849R H1849R H1849R

Figure 3. Persistent current is not increased in H1849R mice. (a-c) The ratio of persistent Na+ current to peak Na+ current in ventricular cardiomyocytes isolated from pWT and H1849R mice. Y-axis is ratio of persistent Na+ current determined using 100 mM Na+ in the extracellular solution to peak Na+ current determined using 3 mM Na+ in the extracellular solution. Data shown with negative values present, set to zero, or removed. Persistent Na+ current was evaluated with a 190-ms depolarization from a holding potential of -110 mV to -30 mV. Values between groups are not statistically significant ordinary one way Anova. N = 8 pWT cardiomyocytes. N = 6 H1849R cardiomyocytes. Dunnett’s multiple comparisons test.

Steady-State Activation

To assess steady-state activation, measurements were made in the presence and absence of TTX in H1849R cardiomyocytes (see Chapter 2). Steady-state activation was 91 not changed in H1849R cardiomyocytes (-53.15 mv vs -54.76 mv in pWT and H1849R respectively) (Figure 4a-b). Comparison of steady-state activation with and without TTX present in the extracellular solution showed that the H1849R mutation did not alter steady-state activation of the TG channels or the endogenous channels in H1849R mice

(Figure 4a-b).

Figure 4. Voltage dependence of activation in H1849R mice. (a) Boltzmann fits of the voltage- dependent activation from cardiomyocytes isolated from H1849R in the absence (black) and presence (red) of 20 nM TTX. (b) V50 values from pWT and H1849R cardiomyocytes in absence (black) and presence (red) of 20 nM TTX. Boltzmann fits of the voltage- dependent activation profile showed comparable V50 in H1849R mice (- 53.15 mv vs -54.76 mv for pWT and H1849R respectively). Steady-state activation of endogenous channels in pWT and H1849R mice were also comparable (-52.21 mv vs -50.09 mv in pWT with 20 nM TTX and H1849R with 20 nM TTX, respectively. Values between groups are not statistically significant by ANOVA multiple comparisons: Tukey’s multiple comparisons test. N=15 pWT cardiomyocytes with and without TTX; N=5 H1849R cardiomyocytes with and without TTX.

Steady-State Inactivation

To assess inactivation properties, we recorded steady-state inactivation and also

92 measured the rate of NaV1.5 inactivation. In adult ventricular cardiomyocytes from

H1849R mice, the V50 of steady-state inactivation was shifted in the depolarizing direction (Figure 5b, -80.55 vs -75.06 mv for pWT and H1849R respectively; P<0.05 by

Anova and Tukey’s multiple comparison test). In the H1849R cardiomyocytes, we observed that the endogenous mouse channels were less available than TG channels and had similar availability as the pWT TG channels, further suggesting that the H1849R mutation increases Na+ channel availability. It is likely that we partially underestimated the degree to which steady-state inactivation is right shifted in H1849R channels compared to pWT channels because of the presence of endogenous channels in these measurements.

Figure 5. H1849R results in gain-of-function. (a) Boltzmann fits of the steady-state inactivation from cardiomyocytes isolated from H1849R in the absence (black) and presence (red) of 20 nM TTX (b) V50 values from pWT and H1849R cardiomyocytes in absence (black) and presence (red) of 20 nM TTX. Boltzmann fits of the voltage-dependent inactivation profile showed significantly shifted V50 in pWT vs H1849R (-80.55 vs -75.06 mv for pWT and H1849R respectively. * P<0.05). H1849R vs H1849R in 20 nM TTX (-75.06 vs –81.71 mv for H1849R vs H1849R in 20 nM TTX respectively. * P <0.05). pWT in 20 nM TTX vs H1849R (-83.59 vs -75.06 mv for pWT in 20 nM TTX vs H1849R respectively. *** P<0.001) N=15 pWT cardiomyocytes with and without TTX. N=5 H1849R cardiomyocytes with and without TTX. Ordinary one way ANOVA Multiple comparison’s: Tukey’s multiple comparisons test. (c) Tau values from pWT and H1849R cardiomyocytes in absence (black) and presence (red) of 20 nM TTX. One- exponential fits of the decay phases for the pWT (left) and H1849R (right). Tau of inactivation prolonged in H1849R. Tau significantly changed in: pWT vs H1849R (1.983 vs 2.548 ms in pWT and H1849R respectively. * p<0.05). N=14 pWT cardiomyocytes with and without TTX. N=5 H1849R cardiomyocytes with and without TTX. Ordinary one way ANOVA: Tukey’s multiple comparisons test 93

We similarly assessed the rate of inactivation of INa. The rate of INa inactivation was significantly slower in H1849R cardiomyocytes (2.16 ms H1849R vs 1.98 ms pWT,

Figure c, * P<0.05 by Anova and Dunnett’s multiple comparison test). In H1849R cardiomyocytes, inactivation was faster in the presence of 20 nM TTX, suggesting that transgenic channels also had delayed inactivation compared to endogenous channels within these mice. Rate of inactivation of endogenous channels was comparable between pWT and H1849R mice. These results similarly suggest that TG channels with the

H1849R mutation drove the overall slowing of inactivation of INa in H1849R cardiomyocytes in comparison to pWT cardiomyocytes (Figure 5c).

H1849R Mutation in Context of Loss of Calmodulin Binding Doesn’t Increase

Persistent Current

One question we were particularly interested in was whether the H1849R mutation would result in increased persistent current in sodium channels with decreased calmodulin binding. However, in H1849R-IQ/AA cardiomyocytes, persistent current was not elevated (Figure 6). As a result, it does not appear as if the absence of FHF expression in HEK cells accounts for the elevated persistent current seen with expression of IQ-1908-1909AA NaV1.5 in these cells [53]. It is possible however, that the H1849R mutation does not accurately replicate loss of FHF binding, as the mutation clearly has other effects on channel function.

a Late Current Ratio (all values included) Figure 6. Persistent current is not 0.10 increased in H1849R –IQ/AA mice. (a-c) + e The ratio of persistent Na current to peak 0.05 + Lat Na current in ventricular cardiomyocytes R isolated from pWT and H1849R-IQ/AA 0.00 mice. Y-axis is ratio of persistent Na+ current + -0.05 determined using 100 mM Na in the extracellular solution to peak Na+ current + pWT determined using 3 mM Na in the H1849R extracellular solution. Data shown with

H1849R-IQ/AA negative values present (a), removed (b) or set to zero (c). Persistent Na+ current was evaluated with a 190-ms depolarization from Late Current Ratio a holding potential of -110 mV to -30 mV in b (Negative values removed) 0.10 the absence and presence of 50 µM ranolazine. The intracellular solution 0.08 + e contained 3 mM Na and the extracellular 0.06 +

Lat solution contained 100 mM Na . R 0.04 0.02

0.00

pWT H1849R

H1849R-IQ/AA

Late Current Ratio c (Negative values zero’d) 0.10

0.08 e 0.06 Lat

R 0.04

0.02

0.00

pWT H1849R

H1849R-IQ/AA

95 Steady State Activation of H1849R-IQ/AA Cardiomyocytes

The combined H1849R and IQ/AA mutations did not lead to changes in the voltage-dependence of activation (Figure 7B). The V50 of steady-state activation in

H1849R-IQ/AA cardiomyocytes was similar to the V50 in pWT cardiomyocytes.

Similarly, within H1849R-IQ/AA cardiomyocytes, steady-state activation of endogenous channels was comparable to that of the total NaV1.5 population, indicating that endogenous and transgenic H1849R-IQ/AA channels had similar steady-state activation

(Figure 7).

Figure 7. Voltage dependence of activation in H1849R-IQ/AA mice. (a) Boltzmann fits of the voltage- dependent activation from cardiomyocytes isolated from H1849R-IQ/AA in the absence (black) and presence (red) of 20 nM TTX. (b) V50 values from pWT, H1849R, and H1849R-IQ/AA cardiomyocytes in absence (black) and presence (red) of 20 nM TTX. N= No significant differences between groups. N=15 pWT cardiomyocytes with and without TTX. N=5 H1849R cardiomyocytes with and without TTX. N=18 H1849R-IQ/AA cardiomyocytes no TTX; 17 with TTX. Ordinary one way ANOVA: Tukey’s multiple comparisons test.

Channel Availability and Inactivation Kinetics

We next analyzed parameters of inactivation in the H1849R-IQ/AA cardiomyocytes.

The effects of this double mutation were of particular curiosity given the opposing effects of the H1849R and IQ/AA mutations. In our experiments, the H1849R mutation seemed to have a dominant effect over the IQ/AA mutation with respect to steady-state 96 inactivation. We observed a significant rightward shift (V50 shifted by 3.5 mV) in steady- state inactivation of INa measured in cardiomyocytes from H1849R-IQ/AA compared to pWT (-90.68 vs -87.16 in pWT and H1849R-IQ/AA respectively; p <0.05 by Anova and

Dunnett’s multiple comparison test).

In the H1849R-IQ/AA cardiomyocytes, we observed that the endogenous mouse channels were less available than TG channels and had similar availability as the pWT and H1849R endogenous channels, further suggesting that the combined H1849R-IQ/AA mutation increases Na+ channel availability. Thus, in the double mutant channels, the gain-of-function properties of the H1849R mutation seemed to exert greater influence than the loss of function properties of the IQ/AA mutation in terms of steady-state inactivation. However, we found that the H1849R-IQ/AA mutation had no effect on the rate of inactivation, which was not significantly different than the pWT TG channels

(Figure 8a). Furthermore, the rate of inactivation of the mutant channels was similar to the endogenous channel. This was in contrast to the slowed rate of inactivation observed with the H1849R mutation alone, and the faster rate of inactivation observed with the

IQ/AA mutation alone.

Figure 8. Increased availability in H1849R-IQ/AA. (a) Boltzmann fits of the steady-state inactivation from cardiomyocytes isolated from H1849R-IQ/AA in the absence (black) and presence (red) of 20 nM TTX. (b) V50 values from pWT, H1849R and H1849R-IQ/AA cardiomyocytes in absence (black) and presence (red) of 20 nM TTX. V50 significantly right shifted in H1849R-IQ/AA vs pWT (-80.55 vs -76.89 mv in pWT and H1849R-IQ/AA respectively. *P<0.05). pWT in 20 nM TTX vs H1849R-IQ/AA (-83.59 vs -76.89 mv in pWT in 20 nM TTX and H1849R-IQ/AA **** P<0.0001). H1849R-IQ/AA in no TTX vs H1849R-IQ/AA in 20 nM TTX (-76.89 vs -82.15 mv in H1849R-IQ/AA no TTX vs H1849R-IQ/AA in 20 nM TTX). N=15 pWT cardiomyocytes with and without TTX. N=5 H1849R cardiomyocytes with and without TTX. N=18 H1849R-IQ/AA cardiomyocytes with and without TTX. Ordinary one-way ANOVA: Tukey’s multiple comparisons test. (c). Tau values from pWT, H1849R and H1849R-IQ/AA cardiomyocytes in presence (black) and absence (red) of 20 nM TTX. One exponential fits of the decay phases for the pWT (left), H1849R (middle) and H1849R-IQ/AA (right). Tau of inactivation not increased in H1849R-IQ/AA compared to pWT. Ordinary one way ANOVA: Tukey’s multiple comparisons test. N=14 pWT cardiomyocytes with and without TTX. N=5 H1849R cardiomyocytes with and without TTX. N=16 H1849R-IQ/AA cardiomyocytes no TTX; 15 with TTX.

DISCUSSION

We have explored the effects of the H1849R mutation on properties of INa within cardiomyocytes. We have addressed its effects in isolation, as well as in combination with the IQ1908-1909AA mutation. We show that the H1849R mutation results in gain- of-function properties including a slowed rate of inactivation and increased channel availability. The QT interval in mice was not prolonged, in contrast to what is observed in patients with H1849R mutations, consistent with the modest effects of the H1849R 98 mutation observed by patch clamp analysis. Interestingly, we see that its gain-of-function properties persist even in the presence of the loss-of-function IQ1908-1909AA mutation, which we have previously showed resulted in a left shift in steady-state inactivation and faster channel inactivation kinetics. Contrary to our hypothesis, persistent current was not increased in cardiomyocytes expressing this double mutant channel.

Given the differences in peak current (10.5 nA in H1849R vs 5.09 nA in pWT, 5.79 nA in H1849R-IQ/AA), one concern is that differences in properties such as voltage dependence of activation, steady-state inactivation or inactivation kinetics could result from altered stoichiometry of interaction with molecular partners. However, none of the parameters assessed - persistent current, voltage dependence of activation, steady-state inactivation, or rate of inactivation - correlated with peak current suggesting this was not the case.

Our investigation of the H1849R mutation in an adult ventricular cardiomyocyte confirms prior investigations of this channel in HEK293 cells and non-adult cardiomyocytes. In 2015, Musa et al. reported that H1849R-NaV1.5 had a reduced rate of inactivation and a depolarizing shift in steady-state inactivation when expressed in

HEK293 cells [62]. They observed similar results in neonatal cardiomyocytes with adenovirus mediated expression of the H1849R channel and Cre recombinase mediated silencing of endogenous channels. The gain-of-function effect of H1849R was previously attributed loss of FHF binding in part because of the necessity of FHF expression with

WT channels in HEK293 cells to observe an effect of the mutation [62]. Knockout of

FHF13 (Fhf2) in mice, in contrast, caused a loss-of-function, most readily observed with elevation of temperature [161]. Thus, it is likely that the H1849R mutation has other

99 effects in addition to the loss of FHF binding.

While H1849R-NaV1.5 has been shown to have significantly reduced binding affinity with FHF’s, the intracellular concentrations of the various FHF’s are not known.

Therefore, the degree to which FHF’s are interacting with and modulating the H1849R-

NaV1.5 channels cannot be determined. It is in fact possible that the expression of FHF’s is sufficient to interact with these mutant channels in a near-normal fashion. As a result, we cannot be certain of whether the altered channel properties stem from the reduced

FHF binding affinity, or from some other change in the channel structure and function, inherent to the specific mutation itself and independent of its impact on interactions with

FHFs.

A number of studies in which FHFs are knocked down in cardiac cells seem to suggest that the effects of NaV1.5 are loss-of-function in nature unlike the gain-of- function effects caused by H1849R [161]. Whether the difference in these findings stems from the degree to which FHF interaction with NaV1.5 is altered (completely eliminated in the knockdown, reduced binding affinity with H1849R) or is instead due to the unique effects of the H1849R mutation on the NaV1.5 channel itself remains to be determined.

One question we were particularly interested in was whether interactions with FHF – which is missing in HEK293 cells – could account for the lack of persistent current seen in our cardiomyocyte-based exploration of the IQ1908-1909AA mutation. While the extent to which interaction with FHF’s is abrogated with the H1849R mutation in cardiomyocytes is not clear, the H1849R mutation nonetheless provided a system in which we could attempt to explore this question. In our investigations, persistent current was non-significantly reduced in H1849R-IQ/AA cardiomyocytes compared to IQ/AA

100 cardiomyocytes, contrary to our expectations that altered FHF interaction could unmask persistent current due to lost calmodulin binding. Despite this finding, we still cannot rule out FHF’s as protecting against persistent current, since H1849R does not completely eliminate interactions with FHF’s.

Our explorations of IQ1908-1909AA NaV1.5 (Chapter 2) and H1849R NaV1.5 demonstrated that they had opposite effects on steady state inactivation as well as kinetics of inactivation. Thus, we were particularly interested in the properties of H1849R-

IQ1908-1909AA – NaV1.5 with both mutations present. Here, we saw an apparent dominant-negative effect of the H1849R mutation as these channels mimicked the

H1849R-NaV1.5 channels. Steady-state inactivation was shifted in a depolarizing direction and the rate of inactivation was slowed.

Another interesting finding in both H1849R and H1849R-IQ/AA cardiac cells is that the endogenous channels displayed normal voltage dependence of activation, steady-state inactivation and kinetics of inactivation. This differed from our findings in IQ/AA cardiac cells where endogenous channels displayed left shifts in both voltage dependence of activation as well as steady-state inactivation as well as a quicker rate of inactivation.

One possibility is that the H1849R mutation and altered FHF interactions either prevented dimerization of the endogenous and mutant channels, or prevented the allosteric functional effects of dimerization.

SUMMARY

The H1849R mutation results in gain-of-function effects in NaV1.5 including slowed rate of inactivation and increased channel availability. These effects occur with or

101 without calmodulin binding. The altered interaction with FHF in H1849R-NaV1.5 does not result in increased persistent current, even when combined with an IQ1908-1909AA mutation to block calmodulin binding.

102 Conclusions

+ NaV1.5, the cardiac voltage gated Na channel, plays a critical role in the cardiac action potential, rapidly activating to initiate the steep upstroke seen in phase zero.

Biophysical defects in NaV1.5 function can occur at any level- from mutations encoded in the genome to post-translational modifications occurring after it has been expressed.

Moreover, proper functioning of NaV1.5 does not merely require successful expression of the alpha and beta subunits of the channel. Rather, NaV1.5 relies on a vast web of molecular partners, which together regulate the channel and ensure its proper functionality. Alterations to the function of NaV1.5 have been shown to result in electrophysiological and structural disease of the heart. My thesis has investigated the pathological consequences of expression of a mutant NaV1.5 and has also explored the effects of auxiliary proteins in modulating its biophysical properties.

In chapter 1, we examined the electrophysiological and structural consequences of a gain-of-function mutation, F1759A in NaV1.5. We showed that F1759A increases the persistent component of Na+ current and increases window current. We showed that TG expression of this mutant channel in the mouse heart is sufficient to cause spontaneously initiated and long-lasting AF, a rare phenotype in mice. We further showed that expression of this mutant channel leads to frequent PVC’s and dilated cardiomyopathy.

Finally, we treated the mice with SEA0400, a specific NCX inhibitor, which resulted in a significant decrease in the AF burden as well as the incidence of PVC’s. This work is consistent with a growing body of evidence suggesting the pathological role played by increased persistent Na+ current. The mouse model phenocopies human AF in important respects, and represents a valuable tool for further investigations of AF.

103 In chapter 2, we shifted our investigations into the role played by auxiliary proteins in regulating the function of NaV1.5. Our focus is on exploring these interactions in cardiac myocytes, rather than heterologous systems through cardiac-specific TG expression. We initially investigated the role played by the ubiquitous calcium sensor, calmodulin. To our surprise, the IQ1908-1909AA mutation –which decreases calmodulin binding affinity - has loss-of-function effects on NaV1.5 function. Channel availability is decreased with steady-state inactivation of INa significantly shifted in a hyperpolarizing direction.

Similarly, the rate of inactivation is accelerated. Contrary to work in heterologous expression systems, we do not see increases in persistent current due to IQ1908-1909AA

NaV1.5. These results thus suggest that apocalmodulin’s effects on NaV1.5 are gain-of- function in nature, resulting in greater channel availability. Interestingly, we also show that endogenous channels are affected by the IQ/AA mutant channels with a similar decrease in channel availability. This suggests a possible role of dimerization in affecting channel properties. The work presented here is consistent with work showing that calmodulin increases availability of NaV1.4 [56].

In chapter 3, we continued our investigation into auxiliary proteins, exploring the role of FHF’s via the H1849R mutation, which decreases FHF binding affinity [62]. We show that H1849R leads to slowed inactivation and a depolarizing shift in steady-state inactivation suggesting that altered FHF interaction with NaV1.5 has gain-of-function effects on NaV1.5. We also show that combined H1849R and IQ1908-1909AA mutations similarly result in a depolarizing shift in steady-state inactivation suggesting a dominant effect of the altered FHF interaction over the loss of calmodulin binding. This work is consistent with previous explorations of H1849R which similarly showed right-shifted

104 steady-state inactivation in neonatal cardiomyocytes. It differs however, from other explorations of knockdown of FHF’s suggesting that while H1849R likely alters FHF interaction, it does not completely prevent it, and H1849R may have effects that are FHF- independent.

In summary, we have demonstrated that increased persistent Na+ current has deleterious effects in cardiac function and is sufficient to cause electrophysiological and structural abnormalities in the mouse heart including AF. We have also shed light on the critical role played by NaV1.5’s molecular partners in regulating its function within its native milieu. In doing so, we have presented a methodology which will enable other groups to readily carry out such investigations within cardiomyocytes.

105 Citations

1. Catterall, W.A., Voltage-gated sodium channels at 60: structure, function and pathophysiology. J Physiol, 2012. 590(11): p. 2577-89. 2. Luo, C.H. and Y. Rudy, A dynamic model of the cardiac ventricular action potential. I. Simulations of ionic currents and concentration changes. Circ Res, 1994. 74(6): p. 1071-96. 3. Wan, E., J. Abrams, R.L. Weinberg, A.N. Katchman, J. Bayne, S.I. Zakharov, . . . S.O. Marx, Aberrant sodium influx causes cardiomyopathy and atrial fibrillation in mice. J Clin Invest, 2016. 126(1): p. 112-22. 4. Piccolino, M., Animal electricity and the birth of electrophysiology: the legacy of Luigi Galvani. Brain Res Bull, 1998. 46(5): p. 381-407. 5. Hille, B., Ion channels of excitable membranes. 3rd ed. 2001, Sunderland, Mass.: Sinauer. xviii, 814 p. 6. Hodgkin, A.L. and A.F. Huxley, A quantitative description of membrane current and its application to conduction and excitation in nerve. J Physiol, 1952. 117(4): p. 500-44. 7. Nerbonne, J.M. and R.S. Kass, Molecular physiology of cardiac repolarization. Physiol Rev, 2005. 85(4): p. 1205-53. 8. Beneski, D.A. and W.A. Catterall, Covalent labeling of protein components of the sodium channel with a photoactivable derivative of scorpion toxin. Proc Natl Acad Sci U S A, 1980. 77(1): p. 639-43. 9. Hartshorne, R.P., B.U. Keller, J.A. Talvenheimo, W.A. Catterall, and M. Montal, Functional reconstitution of the purified brain sodium channel in planar lipid bilayers. Proc Natl Acad Sci U S A, 1985. 82(1): p. 240-4. 10. Goldin, A.L., Resurgence of sodium channel research. Annu Rev Physiol, 2001. 63: p. 871-94. 11. Wang, Q., Z. Li, J. Shen, and M.T. Keating, Genomic organization of the human SCN5A gene encoding the cardiac sodium channel. Genomics, 1996. 34(1): p. 9-16. 12. Noda, M., S. Shimizu, T. Tanabe, T. Takai, T. Kayano, T. Ikeda, . . . et al., Primary structure of Electrophorus electricus sodium channel deduced from cDNA sequence. Nature, 1984. 312(5990): p. 121-7. 13. Guy, H.R. and P. Seetharamulu, Molecular model of the action potential sodium channel. Proc Natl Acad Sci U S A, 1986. 83(2): p. 508-12. 14. George, A.L., Jr., Inherited disorders of voltage-gated sodium channels. J Clin Invest, 2005. 115(8): p. 1990-9. 15. Yang, N. and R. Horn, Evidence for voltage-dependent S4 movement in sodium channels. Neuron, 1995. 15(1): p. 213-8. 16. Stuhmer, W., F. Conti, H. Suzuki, X.D. Wang, M. Noda, N. Yahagi, . . . S. Numa, Structural parts involved in activation and inactivation of the sodium channel. Nature, 1989. 339(6226): p. 597-603.

106 17. Hartmann, H.A., G.E. Kirsch, J.A. Drewe, M. Taglialatela, R.H. Joho, and A.M. Brown, Exchange of conduction pathways between two related K+ channels. Science, 1991. 251(4996): p. 942-4. 18. Armstrong, C.M., Sodium channels and gating currents. Physiol Rev, 1981. 61(3): p. 644-83. 19. Vassilev, P., T. Scheuer, and W.A. Catterall, Inhibition of inactivation of single sodium channels by a site-directed antibody. Proc Natl Acad Sci U S A, 1989. 86(20): p. 8147-51. 20. West, J.W., D.E. Patton, T. Scheuer, Y. Wang, A.L. Goldin, and W.A. Catterall, A cluster of hydrophobic amino acid residues required for fast Na(+)-channel inactivation. Proc Natl Acad Sci U S A, 1992. 89(22): p. 10910-4. 21. Eaholtz, G., T. Scheuer, and W.A. Catterall, Restoration of inactivation and block of open sodium channels by an inactivation gate peptide. Neuron, 1994. 12(5): p. 1041-8. 22. Payandeh, J., T. Scheuer, N. Zheng, and W.A. Catterall, The crystal structure of a voltage-gated sodium channel. Nature, 2011. 475(7356): p. 353-8. 23. Zipes, D.P. and J. Jalife, Cardiac electrophysiology : from cell to bedside. Sixth edition. ed. 2014, Philadelphia, PA: Elsevier/Saunders. xxvi, 1365 pages. 24. Moreno, J.D. and C.E. Clancy, Pathophysiology of the cardiac late Na current and its potential as a drug target. J Mol Cell Cardiol, 2012. 52(3): p. 608-19. 25. Kiyosue, T. and M. Arita, Late sodium current and its contribution to action potential configuration in guinea pig ventricular myocytes. Circ Res, 1989. 64(2): p. 389-97. 26. Undrovinas, A.I., I.A. Fleidervish, and J.C. Makielski, Inward sodium current at resting potentials in single cardiac myocytes induced by the ischemic metabolite lysophosphatidylcholine. Circ Res, 1992. 71(5): p. 1231-41. 27. Saint, D.A., Y.K. Ju, and P.W. Gage, A persistent sodium current in rat ventricular myocytes. J Physiol, 1992. 453: p. 219-31. 28. Noble, D. and P.J. Noble, Late sodium current in the pathophysiology of cardiovascular disease: consequences of sodium-calcium overload. Heart, 2006. 92 Suppl 4: p. iv1-iv5. 29. Wasserstrom, J.A. and J.J. Salata, Basis for tetrodotoxin and lidocaine effects on action potentials in dog ventricular myocytes. Am J Physiol, 1988. 254(6 Pt 2): p. H1157-66. 30. Maltsev, V.A., H.N. Sabbah, R.S. Higgins, N. Silverman, M. Lesch, and A.I. Undrovinas, Novel, ultraslow inactivating sodium current in human ventricular cardiomyocytes. Circulation, 1998. 98(23): p. 2545-52. 31. Maltsev, V.A. and A.I. Undrovinas, A multi-modal composition of the late Na+ current in human ventricular cardiomyocytes. Cardiovasc Res, 2006. 69(1): p. 116-27. 32. Clancy, C.E., M. Tateyama, and R.S. Kass, Insights into the molecular mechanisms of bradycardia-triggered arrhythmias in long QT-3 syndrome. J Clin Invest, 2002. 110(9): p. 1251-62. 33. Zygmunt, A.C., G.T. Eddlestone, G.P. Thomas, V.V. Nesterenko, and C. Antzelevitch, Larger late sodium conductance in M cells contributes to

107 electrical heterogeneity in canine ventricle. Am J Physiol Heart Circ Physiol, 2001. 281(2): p. H689-97. 34. Nemec, J., M. Buncova, V. Bulkova, J. Hejlik, B. Winter, W.K. Shen, and M.J. Ackerman, Heart rate dependence of the QT interval duration: differences among congenital long QT syndrome subtypes. J Cardiovasc Electrophysiol, 2004. 15(5): p. 550-6. 35. Bennett, P.B., K. Yazawa, N. Makita, and A.L. George, Jr., Molecular mechanism for an inherited cardiac arrhythmia. Nature, 1995. 376(6542): p. 683-5. 36. Rivolta, I., H. Abriel, M. Tateyama, H. Liu, M. Memmi, P. Vardas, . . . R.S. Kass, Inherited Brugada and long QT-3 syndrome mutations of a single residue of the cardiac sodium channel confer distinct channel and clinical phenotypes. J Biol Chem, 2001. 276(33): p. 30623-30. 37. Veldkamp, M.W., P.C. Viswanathan, C. Bezzina, A. Baartscheer, A.A. Wilde, and J.R. Balser, Two distinct congenital arrhythmias evoked by a multidysfunctional Na(+) channel. Circ Res, 2000. 86(9): p. E91-7. 38. Vatta, M., M.J. Ackerman, B. Ye, J.C. Makielski, E.E. Ughanze, E.W. Taylor, . . . J.A. Towbin, Mutant caveolin-3 induces persistent late sodium current and is associated with long-QT syndrome. Circulation, 2006. 114(20): p. 2104-12. 39. Cronk, L.B., B. Ye, T. Kaku, D.J. Tester, M. Vatta, J.C. Makielski, and M.J. Ackerman, Novel mechanism for sudden infant death syndrome: persistent late sodium current secondary to mutations in caveolin-3. Heart Rhythm, 2007. 4(2): p. 161-6. 40. Ueda, K., C. Valdivia, A. Medeiros-Domingo, D.J. Tester, M. Vatta, G. Farrugia, . . . J.C. Makielski, Syntrophin mutation associated with long QT syndrome through activation of the nNOS-SCN5A macromolecular complex. Proc Natl Acad Sci U S A, 2008. 105(27): p. 9355-60. 41. Cheng, J., D.W. Van Norstrand, A. Medeiros-Domingo, C. Valdivia, B.H. Tan, B. Ye, . . . M.J. Ackerman, Alpha1-syntrophin mutations identified in sudden infant death syndrome cause an increase in late cardiac sodium current. Circ Arrhythm Electrophysiol, 2009. 2(6): p. 667-76. 42. Wagner, S., N. Dybkova, E.C. Rasenack, C. Jacobshagen, L. Fabritz, P. Kirchhof, . . . L.S. Maier, Ca2+/calmodulin-dependent protein kinase II regulates cardiac Na+ channels. J Clin Invest, 2006. 116(12): p. 3127-38. 43. Lu, Z., Y.P. Jiang, C.Y. Wu, L.M. Ballou, S. Liu, E.S. Carpenter, . . . R.Z. Lin, Increased persistent sodium current due to decreased PI3K signaling contributes to QT prolongation in the diabetic heart. Diabetes, 2013. 62(12): p. 4257-65. 44. Yang, T., Y.W. Chun, D.M. Stroud, J.D. Mosley, B.C. Knollmann, C. Hong, and D.M. Roden, Screening for acute IKr block is insufficient to detect torsades de pointes liability: role of late sodium current. Circulation, 2014. 130(3): p. 224- 34. 45. Shy, D., L. Gillet, and H. Abriel, Cardiac sodium channel NaV1.5 distribution in myocytes via interacting proteins: the multiple pool model. Biochim Biophys Acta, 2013. 1833(4): p. 886-94. 46. van Bemmelen, M.X., J.S. Rougier, B. Gavillet, F. Apotheloz, D. Daidie, M. Tateyama, . . . H. Abriel, Cardiac voltage-gated sodium channel Nav1.5 is 108 regulated by Nedd4-2 mediated ubiquitination. Circ Res, 2004. 95(3): p. 284- 91. 47. Allouis, M., F. Le Bouffant, R. Wilders, D. Peroz, J.J. Schott, J. Noireaud, . . . I. Baro, 14-3-3 is a regulator of the cardiac voltage-gated sodium channel Nav1.5. Circ Res, 2006. 98(12): p. 1538-46. 48. Lemaillet, G., B. Walker, and S. Lambert, Identification of a conserved ankyrin- binding motif in the family of sodium channel alpha subunits. J Biol Chem, 2003. 278(30): p. 27333-9. 49. Wu, L., S.L. Yong, C. Fan, Y. Ni, S. Yoo, T. Zhang, . . . Q.K. Wang, Identification of a new co-factor, MOG1, required for the full function of cardiac sodium channel Nav 1.5. J Biol Chem, 2008. 283(11): p. 6968-78. 50. Babu, Y.S., J.S. Sack, T.J. Greenhough, C.E. Bugg, A.R. Means, and W.J. Cook, Three-dimensional structure of calmodulin. Nature, 1985. 315(6014): p. 37- 40. 51. Peterson, B.Z., C.D. DeMaria, J.P. Adelman, and D.T. Yue, Calmodulin is the Ca2+ sensor for Ca2+ -dependent inactivation of L-type calcium channels. Neuron, 1999. 22(3): p. 549-58. 52. Tan, H.L., S. Kupershmidt, R. Zhang, S. Stepanovic, D.M. Roden, A.A. Wilde, . . . J.R. Balser, A calcium sensor in the sodium channel modulates cardiac excitability. Nature, 2002. 415(6870): p. 442-7. 53. Kim, J., S. Ghosh, H. Liu, M. Tateyama, R.S. Kass, and G.S. Pitt, Calmodulin mediates Ca2+ sensitivity of sodium channels. J Biol Chem, 2004. 279(43): p. 45004-12. 54. Makita, N., M. Horie, T. Nakamura, T. Ai, K. Sasaki, H. Yokoi, . . . A. Kitabatake, Drug-induced long-QT syndrome associated with a subclinical SCN5A mutation. Circulation, 2002. 106(10): p. 1269-74. 55. Ben-Johny, M., P.S. Yang, J. Niu, W. Yang, R. Joshi-Mukherjee, and D.T. Yue, Conservation of Ca2+/calmodulin regulation across Na and Ca2+ channels. Cell, 2014. 157(7): p. 1657-70. 56. Adams, P.J., M. Ben-Johny, I.E. Dick, T. Inoue, and D.T. Yue, Apocalmodulin itself promotes ion channel opening and Ca(2+) regulation. Cell, 2014. 159(3): p. 608-22. 57. Liu, C.J., S.D. Dib-Hajj, M. Renganathan, T.R. Cummins, and S.G. Waxman, Modulation of the cardiac sodium channel Nav1.5 by fibroblast growth factor homologous factor 1B. J Biol Chem, 2003. 278(2): p. 1029-36. 58. Goetz, R., K. Dover, F. Laezza, N. Shtraizent, X. Huang, D. Tchetchik, . . . M. Mohammadi, Crystal structure of a fibroblast growth factor homologous factor (FHF) defines a conserved surface on FHFs for binding and modulation of voltage-gated sodium channels. J Biol Chem, 2009. 284(26): p. 17883-96. 59. Lou, J.Y., F. Laezza, B.R. Gerber, M. Xiao, K.A. Yamada, H. Hartmann, . . . D.M. Ornitz, Fibroblast growth factor 14 is an intracellular modulator of voltage- gated sodium channels. J Physiol, 2005. 569(Pt 1): p. 179-93. 60. Wang, C., J.A. Hennessey, R.D. Kirkton, C. Wang, V. Graham, R.S. Puranam, . . . G.S. Pitt, Fibroblast growth factor homologous factor 13 regulates Na+ channels and conduction velocity in murine hearts. Circ Res, 2011. 109(7): p. 775-82. 109 61. Yang, J., Z. Wang, D.S. Sinden, X. Wang, B. Shan, X. Yu, . . . C. Wang, FGF13 modulates the gating properties of the cardiac sodium channel Nav1.5 in an isoform-specific manner. Channels (Austin), 2016. 10(5): p. 410-420. 62. Musa, H., C.F. Kline, A.C. Sturm, N. Murphy, S. Adelman, C. Wang, . . . P.J. Mohler, SCN5A variant that blocks fibroblast growth factor homologous factor regulation causes human arrhythmia. Proc Natl Acad Sci U S A, 2015. 112(40): p. 12528-33. 63. Wang, C., B.C. Chung, H. Yan, S.Y. Lee, and G.S. Pitt, Crystal structure of the ternary complex of a NaV C-terminal domain, a fibroblast growth factor homologous factor, and calmodulin. Structure, 2012. 20(7): p. 1167-76. 64. Liu, M., K.C. Yang, and S.C. Dudley, Jr., Cardiac Sodium Channel Mutations: Why so Many Phenotypes? Curr Top Membr, 2016. 78: p. 513-59. 65. Wang, Q., J. Shen, I. Splawski, D. Atkinson, Z. Li, J.L. Robinson, . . . M.T. Keating, SCN5A mutations associated with an inherited cardiac arrhythmia, long QT syndrome. Cell, 1995. 80(5): p. 805-11. 66. McNair, W.P., L. Ku, M.R. Taylor, P.R. Fain, D. Dao, E. Wolfel, . . . G. Familial Cardiomyopathy Registry Research, SCN5A mutation associated with dilated cardiomyopathy, conduction disorder, and arrhythmia. Circulation, 2004. 110(15): p. 2163-7. 67. Wakili, R., N. Voigt, S. Kaab, D. Dobrev, and S. Nattel, Recent advances in the molecular pathophysiology of atrial fibrillation. J Clin Invest, 2011. 121(8): p. 2955-68. 68. Colilla, S., A. Crow, W. Petkun, D.E. Singer, T. Simon, and X. Liu, Estimates of current and future incidence and prevalence of atrial fibrillation in the U.S. adult population. Am J Cardiol, 2013. 112(8): p. 1142-7. 69. Go, A.S., D. Mozaffarian, V.L. Roger, E.J. Benjamin, J.D. Berry, W.B. Borden, . . . S. Stroke Statistics, Executive summary: heart disease and stroke statistics-- 2013 update: a report from the American Heart Association. Circulation, 2013. 127(1): p. 143-52. 70. Heijman, J., N. Voigt, S. Nattel, and D. Dobrev, Cellular and molecular electrophysiology of atrial fibrillation initiation, maintenance, and progression. Circ Res, 2014. 114(9): p. 1483-99. 71. Ehrlich, J.R., S. Zicha, P. Coutu, T.E. Hebert, and S. Nattel, Atrial fibrillation- associated minK38G/S polymorphism modulates delayed rectifier current and membrane localization. Cardiovasc Res, 2005. 67(3): p. 520-8. 72. Lemoine, M.D., J.E. Duverger, P. Naud, D. Chartier, X.Y. Qi, P. Comtois, . . . S. Nattel, Arrhythmogenic left atrial cellular electrophysiology in a murine genetic long QT syndrome model. Cardiovasc Res, 2011. 92(1): p. 67-74. 73. January, C.T. and J.M. Riddle, Early afterdepolarizations: mechanism of induction and block. A role for L-type Ca2+ current. Circ Res, 1989. 64(5): p. 977-90. 74. January, C.T. and H.A. Fozzard, Delayed afterdepolarizations in heart muscle: mechanisms and relevance. Pharmacol Rev, 1988. 40(3): p. 219-27. 75. Wehrens, X.H., S.E. Lehnart, S.R. Reiken, and A.R. Marks, Ca2+/calmodulin- dependent protein kinase II phosphorylation regulates the cardiac ryanodine receptor. Circ Res, 2004. 94(6): p. e61-70. 110 76. Brillantes, A.B., K. Ondrias, A. Scott, E. Kobrinsky, E. Ondriasova, M.C. Moschella, . . . A.R. Marks, Stabilization of calcium release channel (ryanodine receptor) function by FK506-binding protein. Cell, 1994. 77(4): p. 513-23. 77. Iwasaki, Y.K., K. Nishida, T. Kato, and S. Nattel, Atrial fibrillation pathophysiology: implications for management. Circulation, 2011. 124(20): p. 2264-74. 78. Comtois, P., J. Kneller, and S. Nattel, Of circles and spirals: bridging the gap between the leading circle and spiral wave concepts of cardiac reentry. Europace, 2005. 7 Suppl 2: p. 10-20. 79. Pandit, S.V. and J. Jalife, Rotors and the dynamics of cardiac fibrillation. Circ Res, 2013. 112(5): p. 849-62. 80. Nattel, S., New ideas about atrial fibrillation 50 years on. Nature, 2002. 415(6868): p. 219-26. 81. Davidenko, J.M., P.F. Kent, D.R. Chialvo, D.C. Michaels, and J. Jalife, Sustained vortex-like waves in normal isolated ventricular muscle. Proc Natl Acad Sci U S A, 1990. 87(22): p. 8785-9. 82. Haissaguerre, M., P. Jais, D.C. Shah, A. Takahashi, M. Hocini, G. Quiniou, . . . J. Clementy, Spontaneous initiation of atrial fibrillation by ectopic beats originating in the pulmonary veins. N Engl J Med, 1998. 339(10): p. 659-66. 83. Chelu, M.G., S. Sarma, S. Sood, S. Wang, R.J. van Oort, D.G. Skapura, . . . X.H. Wehrens, Calmodulin kinase II-mediated sarcoplasmic reticulum Ca2+ leak promotes atrial fibrillation in mice. J Clin Invest, 2009. 119(7): p. 1940-51. 84. Wolff , L., Familial Auricular Fibrillation. New England Journal of Medicine, 1943. 229(10): p. 396-398. 85. Lubitz, S.A., X. Yin, J.D. Fontes, J.W. Magnani, M. Rienstra, M. Pai, . . . E.J. Benjamin, Association between familial atrial fibrillation and risk of new-onset atrial fibrillation. JAMA, 2010. 304(20): p. 2263-9. 86. Brugada, R., T. Tapscott, G.Z. Czernuszewicz, A.J. Marian, A. Iglesias, L. Mont, . . . R. Roberts, Identification of a genetic locus for familial atrial fibrillation. N Engl J Med, 1997. 336(13): p. 905-11. 87. Chen, Y.H., S.J. Xu, S. Bendahhou, X.L. Wang, Y. Wang, W.Y. Xu, . . . W. Huang, KCNQ1 gain-of-function mutation in familial atrial fibrillation. Science, 2003. 299(5604): p. 251-4. 88. Tucker, N.R. and P.T. Ellinor, Emerging directions in the genetics of atrial fibrillation. Circ Res, 2014. 114(9): p. 1469-82. 89. Olson, T.M., A.E. Alekseev, X.K. Liu, S. Park, L.V. Zingman, M. Bienengraeber, . . . A. Terzic, Kv1.5 channelopathy due to KCNA5 loss-of-function mutation causes human atrial fibrillation. Hum Mol Genet, 2006. 15(14): p. 2185-91. 90. Qiu, X.B., Y.J. Xu, R.G. Li, L. Xu, X. Liu, W.Y. Fang, . . . X.K. Qu, PITX2C loss-of- function mutations responsible for idiopathic atrial fibrillation. Clinics (Sao Paulo), 2014. 69(1): p. 15-22. 91. Darbar, D., P.J. Kannankeril, B.S. Donahue, G. Kucera, T. Stubblefield, J.L. Haines, . . . D.M. Roden, Cardiac sodium channel (SCN5A) variants associated with atrial fibrillation. Circulation, 2008. 117(15): p. 1927-35. 92. Benito, B., R. Brugada, R.M. Perich, E. Lizotte, J. Cinca, L. Mont, . . . J. Brugada, A mutation in the sodium channel is responsible for the association of long QT 111 syndrome and familial atrial fibrillation. Heart Rhythm, 2008. 5(10): p. 1434- 40. 93. Johnson, J.N., D.J. Tester, J. Perry, B.A. Salisbury, C.R. Reed, and M.J. Ackerman, Prevalence of early-onset atrial fibrillation in congenital long QT syndrome. Heart Rhythm, 2008. 5(5): p. 704-9. 94. Sossalla, S., B. Kallmeyer, S. Wagner, M. Mazur, U. Maurer, K. Toischer, . . . L.S. Maier, Altered Na(+) currents in atrial fibrillation effects of ranolazine on arrhythmias and contractility in human atrial myocardium. J Am Coll Cardiol, 2010. 55(21): p. 2330-42. 95. Scirica, B.M., D.A. Morrow, H. Hod, S.A. Murphy, L. Belardinelli, C.M. Hedgepeth, . . . E. Braunwald, Effect of ranolazine, an antianginal agent with novel electrophysiological properties, on the incidence of arrhythmias in patients with non ST-segment elevation acute coronary syndrome: results from the Metabolic Efficiency With Ranolazine for Less Ischemia in Non ST-Elevation Acute Coronary Syndrome Thrombolysis in Myocardial Infarction 36 (MERLIN- TIMI 36) randomized controlled trial. Circulation, 2007. 116(15): p. 1647-52. 96. Reiffel, J.A., A.J. Camm, L. Belardinelli, D. Zeng, E. Karwatowska-Prokopczuk, A. Olmsted, . . . H. Investigators, The HARMONY Trial: Combined Ranolazine and Dronedarone in the Management of Paroxysmal Atrial Fibrillation: Mechanistic and Therapeutic Synergism. Circ Arrhythm Electrophysiol, 2015. 8(5): p. 1048-56. 97. de Vos, C.B., R. Pisters, R. Nieuwlaat, M.H. Prins, R.G. Tieleman, R.J. Coelen, . . . H.J. Crijns, Progression from paroxysmal to persistent atrial fibrillation clinical correlates and prognosis. J Am Coll Cardiol, 2010. 55(8): p. 725-31. 98. Wijffels, M.C., C.J. Kirchhof, R. Dorland, and M.A. Allessie, Atrial fibrillation begets atrial fibrillation. A study in awake chronically instrumented goats. Circulation, 1995. 92(7): p. 1954-68. 99. Li, D., S. Fareh, T.K. Leung, and S. Nattel, Promotion of atrial fibrillation by heart failure in dogs: atrial remodeling of a different sort. Circulation, 1999. 100(1): p. 87-95. 100. Yue, L., J. Feng, R. Gaspo, G.R. Li, Z. Wang, and S. Nattel, Ionic remodeling underlying action potential changes in a canine model of atrial fibrillation. Circ Res, 1997. 81(4): p. 512-25. 101. Voigt, N., A. Maguy, Y.H. Yeh, X. Qi, U. Ravens, D. Dobrev, and S. Nattel, Changes in I K, ACh single-channel activity with atrial tachycardia remodelling in canine atrial cardiomyocytes. Cardiovasc Res, 2008. 77(1): p. 35-43. 102. Allessie, M.A., F.I. Bonke, and F.J. Schopman, Circus movement in rabbit atrial muscle as a mechanism of tachycardia. II. The role of nonuniform recovery of excitability in the occurrence of unidirectional block, as studied with multiple microelectrodes. Circ Res, 1976. 39(2): p. 168-77. 103. Allessie, M.A., Atrial electrophysiologic remodeling: another vicious circle? J Cardiovasc Electrophysiol, 1998. 9(12): p. 1378-93. 104. Voigt, N., N. Li, Q. Wang, W. Wang, A.W. Trafford, I. Abu-Taha, . . . D. Dobrev, Enhanced sarcoplasmic reticulum Ca2+ leak and increased Na+-Ca2+ exchanger function underlie delayed afterdepolarizations in patients with chronic atrial fibrillation. Circulation, 2012. 125(17): p. 2059-70. 112 105. Spach, M.S. and P.C. Dolber, Relating extracellular potentials and their derivatives to anisotropic propagation at a microscopic level in human cardiac muscle. Evidence for electrical uncoupling of side-to-side fiber connections with increasing age. Circ Res, 1986. 58(3): p. 356-71. 106. Tan, A.Y., S. Zhou, M. Ogawa, J. Song, M. Chu, H. Li, . . . P.S. Chen, Neural mechanisms of paroxysmal atrial fibrillation and paroxysmal atrial tachycardia in ambulatory canines. Circulation, 2008. 118(9): p. 916-25. 107. Gould, P.A., M. Yii, C. McLean, S. Finch, T. Marshall, G.W. Lambert, and D.M. Kaye, Evidence for increased atrial sympathetic innervation in persistent human atrial fibrillation. Pacing Clin Electrophysiol, 2006. 29(8): p. 821-9. 108. Dobrev, D., N. Voigt, and X.H. Wehrens, The ryanodine receptor channel as a molecular motif in atrial fibrillation: pathophysiological and therapeutic implications. Cardiovasc Res, 2011. 89(4): p. 734-43. 109. Kneller, J., R. Zou, E.J. Vigmond, Z. Wang, L.J. Leon, and S. Nattel, Cholinergic atrial fibrillation in a computer model of a two-dimensional sheet of canine atrial cells with realistic ionic properties. Circ Res, 2002. 90(9): p. E73-87. 110. Hohnloser, S.H., K.H. Kuck, and J. Lilienthal, Rhythm or rate control in atrial fibrillation--Pharmacological Intervention in Atrial Fibrillation (PIAF): a randomised trial. Lancet, 2000. 356(9244): p. 1789-94. 111. Cappato, R., H. Calkins, S.A. Chen, W. Davies, Y. Iesaka, J. Kalman, . . . E. Biganzoli, Updated worldwide survey on the methods, efficacy, and safety of catheter ablation for human atrial fibrillation. Circ Arrhythm Electrophysiol, 2010. 3(1): p. 32-8. 112. Riley, G., F. Syeda, P. Kirchhof, and L. Fabritz, An introduction to murine models of atrial fibrillation. Front Physiol, 2012. 3: p. 296. 113. Adam, O., G. Frost, F. Custodis, M.A. Sussman, H.J. Schafers, M. Bohm, and U. Laufs, Role of Rac1 GTPase activation in atrial fibrillation. J Am Coll Cardiol, 2007. 50(4): p. 359-67. 114. Choi, E.K., P.C. Chang, Y.S. Lee, S.F. Lin, W. Zhu, M. Maruyama, . . . P.S. Chen, Triggered firing and atrial fibrillation in transgenic mice with selective atrial fibrosis induced by overexpression of TGF-beta1. Circ J, 2012. 76(6): p. 1354- 62. 115. Glukhov, A.V., A. Kalyanasundaram, Q. Lou, L.T. Hage, B.J. Hansen, A.E. Belevych, . . . V.V. Fedorov, Calsequestrin 2 deletion causes sinoatrial node dysfunction and atrial arrhythmias associated with altered sarcoplasmic reticulum calcium cycling and degenerative fibrosis within the mouse atrial pacemaker complex1. Eur Heart J, 2015. 36(11): p. 686-97. 116. Watanabe, H., T. Yang, D.M. Stroud, J.S. Lowe, L. Harris, T.C. Atack, . . . D.M. Roden, Striking In vivo phenotype of a disease-associated human SCN5A mutation producing minimal changes in vitro. Circulation, 2011. 124(9): p. 1001-11. 117. Carneiro, J.S., A.S. Bento, D. Bacic, B.D. Nearing, S. Rajamani, L. Belardinelli, and R.L. Verrier, The Selective Cardiac Late Sodium Current Inhibitor GS- 458967 Suppresses Autonomically Triggered Atrial Fibrillation in an Intact Porcine Model. J Cardiovasc Electrophysiol, 2015. 26(12): p. 1364-9.

113 118. Nishida, K., X.Y. Qi, R. Wakili, P. Comtois, D. Chartier, M. Harada, . . . S. Nattel, Mechanisms of atrial tachyarrhythmias associated with coronary artery occlusion in a chronic canine model. Circulation, 2011. 123(2): p. 137-46. 119. Gellens, M.E., A.L. George, Jr., L.Q. Chen, M. Chahine, R. Horn, R.L. Barchi, and R.G. Kallen, Primary structure and functional expression of the human cardiac tetrodotoxin-insensitive voltage-dependent sodium channel. Proc Natl Acad Sci U S A, 1992. 89(2): p. 554-8. 120. Zhang, T., S.L. Yong, X.L. Tian, and Q.K. Wang, Cardiac-specific overexpression of SCN5A gene leads to shorter P wave duration and PR interval in transgenic mice. Biochem Biophys Res Commun, 2007. 355(2): p. 444-50. 121. Sanbe, A., J. Gulick, M.C. Hanks, Q. Liang, H. Osinska, and J. Robbins, Reengineering inducible cardiac-specific transgenesis with an attenuated myosin heavy chain promoter. Circ Res, 2003. 92(6): p. 609-16. 122. Hambleton, M., A. York, M.A. Sargent, R.A. Kaiser, J.N. Lorenz, J. Robbins, and J.D. Molkentin, Inducible and myocyte-specific inhibition of PKCalpha enhances cardiac contractility and protects against infarction-induced heart failure. Am J Physiol Heart Circ Physiol, 2007. 293(6): p. H3768-71. 123. Valencik, M.L. and J.A. McDonald, Codon optimization markedly improves doxycycline regulated gene expression in the mouse heart. Transgenic Res, 2001. 10(3): p. 269-75. 124. Satin, J., J.W. Kyle, M. Chen, P. Bell, L.L. Cribbs, H.A. Fozzard, and R.B. Rogart, A mutant of TTX-resistant cardiac sodium channels with TTX-sensitive properties. Science, 1992. 256(5060): p. 1202-5. 125. O'Connell, T.D., M.C. Rodrigo, and P.C. Simpson, Isolation and culture of adult mouse cardiac myocytes. Methods Mol Biol, 2007. 357: p. 271-96. 126. Nguyen, T.P., D.W. Wang, T.H. Rhodes, and A.L. George, Jr., Divergent biophysical defects caused by mutant sodium channels in dilated cardiomyopathy with arrhythmia. Circ Res, 2008. 102(3): p. 364-71. 127. Hennessey, J.A., E.Q. Wei, and G.S. Pitt, Fibroblast growth factor homologous factors modulate cardiac calcium channels. Circ Res, 2013. 113(4): p. 381-8. 128. Gutbrod, S.R., R. Walton, S. Gilbert, V. Meillet, P. Jais, M. Hocini, . . . I.R. Efimov, Quantification of the transmural dynamics of atrial fibrillation by simultaneous endocardial and epicardial optical mapping in an acute sheep model. Circ Arrhythm Electrophysiol, 2015. 8(2): p. 456-65. 129. Laughner, J.I., F.S. Ng, M.S. Sulkin, R.M. Arthur, and I.R. Efimov, Processing and analysis of cardiac optical mapping data obtained with potentiometric dyes. Am J Physiol Heart Circ Physiol, 2012. 303(7): p. H753-65. 130. Ragsdale, D.S., J.C. McPhee, T. Scheuer, and W.A. Catterall, Molecular determinants of state-dependent block of Na+ channels by local anesthetics. Science, 1994. 265(5179): p. 1724-8. 131. McNulty, M.M., G.B. Edgerton, R.D. Shah, D.A. Hanck, H.A. Fozzard, and G.M. Lipkind, Charge at the lidocaine binding site residue Phe-1759 affects permeation in human cardiac voltage-gated sodium channels. J Physiol, 2007. 581(Pt 2): p. 741-55.

114 132. Carboni, M., Z.S. Zhang, V. Neplioueva, C.F. Starmer, and A.O. Grant, Slow sodium channel inactivation and use-dependent block modulated by the same domain IV S6 residue. J Membr Biol, 2005. 207(2): p. 107-17. 133. Sossalla, S., U. Maurer, H. Schotola, N. Hartmann, M. Didie, W.H. Zimmermann, . . . L.S. Maier, Diastolic dysfunction and arrhythmias caused by overexpression of CaMKIIdelta(C) can be reversed by inhibition of late Na(+) current. Basic Res Cardiol, 2011. 106(2): p. 263-72. 134. Valdivia, C.R., W.W. Chu, J. Pu, J.D. Foell, R.A. Haworth, M.R. Wolff, . . . J.C. Makielski, Increased late sodium current in myocytes from a canine heart failure model and from failing human heart. J Mol Cell Cardiol, 2005. 38(3): p. 475-83. 135. Wagner, E., M.A. Lauterbach, T. Kohl, V. Westphal, G.S. Williams, J.H. Steinbrecher, . . . S.E. Lehnart, Stimulated emission depletion live-cell super- resolution imaging shows proliferative remodeling of T-tubule membrane structures after myocardial infarction. Circ Res, 2012. 111(4): p. 402-14. 136. Kohlhaas, M., T. Liu, A. Knopp, T. Zeller, M.F. Ong, M. Bohm, . . . C. Maack, Elevated cytosolic Na+ increases mitochondrial formation of reactive oxygen species in failing cardiac myocytes. Circulation, 2010. 121(14): p. 1606-13. 137. Iwai, T., K. Tanonaka, R. Inoue, S. Kasahara, K. Motegi, S. Nagaya, and S. Takeo, Sodium accumulation during ischemia induces mitochondrial damage in perfused rat hearts. Cardiovasc Res, 2002. 55(1): p. 141-9. 138. Nuyens, D., M. Stengl, S. Dugarmaa, T. Rossenbacker, V. Compernolle, Y. Rudy, . . . P. Carmeliet, Abrupt rate accelerations or premature beats cause life- threatening arrhythmias in mice with long-QT3 syndrome. Nat Med, 2001. 7(9): p. 1021-7. 139. Ju, Y.K., D.A. Saint, and P.W. Gage, Hypoxia increases persistent sodium current in rat ventricular myocytes. J Physiol, 1996. 497 ( Pt 2): p. 337-47. 140. Luczak, E.D. and M.E. Anderson, CaMKII oxidative activation and the pathogenesis of cardiac disease. J Mol Cell Cardiol, 2014. 73: p. 112-6. 141. Purohit, A., A.G. Rokita, X. Guan, B. Chen, O.M. Koval, N. Voigt, . . . M.E. Anderson, Oxidized Ca(2+)/calmodulin-dependent protein kinase II triggers atrial fibrillation. Circulation, 2013. 128(16): p. 1748-57. 142. Bourgonje, V.J., M.A. Vos, S. Ozdemir, N. Doisne, K. Acsai, A. Varro, . . . G. Antoons, Combined Na(+)/Ca(2+) exchanger and L-type calcium channel block as a potential strategy to suppress arrhythmias and maintain ventricular function. Circ Arrhythm Electrophysiol, 2013. 6(2): p. 371-9. 143. Milberg, P., C. Pott, G. Frommeyer, M. Fink, M. Ruhe, T. Matsuda, . . . L. Eckardt, Acute inhibition of the Na(+)/Ca(2+) exchanger reduces proarrhythmia in an experimental model of chronic heart failure. Heart Rhythm, 2012. 9(4): p. 570- 8. 144. Mann, S.A., M.L. Castro, M. Ohanian, G. Guo, P. Zodgekar, A. Sheu, . . . D. Fatkin, R222Q SCN5A mutation is associated with reversible ventricular ectopy and dilated cardiomyopathy. J Am Coll Cardiol, 2012. 60(16): p. 1566-73. 145. Jalife, J., Mechanisms of persistent atrial fibrillation. Curr Opin Cardiol, 2014. 29(1): p. 20-7.

115 146. Umapathy, K., K. Nair, S. Masse, S. Krishnan, J. Rogers, M.P. Nash, and K. Nanthakumar, Phase mapping of cardiac fibrillation. Circ Arrhythm Electrophysiol, 2010. 3(1): p. 105-14. 147. Dautova, Y., Y. Zhang, I. Sabir, A.A. Grace, and C.L. Huang, Atrial arrhythmogenesis in wild-type and Scn5a+/delta murine hearts modelling LQT3 syndrome. Pflugers Arch, 2009. 458(3): p. 443-57. 148. Guzadhur, L., S.M. Pearcey, R.M. Duehmke, K. Jeevaratnam, A.F. Hohmann, Y. Zhang, . . . C.L. Huang, Atrial arrhythmogenicity in aged Scn5a+/DeltaKPQ mice modeling long QT type 3 syndrome and its relationship to Na+ channel expression and cardiac conduction. Pflugers Arch, 2010. 460(3): p. 593-601. 149. Blana, A., S. Kaese, L. Fortmuller, S. Laakmann, D. Damke, K. van Bragt, . . . L. Fabritz, Knock-in gain-of-function sodium channel mutation prolongs atrial action potentials and alters atrial vulnerability. Heart Rhythm, 2010. 7(12): p. 1862-9. 150. Zhang, T., S.L. Yong, J.K. Drinko, Z.B. Popovic, J.C. Shryock, L. Belardinelli, and Q.K. Wang, LQTS mutation N1325S in cardiac sodium channel gene SCN5A causes cardiomyocyte apoptosis, cardiac fibrosis and contractile dysfunction in mice. Int J Cardiol, 2011. 147(2): p. 239-45. 151. Kohajda, Z., N. Farkas-Morvay, N. Jost, N. Nagy, A. Geramipour, A. Horvath, . . . L. Virag, The Effect of a Novel Highly Selective Inhibitor of the Sodium/Calcium Exchanger (NCX) on Cardiac Arrhythmias in In Vitro and In Vivo Experiments. PLoS One, 2016. 11(11): p. e0166041. 152. Watano, T., Y. Harada, K. Harada, and N. Nishimura, Effect of Na+/Ca2+ exchange inhibitor, KB-R7943 on ouabain-induced arrhythmias in guinea-pigs. Br J Pharmacol, 1999. 127(8): p. 1846-50. 153. Milberg, P., C. Pott, M. Fink, G. Frommeyer, T. Matsuda, A. Baba, . . . L. Eckardt, Inhibition of the Na+/Ca2+ exchanger suppresses torsades de pointes in an intact heart model of long QT syndrome-2 and long QT syndrome-3. Heart Rhythm, 2008. 5(10): p. 1444-52. 154. Kaese, S. and S. Verheule, Cardiac electrophysiology in mice: a matter of size. Front Physiol, 2012. 3: p. 345. 155. Abriel, H., Cardiac sodium channel Na(v)1.5 and interacting proteins: Physiology and pathophysiology. J Mol Cell Cardiol, 2010. 48(1): p. 2-11. 156. Makita, N., N. Yagihara, L. Crotti, C.N. Johnson, B.M. Beckmann, M.S. Roh, . . . A.L. George, Jr., Novel calmodulin mutations associated with congenital arrhythmia susceptibility. Circ Cardiovasc Genet, 2014. 7(4): p. 466-74. 157. Bankston, J.R., K.J. Sampson, S. Kateriya, I.W. Glaaser, D.L. Malito, W.K. Chung, and R.S. Kass, A novel LQT-3 mutation disrupts an inactivation gate complex with distinct rate-dependent phenotypic consequences. Channels (Austin), 2007. 1(4): p. 273-80. 158. Gabelli, S.B., A. Boto, V.H. Kuhns, M.A. Bianchet, F. Farinelli, S. Aripirala, . . . L.M. Amzel, Regulation of the NaV1.5 cytoplasmic domain by calmodulin. Nat Commun, 2014. 5: p. 5126. 159. Winkel, B.G., L. Yuan, M.S. Olesen, G. Sadjadieh, Y. Wang, B. Risgaard, . . . T. Jespersen, The role of the sodium current complex in a nonreferred nationwide

116 cohort of sudden infant death syndrome. Heart Rhythm, 2015. 12(6): p. 1241- 9. 160. Smallwood, P.M., I. Munoz-Sanjuan, P. Tong, J.P. Macke, S.H. Hendry, D.J. Gilbert, . . . J. Nathans, Fibroblast growth factor (FGF) homologous factors: new members of the FGF family implicated in nervous system development. Proc Natl Acad Sci U S A, 1996. 93(18): p. 9850-7. 161. Park, D.S., A. Shekhar, C. Marra, X. Lin, C. Vasquez, S. Solinas, . . . G.I. Fishman, Fhf2 gene deletion causes temperature-sensitive cardiac conduction failure. Nat Commun, 2016. 7: p. 12966. 162. Hennessey, J.A., C.A. Marcou, C. Wang, E.Q. Wei, C. Wang, D.J. Tester, . . . G.S. Pitt, FGF12 is a candidate Brugada syndrome locus. Heart Rhythm, 2013. 10(12): p. 1886-94.

117 Appendix: Supplementary Tables

Paper Cell Type Summary

A calcium sensor in the TSA201 Calmodulin binds to “IQ” domain of Nav1.5 – CTD. sodium channel modulates Calmodulin antagonist caused 6 mv depolarizing shift cardiac excitability. [52] in steady state inactivation only in presence of calcium. Mutation A1924T in IQ motif, and seen in Brugada syndrome family, causes increased rate of fast inactivation. Calmodulin Mediates Ca2+ HEK293 Calmodulin binds to “IQ” motif of both Nav1.5 and Sensitivity of Sodium Nav1.2. IQ/AA mutation of Nav1.5 leads to increased Channels. [53] late current compared to WT when expressed in HEK cells (0.07% in WT, 0.48% in IQ/AA mutant channels). Increased late current is seen in a calcium- independent manner. Conservation of HEK293/ In HEK293 cells, Ca2+ has no effect on Ca2+/Calmodulin Regulation 2+ currents in Nav1.5 whereas in Nav1.4, Ca across Na+ and Ca2+ Channels. Ventricular [55] had an inhibitory effect with a peak decrease Myocyte of 35% and a K1/2 of 1.5 uM. Similar results were seen in cardiomyocyte (Nav1.5) and skeletal myotube (Nav1.4). Chimeric Nav1.4 channels with the CTD of Nav1.5 substituted no longer exhibits calcium dependent inactivation (CDI). IQ/AA mutation in Nav1.4 similarly led to loss of CDI. Apocalmodulin Itself HEK293 At low calcium concentration, Nav1.4 with Promotes Ion Channel IQ/AA mutation resides in a low open Opening and Ca2+ Regulation. [56] probability state compared to WT Nav1.4 Over-expression of CaM leads to return to normal open probability suggesting that apocalmodulin increases open probability. Regulation of the NaV1.5 HEK293 Tested effects of various mutations in or near the 2+ cytoplasmic domain by calmodulin binding site of Nav1.5. In Ca free calmodulin. [158] solution, IQ/AA mutation caused right shift in SSI and prolongation of recovery from inactivation. In 0.5 mM Ca2+, effects of different mutations in calmodulin binding region varied, with Q1909R and K1922A causing right shift in SSI and R1910A causing left shift. The role of the sodium current Clinical Q1909R mutation found in patient with complex in a nonreferred Data/HEK- SIDS. Studies in HEK cells showed it nationwide cohort of sudden 293 infant death syndrome. [159] increased persistent current- 2-fold increased. Time constant of inactivation was slowed. SSA left shifted, SSI unshifted.

Table 1. Overview of prior studies exploring calmodulin regulation of Nav1.5

118

Paper Cell Type Findings SCN5A variant that blocks HEK293/Neonatal SCN5A variant H1849R leads to cardiac fibroblast growth factor cardiomyocyte disease in 5 generation kindred including homologous factor regulation atrial and ventricular arrhythmia. This causes human arrhythmia. [62] mutation alters interaction with FHFs. In HEK cells, Nav1.5 with H1849R mutation has right shift in SSI and delayed time constant of inactivation. In neonatal cardiomyocytes with adenovirally expressed Nav1.5 – H1849R, there is also shift in SSI. Fibroblast Growth Factor Adult ventricular FGF13 is the most abundant FHF in Homologous Factor 13 Regulates myocytes ventricular cardiomyocytes. FGF13 Na Channels and Conduction knockdown leads to reduced Na+ current Velocity in Murine Hearts. [60] density, and decreased channel availability with a left shift in channel steady state inactivation. Both fast and slow components of inactivation were slowed with FGF13 knockdown. FGF12 is a candidate Brugada Rat ventricular FHF mutation was identified in cohort of syndrome locus. [162] cardiomyocyte patients with Brugada syndrome. The mutant FHF had reduced binding to Nav1.5. This resulted in reduced Na+ channel density, and availability, a loss of function finding consistent with the Brugada phenotype. Fhf2 gene deletion causes Cardiomyocytes/HE FHF2 knockdown have abnormal cardiac temperature-sensitive cardiac K293 rhythm when temperatures are elevated. In conduction failure. [161] cardiomyocytes with FHF2 knocked out, voltage dependence of steady state inactivation was shifted in a hyperpolarizing direction. Current densities were lower in knockout mice only when temperature was raised to 37 degrees Celsius. In HEK cells, absence of FHF left-shifted SSI and peak conductance was lower. Modulation of the cardiac sodium HEK293 FHF1B binds to the Nav1.5 –CTD. FHF1B channel Nav1.5 by fibroblast binding produces a hyperpolarizing shift in growth factor homologous factor SSI of ~ 9mV when co-expressed with 1B. [57] Nav1.5 in HEK cells. Furthermore, a known LQT3 mutation, D1790G, diminishes FHF binding consistent with loss of FHF binding promoting gain of function effects.

Table 2. Overview of prior studies exploring FHF regulation of Nav1.5

119