BIOLOGICAL ANTIARRHYTHMICS—SODIUM CHANNEL INTERACTING PROTEINS

ABSTRACT

Voltage gated Na channels (Na_V) are essential for excitation of tissues. Mutations in Na_Vs cause a spectrum of human disease from autism and epilepsy to cardiac arrhythmias to skeletal myotonias. The carboxyl termini (CT) of Na_V channels are hotspots for disease-causing mutations and are richly invested with protein interaction sites. We have focused on the regulation of Na_V by two proteins that bind in this region: calmodulin (CaM) and non-secreted fibroblast growth factors (iFGF or FHF). CaM regulates Na_V gating, mediating Ca²⁺-dependent inactivation (CDI) in a channel isoform-specific manner, while Ca²⁺-free CaM (apo-CaM) binding broadly regulates Na_V opening and suppresses the arrhythmogenic late Na current (I_Na-L). FHFs inhibit CDI, in Na_V isoforms that exhibit this property, and potently suppress I_Na-L, the latter requiring the amino terminus of the FHF. A peptide comprised of the first 39 amino acids of FHF1_A is sufficient to inhibit I_Na-L, constituting a credible specific antiarrhythmic.

PROLOGUE

It is a pleasure and honor for me to present here at the American Clinical and Climatological Association and to have the opportunity to talk about some of the work that is ongoing with collaborators at Hopkins and Columbia. Before I start, I would like to refer to my origin story as a physician-scientist.

My first Chair of Medicine at University of California San Francisco was Holly Smith, and I was struck by what he had to say about health care at the Alan Gregg Memorial Lecture in 1985: “The major health care problem of our time is not its crushing cost or inequitable distribution, serious as these problems are … but rather that we still face many diseases for which we have no answers.” This resonated with me then and I think still defines the challenge and role of the Academic Medical Center and the need for the physician-scientist, even more so now as we most recently experienced during the COVID-19 pandemic. We need an infrastructure that supports biomedical research and a clear message of the value thereof to the public.

INTRODUCTION

Sudden cardiac death (SCD) is a devastating consequence of heritable and acquired heart diseases. There are gaps in our understanding of SCD including its definition, etiology, and rates. Death certificate adjudication overestimates the frequency of SCD with the best estimates of the rates of SCD coming from multisource prospective studies (1). As with most acquired heart diseases, coronary artery disease (CAD) is at the center of the etiology: most patients who experience SCD have CAD. The focus of this discussion is SCD due to ventricular tachyarrhythmias.

The sodium channel (Na_V) is a member of the evolutionarily related superfamily of voltage-dependent ion channels that include K_V, Ca_V, and Na_V channels with similar transmembrane topologies (Figure 1). Na_V channels produce currents that underlie the rapid upstroke of the AP and are a major determinant of conduction velocity in excitable tissue. They are the most plentiful ion channels in excitable cells; for example, there are over 100,000 channels in a ventricular myocyte. Na_V channels are modulated by a host of signaling systems, in a complex and isoform-specific fashion. Disruption of Na_V channel function has been linked to several disorders of excitability such as seizures and myotonia, and in the heart, arrhythmias such as long QT syndrome (LQTS), atrial fibrillation, conduction system disease, and cardiomyopathies (2). A particularly important disturbance of channel function is impaired closure or inactivation leading to an increase in late current (I_Na-L). In the heart, pathologic I_Na-L produces APD prolongation, electrical instability, and increased risk of lethal arrhythmias. The late Na current may be augmented by channel mutations or post translational modifications that are features of cardiomyopathy and heart failure.

Fig. 1.

Schematic of the voltage-gated sodium channel (Na_V). Transmembrane glycoprotein with four internally homologous domains (I–IV), each comprised of six membrane-spanning alpha helices (cylinders), the fourth (orange) being positively charged and serving as a voltage sensor. The loops between the fifth and sixth membrane-spanning helices in each domain form the ion selective pore. There are two classes of long and short variants that differ in the length of the loop connecting the first and second homologous domains. There are a number of ancillary subunits, represented by the single membrane spanning β subunit (red cylinder). The channel is phosphorylated at a number of sites by several kinases, shown here as blue (PKA), pink (CaMKII), and red (PKC) circles. The carboxy terminus is structurally ordered in the region adjacent to the membrane and is comprised of six alpha helices (purple); the sixth contains an IQ motif that binds CaM and FHF binds in the pre-IQ region. Abbreviations: Syn, syntrophin; EFL, EF hand-like motif

The CT of Na_V channels is a hotspot for pathogenic/likely pathogenic mutations associated with altered channel functional expression including increased I_Na-L. There is endogenous variability in I_Na-L. Our data and others suggest that calmodulin (CaM) and fibroblast growth factor homology factors (FHF), two CT binding proteins, modify the magnitude and the arrhythmogenic impact of I_Na-L (3,4).

METHODS AND MATERIALS

The methods and materials have been described in detail in previous publications (3–9). These include molecular biological methods, cell culture, crystallization and structure determination, fluorescence measurements, and electrophysiological recording.

RESULTS

Calcium/CaM Regulation of Na Channels

CaM is the most ubiquitous Ca²⁺-sensing protein in biology. Its structure is highly conserved in nature, and it contains N- and C-lobe EF hand motifs that bind Ca²⁺. CaM binds to the proximal CT of Na_V channels, at a helix containing an isoleucine-glutamine (IQ) motif (10). This region of the channel harbors several mutations that have been associated with heritable arrhythmias such as LQTS and Brugada Syndrome (BrS) and several of these mutations are at the CaM binding interface, near the IQ motif that binds to the C-lobe of CaM (6,11,12).

CaM binds to several ion channels with effects on expression and function. In Na_V channels, it influences membrane trafficking and multiple aspects of channel opening and closure (gating) (13–15). In the case of Ca_V channels, bound CaM is responsible for Ca²⁺-induced channel closure referred to as calcium-dependent inactivation (CDI) (16). A related regulatory phenomenon is present in some Na_V channels where CaM tethered to the CT binds Ca²⁺ and induces channel closure. This process is isoform-specific and completely contained within the CT domain (4,7,17,18). Structural studies suggest the following model of Ca²⁺-dependent CaM binding and subsequent channel inactivation. In the activated state, at low intracellular Ca²⁺ concentration ([Ca²⁺]_i), both the N- and C-lobes of CaM are bound to the channel. As [Ca²⁺]_i increases during electrical systole, the Ca²⁺ bound N-lobe of the channel is released and free to interact with the cytoplasmic face of the channel mediating closure characteristic of CDI. In some channel variants such as the cardiac isoform, Na_V1.5, the calcified N-lobe of CaM is bound to the distal CT of the channel and therefore not available to serve as a pore-blocking inactivation particle (Figure 2).

Fig. 2.

Schematic of CaM and isoform-specific dynamic Ca²⁺ regulation of Na_V channels. The skeletal muscle isoform, Na_V1.4, exhibits Ca²⁺-dependent inactivation (CDI). At low levels of [Ca²⁺], the N-lobe of CaM is tethered to the proximal CT near the EFL, and calcification of the N-lobe promotes unbinding from the EFL, with the N-lobe serving as an inactivation particle. Differences in CaM binding in Na_V1.5 produce tethering of the calcified N-lobe to the distal CT; thus, NaV1.5 does not exhibit CDI. Modified from Yoder, et al. 2014.

Independent of its ability to bind Ca²⁺, CaM behaves as a functionally important channel subunit. Binding of CaM and its disruption by mutations that alter binding influence the probability of channel opening (P_open). In the absence of CaM, enforced by a peptide CaM chelator or mutations near the CaM binding site, P_open is dramatically reduced, an effect that can be reversed by increasing intracellular [CaM] (3).

Disordered CaM binding is also associated with an increase in arrhythmogenic residual or late current, I_Na-L. In order to demonstrate the impact of CaM binding on I_Na-L, we studied two mutations that disrupt CaM binding to the channel: replacement of the isoleucine and glutamine in the IQ motif (I1908A, Q1909A, IQ/AA) (9,19) and a mutation discovered in a patient with a complex arrhythmic phenotype, S1904L. High resolution patch clamp recording revealed an increase in late openings of both IQ/AA and S1904L channels expressed in HEK-293 cells when compared to similarly expressed wild-type (WT) controls. Figure 3 demonstrates the impact of CaM binding mutations on I_Na-L. The WT channel exhibits very rare late channel openings, and the P_open decays to a value well below 1% of the peak current (Figure 3A). The defective CaM binding mutants IQ/AA and S1904L exhibit a significant number of late channel openings producing persistent current (Figures 3D and 3G). Late channel openings could be induced in WT channels if a CaM chelator was added to the intracellular solution (Figure 3B), without a significant change in I_Na-L through the mutant channels (Figures 3E and 3H). Conversely, overexpression of CaM (or apo-CaM) will reduce the persistent current, through the mutant channels without an impact on the WT Na_V1.5 (Figures 3C, 3F, and 3I) (3). The late current is a therapeutic target for several arrhythmias, and ranolazine which is partially selective for I_Na-L has been used in this context. Modulation of the levels of CaM is a difficult therapeutic target; however, there are other interacting proteins in this region that influence I_Na-L.

Fig. 3.

CaM binding tunes the late current. Panels A-C show WT channels expressed in HEK293 cells under control conditions and in the presence of a CaM chelator and CaM. Time-compressed currents are shown as well as single channel currents recorded between 50 and 300 msec after the depolarization. In panel A, the open probability plot, and an amplified inset of the first 100 msec is shown below the currents. Panels D-F and G-I are currents through IQ/AA and S1904L mutants, respectively, in the same format as shown for WT. From Kang, et al. 2021.

FHFs as Specific Modulators of Na Channel Function

FHFs or intracellular fibroblast growth factors (iFGFs) are not secreted and have a limited number of targets, prominently Na_V channels (20,21). The FHFs are encoded by four genes (FHF1/iFGF12, FHF2/iFGF13, FHF3/iFGF11, and FHF4/iFGF14). The diversity of the FHFs is further enhanced by alternative splicing—prominently at the amino terminus (NT) (22). The FHFs bind to the channel in the carboxy terminus (CT) in a loop in the pre-IQ segment which interacts with the β-trefoil core of the FHF; the NT interacts with an intracellular domain serving as an inactivation particle synergizing or competing with the N-lobe of CaM (Figure 1) (11). FHFs are differentially expressed; in the mouse heart FHF2 and in the human heart FHF1_B are the predominant FHFs (22,23).

Transgenic mice harboring the CaM binding mutation IQ/AA remarkably did not exhibit an increase in I_Na-L and the expected LQTS phenotype. The proximity of FHF binding suggested the possibility that FHF may be compensating for CaM and inhibiting I_Na-L in mutant cardiomyocytes. Moreover, a mutation in the CT of NaV1.5 blocks the functional effects of FHF (24) and a mutation in the NT of FHF1 has been associated with BrS (25). Ventricular myocytes from WT and transgenic mice were treated with siRNAs designed to knock down FHF2, which increased late openings in cardiomyocytes from IQ/AA mutant mouse cells without a significant impact on cells from WT mice. This increase in persistent current with knock down of FHF2 in mouse cardiomyocytes is consistent with cross talk between CaM and FHFs in the regulation of the I_Na-L (Figure 4).

Fig. 4.

Summary plot of the effect of siRNA knock down of FHF2 on persistent Na current in mouse ventricular myocytes. In the presence of the targeted siRNA, persistent current was significantly increased in IQ/AA mice but not WT and non-transgenic controls. From Chakouri, et al. 2022.

The effect of FHFs on I_Na-L is splice variant-dependent and requires the presence of the NT, so type B FHFs, which lack the NT, are predicted to have no effect on the late current. To test this explicitly, WT and CaM binding mutant Na channels were expressed in HEK293 cells in the presence of FHF variants. As expected, WT channels had little late current and there was no impact of FHF expression on I_Na-L. Two LQT mutant channels, IQ/AA and Na_V1.5-ΔKPQ which alters the putative inactivation particle on the cytoplasmic surface of the channel, exhibited late openings and increased I_Na-L. Late channel openings were suppressed by A-type FHFs, FHF1_A and FHF3_A, in both IQ/AA and ΔKPQ mutant channels. The B-type variant FHF2_B did not alter I_Na-L; the splice variant FHF2_S which retains a portion of the NT suppresses late openings of IQ/AA and ΔKPQ channels (Figure 5). These data are consistent with a model of FHFs being anchored by its core to the CT of the channel and binding of the amino terminus of type A FHFs to the inner mouth of the channel to inhibit I_Na-L (5).

Fig. 5.

Summary plot of the effect of isoforms of FHF on persistent Na current through mutant NaV1.5 channels. Type A FHFs with an intact amino terminus (NT) inhibit I_Na-L; type B variants do not. FHF1_S, which retains part of the NT, suppresses the late current. From Chakouri, et al. 2022.

The effect of FHFs on native Na_V currents in human cardiomyocytes is retained. Human iPSC-CMs generated from a patient with LQTS3 caused by the ΔKPQ mutation exhibited frequent late openings and a robust I_Na-L. In contrast to murine cardiomyocytes from the IQ/AA mutant mouse, knock down of FHF2 had no significant impact on late openings in human cells (Figure 6). This is consistent with the differences in FHF isoform expression in human and mouse cells. However, FHF1_A dramatically reduced late channel openings, as predicted with the model of type A FHF binding to the channel and the NT inhibiting ion flux (5).

Fig. 6.

Na currents from human iPSC cardiomyocytes reveal FHF inhibition of late current. Cardiomyocytes were generated from a control subject and patient with LQTS due to a ΔKPQ mutation. Knock down of FHF2 had no impact currents in either cell. FHF1_A dramatically reduced late channel openings in the mutant cardiomyocytes. Consistent with isoform-specific FHF expression and the importance of A-type FHFs in tuning I_Na-L. From Chakouri, et al. 2022.

Type A FHFs are protective (i.e., they inhibit the arrhythmogenic I_Na-L). The NT of FHFs is necessary for inhibition of the late current, but is it sufficient? The entire NT and a fragment comprising the first 39 amino acids were as efficient as full-length FHF1_A in inhibiting I_Na-L. Other fragments comprising the first 17 amino acids or residues 9-39 did not have a significant impact on the I_Na-L. The fragment from amino acids1-39, referred to as FixR, was a better inhibitor of I_Na-L than the late Na channel blocker ranolazine (Figure 7A). FixR packaged in an adenovirus that was used to infect hiPSC-CMs containing the ΔKPQ mutation significantly reduced I_Na-L (Figure 7B). There are a number of circumstances in which I_Na-L is increased: different LQTS mutations in Na_V channels and post translational modifications. In addition to the IQ/AA and ΔKPQ variants, several channelopathic mutations of Na_V1.5 exhibit increased I_Na-L and are inhibited by FixR. Channel phosphorylation is another mechanism associated with an increase in I_Na-L, often observed in the failing heart. Incubation of WT channels with catalytically active versions of PKA or CaMKII result in an increase in late current that is inhibited by FixR (Figure 7C) (5).

Fig. 7.

Panel A: Summary plot of the inhibition of persistent current by FHF1_A amino terminal fragments. The fragment comprising amino acids 1-39 is as potent an inhibitor as full length FHF1_A or the complete amino terminal fragment and significantly more potent than ranolazine. Panel B: Adenovirus encoded FixR inhibits late openings of Na_V1.5 ΔKPQ mutant channels in hiPSC-CMs. Panel C: FixR inhibits I_Na-L in several contexts, mutations that cause LQTS, and channel phosphorylation by PKA and CaMKII known to be associated with structural heart disease. Modified from Chakouri, et al. 2022.

DISCUSSION

Life-threatening arrhythmias and sudden death remain a significant complication of heritable and acquired heart diseases. Altered ion channel function is a frequent contributing mechanism to serious, often life-threatening arrhythmias. However, ion channels are challenging pharmacological targets, and channel modifying drugs often lack specificity and have dangerous liabilities.

Ancillary proteins are important regulators of voltage gated ion channels. The CT of Na_V channels is richly invested with accessory proteins and in the case of the cardiac channel is a hotspot for mutations associated with arrhythmias and cardiomyopathies. CaM and FHFs are neighboring binding proteins in the channel CT that modulate Na_V channel function. Indeed, the impact of CaM and FHF regulation help to explain some of the phenotypic variability in Na_V function caused by arrhythmogenic channel mutations and post translational modifications associated with structural heart disease (8,26,27). Enhanced I_Na-L is a feature of heritable and acquired arrhythmias (2) and is a compelling therapeutic target that is modulated by CaM and FHF binding.

Mutation-induced disruption of CaM binding produces two seemingly paradoxical effects on channel function: a reduction in the peak P_open and an increase in I_Na-L, or a loss and gain of function, respectively. The effect of mutations that impact CaM binding can be overcome by overexpression of CaM or apo-CaM, a variant that does not bind Ca²⁺(3). Remarkably, type A FHFs can also mitigate the increase in I_Na-L produced by CaM binding deficient variants. Unlike CaM, FHFs have a more restricted panel of binding partners, thus the possibility of greater specificity of their functional effects. Another virtue of FHFs as a therapeutic target is that they are remarkably effective at inhibiting I_Na-L produced by diverse mechanisms (Figure 7C). In fact, FHFs are at least as effective at inhibiting I_Na-L as contemporary therapy, local anesthetic antiarrhythmics. FixR, an amino terminal peptide fragment, is as potent as full-length FHF1_A and nearly an order of magnitude more potent an inhibitor of I_Na-L compared to ranolazine (5). FixR and its analogs represent an alternative class of I_Na-L inhibitors with broad therapeutic implications, as enhanced late currents have also been linked to seizure disorders, myotonia, and pain syndromes (2). The understanding of ancillary protein-based regulation of Na_V function provides insight into the mechanisms of cardiac arrhythmias and other diseases of excitability as well identification of novel targets for treatment of these disorders.

DISCUSSION

Anderson, Chicago: Gordon, thanks for a terrific talk; you covered a lot in a short period of time. I think it emphasizes the idea that going deeply into mechanistic aspects of rare genetic diseases gives insight into more common diseases. You teased us a little bit about how physical readouts or the sodium current in these genetic diseases were, in fact, mimicked by common diseases and the base of your pyramid was coronary artery disease. Is there something about FGF regulation that is a mimic or that there is not enough of in diseased hearts or after ischemia or myocardia infraction? Could it be because there are other disease-related signals that affect the sodium current similarly? Is it part of the disease pathway that teaches us about how hearts become electrically unstable with pathological stress?

Tomaselli, New York City: Thanks, Mark. That's a great question, and I think a couple of factors come into play. There may be not only changes in levels of FGF, and we've actually looked at this in one of our animal models pacing tachycardia induced heart failure in the dog, where FGF goes down, but it was also a little difficult for us to explain frankly because most of what went down was a type-B FGF or FHF so that's not the full explanation. It may be that this part of the channel and other parts of the channel are also phosphorylated and differentially phosphorylated in acquired forms of heart disease that may alter calmodulin binding and FHF interactions. Suffice it to say regardless of which way we did this in vitro model we could overcome the increase in late current by overexpressing a type-A FHF, so you're absolutely right. Many factors are contributing and calmodulin and FHF also have other effects on currents particularly sodium currents that I didn't have time to describe. It seems to me that if we're targeting this late current, this is a strategy that at least has some promise.

Arnaout, Boston: So, there are two isoforms of FHF: one with an amino-terminal and one without. Did I understand that there is more of the isoform that doesn't have the amino terminus in human heart versus the mouse?

Tomaselli, New York City: Correct. These are in samples that we've acquired from explanted, transplanted human hearts. Not just us but others have shown that FHF, FHF-2-1B seems to be the predominant form at least in hearts that have been sampled. Now whether or not that changes as a function of pathology, for example, is unclear because most of the samples that have demonstrated this are not normal tissue. Obviously, they're tissue from failing hearts. So, yes, you understood it correctly.

Arnaout, Boston: So, it may be that the mouse with 400 heart beats per minute requires the amino terminal isoform of FHF to protect against fatal arrhythmias.

Tomaselli, New York City: That's absolutely a great observation. The channel has to close robustly and quickly in order to maintain a heart rate of an ambulatory heart for a mouse of 400 beats per minute.