Cardiac Kir2.1 and Na_V1.5 Channels Traffic Together to the Sarcolemma to Control Excitability

Abstract

Rationale

In cardiomyocytes, Na_V1.5 and Kir2.1 channels interact dynamically as part of membrane bound macromolecular complexes.

Objective

To test whether Na_V1.5 and Kir2.1 preassemble during early forward trafficking and travel together to common membrane microdomains.

Methods and Results

In patch-clamp experiments, co-expression of trafficking deficient mutants Kir2.1^Δ314–315 or Kir2.1^R44A/R46A with wildtype (WT) Na_V1.5^WT in heterologous cells reduced I_Na, compared to Na_V1.5^WT alone or co-expressed with Kir2.1^WT. In cell surface biotinylation experiments, expression of Kir2.1^Δ314–315 reduced Na_V1.5 channel surface expression. Glycosylation analysis suggested that Na_V1.5^WT and Kir2.1^WT channels associate early in their biosynthetic pathway, and fluorescence recovery after photobleaching experiments demonstrated that co-expression with Kir2.1 increased cytoplasmic mobility of Na_V1.5^WT, and vice versa, whereas co-expression with Kir2.1^Δ314–315 reduced mobility of both channels. Viral gene transfer of Kir2.1^Δ314–315 in adult rat ventricular myocytes and human induced pluripotent stem cell-derived cardiomyocytes (hiPSC-CMs) reduced I_K1 and I_Na, maximum diastolic potential (MDP) and action potential depolarization rate, and increased action potential duration (APD). Upon immunostaining, the adaptor protein complex 1 (AP1) colocalized with Na_V1.5^WT and Kir2.1^WT within areas corresponding to t-tubules and intercalated discs. Like Kir2.1^WT, Na_V1.5^WT co-immunoprecipitated with AP1. Site-directed mutagenesis revealed that Na_V1.5^WT channels interact with AP1 through the Na_V1.5^Y1810 residue, suggesting that, like for Kir2.1^WT, AP1 can mark Na_V1.5 channels for incorporation into clathrin-coated vesicles at the trans-Golgi. Silencing the AP1 ϒ-adaptin subunit in hiPSC-CMs reduced I_K1, I_Na and MPD, and impaired rate dependent APD adaptation.

Conclusions

The Na_V1.5-Kir2.1 macromolecular complex preassembles early in the forward trafficking pathway. Therefore, disruption of Kir2.1 trafficking in cardiomyocytes affects trafficking of Na_V1.5, which may have important implications in the mechanisms of arrhythmias in inheritable cardiac diseases.

INTRODUCTION

In the human heart, three strong inward rectifying channels (Kir2.1, Kir2.2 and Kir2.3) contribute to the inward rectifier potassium current (I_K1)¹. I_K1 is the major current responsible for maintenance of the maximum diastolic potential (MDP) in cardiomyocytes. It modulates excitability and influences the shape of the action potential (AP)^2–5. Kir2.1 is the major channel isoform underlying I_K1 in human ventricles, with Kir2.2 expressed to a lesser extent. In human atrial cells Kir2.3 is the main carrier of I_K1^{6, 7}.

An important aspect of Kir2.x channel regulation is precise forward trafficking to specific membrane subdomains^{2, 7–9}. The Golgi acts as a central biosynthetic sorting station, responsible for directing newly synthesized membrane proteins to their appropriate subcellular destination^{10, 11}. Signal-dependent Golgi export processes control membrane surface density of Kir2.x channels¹². Transport of Kir2.1 channels from the Golgi is dictated by residues embedded in the NH₂ and COOH domains of the channel. These residues form a recognition site for interaction with adaptor protein complex 1 (AP1), thereby marking Kir2.1 for incorporation into clathrin-coated vesicles at the trans-Golgi¹².

Adaptor protein complexes are heterotetrameric protein complexes that mediate intracellular membrane trafficking. Of the five different adaptor protein complexes, AP1, AP2 and AP3 are clathrin-associated, whereas AP4 and AP5 are not¹³. The adaptor protein complexes bind to sorting signals on cytoplasmic tails of cargo proteins. They also recruit clathrin and other accessory proteins and concentrate cargo proteins into vesicular carriers that transport cargo proteins from donor to target membranes^13–15.

The voltage-gated cardiac sodium channel consists of a pore-forming alpha subunit (Na_V1.5) and one or more beta subunits that modulate, but are not crucial, for functional Na_V1.5 expression¹⁶. Na_V1.5 channels permeate inward sodium current (I_Na), which is the main depolarizing current in cardiomyocytes, and thus is critical for normal electrical conduction^{16, 17}. Regulation of Na_V1.5 channel transport from Golgi to plasma membrane is poorly understood. Examination of Na_V1.5 glycosylation suggests the Na⁺ channel can reach the plasma membrane via two different transport routes: the classical secretory transport route, or a Golgi-independent pathway¹⁸.

Cellular processes that regulate trafficking and expression of Kir2.1 and Na_V1.5 channels are of interest as potential mediators of proarrhythmic channelopathies. Kir2.1 and Na_V1.5 can interact through common partners such as scaffolding, anchoring/adapter proteins, enzymes, and regulatory proteins^19–26. We previously demonstrated that both channels belong to multi-protein complexes, enabling one to regulate the other’s expression^{22, 26, 27}. However, it remains unknown whether Kir2.1 and Na_V1.5 pre-assemble or interact during early trafficking steps of their secretory pathways.

Here we investigate whether Kir2.1 and Na_V1.5 associate early in their respective biosynthetic pathways, and whether they share a coupled forward trafficking process prior to formation of the Na_V1.5-Kir2.1 macromolecular complex at the cell membrane. We use trafficking-deficient mutant channels Kir2.1^Δ314–315 and Kir2.1^R44A/R46A that disrupt Golgi trafficking of Kir2.1, thus impairing its surface expression^{12, 28–31}. We hypothesize that expression of these trafficking-deficient Kir2.1 mutants will reciprocally disturb trafficking and functional expression of associated Na_V1.5 channels, consequently altering cellular excitability and establishing a substrate for arrhythmogenesis.

METHODS

The authors declare that all supporting data are available within the article [and its online supplementary files].

Expanded methodology can be found in Online Supplement.

Isolation and culture of cardiac myocytes

Calcium-tolerant adult rat ventricular myocytes (ARVMs) were isolated from hearts of normal adult male Sprague-Dawley rats. Cells were plated on laminin coated tissue culture glass coverslips/dishes and maintained in M199 for viral/non-viral transfer and patch clamping as previously described³².

Cell culture and transfections

Human Embryonic Kidney (HEK-293T) cells were cultured in DMEM (Gibco) supplemented with fetal bovine serum (FBS), penicillin, and streptomycin at 37°C. HEK-293 cells stably expressing Na_V1.5+β1 channels were grown under the same conditions with the addition of Geneticin for selection of cells expressing the resistance marker. For patch-clamp experiments, cells were grown in 60 mm dishes and transfected or co-transfected using Lipofectamine.

iCell hiPSC-CM monolayers

Cryopreserved vials of human iCell^® Cardiomyocytes were obtained from Cellular Dynamics International, Inc (Madison, WI). After 48 to 72 h of adenoviral overexpression electrophysiological phenotype analysis was performed.

Biotinylation assay

Forty-eight hours after transfection, HEK-293T cells were washed with phosphate-buffered saline and biotinylated for 60 min then incubated with glycine. Biotinylated proteins were analyzed by Western blot and immunoprecipitation assays.

Deglycosylation assay

Total membrane homogenates were solubilized and incubated with mouse anti-Kir2.1 antibody for immunoprecipitation. Samples were incubated with either PNGase-F, Endo-H or phosphate buffer and used for SDS-PAGE analysis.

I_Na, I_K1 and AP recordings in ARVMs, hiPSC-CMs and HEK-293T cells

Voltage-clamp recordings were performed at room temperature and APs were recorded using current-clamp at 37°C.

Fluorescence recovery after photobleaching (FRAP)

Na_V1.5^WT and Kir2.1^WT channels were bound to mTurquoise (^mTQ) and venus (^Venus) fluorophores for FRAP experiments in COS-7 cells. Analysis was performed using Fiji/ImageJ and Matlab.

Statistics

All data are expressed as mean±SEM. Two-tail Student’s t-test, One or Two-way ANOVA with Bonferroni’s multiple comparison tests performed with Prism 6 were used. p<0.05 was considered significant.

RESULTS

Trafficking-deficient mutant Kir2.1^Δ314–315 impairs functional expression of Na_V1.5

We conducted patch-clamp experiments in transiently transfected HEK-293T cells. For Kir2.1^WT, voltage steps elicited the expected large inward and small outward currents of a strong inward rectifier channel. In contrast, no potassium current was detected for the trafficking-deficient mutant Kir2.1^Δ314–315 (Online Figure I A,B), as previously reported²⁸. In addition, we confirmed previous data showing cooperative interaction between Kir2.1^WT and Na_V1.5^WT ion channel expression^{22, 26, 27}. We observed a synergistic effect on I_Nav1.5 density when Kir2.1^WT channels were co-expressed (1:1) with Na_V1.5^WT channels, resulting in increased I_Nav1.5. Co-expression of Kir2.1^WT channels with Na_V1.5^WT channels resulted in larger peak I_Nav1.5 density at −25 mV compared with Na_V1.5^WT alone (−830±65 pA/pF versus −562±57 pA/pF, respectively; p=0.001, Online Figure I C,D). We also assessed functional consequences on I_Kir2.1 of co-expressing Na_V1.5^WT with the trafficking-deficient Kir2.1^Δ314–315 channel. No I_Kir2.1 density was generated by Kir2.1^Δ314–315 channels when co-expressed with Na_V1.5^WT channels. Co-expression of Kir2.1^WT, Kir2.1^Δ314–315 and Na_V1.5^WT (0.5:0.5:1) yielded an I_Kir2.1 density that was 92% smaller than when Kir2.1^WT was expressed alone (Online Figure I A,B).

We then co-expressed Na_V1.5^WT with Kir2.1^Δ314–315 and measured Na_V1.5^WT channel function. I_Nav1.5 density in cells with Na_V1.5^WT channels expressed alone was larger than in cells that co-expressed Kir2.1^Δ314–315, with peak currents at −25 mV of −562±57 versus −275±38 pA/pF, respectively (p<0.0001, Online Figure I C,D). Correspondingly, I_Nav1.5 density in cells with Na_V1.5^WT channels expressed alone was larger than in cells that co-expressed Kir2.1^WT, Kir2.1^Δ314–315 and Na_V1.5^WT subunits (peak current at −25 mV, −562±57 versus −178±34 pA/pF, respectively, p<0.0001, Online Figure I C,D).

To rule out the possibility that decreased I_Nav1.5 density in the presence of the trafficking-deficient mutant Kir2.1^Δ314–315 was due to variability in expression from transfection, we used a stable HEK-293 cell line expressing Na_V1.5+β1. Cells transfected with Kir2.1^Δ314–315 had a lower I_Nav1.5 than control GFP-transfected cells. Cells transfected with Kir2.1^Δ314–315 exhibited smaller I_Nav1.5 density −244±37 pA/pF than control GFP-transfected cells −373±50 pA/pF at −25 mV, p<0.003. On the other hand, cells transfected with Kir2.1^WT had a larger mean I_Na peak density −520±69 pA/pF at −25 mV than control cells, p<0.0001, Online Figure II A,B and Online Table I. We observed no differences in voltage-dependence of activation or inactivation, or recovery from inactivation between groups (Online Figure III A-D and Online Table I).

Thus, trafficking deficiency of one channel component of the Na_V1.5-Kir2.1 macromolecular complex (in this case Kir2.1^Δ314–315) negatively influenced functional expression of the other channel (Na_V1.5), possibly by disturbing channel trafficking to the plasma membrane.

Double mutant Kir2.1^R44A/R46A reduces I_Nav1.5 density in CHO cells

To explore whether other trafficking-deficient Kir2.1 channels can produce similar effects on I_Nav1.5 density, we studied the NH₂ double mutant Kir2.1^R44A/R46A, which accumulates in the Golgi resulting in decreased membrane expression³³. Surprisingly, patch-clamp experiments in HEK-293T and CHO cells showed that the Kir2.1^R44A/R46A mutant produced an I_Kir2.1 density similar to Kir2.1^WT channels (Online Figure IV A,B). However, we observed that in transfected CHO cells, I_Nav1.5 density of Na_V1.5^WT channels expressed alone was larger than in cells that co-expressed Kir2.1^R44A/R46A (Online Figure II C,D). Thus, decreased I_Nav1.5 density was not unique to Kir2.1^Δ314–315 channels, as other Kir2.1 trafficking-deficient mutants in a different cell line produced similar effects.

Kir2.1^Δ314–315 expression reduces I_Na density in adult rat ventricular myocytes (ARVMs)

We next studied the functional consequences of expressing Kir2.1^Δ314–315 in adult mammalian cardiac cells. Using an adenoviral construct, we overexpressed Ad-Kir2.1^Δ314–315 in ARVMs; control myocytes were infected with GFP (Ad-GFP). Cell capacitance was similar in myocytes infected with Ad-Kir2.1^Δ314–315 and control Ad-GFP myocytes (100±5 versus 107±2 pF; n=27, n=23; respectively; p=0.4). Representative Ba²⁺ sensitive I_K1 traces from Ad-GFP and Ad-Kir2.1^Δ314–315 are shown in Figure 1A. Figure 1B shows the I_K1 current-voltage relationship for Ad-GFP and Ad-Kir2.1^Δ314–315. As expected, ARVMs transfected with Ad-Kir2.1^Δ314–315 generated smaller inward −3±0.4 versus −10±8 pA/pF at −120 mV; p<10⁻⁷ and outward 0.15±0.01 versus 0.5±0.06 pA/pF at −60 mV, p<10⁻⁵ currents than control. We next investigated functional effects of Ad-Kir2.1^Δ314–315 overexpression on I_Na in myocytes. Representative I_Na traces from Ad-GFP and Ad-Kir2.1^Δ314–315 are in Figure 1C. Overexpression of Kir2.1^Δ314–315 decreased I_Na density compared with Ad-GFP, Figure 1D. Mean peak current density at −35 mV was −39±3 versus −52±4 pA/pF, respectively, p=0.05.

Figure 1. Adenoviral gene overexpression of GFP and Kir2.1^Δ314–315 in ARVMs at 48 h.

A. I_K1 traces from ARVMs infected with Ad-GFP or Ad-Kir2.1^Δ314–315. B. I_K1 I/V relationships for Ad-GFP and Ad-Kir2.1^Δ314–315 (N=4, n=12 and N=6, n=13, respectively). C. I_Na traces from ARVMs infected with Ad-GFP or Ad-Kir2.1^Δ314–315. D. I_Na, I/V relationships for Ad- GFP, Ad-Kir2.1^Δ314–315 (N=3, n=12 and N=4, n=10, respectively). *p<0.05, mean±SEM, (t-test). N, number of animals; n, number of cells. Insert describes voltage pulse protocols.

To evaluate whether overexpression of Kir2.1^Δ314–315 in ARVMs could affect other currents, like transient outward potassium current (I_to), we compared the difference between initial peak and sustained K⁺ background current density of myocytes infected with Ad-Kir2.1^Δ314–315 versus Ad-GFP cells. Peak currents at 30 mV were 8.1±1 versus 7.5±1 pA/pF, respectively, p=0.79. In contrast with the decrease in I_K1, Ad-Kir2.1^Δ314–315 expression did not modify I_to (Online Figure V).

Kir2.1^Δ314–315 expression modifies ARVM cardiac action potential

We studied cellular electrophysiological consequences of Kir2.1^Δ314–315 expression by recording APs in Ad-Kir2.1^Δ314–315-infected ARVMs. As expected, the MDP was reduced compared to control (−66±3 versus −79±1 mV; p=0.0028; Figure 2A,C). MDP decrease was consistent with I_K1 reduction (Figure 1A,B). In some Ad-Kir2.1^Δ314–315-infected ARVMs, MDP reduction resulted in phase-4 depolarization and generation of spontaneous APs (Online Figure VI). In addition, maximum upstroke velocity (dV/dt_max) was reduced in Ad-Kir2.1^Δ314–315 myocytes compared to control 91±2 versus 184±1 V/s, respectively; p=0.012, Figure 2D. These findings were in agreement with reduction in both sodium channel availability during the AP upstroke and I_Na density in Ad-Kir2.1^Δ314–315 myocytes (Figure 1C,D). Moreover, APD was prolonged in Ad-Kir2.1^Δ314–315 (Figure 2E). Finally, we observed no difference in AP overshoot between control and Ad-Kir2.1^Δ314–315 myocytes (Figure 2F).

Figure 2. Effect Ad-Kir2.1^Δ314–315 overexpression on action potential characteristics of ARVMs.

A. Representative action potential recordings from ARVMs infected with adenoviruses encoding Ad-GFP and Ad-Kir2.1^Δ314–315 at 1 Hz. B,C,D, E, F Average amplitude (B), maximum diastolic potential (C), maximal dV/dt (D), action potential duration to 30, 60 and 90% of repolarization (E) and overshoot (F) (N=3, n=8 and N=3, n=8, respectively, where N is the number of animals and n is the number of cells) *p<0.05, mean±SEM, (t-test).

Kir2.1^Δ314–315 mutant decreases I_K1 and I_Na in hiPSC-CMs

We determined the consequences of expression of Kir2.1^Δ314–315 in hiPSC-CMs by adenoviral infection; control cells were infected with Ad-GFP. Figure 3A shows representative Ba²⁺ sensitive I_K1 traces for Ad-GFP and Ad-Kir2.1^Δ314–315. Control I_K1 current-voltage relationship is plotted with the Ad-Kir2.1^Δ314–315 curve. As expected, hiPSC-CMs infected with Ad-Kir2.1^Δ314–315 generated smaller currents than control −0.8±0.3 versus −3.6±0.6 pA/pF at −120 mV, p=0.006, Figure 3B. We next investigated the functional effects of Ad-Kir2.1^Δ314–315 overexpression on I_Na in human myocytes. Representative I_Na traces are shown in figure 3C. In figure 3D, overexpression of Kir2.1^Δ314–315 decreased I_Na density at several test voltages compared with Ad-GFP. Peak currents at −20 mV were −14±1 versus −21±2 pA/pF, respectively, p=0.014.

Figure 3. ATS1 mutant Kir2.1^Δ314–315 decreases I_K1 and I_Na density on human induced pluripotent stem cell-derived cardiomyocytes (hiPSC-CMs).

A. I_K1 traces from hiPSC-CMs infected with Ad-GFP or Ad-Kir2.1^Δ314–315. B. Superimposed I_K1 I/V relationships for Ad-GFP and Ad-Δ314–315 (N=3, n=6 and N=3, n=6, respectively). C. I_Na traces from hiPSC-CMs infected with Ad-GFP or Ad-Kir2.1^Δ314–315. D. I_Na, I/V relationships for Ad-GFP, Ad-Kir2.1^Δ314–315 (N=3, n=7 and N=3, n=17, respectively). *p<0.05, mean±SEM, (t-test). N, number of hiPSC-CMs batches infected, n, number of cells. Insert describes voltage pulse protocols.

Kir2.1^Δ314–315 does not alter total Na_V1.5 protein levels but impairs Na_V1.5 plasma membrane expression

We observed no change in total Na_V1.5^WT protein expression in whole-cell extracts of HEK-293 (Na_V1.5+β1) transfected with either Kir2.1^WT or mutant Kir2.1^Δ314–315 protein, p=0.14. Similar results were obtained in HEK-293T cells transiently transfected with Na_V1.5 alone or co-transfected with either Kir2.1^WT or mutant Kir2.1^Δ314–315, p=0.25, Online Figure VII B,E. Moreover, as previously, we observed less total protein production of Kir2.1^Δ314–315 than Kir2.1^WT channel^{12, 28, 34} (Online Figure VII C,F). Therefore, to determine whether the decrease in I_Na density in cells co-expressing Kir2.1^Δ314–315 was due to a decrease in Na_V1.5^WT surface membrane expression, we performed cell surface biotinylation experiments in HEK-293T cells after transient co-transfection 1:1 of Na_V1.5^WT with Kir2.1^WT or Kir2.1^Δ314–315 channels. Cell lysates were subjected to immunoprecipitation with an antibody against the COOH-domain of Na_V1.5 and analyzed by Western blotting. The presence of the trafficking-deficient mutant Kir2.1^Δ314–315 resulted in a reduction in Na_V1.5^WT protein at the surface membrane compared with Kir2.1^WT channel, p=0.009, Figure 4A,B.

Figure 4. The trafficking deficient mutant Kir2.1^Δ314–315 decreases plasma membrane expression of Na_V1.5 channels in HEK-293T cells.

A. Representative Western blots of cell surface biotinylation studies. Samples were immunoprecipitated with rabbit anti-Na_V1.5 and detected by immunobloting with mouse anti-Na_V1.5 and streptavidin HRP conjugated. B. Graphic of quantification of Na_V1.5 cell surface expression. Cell surface membrane expression of Na_V1.5 channels was markedly reduced in the presence of Kir2.1^Δ314–315 compared with Kir2.1^WT. *p<0.05, mean±SEM, (t-test), (Na_V1.5^WT+Kir2.1^WT (n=7) and Na_V1.5^WT+Kir2.1^Δ314–315 (n=8), respectively. C. Whole-cell inputs of HEK-293T co-transfected 1:1 with Na_V1.5^WT and Kir2.1^WT incubated in the absence (control samples) or in the presence of endoglycosidase (Endo-H) or peptide N-Glycosidase-F (PNGase-F). Left, Na_V1.5^WT has populations that are Endo-H and PNGase-F sensitive. Right, Triton solubilized Na_V1.5^WT were first immunoprecipitated with Kir2.1^WT and then deglycosylated with Endo-H and PNGase-F. Equal amounts of sample were treated and subjected to 7.5% SDS-PAGE gel electrophoresis. Deglycosylation assay experiments were performed in five different batches of cells transfected at different days. MW, molecular weight.

Na_V1.5 and Kir2.1 associate early in their biosynthetic pathway

Na_V1.5 and Kir2.1 channels are present as complexes together with scaffolding proteins^19–26, and co-immunoprecipitate in both forward and reverse co-immunoprecipitation reactions²². However, the spatial and temporal coordination of their association remains unclear regarding Na_V1.5 and Kir2.1 coupling before, during or after insertion into the membrane. To determine whether association occurs before the channels reach the plasma membrane we examined their sensitivity to Endoglycosidase-H (Endo-H), which cleaves high mannose carbohydrates and indicates association with the endoplasmic reticulum (ER)-Golgi apparatus³⁵. Whole-cell homogenates and immunoprecipitates from HEK-293T cells co-transfected 1:1 with Na_V1.5^WT and Kir2.1^WT channels were analyzed by Western blot after treatment with Endo-H or Peptide-N-Glycosidase-F (PNGase-F), which enzymatically removed glycans added in the ER or all N-linked carbohydrates, respectively (Figure 4C). Na_V1.5^WT channels that co-immunoprecipitated with Kir2.1^WT channels were partially Endo-H-sensitive, as evidenced by the appearance of a lower molecular weight band after Endo-H treatment. The positive association between Endo-H-sensitive Na_V1.5^WT and Kir2.1^WT channels suggested that the Na_V1.5-Kir2.1 complex assembly occurred at least partially when the two channels trafficked through the ER-Golgi. Moreover, in co-immunoprecipitation experiments using an antibody against Nav1.5 in cells co-transfected with Na_V1.5^WT and Kir2.1^WT or Kir2.1^Δ314–315, Nav1.5^WT was able to pull down both Kir2.1^WT and mutant Kir2.1^Δ314–315 (Online Figure VIII). Taken together, these data indicate that Na_V1.5 and Kir2.1 channels can associate early in their biosynthetic pathway.

Kir2.1^WT increases while Kir2.1^Δ314–315 impairs Na_V1.5 mobility toward plasma membrane

We conducted fluorescence recovery after photobleaching (FRAP) experiments to assess whether Kir2.1 and Na_V1.5 traffic together along a common pathway. We subcloned each channel gene with one of two separate fluorophores (SCN5A with mTq2, and KCNJ2 or KCNJ2^Δ314–315 with Cp173venus), then used confocal microscopy and FRAP to quantify channel mobility when transfected alone or co-transfected together into COS-7 cells. After photobleaching a small region of interest (ROI = 5 μm diameter), the rate of fluorescence recovery was measured as an index of movement of unbleached channels entering the ROI. Online Figure IX A shows FRAP data for Kir2.1^{WT Venus} alone and for Kir2.1^WT co-transfected with Na_V1.5^{WT mTQ}. Presence of Na_V1.5^{WT mTQ} channels did not modify the fraction of mobile Kir2.1^{WT Venus} channels (Online Figure IX B), but accelerated the fluorescence recovery within the ROI t_1/2= 2.3±0.3 min (n=8) versus 4.7±0.6 min (n=7); p=0.002. These changes are better appreciated in Online Figure IX C, which shows Kir2.1^{WT Venus} recovery normalized to the maximum fluorescence recorded after photobleaching in the absence and the presence of Na_V1.5^{WT mTQ}. The mobile fraction of Na_V1.5^{WT mTQ} channels increased when co-transfected with Kir2.1^{WT Venus} (Online Figure IX D-F), although its t_1/2 was not modified t_1/2=2.4±0.2 min (n=11) versus 2.0±0.2 min (n=10), respectively; p=0.17. The mobile fraction and t_1/2 for both Na_V1.5^{WT mTQ} and Kir2.1^{WT Venus} channels were nearly identical when measurements were performed in the presence of the other channel (i.e. Na_V1.5^{WT mTQ} and Kir2.1^{WT Venus}, and vice versa; Online Figure IX G-I). Furthermore, co-transfection with Kir2.1^{Δ314–315Venus} decreased the mobile fraction, p<0.005, Online Figure IX E and delayed recovery of Na_V1.5^{WT mTQ2} channels t_1/2=3.1±0.4 min, n=9, p=0.11, Online Figure IX D-F. Altogether, the results strongly suggest that Kir2.1^WT-Na_V1.5^WT complexes travel more efficiently than Na_V1.5^{WT mTQ2} channels alone and faster than Kir2.1^{WT Venus} channels alone, and that the presence of Kir2.1^{Δ314–315Venus} reduces both the amount and rate of Na_V1.5^{WT mTQ} channel protein trafficking.

Localization of AP1 ϒ-adaptin subunit in ARVMs

Forward protein trafficking along intracellular compartments is accomplished by transport vesicles that form at sites where coat proteins have been recruited^{15, 36}. Previously it was shown that an interaction with the AP1 ϒ-adaptin subunit marks the incorporation of Kir2.1^WT channels into clathrin coated vesicles at the trans Golgi network¹². Also previous studies have shown that Kir2.1 and Na_V1.5 channels present similar subcellular distributions at t-tubules and the intercalated discs^{7, 20–22, 26, 37, 38} We therefore sought to determine whether in addition to Kir2.1, AP1 ϒ-adaptin also interacts with Na_V1.5 in ARVM. On the top panel of Figure 5, immunostaining with an antibody against AP1 ϒ-adaptin, shows a striated and punctate pattern distributed within regions corresponding to the t-tubules and the intercalated discs. Magnifications of the box region (panels a-g) shows that AP1 ϒ-adaptin colocalizes with Na_V1.5 and Kir2.1 channels at both the t-tubular regions and more intensely the intercalated discs (panels d,f and g).

Figure 5. Localization of AP1 ϒ-adaptin, Na_V1.5 and Kir2.1 in ARVMs.

Top, Overlapping immunostaining for ϒ-adaptin (green), Na_V1.5 (red) and Kir2.1 (cyan)., a,d, ϒ-adaptin expresses in the t-tubular regions and the intercalated discs. Intense punctate signal densities were also found randomly distributed within the cytoplasm. Na_V1.5 and Kir2.1 show similar subcellular distribution at the t-tubules and intercalated discs, b and c, respectively. ϒ-adaptin, colocalization with Na_V1.5 (f) and Kir2.1 (g). d. Colocalization of the three proteins (Scale bar: 10 μm).

Na_V1.5 Golgi export depends on interaction with the AP1 ϒ-adaptin subunit clathrin adaptor complex

Next, we conducted co-immunoprecipitation experiments to investigate whether, similar to Kir2.1^WT, Na_V1.5^WT protein interacts with the AP1 ϒ-adaptin subunit. We used an antibody against the COOH-domain of Na_V1.5 for immunoprecipitation from whole-cell homogenates of HEK-293T cells transfected with Na_V1.5^WT alone or co-transfected 1:1 with Kir2.1^WT or Kir2.1^Δ314–315. Immunoprecipitated proteins were prepared for Western blot analysis and probed with anti-AP1 ϒ-adaptin monoclonal antibody. As illustrated in Online Figure X A and B, Na_V1.5 channels pulled down AP1 ϒ-adaptin; the reverse interaction demonstrated that AP1 was also able to pull down Na_V1.5 in immunoprecipitation experiments. Since selective recruitment of the coat protein is based on sequence information within the cytoplasmic domains of the respective cargo protein^{33, 39–44}, we next investigated whether the Na_V1.5 COOH and/or NH₂ domains contained the typical consensus motif (YXXΦ; X being any amino acid and Φ being a bulky, hydrophobic residue) for adaptin binding. In the COOH-domain of Na_V1.5 channels we found four consensus motifs for AP1 binding, beginning at amino acid positions Y1794, Y1810, Y1888 and Y1976. In the NH₂ domain one consensus motif for AP1 begins at amino acid position Y68. We therefore subjected the Tyrosine (Y) residues of the consensus motifs to sequential Alanine (A) replacement mutagenesis and assessed whether the mutant Na_V1.5 channels interacted with the AP1 ϒ-adaptin subunit and whether they were functional. Immunoblots of all Alanine-substitutions (Na_V1.5^Y68A, Na_V1.5^Y1810A, Na_V1.5^Y1888A and Na_V1.5^Y1976A) showed a band at ~250 KDa in all inputs and co-immunoprecipitation samples from Na_V1.5 mutants and Na_V1.5^WT channels (Figure 6A). No change in total protein expression was observed between AP1 binding mutant consensus motifs and the Na_V1.5^WT channel (Online Figure XI A,B). Moreover, all Alanine-substitutions expressed functional Na_V1.5 channels (Figure 6B). However, I_Na density was reduced for the Na_V1.5^Y1810A mutant (Figure 6B,C). This result was consistent with the observed loss of interaction between Na_V1.5^Y1810A and the AP1 ϒ-adaptin subunit (Figure 6A). Taken together, the data suggest that similar to Kir2.1, a significant proportion of Na_V1.5 channels interact with AP1 and are selected for export from the Golgi in a signal dependent manner through an AP1 clathrin adaptor interaction.

Figure 6. Alanine-scanning mutagenesis of Na_V1.5 channels for typical consensus motif to adaptin binding.

A. Representative Western blot after immunoprecipitation (IP) with ϒ-adaptin specific antibody. Detected by immunoblot were Na_V1.5 and ϒ-adaptin proteins on HEK-293T cells transfected with Na_V1.5^WT, Nav1.5^Y68A, Nav1.5^Y1810A, Nav1.5^Y1888A and Nav1.5^Y1976A (n=3). Botton (ϒ-adaptin #) shows a longer exposure of the same immunoblot shown in Middle. B. Current traces from Na_V1.5^WT, Nav1.5^Y68A, Nav1.5^Y1810A, Nav1.5^Y1888A and Nav1.5^Y1976A. C. Peak I_Nav1.5 density at −25 mV of Na_V1.5^WT, Nav1.5^Y68A, Nav1.5^Y1810A, Nav1.5^Y1888A and Nav1.5^Y1976A (n=16, n=7, n=16, n=10 and n=12, respectively). All constructs/combinations were tested at least in three different batches of cells transfected in at least three different days. MW, molecular weight. *p<0.05, Na_V1.5^WT vs Na_V1.5^Y1810A, ^# p<0.05, Na_V1.5^WT vs Na_V1.5^Y1976A, mean±SEM, Two-way ANOVA.

Silencing the AP1 ϒ-adaptin subunit reduces I_Na, I_K1 and modifies the action potential characteristics of hiPSC-CMs

We used adenoviral transfer technology to knock down the expression of AP1 ϒ-adaptin subunit and measured I_Na and I_K1 densities in hiPSC-CMs. We cultured hiPSC-CM monolayers for 7 days on 96-well plates as describe previously.⁴⁵ hiPSC-CM monolayers were infected with shRNA for AP1 ϒ-adaptin (Ad-shϒ-adaptin) at varying MOIs to determine the level needed for knockdown; samples were collected 3 days postinfection. Separate groups of hiPSC-CM monolayers were infected with control Scramble-GFP virus (Ad-Scr-GFP), Online Figure XII A,B. On Western Blot analysis, 200 MOI reduced relative levels of AP1 ϒ-adaptin protein by ~63 % without suppressing total Na_V1.5^WT or Kir2.1^WT protein levels (Figure 7A,B). I_K1 density was smaller in hiPSC-CMs infected with Ad-shϒ-adaptin compared with control GFP infected hiPSC-CMs (−9.3±1.7 versus −3.8±0.5 pA/pF at −120 mV, p=0.01, Figure 7C,D). Similar to I_K1, infection with Ad-shϒ-adaptin also reduced I_Na density compared to Ad-Scr-GFP. Peak currents at −20 mV were −18±1 versus −24±1 pA/pF, respectively, p=0.004, Figure 7E,F.

Figure 7. Silence AP1 ϒ-adaptin subunit reduce I_Na and I_K1 on hiPSC-CMs.

A, B Ad-shϒ-Adaptin infection silence ϒ-Adaptin subunit on hiPSC-CMs by day 3 in culture monolayers compared with control Ad-Scr-GFP without changing Na_V1.5^WT, Kir2.1^WT and β-cardiac myosin heavy chain (βMHC). A. Postinfection Western blot of whole cell preparation of hiPSC-CMs monolayers and human ventricle tissue (100 μg) showing immunodetection of βMHC, Na_V1.5^WT, ϒ-Adaptin, Kir2.1^WT and GAPDH (loading control). B. Quantification of ϒ-Adaptin, Na_V1.5^WT, Kir2.1^WT and βMHC, protein expression on whole cell extract from A. C. I_K1 traces from hiPSC-CMs infected with Ad-Scr-GFP or Ad-shϒ-Adaptin, respectively. D. Superimposed I_K1 I/V relationships for Ad-GFP and Ad-shϒ-Adaptin (N =3, n=13 and N=3, n=10, respectively). E. I_Na traces from hiPSC-CMs infected with Ad-Scr-GFP or Ad-shϒ-Adaptin. F. I_Na, I/V relationships for Ad-Scr-GFP, Ad-shϒ-Adaptin (N=3, n=45 and N=3, n=42, respectively). *p<0.05, mean±SEM, (t-test). N, number of hiPSC-CMs batches infected, n, number of cells. Insert describes voltage pulse protocols. MW, molecular weight.

Next, we determined the effect of silencing the AP1 ϒ-adaptin subunit on the action potential characteristics of hiPSC-CMs. In Figure 8F, MDP was reduced in hiPSC-CMs treated with Ad-shϒ-adaptin relative to control hiPSC-CMs infected with Ad-Scr-GFP. All cells were paced at 2.5 and 5 Hz. MDPs were −68±1 (2.5 Hz) and −68±1 mV (5 Hz) in Ad-shϒ-adaptin hiPSC-CMs, versus −79±2 (2.5 Hz) and −78±2 mV (5 Hz) in Ad-Scr-GFP iPSC-CMs (p=0.01 in both cases). Action potential amplitude, APD₅₀, APD₉₀, dV/dt_max and overshoot, were similar in both hiPSC-CMs groups (Figure 8 B,C,D,G,H). Silencing the AP1 ϒ-adaptin subunit impaired the rate dependent APD adaptation of hiPSC-CMs, ΔAPD₉₀, ΔAPD₅₀ were 2±4, 1±4 ms, respectively, in iPSC-CMs Ad-shϒ-adaptin versus 20±4, 18±4 ms in iPSC-CMs Ad-Scr-GFP, respectively, p=0.04 for both ΔAPD, Figure 8E.

Figure 8. Silence AP1 ϒ-adaptin subunit reduce MDP and impairs rate dependent APD adaptation of hiPSC-CMs.

A. Representative action potential recordings from hiPSC-CMs infected with adenoviruses encoding Ad-Scr-GFP and Ad-shϒ-Adaptin at 2.5, 5 Hz (black and blue), respectively. B,C,D, F, G, H Plot individual values, amplitude (B), action potential duration to 50 and 90% of repolarization (C), (D), maximum diastolic potential (F), maximal dV/dt (G), and overshoot (H). Delta APD₅₀ and APD₉₀ repolarization between 2.5 and 5 Hz (ΔAPD5_{50, 90}) respectively (N=2, n=4 and N=2, n=4, respectively, N, number of hiPSC-CMs batches infected, n, number of cells. *p<0.05, Ad-Scr-GFP vs Ad-shϒ-Adaptin mean±SEM, (t-test). Action potential amplitude, overshoot, dV/dt_max, APD₅₀ and APD₉₀ were similar in both hiPSC-CMs groups (112±2, 35±0.6 mV, 229±31 V/s, 78±9 and 81±7 ms in Ad-shϒ-adaptin treated iPSC-CMs versus 110±4, 31±2 mV, 142±22 V/s, 85±8 and 88±9 ms in Ad-Scr-GFP treated iPSC-CMs, p>0.05 for Action potential amplitude, overshoot, dV/dt_max, APD₅₀ and APD₉₀ at 2.5 Hz and (108±1, 32±0.2 mV, 197±18 V/s, 77±1 and 79±9 ms in treated Ad-shϒ-adaptin iPSC-CMs versus 107±2, 29±0.6, 127±14 V/s, 66±4 and 72±5 ms, respectively, in Ad-Scr-GFP treated iPSC-CMs, p>0.05 for Action potential amplitude, APD₅₀ and APD₉₀, p=0.04, p=0.004 for dV/dt_max and overshoot, respectively at 5 Hz.

DISCUSSION

We used two trafficking-deficient mutant Kir2.1 channels to provide novel insight into mechanisms that regulate the Na_V1.5-Kir2.1 macromolecular complex. Our data reveal that co-expression with Kir2.1^Δ314–315 reduces expression of Na_V1.5 at the surface membrane without modifying biophysical properties of the channel or total protein production. We show that Na_V1.5 and Kir2.1 channels can associate early in their biosynthetic pathway, and that the Na_V1.5-Kir2.1 complex traffics more efficiently than Na_V1.5 alone, but Kir2.1^Δ314–315 impairs Na_V1.5 trafficking. Correspondingly, our patch-clamp data demonstrate that reduced Na_V1.5 surface membrane expression results in decreased I_Na density. In both ARVMs and hiPSC-CMs expression of mutant Kir2.1^Δ314–315 channel reduced I_K1 and I_Na, MDP and dV/dt_max, and increased APD. In immunostaining experiments AP1 ϒ-adaptin shows a t-tubular distribution and is also concentrated at the intercalated disc colocalizing with both Na_V1.5 and Kir2.1. Site-direct mutagenesis revealed that Na_V1.5^WT interacts with AP1 ϒ-adaptin through the Na_V1.5^Y1810 residue. Finally, silencing the AP1 ϒ-adaptin subunit in hiPSC-CMs reduced I_K1, I_Na and MDP, and impaired rate dependent APD adaptation. These data suggest strongly that in cardiomyocytes, Na_V1.5 and Kir2.1 channels may travel together to their eventual micro-domains, whether at the lateral membrane, t-tubule or intercalated disc; disruption of Kir2.1 trafficking affects trafficking of Na_V1.5; and Golgi export of a proportion of Na_V1.5^WT channels depends on interaction with the AP1 ϒ-adaptin subunit clathrin adaptor complex. The Kir2.1^Δ314–315 mutant is expressed and processed into mature channels but at much lower efficiency than Kir2.1^{WT;12, 28, 34} we observed less total protein production for Kir2.1^Δ314–315 than Kir2.1^WT (Online Figure VII C,F). Nevertheless, Kir2.1^Δ314–315 was sufficient to affect trafficking of Na_V1.5 to the cell surface whether it was expressed to form homotetramers, or to form heterotetramers with Kir2.1^WT (Figure 1C,D, Figure 3C,D, Online Figure I C,D and Online Figure II B). Further, when we co-transfected Kir2.1^WT, Kir2.1^Δ314–315 and Na_V1.5^WT, the I_Kir2.1 density was 92% smaller than the I_Kir2.1 density of Kir2.1^WT when expressed alone. These results agree with previous reports showing that, while Kir2.1^Δ314–315 forms channels that do not traffic to the membrane surface, they retain their ability to co-assemble with, and exert a dominant negative effect on the Kir2.1^WT channels.²⁸ Therefore, it is possible that the I_Kir2.1 density recorded is elicited by Kir2.1^WT subunits co-assembled as homotetramers rather than Kir2.1^WT or Kir2.1^Δ314–315 subunits rescued by the expression by Na_V1.5^WT channels, Online Figure I A,B.

Expression of the trafficking-deficient mutant Kir2.1^Δ314–315 channel significantly reduced I_Na density in HEK-293T cells, ARVMs and hiPSC-CMs. However, a significant amount of residual current was maintained in all cases, implying that some Na_V1.5 channels were able to reach the plasma membrane. Data in the literature suggest that HEK-293 cells, cardiomyocytes and developing neurons contain multiple intracellular storage pools of Na_V1.5 channel protein that can be mobilized following a physiologic stimulus^46–48. Those reports together with our data suggest that different pools of Na_V1.5 channels reach the plasma membrane from distinct origins (Golgi apparatus, unconventional Golgi pathway or storage pools). Based on the report by Mercier and coworkers¹⁸, as well as our results, it is likely that the Na_V1.5 channels affected by Kir2.1^Δ314–315 are nascent Na_V1.5 channels that use the classical Golgi secretory route to reach the plasma membrane. The percentage of Na_V1.5 channels that use one or the other pathway remains a topic of further experimentation. Even so, the fact that a fraction of Na_V1.5^WT channels that associate with Kir2.1^WT are Endo-H-sensitive, strongly suggests that Na_V1.5^WT and Kir2.1^WT channels can interact early in their biosynthetic pathway.

Our understanding of how newly synthesized plasma membrane proteins are sorted to the cell membrane is still limited. Critical regulatory steps that control the surface expression of ion channels are likely to be involved starting at the endoplasmic reticulum and Golgi. Considerable evidence indicates that targeting or trafficking of ion channels within the cell is frequently mediated by interaction with other proteins through specific amino acid motifs present in the channel sequence^{20–23, 26, 38, 49}. Moreover, evidence suggests the existence of retention and export trafficking signals that regulate functional expression of transmembrane proteins^{33, 39–44}. Still, the precise export signals that regulate exit of Na_V1.5 from the Golgi and transit to the plasma membrane are not fully understood. The AP1 clathrin adaptor is one of the best known molecules to facilitate export of cargo proteins from the trans-Golgi network⁵⁰. We show the distribution of AP1 protein and its colocalization with Na_V1.5 and Kir2.1 in cardiac cells and explored whether Na_V1.5 contained the typical amino acid consensus motif sequence (YXXΦ) for adaptin binding. Based on the amino acid sequence of the channel we identified four probable consensus binding motifs for the AP1 complex on the COOH domain (Y1794, Y1810, Y1888, Y1976) and one possible motif on the NH₂ domain (Y68). Furthermore, our co-immunoprecipitation experiments demonstrate that Na_V1.5 interacts with the AP1 complex. Yet Na_V1.5^Y1810A did not co-immunoprecipitate with AP1 protein, consistent with a decrease in I_Na density flowing through Na_V1.5^Y1810A channels. This suggests that, similar to Kir2.1, some Na_V1.5 channels may be exported from the Golgi in a signal-dependent manner through an AP1 clathrin adaptor interaction. We observed that Na_V1.5^Y1976A channels had an increased I_Na density relative to control. Interestingly the Y1976 residue locates within the PY motif of the Na_V1.5 channel,^{51, 52} and the Na_V1.5^Y1976A mutant has been shown to impair the Nedd4-2 binding motif⁵³. Therefore, it is likely that the increase in I_Na density observed for the Na_V1.5^Y1976A mutant is the result of decreased ubiquitination followed by internalization and/or degradation of Na_V1.5^Y1976A channels. Future studies will be necessary to confirm that hypothesis.

It has been proposed that the YXXΦ consensus motif recruits specific coat protein complexes that induce formation of transport vesicles, not only destined for certain subcellular localization, but also preferentially loaded with respective cargo proteins⁵⁴. Thus, based on our data showing that both channels interact with the AP1 complex, we postulate that Kir2.1 and Na_V1.5 channels may be loaded in joint transport vesicles to be delivered to common subdomains at the plasma membrane. These data support previous reports on the existence of different pools of Na_V1.5 channel trafficking vesicles¹⁸ and suggest that trafficking vesicles may carry varying compositions of load proteins. However, even though our data show Na_V1.5 can interact with AP1 for transportation, we cannot exclude participation of GGA1, a protein also known to facilitate export cargo from the trans-Golgi network⁵⁰.

Moreover, we observed that silencing the AP1 ϒ-adaptin subunit in hiPSC-CMs reduced I_K1, I_Na and the MDP and impaired the rate dependent APD adaptation. However, significant amounts of residual I_K1 and I_Na were maintained, implying that some Kir2.1 and Na_V1.5 channels reached the plasma membrane. These data suggest that, like Na_V1.5 channels, Kir2.1 channels reach the plasma from distinct origins that may include, in addition to the Golgi apparatus, an unconventional Golgi pathway or storage pools and that other proteins may be involved in the export of Kir2.1 and Na_V1.5 channels from the trans-Golgi network. The observation that AP1 ϒ-adaptin silencing impaired the rate dependence of APD adaptation in hiPSC-CMs, suggests that AP1 ϒ protein may regulate the expression of ion channels and proteins involved in calcium homeostasis^55–58.

The intracellular COOH terminal of Na_V1.5 contains several protein-protein interaction motifs. The most well-characterized interaction sites are the calmodulin-biding IQ motif, the PY motif, and a PDZ domain-biding motif^{59, 60}. In contrast, less is known about proteins that interact with the NH₂ domain of Na_V1.5. The Na_V1.5 NH₂ domain is critical for expression density, although the specific mechanism is not clear. A putative Na_V1.5 NH₂ fragment enhances Na_V1.5^WT channel trafficking by possibly acting as a decoy, allowing Na_V1.5^WT channels to bypass a regulatory system and reach the plasma membrane⁶¹. More recently, Matamoros et al. demonstrated that the NH₂ terminus of Na_V1.5 has a PDZ-like binding domain that mediates binding to α1-syntrophin, and can act like a chaperone to increase I_Na and I_K1 through enhanced expression of Na_V1.5, Kir2.1, and Kir2.2 but not Kir2.3 proteins²⁶.

Our data demonstrate for the first time that Na_V1.5 interact with AP1 involved in the trans-Golgi network of cargo export proteins. Similar to the Kir2.1-AP1 complex¹², the Na_V1.5-AP1 complex interaction may participate in regulation of functional expression of Na_V1.5, and contribute to formation of the Na_V1.5-Kir2.1 channelosome²²

Clinical relevance

The interplay between I_K1 and I_Na is crucial to normal cardiac electrical function. By controlling resting membrane potential, I_K1 modifies sodium channel availability and therefore cell excitability, APD, and conduction velocity. I_K1 and I_Na interactions are key determinants of electrical rotor dynamics which are responsible for abnormal, often lethal, cardiac reentrant activity⁶². Data presented here demonstrate that expression of the trafficking-deficient mutant Kir2.1^Δ314–315 in myocytes not only reduces native I_K1, but also I_Na by disrupting functional expression of both channels at the plasma membrane. Previous simulation studies predicted that a reduction in I_K1 might have unique effects on AP configuration and arrhythmia susceptibility³⁰. It was anticipated that a reduction in I_K1 would prolong the terminal repolarization phase of the cardiac AP by reducing the amount of repolarizing current during the terminal phase of the AP, resulting in QT prolongation in affected individuals³⁰. It is well known that AP prolongation is a prerequisite for early afterdepolarizations that can trigger spontaneous APs. In ARVMs we observed that expression of Kir2.1^Δ314–315 resulted in membrane depolarization, marked APD prolongation and spontaneous APs, validating the aforementioned numerical predictions.

Andersen-Tawil Syndrome type 1 (ATS1) is a rare disorder caused by mutations in the KCNJ2 gene coding Kir2.1. Clinical manifestations of ATS1 include periodic paralysis, cardiac arrhythmias, and dysmorphic features. The cardiac manifestations of ATS1 patients with KCNJ2 mutations include QT prolongation, premature ventricular contractions, complex ventricular ectopy, bigeminy, and polymorphic or bidirectional ventricular tachycardia. It has been difficult to correlate functional consequences of individual Kir2.1 mutation12 s with corresponding clinical phenotypes^{28–30, 63}. Our data provide novel insights into the ionic basis of arrhythmia susceptibility in ATS1 patients. We postulate that in addition to reduced I_K1, ATS1 patients also have reduced I_Na as a consequence of disruption of the Kir2.1-Na_V1.5 complex by a trafficking-deficient Kir2.1 mutation. Consequently, co-expression of the trafficking-deficient mutant Kir2.1 channel with Na_V1.5^WT is expected to contribute to arrhythmogenesis in ATS1 patients by reducing MDP and dV/dt_max via two different mechanisms: the voltage dependent mechanism, and effects of physical intermolecular interactions between the two channels. Our data highlight the importance of considering effects of a given mutation on functional expression of other ion channels or proteins that may be a part of the macromolecular complex within which Kir2.1 interacts.

Our study focused on the interplay between Kir2.1 and Na_V1.5. However, it is likely that other proteins, including ion channels, are part of the same macromolecular complex. The presence of a trafficking-deficient mutant channel may have the potential to affect functional expression of another channel within its macromolecular complex. Notably, we selected Kir2.1^Δ314–315 for its ability to retain the channel in the Golgi and its strong dominant negative effect. We also explored functional consequences of co-expressing Na_V1.5 with the double mutant Kir2.1^R44A/R46A, whose surface expression is reduced³³. We observed that I_Nav1.5 density of cells that co-expressed Kir2.1^R44A/R46A was reduced with respect to cells expressing Na_V1.5^WT channels alone (Online Figure II C,D). Unexpectedly, our functional analysis showed that mutant Kir2.1^R44A/R46A channels generated a current density that was similar to I_Kir2.1^WT. Future studies will be necessary to further elucidate the molecular mechanism involved in the regulation of functional expression of Kir2.1 channels.

These data support our contention that a common early anterograde trafficking mechanism is involved in transport of Na_V1.5 and Kir2.1 channels to the plasma membrane. Therefore, it seems reasonable to speculate that Kir2.1 mutants that affect trafficking from the endoplasmic reticulum have similar effects. However, while our data uncovers a common early anterograde trafficking pathway for Na_V1.5 and Kir2.1 channels, it is premature to expect that all Kir2.1 trafficking-deficient channels will induce a decrease in I_Na density. In this regard, an unexpected finding was that Na_V1.5^WT channel failed to rescue mutant Kir2.1^Δ314–315 plasma membrane expression. However, it is possible that for trafficking deficient Kir2.1 mutants with weaker trafficking properties, the presence of Na_V1.5 channels might rescue and promote surface expression. It is also possible that trafficking-deficient Na_V1.5 mutations affect functional expression of Kir2.1 and other membrane channels. In this regard, it has been shown previously that the α-subunit of I_Ks can interact with and modify the localization and current-carrying properties of the α-subunit of I_Kr.⁶⁴ Clearly additional work is needed to unravel the details of regulation of early trafficking of cardiac ion channels and their incorporation into macromolecular complexes at the cell membrane.

Potential limitations

Our study has potential limitations. First, some experiments were performed in heterologous HEK-293T, COS-7 or CHO cell expression systems, which make interpretation of trafficking studies difficult. However, we validated functional expression results by recording native I_Na and I_K1 as well as AP characteristics in ARVMs and hiPSC-CMs. Second, based on our data we are unable to differentiate whether the observed trafficking disruption in Na_V1.5 on Kir2.1^Δ314–315 expression was due to a delay in expression kinetics, or whether all newly synthetized Na_V1.5 channels were retained in the Golgi. Third, we cannot rule out that accumulation of trafficking-deficient mutant Kir2.1^Δ314–315 at the Golgi evokes a stress state, and the response to adjust Golgi capacity to myocyte demand might be compromised. Finally, we did not determine whether preferential reduction of Na_V1.5 channel was localized to the lateral membrane, T tubules or intercalated discs.

The above limitations notwithstanding, our results provide strong evidence that the Na_V1.5-Kir2.1 macromolecular complex preassembles early in its forward trafficking pathway, and that trafficking deficiency and retention of one channel at the Golgi affects trafficking of both channels. Our data also show that Na_V1.5 channels may be selected as cargo into Golgi export carriers in a signal dependent manner through an AP1 clathrin adaptor interaction. Altogether, the data provide novel insights into molecular mechanisms underlying arrhythmogenesis in ATS1 patients.

NOVELTY AND SIGNIFICANCE.

What Is Known?

• The main cardiac sodium channel (Na_V1.5) and the strong inward rectifier potassium channel (Kir2.1) reciprocally modulate their functional expression, an interplay that is crucial to normal cardiac electrical function. By controlling resting membrane potential, IK1 modifies sodium channel availability and therefore cell excitability, APD, and conduction velocity.

• Na_V1.5 and Kir2.1 channels physically interact together with multiple other protein partners from early stages of protein trafficking, targeting through membrane anchoring, recycling and degradation.

What New Information Does This Article Contribute?

• A pool of Na_V1.5 and Kir2.1 channels preassemble during early forward trafficking and travel together to common membrane microdomains.

• Similar to Kir2.1, some Na_V1.5 channels can exit the Golgi apparatus by an adaptor protein complex 1 (AP1)-dependent trafficking process.

• Trafficking deficiency of Kir2.1 channels affects trafficking of both Kir2.1 and Na_V1.5 channels complex components, which may have important implications for the mechanisms of arrhythmias in inheritable and possibly other cardiac diseases.

Cardiac Na_V1.5 and Kir2.1 function within macromolecular complexes to control cardiac excitability. Cellular processes that regulate trafficking and expression of Kir2.1 and Na_V1.5 channels are of interest as potential mediators of proarrhythmic channelopathies and arrhythmogenesis in other cardiac diseases, including heart failure. However, it remains unknown whether Kir2.1 and Na_V1.5 pre-assemble or interact during early anterograde trafficking steps of their secretory pathways. Here, we demonstrate that the Na_V1.5 and Kir2.1 form a complex that preassembles early in its forward trafficking pathway. Further, we found that Na_V1.5 channels may be selected as cargo into Golgi export carriers in a signal dependent manner through an AP1 clathrin adaptor interaction. Moreover, trafficking deficiency and retention of Kir2.1 channel at the Golgi apparatus affects trafficking of Na_V1.5 channels. Our data highlight the importance of considering effects of trafficking defective mutations on functional expression of other ion channels or proteins that may be a part of the macromolecular complex within which the proteins interact. Our study offers a novel paradigm regarding the molecular mechanism of arrhythmia susceptibility in monogenic ion channels diseases such as the Andersen Tawil Syndrome and potentially other diseases, including Brugada syndrome.

Nonstandard Abbreviations and Acronyms

AP

Action potential

AP1

Adaptor protein complex 1

AP2

Adaptor protein complex 2

AP3

Adaptor protein complex 3

AP4

Adaptor protein complex 4

AP5

Adaptor protein complex 5

APD

Action potential duration

ARVM

Adult rat ventricular myocytes

ATS1

Andersen-Tawil Syndrome type 1

CHO

Chine hamster ovary

COS-7

CV-1 (simian) in Origin carrying the SV40 genetic material

EAD

Early afterdepolarizations

Endo-H

Endoglycosidase-H

ER

Endoplasmic reticulum

FBS

Fetal bovine serum

FRAP

Fluorescence-recovery-after-photobleaching

HEK

Human embryonic kidney

hiPSC-CMs

Human induce pluripotent stem cell derived cardiomyocytes

I_K1

Inward rectifier potassium current

I_Na

Inward sodium current

I_to

Transient outward potassium current

Kir

Inward rectifier potassium channel

MDP

Maximum diastolic potential

Na_V1.5

Voltage-gate sodium channel

PBS

Phosphate buffer saline

PNGase-F

Peptide-N-Glycosidase-F

PVCs

Premature ventricular contraction

ROI

Regions of interest

VT

Ventricular tachycardia

WT

Wildtype