Structural and Functional Characterization of A Na_v1.5-Mitochondrial Couplon

Abstract

Rationale:

The cardiac sodium channel Na_v1.5 has a fundamental role in excitability and conduction. Previous studies have shown that sodium channels cluster together in specific cellular subdomains. Their association with intracellular organelles in defined regions of the myocytes, and the functional consequences of that association, remain to be defined.

Objective:

To characterize a subcellular domain formed by sodium channel clusters in the crest region of the myocytes, and the subjacent subsarcolemmal mitochondria (SSM).

Methods and Results:

Through a combination of imaging approaches including super-resolution microscopy and electron microscopy we identified, in adult cardiac myocytes, a Na_v1.5 subpopulation in close proximity to SSM; we further found that SSM preferentially host the mitochondrial Na⁺/Ca²⁺ exchanger (NCLX). This anatomical proximity led us to investigate functional changes in mitochondria resulting from sodium channel activity. Upon TTX exposure, mitochondria near Na_v1.5 channels accumulated more Ca²⁺ and showed increased ROS production when compared to interfibrillar mitochondria. Finally, crosstalk between Na_v1.5 channels and mitochondria was analyzed at a transcriptional level. We found that SCN5A and SLC8B1 (which encode Na_v1.5 and NCLX, respectively) are negatively correlated both in a human transcriptome dataset (GTEx) and in human-induced pluripotent stem cell-derived cardiac myocytes deficient in SCN5A.

Conclusions:

We describe an anatomical hub (a couplon) formed by sodium channel clusters and SSM. Preferential localization of NCLX to this domain allows for functional coupling where the extrusion of Ca²⁺ from the mitochondria is powered, at least in part, by the entry of sodium through Na_v1.5 channels. These results provide a novel entry-point into a mechanistic understanding of the intersection between electrical and structural functions of the heart.

Graphical Abstract

INTRODUCTION

The voltage-gated sodium channel (Na-channel) is central to cardiac electrogenesis. Its dysfunction can lead to lethal arrhythmias both in acquired (ischemia;^{1, 2} heart failure²), and genetic disorders.^3-5 Rare and common genetic variants in SCN5A, encoding the pore-forming Na-channel subunit Na_v1.5, are strongly associated with life-threatening arrhythmias^{4, 6} and with structural heart disease.^7-9 Yet, insight into Na-channel function, particularly as it extends beyond electrogenesis, is limited.

Na-channels form macromolecular complexes that cluster at specific subcellular domains (or “pools”).^10-12 An emerging concept is that certain Na-channel partners localize only to one subcellular domain, thus endowing the functional complex with region-specific properties that, if disrupted, facilitate arrhythmias.^13-17

Subcellular organelles such as mitochondria also organize in region-specific subdomains. Mitochondria in myocytes are highly organized, spatially separated into at least three subpopulations: one just beneath the surface sarcolemma (subsarcolemmal mitochondria, SSM), a second one (the largest fraction) between myofibrils in longitudinal chains (interfibrillar mitochondria, IFM), and a third one, occupying the perinuclear subdomain (PNM). Data show that these subpopulations possess distinct biochemical properties, morphology, Ca²⁺ handling, and that they may interact differently with other intracellular structures, causing functional specificity.^18,19

The relationship between Na-channel and mitochondrial function is well established in one direction only: mitochondrial function and/or molecular composition can alter sodium current (I_Na).^20,21 Recently, however, Zhang et al demonstrated that reduced levels of Scn5a expression lead to accumulation of myocardial reactive oxygen species (ROS),²² and Wan et al documented that expression of Na_v1.5 mutant F1759A leads to mitochondrial injury.²³ The mechanism by which sodium channel expression or activity can affect mitochondrial function or integrity is not known.

We investigated the anatomical and functional relation between mitochondria and Na_v1.5 in adult cardiac myocytes. Advanced imaging methods allowed us to observe that a subpopulation of membrane-resident Na_v1.5 clusters sits in close apposition to subjacent SSM, and that it is in this mitochondrial subpopulation where the mitochondrial Na⁺/Ca²⁺ exchanger (NCLX; encoded by SLC8B1) is primarily located. This anatomical proximity led us to investigate functional changes in mitochondria resulting from Na-channel activity. In particular, exposing adult cardiac myocytes to the Na-channel blocker tetrodotoxin (TTX) caused accumulation of mitochondrial Ca²⁺ and ROS production, predominantly in SSM. We speculate that proximity to functional Na_v1.5 clusters facilitates sodium-dependent activity of NCLX to extrude mitochondrial Ca²⁺, whereas reduced sodium entry impairs Ca²⁺ extrusion. In addition, we observed that the relation between Na_v1.5 and NCLX is not only local. Indeed, human transcriptome analysis shows that SCN5A negatively correlates with SLC8B1. Experimental confirmation was obtained from human-induced pluripotent stem cell-derived cardiomyocytes (hiPSC-CM) in which SCN5A was knocked down. We conclude that not only mitochondrial metabolism can affect I_Na but Na-channels can, in turn, positively affect mitochondrial function. Our data lead us to propose the existence of a Na-channel subpopulation that clusters with subjacent mitochondria to create an anatomical couplon for electro-metabolic coupling.

METHODS

Data Availability.

The authors declare that all supporting data are available within the article and its Online supplement files.

Expanded Methods available in Online Data Supplement.

Mice.

Wild-type black mice C57BL/6N strain 027 (Charles River Laboratories) were treated in accordance to the Guide for Care and Use of Laboratory Animals published by the US National Institutes of Health. Procedures were approved by the NYU IACUC committee, protocol 160726. Both genders were included and animals were between 3-6 months of age.

Focused Ion Beam-Scanning Electron Microscopy (FIB-SEM).

Protocol modified from Wilke et al. ²⁴ Tissue was processed and en bloc lead staining performed to enhance membrane contrast.²⁵ Acquisition was done using Auto Slice and View G3 software.²⁶

Inmunostaining.

For immunolocalization, HL-1 cells were incubated with Mitotracker and NCLX antibody, as described in Online Supplement Data.

Cardiomyocyte dissociation.

Adult mouse ventricular myocytes were obtained by enzymatic dissociation, as described in Online Supplement Data.

Single-molecule localization microscopy (SMLM) by stochastic optical reconstruction microscopy (STORM).

Ventricular cardiomyocytes were incubated with Mitotracker before fixation and immunostained with NCLX, Na_v1.5 and α-actinin antibodies. Interaction factor analysis (details in Supplemental Data) provided a measure of the degree to which mitochondria and Na_v1.5 proximity can be the result of random (vs deterministic) distribution.

Mitochondrial Ca²⁺ dynamics.

As described,²⁷ mitochondrial Ca²⁺ was measured with Rhod-2-AM dye. Analysis and quantification were done using LASX and ImageJ software. To evaluate mitochondrial Ca²⁺ dynamics in response to drugs, we monitored Rhod-2 fluorescence signal before (F₀) and after (F) adding: TTX, carbonyl cyanide 4-(trifluoromethoxy) phenylhydrazone (FCCP) + oligomycin A and CGP-37157.

Mitochondrial superoxide detection.

Superoxide production was measured using MitoSOX, per manufacturer's instructions. To evaluate mitochondrial superoxide changes in response to TTX, we measured MitoSOX fluorescence signal before (F₀) and after (F) TTX.

SCN5A correlation analysis in human left ventricle.

RNAseq raw data from human left ventricular tissue were downloaded from the Genotype-Tissue Expression v8 (GTEx) consortium (https://www.gtexportal.org/home/datasets). Analysis done using Rstudio.

hiPSC-CM model of SCN5A knockout.

hiPSCs were maintained as previously described.²⁸ CRISPR/Cas9 was used to generate the SCN5A knock-out line. Differentiation into hiPSC-CMs was modified from a previously described protocol.^29,30

Statistics.

Numerical results are given as means±standard error of the mean (S.E.M.). Statistical tests were used as stated. Differences were considered significant at p<0.05.

RESULTS

Mitochondria are in close apposition to the sarcolemma.

Figure 1A shows an electron microscopy image of mouse left ventricle revealing the alignment of mitochondria along the interfibrillar space. A complete volumetric image of cardiac tissue acquired using FIB-SEM (Fig. 1B) showed mitochondria “nested” immediately under the sarcolemma in the dome-shaped surface spanning the distance between z-disks (the “crest”³¹), with both structures (sarcolemma and SSM) reaching points of close apposition (red arrows; small panels show different planes of the same preparation). A zoom-in of Figure 1B-A, with detailed description of structures observed in the left end of the image, is presented in Supplemental Fig.I. Segmentation analysis of FIB-SEM images facilitated visualization (Fig. 1C) and revealed multiple connections of SSM with deeper mitochondria both immediately subjacent and, through mitochondria that bridge across sarcomeres, to other interfibrillar chains in the cell, hence expanding into a complete mitochondrial reticulum³¹ (Fig. 1C and Supplemental Fig.II). Approximately 13% of the mitochondrial surface in the subsarcolemmal space was found <20 nm apart from the membrane (Supplemental Fig.III).

Figure 1. Mitochondria in close apposition to sarcolemma.

A. Focused Ion Beam-Scanning Electron Microscopy (FIB-SEM) image of ventricular myocytes. Notice the presence of mitochondria between z-disks (red box). Scale bar 2 μm. B. FIB-SEM single plane of complete 3D-volumetric image of cell membrane and subjacent mitochondria. Notice mitochondria under the crest. Enlarged images in bottom frames: sequential z-sections showing proximity of membranes and points where membranes come in close apposition (red arrows). C. Segmentation analysis of subsarcolemmal mitochondria connecting to mitochondrial reticulum. Top left: Mitochondria in bottom images of panel B, highlighted in blue. Top right: Segmentation of layers below shows subsarcolemmal mitochondria coupling to first chain of interfibrillar mitochondria. Dotted circles identify points of proximity. Lower panel in left shows subsarcolemmal mitochondria (purple) that bifurcates to reach two interfibrillar chains. Bottom right: bifurcating mitochondrial from a different angle. Through these connections, mitochondria form a complete reticulum, as documented by Glancy et al.^37,38

Na_v1.5 clusters adjacent to SSM in a non-random organization.

In separate experiments, we used dual-color STORM to localize Na_v1.5 in the membrane surface of adult ventricular myocytes. α-Actinin was used to mark sarcomeric edges (z-disks). As shown in Figure 2A, we observed large Na_v1.5 clusters between z-disks, in agreement with previous functional data.¹⁴ Negative controls are shown in Supplemental Fig.IV. Na_v1.5 distribution led us to speculate that a subpopulation of Na-channels may localize in close proximity to SSM, creating a macromolecular complex.

Figure 2. Na_v1.5 is found between z-disks close to mitochondria.

A. Stochastic optical reconstruction microscopy (STORM) image of adult ventricular myocyte. Blue: α-actinin as marker of z-disk. Green: Na_v1.5. White-boxed area enlarged in bottom panel. Scale bars 5 μm. Inset scale bar 1 μm. Apparent misalignment of α-actinin lines is due to cell surface curvature. Notice that large clusters of Na_v1.5 fall not on α-actinin (the expectation for costameric localization), but in-between, corresponding to the crest. B-C. STORM-acquired images of Mitotracker (red), Na_v1.5 (green) and α-Actinin (only in panel C, blue) in single myocytes dissociated from wild-type mice. White-boxed areas enlarged in respective bottom panels. Scale bars 5 μm. Inset scale bar 2 μm. Notice alignment of Na_v1.5 with SSM at cell surface, spaced by marker of z-disk.

Figure 2B presents an image of a single, isolated adult cardiac myocyte labeled for Na_v1.5 (green) and Mitotracker (red) to localize mitochondria. At the cell lateral membrane (distinguishable as the edge framing the long axis of the cell) a seemingly preferential proximity of mitochondria and Na_v1.5 was noted (white rectangles indicating area enlarged in the bottom of each panel). For some experiments, a third color was added (blue) to mark α-actinin (Fig. 2C). Data are consistent with the localization of Na_v1.5 and of mitochondria separately detected by FIB-SEM and by STORM (Figs. 1 and 2A, respectively). Moreover, the organization of both resembles the shape of the crest, with a Na_v1.5 cluster resting on top of the mitochondria, and between z-disks. Na_v1.5 cluster properties are indicated in Supplemental Fig.V.

We developed an analytical tool to address the question of whether proximity is random or deterministic. A line traced over the aligned SSM was used as reference (Fig. 3A). Each Na_v1.5 cluster was defined as an ellipse. Two parameters were defined: the distance between the centroid of the ellipse and the reference line, and the angle formed between the long axis of the ellipse, and the reference line (details on online supplement). Parameters were correlated for each cluster on an xy plot (Fig. 3B). Experimental data were compared to a simulated image where mitochondria were kept as in the experimental data but Na_v1.5 clusters were distributed randomly, though constrained by the edge of the cell (Fig. 3C-D). As shown in Figure 3B, most experimental Na_v1.5 clusters assembled in a small window of 10 to 300 nm in distance and a 0-60° angle to the reference line. However, in simulated images, green clusters were scattered along the plot (Fig. 3D). The method was further validated when comparing the relation between SSM and a line formed by immunostaining of α-actinin, which traces the z-disk perpendicular to the lateral surface of the membrane. In this case, almost all α-actinin clusters were gathered in a very small window, with a 500 to 2000 nm distance and a 70-90° angle to the mitochondria long axis, as expected (Fig. 3E-H). Overall, our results support the concept that the Na-channel complex and mitochondria form a structural macromolecular complex. Whether the association has a functional correlate was investigated next.

Figure 3. Na_v1.5 localizes on top of SSM in a deterministic matter.

A-B: STORM-acquired image of Mitotracker (red) and Na_v1.5 (green) from the lateral membrane of an adult murine myocyte (“Experimental image”; A) and data analysis (B). Reference line was traced over the first string of mitochondria under the cell membrane (SSM). This line was used as reference to measure distance to centroid of the Na_v1.5 cluster (abscisae in B) and its inclination, namely, the angle of the long axis of the ellipse defining the cluster (ordinates in B). A window of 300 nm and up to 60° in inclination captured 80% of the clusters. Distribution was compared to that obtained by modeling a random distribution of green clusters over the same region of interest (C; “randomized image”). In that condition, only 15% of clusters fell within the same window (D). To validate the model, results were compared to those obtained for a domain established by clusters of known architecture, namely, α-actinin (E: “Experimental image” of Mitotracker (red) and α-Actinin (blue). Analysis in F. In this case, to enhance visualization, the angle of inclination is reported as 90-angle (so perpendicular lines cluster around zero). A similar experiment in the “randomized image” is presented in G-H. As predicted, α-actinin is tightly organized in relation to mitochondria, and such organization is lost when clusters distribute randomly. Distance to reference line measured in nm. Similar results were obtained in n=28 cells from N=4 mice.

Mitochondrial NCLX is predominantly found in SSM near Na_v1.5 clusters.

NCLX plays an important role in maintaining mitochondrial function. Sodium concentration, in turn, is key to the ability of NCLX to extrude Ca²⁺ from the mitochondrial space. Given the found relation between Na_v1.5 and mitochondria, we examined whether NCLX had a preferential localization in the SSM space.

Figure 4 shows that the immunofluorescent signal for NCLX (red) was more abundant in the SSM than in mitochondria localized in the intracellular space; we also show significant co-localization of NCLX with Na_v1.5-positive clusters (green). Indeed, ~80% of Na_v1.5-NCLX clusters colocalized (i.e., were located within 20 nm of each other) at the lateral membrane vs ≈30% in the cell interior (Supplemental Fig. VI). To further validate NCLX antibody specificity for mitochondrial NCLX, we immunolocalized mitochondria with NCLX in HL-1 cells. Supplemental Fig.VII shows colocalization of the two structures and a round-punctuate pattern, characteristic of mitochondrial structures.

Figure 4. Mitochondrial NCLX is mainly found in SSM near Na_v1.5 clusters.

Dual-color stochastic optical reconstruction microscopy (STORM)-acquired images of NCLX (red) and Na_v1.5 (green) in single myocytes dissociated from wild-type mice. White-boxed area enlarged in right image. Notice abundance of NCLX signal at lateral membrane and proximity to Na_v1.5 clusters. Scale bar 5 μm. Inset scale bar 1 μm.

Given the proximity of NCLX to Na_v1.5, we postulated that the presence or absence of I_Na may impact on NCLX function and as such, on the extent of Ca²⁺ accumulation in the mitochondrial space.

Mitochondrial Ca²⁺ dynamics: differences between SSM and IFM.

To assess the relation between NCLX and Na-channel function we used a mitochondrial Ca²⁺ indicator (Rhod2-AM; Fig. 5A; see also Supplemental Fig.VIII.A, “Methods” and²⁷). Supplemental Fig.IX shows preservation of striated cell morphology after treatment. Rhod2-AM intensity was measured before (F₀) and after drug treatment (F) and measurements of the ratio F/F₀ in SSM were compared to those obtained in IFM. Measurements in control conditions (no drugs added) are shown as the first two bars of the graph in Figure 5B, corresponding to data acquired from SSM (red) and IFM (gray), respectively. Since cytoplasmic Na⁺ concentration is a major driver of NCLX forward movement, we first tested whether blocking I_Na with TTX would affect mitochondrial Ca²⁺. Rhod2-AM fluorescence was first recorded in a solution containing 148 mmol/L Na⁺; then, TTX 10 μmol/L was added. The bar graph (bars with vertical lines) in Figure 5B shows that, upon TTX exposure, SSM accumulated significantly more Ca²⁺ (detected as increase in Rhod2 fluorescence) than IFM (F/F_o 1.1±0.03 and 0.99±0.03 for SSM and IFM, respectively; p<0.05). This was consistent with the notion that reduced entry of sodium via Na_v1.5 limits NCLX activity, and that this effect is more noticeable in the area where the Na_v1.5/NCLX complex is abundant (see Supplemental Fig.VIII.B). Of note, when low concentrations of TTX (100 nmol/L) were used, we observed a non-statistically significant tendency for increased Rhod2 intensity in SSM (Supplemental Fig.X).

Figure 5. Mitochondrial Ca²⁺ in relation to sodium channel activity.

A. Confocal 2D images of cardiomyocytes loaded with Rhod 2-AM at 4°C and then incubated at 37°C for 3-5 hours (see Methods). Two regions were defined, SSM (first string of mitochondria immediately below cell surface) and IFM (in the interior of the cell). B. Quantitative analysis of F/F₀ (F corresponds to Rhod2-AM intensity at rest, and F₀ to Rhod2-AM intensity after treatment) ratio under conditions described in the table below. Notice increased fluorescence emission units (mitochondria) after tetrodotoxin (TTX), TTX+FCCP+Oligomycin and CGP-37157 treatment, particularly in SSM. The ratio F/F₀ of Rhod-2 fluorescence intensities was compared between SSM and IFM: higher ratio reflects more mitochondrial Ca²⁺ accumulation. Scale bar 10 μm. +TTX: 15 cells from 8 mice; +TTX+FCCP+Oligomycin: 7 cells from 4 mice; +TTX+FCCP+Oligomycin+CGP-37157: 7 cells from 3 mice; CGP-37157: 6 cells from 3 mice; Negative control: 17 cells from 5 mice. Student’s t-test: between SSM and IFM for each treatment: *p<0.05 vs SSM, **p<0.01 vs SSM, ***p<0.001 vs SSM; between treatment vs negative control: #p<0.05 vs negative control, ###p<0.001 vs negative control.

We next compared Ca²⁺ dynamics in SSM vs IFM in the presence of TTX and after mitochondrial depolarization with 1 μmol/L FCCP + 2 μmol/L oligomycin. FCCP alone can induce mitochondrial ATP depletion via the reverse of F₀/F₁-ATPase, which is prevented by adding the F₀/F₁-ATPase inhibitor oligomycin. FCCP induces a mitochondrial depolarization of around 100 mV, enough for NCLX to function in reverse mode (Ca²⁺_in/ Na⁺_out of the mitochondria).³² Consistent with this notion, FCCP+oligomycin augmented Rhod2 fluorescence exclusively at the SSM (Fig. 5B, bars with horizontal lines). Most importantly, adding the NCLX inhibitor CGP-371257 attenuated the mitochondrial Ca²⁺ accumulation in SSM and no significant difference in Rhod2 fluorescence was observed between SSM and IFM (Fig. 5B, bar with dots).

We analyzed Rhod2 fluorescence in cells treated only with 2 μmol/L CGP-37157, a selective NCLX inhibitor. As expected, after adding CGP-37157, Rhod2 fluorescence increased in SSM (ratio 1.10±0.024 vs IFM ratio 0.98±0.01, p<0.001, Fig. 5B bars with crosses) indicating that Ca²⁺ influx through the mitochondrial calcium uniporter (MCU) cannot be compensated by extrusion since NCLX is inhibited. Rhod2 fluorescence was practically unchanged after CGP-37157 addition in IFM, suggesting that IFM have another way of extruding Ca²⁺ that is NCLX- and Na⁺-independent.

Finally, we measured mitochondrial Ca²⁺ dynamics at different pacing rates (3 and 5Hz) alone or in the presence of either TTX or CGP-37157. Rapid pacing led to mitochondrial Ca²⁺ accumulation in SSM and IFM (vs rest #p<0.05). When TTX was added (10 μmol/L), there was a tendency for SSM to accumulate more Ca²⁺ in response to pacing (3Hz vs 3Hz+ TTX p=0.06; 5Hz vs 5Hz+ TTX p=0.07) and when comparing the two mitochondrial subpopulations, mitochondrial Ca²⁺ increased significantly more in SSM than in IFM, *p<0.05). Similar results were obtained with CGP-37157. Details in Supplemental Fig.XI. Altogether, the data suggest functional coupling between Na_v1.5 and NCLX in the sarcolemmal-subsarcolemmal space in the crest region of adult cardiac myocytes, with functional implications to the adaptation necessary to meet demand at increased rates.

Increased superoxide production after TTX addition.

There is a reciprocal interaction between Ca²⁺ and ROS which, in physiological conditions, plays an important role in regulating cellular signaling networks.^33,34 Pathological changes in one can affect the other, leading to more damage. Since TTX increases mitochondrial Ca²⁺, we wondered if this could lead to increased ROS production.

We used the fluorescent dye MitoSOX Red, which detects mitochondrial superoxide, to measure ROS in live myocytes before and after TTX. When in mitochondria, MitoSOX exhibits red fluorescence. As seen in Figure 6, at resting conditions (F₀) ventricular myocytes showed low MitoSOX fluorescence, higher in SSM, as previously reported.³⁵ Co-staining with MitoTracker confirmed MitoSox mitochondrial localization (Supplemental Fig.XII).

Figure 6. Increased superoxide production after TTX.

A. Confocal 2D images of cardiomyocytes loaded with MitoSOX (see Methods). Two images from same cell are shown: at rest (F₀; left), and after 10 μmol/L addition of TTX (F; right). Notice brighter fluorescence at SSM at resting conditions compared to IFM. Scale bar 10 μm. B. Ratio F_TTX/F₀ of MitoSOX fluorescence intensities was compared between SSM and IFM: higher ratio reflects more ROS produced. Average was then plotted. N=3 mice, n=12 cells. No statistical differences (Student’s t-test).

After exposure to TTX 10 μmol/L (pacing at 0.5Hz), MitoSOX fluorescence increased in both SSM and IFM, the latter being less pronounced (Paired Student’s t-test: p<0.001 +TTX vs rest in SSM; p<0.05 +TTX vs rest in IFM). This result indicates that blocking Na⁺ entry with TTX leads to increased mitochondrial ROS production, though in this case there was no predominant effect in the SSM, suggesting that the role of I_Na on mitochondrial function may go beyond its local interaction with NCLX.

Correlation analysis of human left ventricle transcriptome in relation to SCN5A expression.

We used the Genotype-Tissue Expression repository (GTEx; v8) of 386 samples of left ventricle tissue from deceased human donors to compare the abundance of SCN5A transcript against SLC8B1 and SLC8A1, which encode the mitochondrial NCLX and the plasma membrane NCX, respectively. Resulting plots are shown in Figure 7A-B. Data show that SLC8A1 is positively correlated with SCN5A (regression coefficient β= 0.62, log-P-value= 4E-58) whereas SLC8B1 is negatively correlated with SCN5A (regression coefficient β= −0.17, log-P-value= 1.66 E-30). This observation is only correlational. We therefore tested whether reduced abundance of one transcript would affect the abundance of the other. To this aim, we used an hiPSC-CM line deficient in SCN5A (see “Methods” and Supplemental Figs.XIII and XIV). As shown in Figure 7C the abundance of SLC8B1 was significantly increased in cells where SCN5A was knocked down, supporting the notion of a transcriptional relation between the two gene products.

Figure 7. SLC8B1 (encoding NCLX), negatively correlates with SCN5A.

A-B. Association analysis of human left ventricle transcriptome in relation to SCN5A expression: correlation between expression of SLC8B1 (panel A) or SLC8A1 (panel B) vs SCN5A expression. Each point represents the normalized level (VST, variance stabilizing transformation) of SCN5A vs that of SLC8B1 (A) or SLC8A1 (B) for one individual in the dataset. Notice the positive slope (=positive correlation) between SCN5A and SLC8A1 (plasma membrane NCX) and negative correlation with SLC8B1 (mitochondrial NCLX), suggesting that SCN5A transcript levels correlate with expression of both in human heart. C. qPCR for SCN5A, SLC8A1 and SLC8B1 in hiPSC-CMs wild-type (WT) and SCN5A-KO. Notice drastic reduction in SCN5A and increase in SLC8B1 in hiPSC-CM SCN5A-KO. N=7. Statistical test: Mann Whitney U-test between SCN5A-KO and WT for each gene. ***p<0.001 vs WT.

DISCUSSION

Our data show that SSM and a subpopulation of Na_v1.5 at the lateral membrane are in close apposition and that in ventricular cardiomyocytes, not only mitochondria can regulate I_Na,^20,21 but also I_Na can regulate NCLX activity. NCLX preferentially localized at SSM near Na_v1.5 clusters. We speculate that such distribution creates a high-Na⁺ microdomain that is energy-efficient, comparable to the high-Ca²⁺ microdomain formed between MCU and the sarcoplasmic reticulum (SR). Many diseases have been linked to changes in intracellular Na⁺ concentration³⁶ which, in turn, could affect NCLX function, therefore affecting ATP production and ROS generation. Overall, we show the presence of a Na-channel-mitochondria complex and suggest that Na_v1.5-NCLX proximity may have functional consequences to metabolic homeostasis via crosstalk with the mitochondrial reticulum.

Cardiomyocytes are rich in mitochondria, which occupy a third of total cell volume.¹⁸ Although mitochondria form a network reaching throughout the cell^37,38, they are divided into at least three subpopulations with distinct morphology, composition, biochemical properties and function, contributing to their capacity for adaptation.^18,19,35 Here, we focus on SSM. As others,³⁹ we found SSM nested in-between z-disks. A nanometric-resolution cell-surface scan shows a pattern of crests and valleys (or “grooves”) with ~2 μm periodicity, that is, the length of a sarcomere.³¹ In this pattern, grooves match the position of the z-disk (prompting the name “z-grooves”), with crests being dome-shaped surfaces spanning the distance between two z-grooves.³¹ Previously, using scanning patch clamp, we measured larger I_Na in the crest compared to grooves.¹⁴ This matches our data showing Na_v1.5 clusters preferentially aligned with SSM at the crest. We also show that the cell membrane in the crest region reaches close proximity with the outer membrane of the mitochondria, in a manner reminiscent of the T-tubule-junctional SR dyad (see, e.g.^40,41) or between cell membranes in the perinexus.⁴² In these structures, the narrow intermembrane space allows functional coupling between channels residing in opposing membranes. The molecular mechanisms of formation of this Na_v1.5-mitochondrial complex remain undefined. Yet, we show that the relation between mitochondria and Na_v1.5 is deterministic and of a particular topology.

We identify NCLX, a molecule key for Na⁺ and Ca²⁺ homeostasis, as resident of the Na_v1.5-mitochondria hub. NCLX (encoded by SLC8B1) is proposed to be the main pathway for mitochondrial Ca²⁺ extrusion in excitable cells, while MCU is responsible for Ca²⁺ transport into mitochondria.^43,44 NCLX transport is voltage-dependent and electrogenic, with stoichiometry of 3-4 Na⁺ for 1 Ca²⁺.⁴⁵ Unlike plasma membrane NCX, unique residues confer NCLX Li⁺ selectivity.⁴⁶ The Ca²⁺ extrusion rate of NCLX is ≈100 times slower than MCU-Ca²⁺ influx rate. Therefore, NCLX activity is the rate-limiting step of mitochondrial Ca²⁺ cycling.^44,47,48

NCLX is fundamental to life. A conditional knockout of NCLX quickly leads to heart failure associated with mitochondrial Ca²⁺ overload and consequent increase in ROS generation.⁴³ In contrast, NCLX overexpression in mouse heart increases mitochondrial Ca²⁺ extrusion preventing pathogenic Ca²⁺ accumulation and ROS generation.⁴³ Whether loss or over-abundance of NCLX affects I_Na or subcellular localization of Na_v1.5 remains unclear.

NCLX is very sensitive to changes in cytoplasmic [Na⁺],^32,48,49 given that the Km of association is similar to resting [Na⁺]_i (K_m ~7-10 mmol/L). Moreover, studies from Skogestad et al show that [Na⁺]_i in an active, excitable myocyte, is not homogeneous. Rather, it is highest in the subsarcolemma and distributes in pools, likely, corresponding to the location of Na⁺-channel clusters.⁵⁰ We propose that NCLX is close to Na_v1.5 channels to rapidly sense Na⁺ signals and properly extrude Ca²⁺ out of mitochondria, in a way similar to how MCU’s low affinity for Ca²⁺ is compensated by proximity to the SR (diagram in Supplemental Fig.VIII).^51,52 Several studies describe an interaction of NCLX with plasma membrane transporters: a) In pancreatic β-cells, glucose-dependent depolarization leads to cytosolic Na⁺ influx via voltage-gated Na⁺ channels, which promotes NCLX activity, maintaining ATP production and controlling insulin secretion.^53,54 b) The nociceptive noxious heat-activated receptor (TRPV1) triggers a Na⁺-dependent depolarization that controls NCLX Ca²⁺ extrusion, which at the same time regulates TRPV1 conductance.^55,56 c) In macrophages and melanoma cells, activation of a splice variant of Na_v1.6 causes release of Na⁺ from vesicular compartments and Na⁺-uptake-Ca²⁺release by NCLX.⁵⁷

Through human transcriptome analysis, we found SLC8B1 negatively correlating with SCN5A. Though the common denominator of this relation remains unknown, its presence may have important functional consequences. Indeed, reduced expression of SCN5A may be compensated by increased expression of SCL8B1 to increase the ability to capture Na⁺ entering the cell. The reverse would also have functional effects: an excess Na⁺ entry could be compensated by reduced expression of the exchanger, maintaining mitochondrial homeostasis. This is consistent with the notion that NCLX activity is strongly regulated by even slight changes in cytosolic Na⁺.^48,58-63 Their physical proximity (shown here) and transcriptional inverse relation would facilitate maintenance of this delicate equilibrium.

In resting Na⁺ and Ca²⁺ conditions and with an inner mitochondrial membrane potential of ~−180 mV, NCLX operates in forward mode, exporting Ca²⁺ out of the mitochondrial matrix in exchange for cytosolic Na⁺. However, in depolarized mitochondria, NCLX functions in reverse mode, transporting Ca²⁺ into the mitochondria and extruding Na⁺.⁶⁴ Studies have shown that increasing cytosolic Na⁺ accelerates mitochondrial Ca²⁺ decay due to increased NCLX forward mode activity,^63,64 but these studies did not compare SSM to IFM. Our mitochondrial Ca²⁺ dynamic experiments show that NCLX is predominantly located at SSM, confirming our SMLM experiments, and TTX leads to mitochondrial Ca²⁺ accumulation more prominently in SSM, connecting Na_v1.5 and NCLX function. The Ca²⁺ accumulation after TTX+FCCP+Oligomycin exposure in SSM seen in our experiments could be explained by 1) active uptake of calcium by the reverse NCLX, or 2) increased mitochondrial uptake via MCU. However, MCU open probability decreases with mitochondrial depolarization (i.e. with FCCP).⁶⁴ In fact, IFM show a slight decrease in Rhod2 fluorescence after TTX+FCCP+Oligomycin which could be accounted for an inhibited MCU and lack of reverse mode NCLX-Ca²⁺ uptake in IFM. We used the benzothiazepine compound CGP-37157 to selectively block NCLX (IC₅₀= 0.36 μM).^59,62 At least a 10-fold higher concentration of CGP-37157 is needed to affect plasma membrane NCX, SERCA and RyR2.⁵⁹ It is known that cytosolic Na⁺ is required for powering NCLX to activate mitochondrial Ca²⁺ efflux. Therefore, our mitochondrial Ca²⁺ dynamics data suggest that a healthy I_Na is needed to maintain mitochondrial Ca²⁺ homeostasis. However, and compared to Ca²⁺ imaging, studying cellular Na⁺ dynamics is challenging, particularly due to the relatively small changes of intracellular Na⁺ concentration.^53,58

There is mutual interaction between ROS and Ca²⁺, where ROS can regulate Ca²⁺ signaling and Ca²⁺ signaling is essential for ROS production.^33,43 Physiologically, ROS are generated as by-product of mitochondrial respiratory chain activity or by specialized ROS-generating enzymes to regulate signaling functions such as cell proliferation, migration, and vascular tone; but toxic levels associate with pathology. Mitochondrial Ca²⁺ accumulation leads to ROS production. Our results show increased ROS production after TTX exposure in both SSM and (less pronounced) IFM, even though only SSM showed increased mitochondrial Ca²⁺ after TTX. It is worth noting that, even though they are part of distinct mitochondrial subpopulations, SSM and IFM are connected, forming a mitochondrial reticulum in which ROS can propagate.^37,38,65 Most importantly, ROS production leads to decreased I_Na.⁶⁶ ROS can alter the oxidative state of PKA, PKC, CaMKII and NAD(H), which in turn regulates I_Na.^33,67 Genetically engineered Scn5a heterozygous mice show increased ROS accumulation.⁶⁸ This, together with our results, suggests that there could be an amplification loop: increased ROS decreases I_Na, leading to less Na⁺ entering the cytoplasm, which in turn affects NLCX activity which leads to mitochondrial Ca²⁺ accumulation, leading to more ROS production.

One question that arises is how do IFM extrude calcium if NCLX is mainly found in SSM? Even though NCLX is the major transporter for Ca²⁺ extrusion, mitochondria contain both Na⁺-dependent and Na⁺-independent mechanisms for Ca²⁺ extrusion.^45,64 This may include leucine zipper EF-hand-containing transmembrane protein 1 (LETM1;^45,64,69). Somehow controversial, it has been proposed that transient mitochondrial permeability transition pore openings could also help discharge mitochondrial Ca²⁺ load.^45,47,70 Other studies show Cx43 hemichannels in mitochondria⁷¹ and particularly in SSM⁷² which could act as mitochondrial Ca²⁺ gateway. Overall, our data are consistent with previous descriptions of proteins that distribute and/or work differently in various sub-populations of mitochondria (see, e.g.^73,74).

Our data suggest the need of healthy I_Na for (proper) mitochondrial Ca²⁺ balance. Many diseases associate with cytoplasmic unbalanced Na⁺ concentration. A pathogenic increase in cytoplasmic Na⁺ accelerates mitochondrial Ca²⁺ efflux trough NCLX, which inhibits mitochondrial metabolic rate and ATP production, as seen in heart failure, ischemia, and seizure-like events.^{48,59,63,75,76} In melanoma cells, an increase in Na⁺ mediated by vesicular Na_v1.6 activates NCLX, which accelerates invasiveness of these tumor cells.⁵⁷ Conversely, hypoxia is known to reverse NCLX mode, which leads to mitochondrial Ca²⁺ overload and ROS production, in turn accelerating cytoplasmic Na⁺ accumulation.⁷⁷ Future studies need to be done to analyze NCLX activity in pathologies where I_Na is reduced (e.g. Brugada syndrome).

NOVELTY AND SIGNIFICANCE.

What Is Known?

• The voltage-gated sodium channels, formed by the protein Na_v1.5, carry the excitatory inward current (the sodium current) that is responsible for the action potential upstroke in adult cardiac ventricular myocytes.

• Na_v1.5 channels localize to different subdomains within the myocyte, associating with other molecules in a domain-specific manner.

• Mitochondrial function can alter the magnitude and properties of the sodium current.

What New Information Does This Article Contribute?

• Na_v1.5 clusters that localize to the membrane region between two z-lines (the “crest” of the sarcomere) are in close apposition to subsarcolemmal mitochondria and to the mitochondrial Na⁺/Ca²⁺ exchanger.

• This anatomical proximity leads to a functional relation between Na_v1.5 and the mitochondrial Na⁺/Ca²⁺ exchanger.

• Na_v1.5 channel malfunction leads to unbalanced mitochondrial calcium dynamics, potentially impacting cell metabolism.

Na-channels form macromolecular complexes that, in adult ventricular myocytes, cluster at specific subdomains. Dysfunction of the channel itself, or of its partners, can lead to arrhythmias that cause sudden cardiac death. The relation between Na-channels and mitochondrial function has been described in one direction only: mitochondrial byproducts (such as ROS) affect sodium current. Here we show that this relation is reciprocal, namely, Na-channel activity influences mitochondrial function. Using imaging techniques, we describe the deterministic localization of Na-channels and subsarcolemmal mitochondria, where we primarily locate the mitochondrial Na⁺/Ca²⁺ exchanger. Functionally, the blockade of Na-channels leads to a local mitochondrial Ca²⁺ accumulation and ROS production. We propose that NCLX is powered by the entry of Na⁺ through Na-channels to properly extrude Ca²⁺. This proposed model for electro-metabolic coupling redefines Na_v1.5 as a key regulator of cell metabolism and opens a new window to understand electrical and structural disorders of the heart.

Nonstandard Abbreviations and Acronyms:

FIB-SEM

Focused Ion Beam-Scanning Electron Microscopy

GTEx

Genotype-Tissue Expression

IFM

InterFibrillar Mitochondria

SSM

SubSarcolemmal Mitochondria

NCLX

Mitochondrial Na⁺/Ca²⁺/Li⁺ Exchanger

ROS

Reactive Oxygen Species

SMLM

Single-Molecule Localization Microscopy

STORM

Stochastic Optical Reconstruction Microscopy

TTX

Tetrodotoxin