Caveolae-Mediated Activation of Mechanosensitive Chloride Channels in Pulmonary Veins Triggers Atrial Arrhythmogenesis.

Background Atrial fibrillation often occurs in the setting of hypertension and associated atrial dilation with pathologically increased cardiomyocyte stretch. In the setting of atrial dilation, mechanoelectric feedback has been linked to the development of ectopic beats that trigger paroxysmal atrial fibrillation mainly originating from pulmonary veins (PVs). However, the precise mechanisms remain poorly understood. Methods and Results We identify mechanosensitive, swelling-activated chloride ion channels (ICl,swell) as a crucial component of the caveolar mechanosensitive complex in rat and human cardiomyocytes. In vitro optical mapping of rat PV, single rat PV, and human cardiomyocyte patch clamp studies showed that stretch-induced activation of ICl,swell leads to membrane depolarization and decreased action potential amplitude, which trigger conduction discontinuities and both ectopic and reentrant activities within the PV. Reverse transcription quantitative polymerase chain reaction, immunofluorescence, and coimmunoprecipitation studies showed that ICl,swell likely consists of at least 2 components produced by mechanosensitive ClC-3 (chloride channel-3) and SWELL1 (also known as LRRC8A [leucine rich repeat containing protein 8A]) chloride channels, which form a macromolecular complex with caveolar scaffolding protein Cav3 (caveolin 3). Downregulation of Cav3 protein expression and disruption of caveolae structures during chronic hypertension in spontaneously hypertensive rats facilitates activation of ICl,swell and increases PV sensitivity to stretch 10- to 50-fold, promoting the development of atrial fibrillation. Conclusions Our findings identify caveolae-mediated activation of mechanosensitive ICl,swell as a critical cause of PV ectopic beats that can initiate atrial arrhythmias including atrial fibrillation. This mechanism is exacerbated in the setting of chronically elevated blood pressures.

Clinical Perspective

What Is New?

We link specialized cardiac cell membrane structures, caveolae, to the activation of mechanosensitive chloride ion channels and the development of arrhythmogenic ectopic beats from pulmonary veins, the most common source of paroxysmal atrial fibrillation.

Downregulation of caveolae structures during chronic hemodynamic overload of the atria facilitates activation of chloride ion channels and increases sensitivity of pulmonary veins to stretch.

What Are the Clinical Implications?

Our study provides conceptual innovation on cardiac mechanosensing, which critically contributes to atrial arrhythmogenesis associated with elevated blood pressure, atrial dilation, and pathologically increased cardiomyocyte stretch.

It forms a mechanistic basis for development of novel and effective therapeutic approaches targeted to treat stretch‐induced atrial arrhythmogenesis by preventing the degradation or promoting the restoration of cardiac cytoarchitecture.

Introduction

Atrial fibrillation (AF) is the most common cardiac rhythm disorder and is often associated with hypertension, heart failure, and valvular heart disease.1, 2 These conditions cause hemodynamic pressure and volume overload of the atria, resulting in stretching of the atrial cardiomyocytes. Though AF maintenance is most likely a reentrant phenomenon conditioned by elevated electrophysiological and/or structural heterogeneity, the precise pathophysiological basis for triggers that initiate AF has not been resolved. Regions within the atria around the pulmonary vein (PV) ostia are subject to the most significant dilation and are known to be the most common regions to develop arrythmogenic ectopic foci.3 The PVs are often responsible for AF initiation in patients and thus are commonly targeted during catheter or surgical AF ablation procedures, which cure 75% of paroxysmal AF cases.4, 5 Animal and human studies demonstrate that acute stretch facilitates PV arrhythmogenesis by increasing dispersion of refractoriness, leading to heterogeneous intra‐PV conduction slowing and triggered automaticity inside the PV.6, 7 Based on the relationship between stretch and arrhythmogenesis, mechanoelectric feedback has been proposed as the principal mediator of the disease,8, 9 although the precise mechanisms remain poorly understood.

Specialized surface membrane structures, referred to as caveolae, represent small (50‐ to 100‐nm) invaginations of the plasma membrane and are enriched by cholesterol, sphingolipids, and scaffolding proteins called caveolins.10, 11 Caveolae represent a reserve source of “extra” cell membrane and are implicated in mechanotransduction by buffering mechanical forces and contributing to cell volume regulation.10 It is estimated that, in the rat ventricular myocyte, 22% of the cell membrane resides in caveolae12 and that stretch of the cell incorporates this reserved membrane into the surface membrane via flattening of caveolae.10, 11

In addition to their ability to influence cell volume, it has been shown in various cell types that caveolae are involved in mechanoelectrical transduction,13, 14 possibly by promoting the activation of mechanosensitive ion channels, including swelling‐activated chloride ion (Cl−) channels (I Cl,swell).10 The presence of I Cl,swell has been shown in myocytes isolated from rabbit PVs.9 In guinea pig ventricular myocytes, activation of I Cl,swell depolarizes the membrane resting potential (RP),15 which may slow electrical impulse conduction. In Langendorff‐perfused isolated mouse hearts, pressure loading significantly decreases the ventricular conduction velocity, along with a concomitant decrease in caveolae density, both of which are restored on pressure unloading.11 Pathological conditions associated with elevated chronic stretch, including AF, hypertension, and hypertrophy, are linked to a decrease in cardiomyocyte caveolae density and downregulation of the muscle‐specific caveolae scaffolding protein Cav3 (caveolin 3).16, 17, 18 These findings suggest a crucial role of caveolae and caveolar ion channels in stretch‐induced electrophysiological changes facilitating PV arrhythmogenesis.

In this study, we show that volume‐activated Cl− channels I Cl,swell are localized in the caveolae microdomains and can be activated on mechanical stretch. I Cl,swell likely consists of at least 2 components produced by mechanosensitive ClC‐3 (chloride channel‐3) and SWELL1 (also known as LRRC8A [leucine rich repeat containing protein 8A]) channels that form a mechanosensitive macromolecular complex with caveolar scaffolding protein Cav3. Activation of I Cl,swell results in an inward current in the distal part of the PV (PVdis). This, in turns, leads to tension‐dependent depolarization of the membrane RP suppressing action potential (AP) amplitude (APA) and resulting in intra‐PV conduction discontinuities as well as ectopic and reentrant activities within the PV. Downregulation of Cav3 expression and disruption of caveolae structures during hypertension facilitates activation of I Cl,swell and increases sensitivity to stretch 5‐ to 10‐fold, promoting the development of AF. Our findings identify caveolae‐mediated activation of mechanosensitive I Cl,swell as a critical cause of the triggering impulses that can initiate atrial arrhythmias including AF, and this mechanism is exacerbated in the setting of chronically elevated blood pressures (BPs). These findings identify new therapeutic targets for atrial arrhythmias.

Methods

A detailed materials and methods section is found in Data S1. The data that support the findings of this study are available from the corresponding author on reasonable request.

Animals and Preparations

All methods and protocols used in these studies were approved by the animal care and use committee of the Cardiology Research Center (Moscow, Russia) and the University of Wisconsin–Madison following the Guidelines for Care and Use of Laboratory Animals published by the National Institutes of Health (NIH; publication no. 85‐23, revised 1996). All animals used in this study received humane care in compliance with the Guide for the Care and Use of Laboratory Animals. Human heart collection protocols were approved by the University of Wisconsin institutional review board.

Adult (8–12‐month‐old) normotensive Wistar rats (n=62) and spontaneously hypertensive rats (SHRs; n=17) of both sexes were used. The PV preparations were isolated as previously described.19 Briefly, the left atrium (LA), together with the LA appendage and PV region, was dissected from the ventricles, right atrium, and interatrial septum. The central PV was cleaned, cut open, and placed in a tissue bath with the endocardial side facing upward (Figure 1A). The pacing electrode was placed on the edge of the LA appendage. A small portion of the end‐distal part of the PV was not cut open and was used to weave a silk suture (4‐0) with a weight applied (Figure 1A).

Figure 1

Effect of pathological stretch on the pulmonary vein (PV) myocardium. A, Photograph of the intact rat PVs (left). Central PV was isolated, cleaned, cut open, and placed in a tissue bath with the endocardial side facing upward (right). Anatomical regions selected for microelectrode action potential (AP) recordings are shown: PV ostium (PV ost) and PV distal (PV dis). B, Representative examples of APs simultaneously recorded in the PV dis (top recordings) and PV ost (bottom recordings) under different pathological stretch. C and D, Membrane resting potential (RP) and AP amplitude (APA) changes for individual rats (light blue lines indicate PV ost, and red lines indicate PV dis) and averaged (solid lines). E, Probability (in percentage from total preparations tested) of intra‐PV conduction block at different tensions. Data include 2 series of experiments: n=8 for tensions <10.5 g, and n=15 for tensions >10.5 g. *P<0.05, **<0.01, ***<0.001 within the same group vs baseline, and # P<0.05, ##<0.01, ###<0.001 for PV dis vs PV ost by repeated‐measurements 2‐way ANOVA with Bonferroni correction. F, Fluorescent optical mapping of electrical activity during the development of intra‐PV conduction block under pathological stretch. At left, PV activation maps are shown at baseline (no stretch applied; top) and during the development of intra‐PV conduction block (stretched; bottom). In the middle, superimposed upstrokes of optical action potentials (Vm) along the PV (PV ost in blue, PV middle in green, and PV dis in red) are shown for each condition. At right, distribution of conduction velocity along the PV and relative PV distance (in percentage from the PV length from a brightfield image) activated at baseline and during stretch. P value was determined by paired Student t test. G and H, Microreentry within the PV occurred under pathological stretch and application of norepinephrine (1 μmol/L). LA indicates left atrium; LAA, left atrial appendage.

Microelectrode Recordings

Transmembrane APs were simultaneously recorded from the endocardial surface of the proximal (PV ostium [PVost]) and PVdis preparations by using 2 glass microelectrodes filled with 3.0 mol/L KCl (tip resistance ≈15–40 MΩ).19

Optical Mapping

PV preparations were stained with voltage‐sensitive dye RH237 (10 μmol/L; Thermo Fisher Scientific) and optically mapped with a MiCAM Ultima‐L CMOS camera (SciMedia USA) from the endocardial field of view ranging from 10×10 to 16×16 mm2, sampled at 1000 to 3000 frames/s.20

Patch Clamp Studies

Patch clamp measurements of I Cl,swell current and spontaneous electrical activity were performed on cardiomyocytes isolated from PVdis and PVost, and on human induced pluripotent stem cell–derived cardiomyocytes (hiPS‐CMs) generated from the well‐characterized DF19‐9‐11T line.21

Immunohistochemistry

Masson's trichrome staining and double‐immunolabeling for ClC‐2 (goat polyclonal; SAB2501373, Sigma‐Aldrich), ClC‐3 (rabbit polyclonal; ACL‐001, Alomone Labs), SWELL1 (anti‐LRRC8A, rabbit polyclonal; AAC‐001, Alomone Labs), and Cav3 (mouse monoclonal; 610421, BD Biosciences; rabbit polyclonal; ab2912, Abcam) were performed on paraffin‐embedded sections of unstretched PV preparations. Images were collected using a Leica SP5 confocal microscope system under ×63 oil‐immersion objective and analyzed using the NIH ImageJ and Matlab software.

Immunoprecipitation

Rat PV and healthy human LA tissue lysates were used, and immunoprecipitations were carried out using anti‐Cav3 or anti–ClC‐2, anti–ClC‐3, or anti‐SWELL1 antibodies; control IgGs (respective to sample species) were used at the same concentrations as the specific antibodies. Immune complexes were analyzed by Western blot by probing with antibodies specific for ClC‐2, ClC‐3, SWELL1, Cav3, and GAPDH.

Quantitative Reverse Transcription Polymerase Chain Reaction Analysis

Total RNA was extracted from rat PVdis, PVost, and LA using TriZol reagent (Invitrogen). Quantitative reverse transcription polymerase chain reaction for ClC‐2, ClC‐3, LRRC8A, ANO1 (anoctamin 1), Cav3, and GAPDH (probes are listed in Table S1) were performed using TaqMan Fast Advanced Master Mix (Thermo Fisher Scientific). Levels of mRNA were quantified using the ΔΔCt method and normalized to housekeeping gene GAPDH.

Statistical Analysis

Student t test was used in 2‐group comparisons (paired for comparisons between the same participant and unpaired for 2 groups of different participants). Multiple groups of normally distributed data of similar variance were compared by 1‐ or 2‐way ANOVA (for nonrepeated measurements) or repeated‐measurements 2‐way ANOVA. For multiple comparisons, the Bonferroni corrected P value is shown. All statistical analyses were performed using GraphPad Prism 5 (GraphPad Software) or Origin v6.1 (OriginLab). P<0.05 was considered statistically significant. Values were presented as mean±SEM.

Results

Electrophysiological Changes in the PV Under Stretch

We modeled the effect of physiological and pathological stretch on Wistar rat PV preparations by applying weights across 2 ranges: 0 to 1.5 g of weight corresponding to a physiological pulmonary venous pressure range from 0 to ≈4 mm Hg (calculated as applied weight times gravity constant divided by cross‐section area of the PV preparation) and 2.5 to 10.5 g of weight corresponding to a pathological pressure range from ≈6 to 26 mm Hg. We used simultaneous microelectrode recordings of transmembrane AP from PVdis (near the lung) and PVost (near the LA) to measure PV response to stretch (Figure 1A). Physiological stretch induced a heterogeneous prolongation of both AP duration and functional refractory period along the PV, with no effect on RP and APA (Figures S1 through S3). It resulted in a significant increase in the dispersion of refractoriness along the PV (from 3±2 ms at baseline up to 26±5 ms at 1.5 g; P<0.01) and the development of spontaneous activity as well as the induction of early after‐depolarizations in the PVdis (Figure S1).

In contrast to physiological stretch, pathological stretch induced dramatic changes in RP and APA exclusively in the PVdis and was accompanied by the subsequent development of intra‐PV conduction block (Figure 1B–1D), correlating with the degree of tension applied (Figure 1E). Consistent with our microelectrode measurements, high‐resolution fluorescent optical mapping of electrical activity revealed intra‐PV conduction block in all Wistar rats (n=4) under pathological stretch (Figure 1F). Conduction velocity slowing and conduction block were found distally in the PV (at ≈60–80% of the PV length), which is consistent with previous studies in both canine22 and human PVs.7 As seen from optical mapping images, pathological stretch that was required to induce intra‐PV conduction block and led to ≈30% to 40% lengthening of the PV tissue. Importantly, tension was applied homogeneously along the PV, as estimated from the cardiomyocyte sarcomere length measured from immunofluorescent images for α‐actinin staining (for maximal tension, sarcomere length increased from 1.19±0.03 μm to 1.74±0.05 in PVost and from 1.20±0.04 to 1.65±0.06 μm in PVdis, not significant between regions; Figure S4).

Reentrant Arrhythmias Induced by Stretch in the PVs

Application of 1 μmol/L norepinephrine always triggered spontaneous automaticity from the PVs. However, at depolarized RP during pathological stretch, norepinephrine led to more regular and faster PV automaticity compared with baseline (the fastest spontaneous beating rate was 118±28 beats/min at 6.2±1.5 g versus 50±13 beats/min at baseline; P=0.049; Figure S5). In some preparations (2/8 at 0.1 g, 1/8 at 2.5 g, 1/8 at 4.5 g, 2/8 at 6.5 g, 2/8 at 8.5 g, and 1/8 at 10.5 g), norepinephrine‐induced paroxysms of spontaneous activity were associated with echo beats (Figure 1G and 1H; Figure S6). In the example shown in Figure 1G, spontaneous AP generated in the PVdis during norepinephrine perfusion with 4.5‐g tension (top recording) propagated to the PVost (bottom recording) and induced an AP that subsequently reexcited the PVdis and led to the echo extrabeat (note the shorter coupling interval between the echo extrabeat and the previous spontaneous AP as well as the different extrabeat AP morphology). Another example shows one and a half reentry circle captured in the PV preparation under 6.5‐g tension and norepinephrine perfusion (Figure 1H). The echo extrabeat from the PVdis reexcited the PVost and induced a nonpropagating response.

Impact of Volume‐Activated Cl− Channels on Stretch‐Induced PV Arrhythmogenesis

To ascertain the impact of mechanosensitive Cl− channels on stretch‐induced changes in PV electrophysiology, we tested I Cl,swell inhibitors on the intra‐PV conduction block. First, the applied tension was adjusted to induce PV RP depolarization and conduction block (characterized by RP above −60 mV and APA <20 mV). The tensions varied from 10.5 g (75% probability of conduction block) to 13.5 g (94%; Figure 1E). Application of nonselective inhibition of I Cl,swell by DIDS (4,4′‐diisothiocyanato‐2,2′‐stilbenedisulfonic acid disodium salt; 100 μmol/L, n=7) restored conduction in all preparations tested, significantly (P<0.001) hyperpolarizing the RP and increasing the APA (Figure 2A). Similarly, selective inhibition of I Cl,swell by DCPIB (4‐[(2‐butyl‐6,7‐dichloro‐2‐cyclopentyl‐2,3‐dihydro‐1‐oxo‐1H‐inden‐5‐yl)oxy]butanoic acid; 100 μmol/L, n=9) quickly (<5 minutes) hyperpolarized the RP and restored conduction in the PVdis (Figure 2B). Washout of DCPIB abolished its effects, and conduction block returned.

Figure 2

Inhibition of volume‐activated chloride ion (Cl−) channels hyperpolarizes resting potential (RP) in pulmonary veins distal (PV dis) and restores intra‐PV conduction in stretched PVs. Two simultaneous microelectrode recordings from PV dis (top recording in red) and PV ostium (PV ost; bottom recording in blue) are shown during conduction block and after application of nonselective chloride current blocker DIDS (A) or selective swelling‐activated chloride ion channel ( l,swell) blocker DCPIB (B) (shown by arrows). Selected time windows (gray rectangles) are shown enlarged in corresponding panels. Left atrium (LA) was constantly paced with S1S1=300 ms. Notice progressive hyperpolarization of RP (shown by arrows) during conduction recovery. Transmembrane potential levels of −60 and −80 mV are shown by dotted lines for PV dis recording. At right, action potential amplitude and RP are shown at baseline (no stretch applied), during intra‐PV block (Block) and after application of corresponding l,swell antagonists. n=7 for DIDS and n=9 for DCPIB. *P<0.05, ***<0.001 vs baseline, and # P<0.05, ###<0.001 for PV dis vs PV ost by repeated‐measurements 2‐way ANOVA with Bonferroni correction. DCPIB indicates 4‐[(2‐butyl‐6,7‐dichloro‐2‐cyclopentyl‐2,3‐dihydro‐1‐oxo‐1H‐inden‐5‐yl)oxy]butanoic acid; DIDS, 4,4′‐diisothiocyanato‐2,2′‐stilbenedisulfonic acid disodium salt.

To test whether calcium‐activated Cl− channel ANO1 (anoctamin 1; also known as TMEM16A [transmembrane member 16A]), which might be activated by a possible intracellular [Ca2+] elevation on stretch, contributes to the observed stretch‐induced intra‐PV conduction dissociation, we applied the ANO1‐selective inhibitor NPPB (5‐nitro‐2‐[3‐phenylpropylamino] benzoic acid ; 10 μmol/L) and found no effect on either PVdis RP or intra‐PV conduction block within 30 minutes of treatment (data not shown).

To correlate the observed effects with the underlying activation of I Cl,swell, cardiomyocytes were enzymatically isolated separately from the rat PVdis and PVost. I Cl,swell was activated by osmotic swelling from relative osmolality of 1T to 0.7T (Figure 3A). Patch clamp studies revealed swelling‐activated, DCPIB‐sensitive Cl− current specifically in PVdis but not in PVost cardiomyocytes; this finding correlates well with the electrophysiological effects of stretch observed on the intact PV (Figures 1 and 2). These results are in agreement with a functional presence of DIDS‐sensitive chloride current shown in isolated rabbit PV myocytes in response to both axial cell stretch and hypotonic cell swelling.9

Figure 3

Activation of swelling‐activated chloride ion channels ( l,swell) leads to resting potential depolarization and inhibition of spontaneous electrical activity in single cardiomyocytes. A, (Left) Whole‐cell l,swell recorded from single myocytes separately isolated from the pulmonary veins distal (PV dis; n=8 for isotonic [Iso], n=8 for hypotonic [Hypo], and n=6 for Hypo plus DCPIB 10 μmol/L; upper panel) and PV ostium (PV ost; n=4 for Iso and Hypo; lower panel), respectively, under Iso and Hypo conditions. DCPIB‐sensitive activation of l,swell was observed during Hypo condition exclusively in PV dis myocytes. (Right) Corresponding current‐voltage (I‐V) curves for l,swell from PV dis (upper panel) and PV ost (lower panel) myocytes. B, Activation of l,swell current by Hypo cell swelling led to depolarization of the membrane resting potential (RP) and inhibition of spontaneous electrical activity in single human induced pluripotent stem cell‐derived cardiomyocytes (iPS‐CMs). DIDS (100 μmol/L) recovered the automaticity and RPs in the single iPS myocytes. ***P<0.001 vs Iso condition by 1‐way ANOVA with Bonferroni correction. C, Similar to PV dis myocytes, the effect was observed in single human iPS‐CMs. *P<0.05, **0.01, ***0.001 vs treatment by 1‐way ANOVA with Bonferroni correction. DCPIB indicates 4‐[(2‐butyl‐6,7‐dichloro‐2‐cyclopentyl‐2,3‐dihydro‐1‐oxo‐1H‐inden‐5‐yl)oxy]butanoic acid; DIDS, 4,4′‐diisothiocyanato‐2,2′‐stilbenedisulfonic acid disodium salt.

To determine whether our findings apply to human cardiomyocytes, we used hiPS‐CMs and modeled cellular stretch by osmotic swelling (Figure 3B and 3C).9, 23 Prior electrophysiology studies indicated that iPS cells have a capacity for differentiation into nodal‐, atrial‐, and ventricular‐like phenotypes based on AP characteristics.21 However, no protocols are currently available to specifically differentiate iPS cells to PV cardiomyocytes or to isolate such a population of cells, so we chose experimental conditions in which most of the hiPS‐CMs exhibited immature ventricular‐like APs characterized by stable spontaneous activity and slightly depolarized maximum diastolic potential (−65±3 mV) similar to what was recorded previously from isolated canine PV myocytes (−66±1 mV in PV versus −74±1 mV in LA; P<0.001).24 Activation of I Cl,swell current by hypotonic cell swelling led to depolarization of the membrane RP and inhibition of spontaneous electrical activity in a single hiPS‐CM (Figure 3B). Similar to tissue level changes (Figure 2), inhibition of I Cl,swell by 100 μmol/L DIDS in the setting of hypotonic swelling hyperpolarized RP to baseline values and restored the spontaneous activity in hiPS‐CMs. In addition, whole‐cell patch clamp studies of hiPS‐CMs demonstrated a DIDS‐sensitive, swelling‐activated Cl− current (Figure 3C). These findings are consistent with human cardiomyocytes exhibiting electrophysiology similar to that observed in the rat PV cardiomyocytes.

Caveolar Mechanosensitive Complex

I Cl,swell current has been linked to the activation of ClC‐2,25 ClC‐3,26 and SWELL127 isoforms of sarcolemmal mechanosensitive Cl− channels. Various studies associated mechanosensitive I Cl with caveolae structures10, 13, 28 in which these channels can be associated with the caveolar scaffolding proteins caveolins.29 Our immunohistochemical analysis showed that ClC‐2, ClC‐3, and SWELL1 are all highly colocalized with myocyte‐specific caveolar scaffolding protein Cav3 in both rat PV and human left atrial myocardium (Figure 4; negative controls are shown in Figure S7). Immunoprecipitation performed on rat PV myocardial lysate revealed that only ClC‐3 and SWELL1, but not ClC‐2, are associated with Cav3 (Figure 5). In contrast, both ClC‐3 and ClC‐2 are found to be associated with Cav3 in human left atrium (Figure S8), suggesting species dependence. Our comparative quantitative reverse transcription polymerase chain reaction analysis demonstrated that SWELL1 and ClC‐3 channels demonstrate an equal abundance in both PV and LA myocardium, whereas ClC‐2 channels (as well as calcium‐activated Cl− channel ANO1) show a 10‐fold lower mRNA expression level compared with SWELL1 and ClC‐3 channels.

Figure 4

Caveolar macromolecular mechanosensitive complex. Immunofluorescent analysis of colocalized expression of ClC‐3 (left), ClC‐2 (middle), and SWELL1 (right) chloride channels with caveolae scaffolding protein Cav3 (caveolin 3) in nonstretched rat pulmonary vein distal (PV dis; A) and human left atrium (B) myocardium. For colocalization analysis, intensity level of 30% was used as a threshold. Cav3 indicates caveolin 3; ClC, chloride channel; SWELL1, also known as LRRC8A (leucine rich repeat containing protein 8A).

Figure 5

Co‐IP Western blots showing degree of association of Cav3 and chloride channels ClC‐2, ClC‐3, and SWELL1 in rat pulmonary vein (PV; n=2). IB indicates immunoblot; IP, immunoprecipitation. Cav3 indicates caveolin 3; ClC, chloride channel; SWELL1, also known as LRRC8A (leucine rich repeat containing protein 8A).

Western blot analysis did not reveal any difference in the protein expression level for ClC‐2, ClC‐3, and SWELL1 along the PV, whereas the Cav3 expression level was higher in PVdis versus PVost (Figure 6). This may indicate higher membrane reserve and thus more substantial I Cl,swell activation on stretch in PVdis than in PVost, as it was observed in both the intact vein (Figures 1 and 2) and isolated PV cells (Figure 3). Together with a significantly lower expression level of the RP stabilizing inwardly rectifying K+ current (I K1) protein in the PV (Figure 6A and 6B; as also demonstrated previously30), these results may provide an explanation for the observed higher sensitivity of the PVdis to caveolae‐mediated stretch‐induced depolarizing Cl− current.

Figure 6

Expression of chloride channels ClC‐2, ClC‐3, and SWELL1, caveolar scaffolding protein caveolin 3 (Cav3), and potassium channel Kir2.1 along the pulmonary vein (PV) vs left atrium (LA). A, Protein expression levels of Kir2.1, ClC‐2, Cav3, ClC‐3, and SWELL1 measured in PV distal (PV dis), PV ostium (PV ost), and LA from the same rat (n=4). B through F, Corresponding protein expression levels normalized to GAPDH (n=4 per group). **P<0.01 by 1‐way ANOVA with Bonferroni correction. G, Comparative analysis of mRNA expression levels for sarcolemmal chloride ion channel isoforms normalized to GAPDH. n=6 per region for SWELL1 and n=5 per region for ClC‐2, ClC‐3, and ANO1. ***P<0.01 vs PV dis for SWELL1; ## P<0.01, ### P<0.001 vs SWELL1 for PV dis; $$ P<0.01 vs ClC‐3 for PV dis; and && P<0.01 vs ClC‐3 for LA by 2‐way ANOVA with Bonferroni correction. ANO1 indicates anoctamin 1; Cav3, caveolin 3; ClC, chloride channel; SWELL1, also known as LRRC8A (leucine rich repeat containing protein 8A).

Hypersensitivity of SHR PVs to Stretch

Hypertension is one of the main risk factors of AF.1 Clinical studies show that an acute increase in atrial pressure and sustained atrial dilatation can result in the development of AF.2 Several studies using SHRs show increased incidence of AF and atrial tachyarrhythmias compared with normotensive rats.31 In this study, we applied stretch protocols to SHR PV preparations (echo data indicating diastolic dysfunction and progress toward cardiac hypertrophy in SHRs are shown in Figure S9) and found that the tension required to induce intra‐PV conduction block in SHRs is inversely correlated with the rat's BP (measured using tail cuff from lightly anesthetized rats). SHRs with BP >200 mm Hg (average BP of 219±6 mm Hg, average tension to induce conduction block was 0.30±0.03 g, n=3) developed conduction block at much smaller tensions, even within the physiological range, compared with SHRs with BP <200 mm Hg (average BP of 163±10 mm Hg, average tension to induce block was 1.6±0.7 g, n=3, P<0.01; Figure 7A, top). Both groups, however, had much higher sensitivity to developing intra‐PV conduction block than Wistar rats (Figure 7A, bottom). In comparison to Wistar rats, SHRs showed significantly smaller APA in both PVost and PVdis, mild RP depolarization, and a significantly prolonged functional refractory period (Figure S10), which may indicate the presence of constitutively active I Cl,swell that has been previously reported in both failing canine ventricular myocytes23 and human atrial myocytes obtained from patients with right atrial enlargement and/or elevated left ventricular end‐diastolic pressure.32

Figure 7

Effect of stretch on electrical activity of the pulmonary vein (PV) in spontaneously hypertensive rats (SHRs). A, Development of intra‐PV conduction block in SHRs. (Top) Negative exponential dependence between the aortic blood pressure (BP) and the maximal tension required to induce intra‐PV conduction block in normotensive rats (Wistar [WT]; black dots) and SHRs with BP <200 mm Hg (blue dots) and BP >200 mm Hg (red dots). (Bottom) Probability of intra‐PV conduction block estimated at different tensions applied to WT (n=8) and SHRs with BP <200 mm Hg and BP >200 mm Hg. B, (Top) PV activation maps and superimposed upstrokes of optical action potentials (Vm) recorded along the PV are shown for baseline (left) and pathological stretch (right) conditions in SHRs. (Bottom) Distribution of conduction velocity along the PV (left) and relative PV distance (in percentage from the PV length; right) activated at baseline, during stretch, and after application of swelling‐activated chloride ion channel ( l,swell) blocker 9AC (9‐anthracenecarboxylic acid) are shown (n=4 per group). *P values were determined by 2‐way ANOVA with Bonferroni correction. C through F, Stretch‐induced changes in active potential (AP) parameters are shown for 2 groups of SHRs based on their BP. *P<0.05, **P<0.01, **P<0.01 vs SHR with BP <200 mm Hg, and # P<0.05, ## P<0.01, ### P<0.001 for PV distal (PV dis) vs PV ostium (PV ost) by repeated measurements 2‐way ANOVA with Bonferroni correction. APA indicates action potential amplitude; APD, action potential duration; FRP, functional refractory period.

Optical mapping of electrical activity revealed a more pronounced intra‐PV conduction dissociation in SHRs, with conduction block occurring more proximally to the PVost than in Wistar rats (Figure 7B). Similar to Wistar rats, inhibition of I Cl,swell by 9AC (9‐anthracenecarboxylic acid; 100 μmol/L) recovered the AP propagation toward the PVdis, confirming the contribution of mechanosensitive I Cl,swell in the stretch‐induced intra‐PV conduction dissociation.

For detailed analysis of PV mechanosensitivity associated with varied severity of hypertension, we separated SHRs into 2 groups with BP ≥200 and <200 mm Hg, respectively. SHRs with BP <200 mm Hg had significantly shorter and stretch‐insensitive AP duration and functional refractory period compared with SHRs with BP >200 mm Hg (Figure 7C and 7D). In SHRs with BP <200 mm Hg, PVdis was more sensitive to stretch than PVost, which was similar to that observed in Wistar rats (Figure 7E and 7F). In SHRs with BP >200 mm Hg, the entire PV was highly sensitive to stretch‐induced changes in RP and APA, indicating more profound remodeling occurring along the PV in rats with severe hypertension versus either mild hypertension or normotension.

Stretch‐Induced Conduction Abnormalities Are Associated With Downregulation of Cav3

To determine whether hypersensitivity of SHR PV is associated with changes in ClC expression profile, we measured ClC‐2, ClC‐3, and SWELL1 protein expression and found no significant changes in levels in SHRs (Figure 8A). In contrast, SHRs demonstrated significant downregulation of Cav3 protein expression level compared with Wistar rats (Figure 8A). Downregulation of Cav3 was associated with a prominent decrease in caveolae density in SHRs assessed by transmission electron microscopy of PVdis tissues (Figure 8B and 8C). Furthermore, SHR PV myocytes also exhibited changes in the lateral sarcolemma ultrastructure. Whereas Wistar myocytes had a relatively convoluted surface, with caveolae clustering within outcroppings of membrane (see electron micrographs in Figure 8B, solid red line), SHR myocytes displayed straighter sarcolemma characterized by a smaller “convolution index” (L/Lo−1, where L is the length of membrane contour, and Lo is the shortest length connecting end points of membrane segment; Figure 8D),33 which indicates caveolae flattening in response to elevated stretch in SHRs.

Figure 8

Molecular and structural remodeling of caveolar mechanosensitive complex. A, Immunoblots for ClC‐2, ClC‐3, SWELL1, and Cav3 from Wistar (WT; n=4) and spontaneously hypertensive rat (SHR; n=4) pulmonary vein distal (PV dis) tissues. Expression levels were normalized to GAPDH. P values were determined by unpaired Student t test. B, Downregulation of Cav3 correlates with elevated membrane tension and disruption of caveolae structures during hypertension. Representative composite electron micrographs showing the lateral sarcolemmal membranes of cardiomyocytes from nonstretched WT and SHR PV dis tissues. Blue arrowheads denote caveolae connected to plasma membrane (scale bars=200 nm). C and D, Quantification of caveolae density, normalized to membrane length (C), and membrane convolution index (L/Lo−1) (D), n=40 cells per group. Box plots show medians with interquartile range; whiskers represent fifth and 95th percentile; each point represents 1 micrograph. P values were determined by unpaired Student t test. Cav3 indicates caveolin 3; ClC, chloride channel; L, length of membrane contour; Lo, shortest length connecting end points of membrane segment; SWELL1, also known as LRRC8A (leucine rich repeat containing protein 8A).

Discussion

Cardiac stretch has previously been shown to modulate several vulnerable electrophysiological parameters in PV myocytes, potentially contributing to the development of AF.6, 7 However, the molecular mechanisms of cardiac mechanoelectrical transduction remain poorly understood. In this study, we provided conceptual innovation of mechanosensing in the heart, linking specialized cell structures, caveolae, and their response to stretch to the activation of mechanosensitive I Cl,swell. We showed that pathological activation of I Cl,swell modulates electrical activity in the PVs, promoting the development of ectopic foci and triggering AF.

The role of caveolae in mechanosensing has been demonstrated in various cell types, including cardiomyocytes. Caveolae flattening and the incorporation of caveolar membrane into the plasma membrane in response to stretch plays a crucial role in cell volume regulation. Our results showing decreased mechanoprotection of SHR PV myocytes agrees with findings that caveolae act as membrane convolutions that can flatten in response to increased membrane tension, thereby serving as a buffer to prevent membrane rupture.33, 34 In addition, evidence has accumulated for the existence of active signaling from caveolae in response to changes in mechanical forces; for example, stretch activation of caveolae‐associated I Cl,swell was linked to the regulation of cellular contractility, thus adjusting cardiac performance on changes in blood volume of the heart.10, 15 In addition, caveolae are essential for angiotensin II type 1 receptor–mediated atrial natriuretic peptide secretion35 and have been shown to be involved in the production of reactive oxygen species and nitric oxide via activation of NADPH oxidase subunits and eNOS.36 In the healthy heart, these signaling pathways provide balancing feedback to regulate cardiac performance on increased BP. However, under disease conditions associated with pathological stretch, disruption of caveolae and activation of caveolar signaling pathways might be proarrhythmic, as demonstrated in the present study.

Pathological stretch‐induced activation of I Cl,swell depolarized RP toward the equilibrium potential for Cl− (between −65 and −40 mV37; Figure 2). Such depolarization was most prominent in PVdis due to ≈40% decreased density of I K1 compared with the LA myocardium.30 I Cl,swell‐induced RP depolarization up to −60 mV shifts a significant proportion of fast sodium channels into an inactivated state, which in turn suppresses conduction velocity and leads to the development of conduction block proximally in the PV. Therefore, conduction block is induced by stretch more proximally in the PV, where Cl− currents cannot be compensated by the I K1. Acute (7.2±1.6 mm Hg) atrial stretch–induced conduction slowing across the PV‐LA junction and enhanced electrogram signal complexity, which likely suggests the presence of intra‐PV conduction blocks, were observed in humans who underwent evaluation of the right superior PV–LA junction using an epicardial mapping plaque.7 Similarly, pronounced repolarization heterogeneity as well as nonsustained reentrant activity was observed by Arora et al in coronary‐perfused, isolated, whole‐atrial canine preparations using optical mapping.22 The authors highlighted that reentry occurred more distally in the PV, whereas focal activity seemed to occur more proximally, with which the results observed in our study agree. A recent case report on a patient who underwent a radiofrequency ablation of drug‐refractory paroxysmal AF described intra‐PV echo beats characterized by PV automaticity with a delayed, stably coupled second component suggesting a small reentry within the vein.38

The molecular composition responsible for cardiac I Cl,swell remains questionable and highly controversial. This also seems to depend on the type of cardiac tissues studied. Xiong et al used cardiac‐specific, inducible ClC‐3 gene deletion, which eliminated a native volume‐sensitive chloride current in both atrial and ventricular mouse myocytes.26 Meanwhile, the same group demonstrated a functional presence of ClC‐2 chloride inward rectifier channels in cardiac sinoatrial nodal guinea pig pacemaker cells.25 Although no experimental data are available to date on a functional presence of ClC‐2 or ClC‐3 channels in PV myocytes, one could suggest that, considering that PV myocytes represent electrophysiological properties of both atrial and pacemaker cells (the presence of spontaneous electrical activity19, 39, 40 and a low membrane RP stabilizing K+ current I K1 24, 40 associated with a decreased expression of Kir2.x channels30), both ClC‐2 and ClC‐3 channels may contribute, at different proportions, to PV I Cl,swell.

We also detected, for the first time, the presence of SWELL1 channels in the rat and human myocardium, at both mRNA and protein levels (Figures 4, 5, 6A, and 6G). SWELL1 is a recently characterized chloride channel and may represent an essential component of volume‐regulated anion current in noncardiac cells.27 SWELL1 mRNA was detected broadly in mouse tissues.27 Although SWELL1 mRNA expression has been detected in the heart, neither protein expression or localization nor functional contribution of these channels into cardiac I Cl,swell were reported. Our comparative reverse transcription quantitative polymerase chain reaction analysis showed that that SWELL1 and ClC‐3 channels demonstrate an equal abundance in both PV and LA myocardium (Figure 6G). Therefore, it is possible that SWELL1 channels could also functionally contribute to PV I Cl,swell. Based on our coimmunoprecipitation (Figure 5) and coimmunofluorescence (Figure 4) analysis, SWELL1 channels might be involved in caveolar mechanosensitive complex and thus participate in stretch‐induced PV arrhythmogenesis and hypertensive‐associated remodeling.

Compared with biophysical properties of ClC‐2, ClC‐3 and SWELL1 channels, cardiac I Cl,swell recorded in native PV myocytes and hiPS‐CMs (Figure 3) demonstrate outward rectification similar to both ClC‐326 and SWELL127 channels. In contrast, ClC‐2 channels show a prominent inward rectification which was not seen in the I Cl,swell recorded in the present study. Assuming a 10‐fold smaller ClC‐2 mRNA expression level compared with that for ClC‐3 and SWELL1 (Figure 6G), it is unlikely that ClC‐2 current, if functionally present, plays an important role in the PV I Cl,swell.

In the present study, higher sensitivity of SHR PV to stretch‐induced RP depolarization and intra‐PV conduction dissociation was associated with decreased Cav3 protein expression level and downregulation of caveolae structures (Figure 8). Although the exact mechanisms of activation of I Cl,swell in response to stretch remain unknown, their interaction with caveolae appears to be critical for channel activation.10 It has been previously shown that ClC‐2 channels are concentrated in lipid rafts in basal condition and relocalize to the cell surface membrane on caveolae disruption via cholesterol depletion, with a subsequent increase in their activity.28 The same might be true for ClC‐3 and SWELL1 channels. This is supported by our data on basal changes in AP morphology in SHR PVs (Figure S10) and their higher sensitivity to stretch which could indicate constitutively active I Cl,swell and/or greater I Cl,swell activated by a lower level of stretch in SHRs versus Wistar rats.

We found that SHR sensitivity to stretch depended on the rat's BP. Although all SHRs had a significantly higher sensitivity to stretch than the Wistar group, rats with BP >200 mm Hg showed RP depolarization and intra‐PV conduction block at much smaller weights compared with SHRs with BP <200 mm Hg (Figure 7). It has been shown that SHRs develop hypertension at 2 to 3 months of age and that it peaks from 6 months of age onward with increasing compensatory concentric cardiac hypertrophy (12–15 months) before the onset of heart failure at the age older than 18 months.41 Significant atrial remodeling, including biatrial hypertrophy, elevated atrial fibrosis, and increased atrial arrhythmogenesis, has been observed in SHRs.42 Importantly, as it was described in a study by Bing et al, ≈60% of SHRs developed evidence of cardiac decompensation, another 13% survived to 24 months and did not have evidence of heart failure, and the rest died for noncardiac reasons (eg, stroke, debilitation, tumor).41 Because the magnitude and duration of hypertension are important determinants of the degree of hypertrophy and the functional status of ventricles, it may result in the different degrees of atrial remodeling and explain the observed sensitivity in 2 groups of SHRs. Therefore, it may result in the different degree of atrial and PVs remodeling and explain the observed distinct sensitivity in 2 groups of SHRs. Further studies are required to explore the cellular and molecular mechanisms behind the observed phenotypes, which might be related to the differences in ion channel expression repertoire and/or calcium handling.

Clinical Perspectives

Although some forms of AF can be successfully treated via PV isolation, the success rate of such procedures for individuals with persistent AF is poor (≈50%).43 The relatively poor response to conventional treatment is attributed to the extensive remodeling of the atrial myocardium, which creates multiple ectopic foci and confounds strategies for identifying ablation targets.44 Furthermore, catheter ablation empirically destroys viable atrial tissue rather than directly addressing the underlying triggering mechanisms of AF. To improve outcomes, both pharmacological and gene therapy should be considered as alternative approaches for AF treatment and prevention. Our study introduces a novel paradigm by which electrophysiological changes occur secondary to cellular structural remodeling. This extends beyond the classical concept of electrical remodeling and adds a new dimension to cardiovascular disease, providing a basis from which to develop novel and effective therapeutic approaches targeted to treat stretch‐induced arrhythmogenesis and AF by preventing the degradation or promoting the restoration of cardiac cytoarchitecture.

Study Limitations

Our study demonstrates that I Cl,swell likely consists of at least 2 currents produced by mechanosensitive ClC‐3 and SWELL1 channels, and ClC‐2 current, if functionally present, plays an important role in the PV I Cl,swell. Unfortunately, isoform‐selective pharmacological inhibitors of mechanosensitive chloride channels are not currently available, and cardiac‐specific and isoform‐selective knockout animals are required to dissect the exact molecular composition of I Cl,swell current measured in PVs. These experiments are outside of the scope of the present study, which reports, for the first time, important findings on the role of caveolae‐associated stretch‐activated I Cl,swell in PV arrhythmogenesis, and will be reported in our follow‐up studies.

Mechanosensitivity has been reported for multiple ion channels, including specific K+‐selective stretch‐activated channels such as TREK and TREK‐like channels, BK channels, cation nonselective stretch‐activated channels (Piezo and transient receptor potential channels), and Cl− channels.9, 45, 46, 47 In addition, stretch can directly modulate biophysical properties and functioning of multiple voltage‐ and ligand‐gated ion channels including Kv and Kir potassium channels, Nav1.5 sodium channels, and Cav1.2 calcium channels, thus modulating AP morphology.45 Therefore, it is possible that other ion channels may contribute to stretch‐induced AP changes in the PV myocardium. However, as demonstrated in our study, I Cl,swell has a key role in RP depolarization and conduction dissociation in PV, which could be abolished via selective pharmacological inhibition of I Cl,swell by DCPIB, DIDS, and 9AC. It is unlikely that all of these agents could augment outward potassium current to counteract I Cl,swell and hyperpolarize RP.

Sources of Funding

This work was supported by the Russian Foundation for Basic Research Grants 14‐04‐01781 and 17‐04‐01634 (to Rosenshtraukh), National Institutes of Health (NIH) 1RO1HL141214‐01, American Heart Association (AHA) 16SDG29120011, and the Wisconsin Partnership Program 4140 (to Glukhov), and AHA Fellowship 17POST33370089 (to Lang). Turner would like to acknowledge NIH predoctoral training grant T32GM008688.

Disclosures

Schmuck and Raval have ownership and financial interest in Cellular Logistics Inc. The remaining authors have no disclosures to report.

Data S1. Supplemental methods.

Table S1. List of Taqman Probes for Reverse Transcription Quantitative Polymerase Chain Reaction Analysis of Sarcolemmal Chloride Channel Isofors and Caveolar Scaffolding Protein Cav3 (Caveolin 3) mRNA Expression in Rat Hearts

Figure S1. Effect of physiological stretch on pulmonary vein electrical activity.

Figure S2. (A) Representative action potential duration (APD) restitution curves are shown for pulmonary vein distal (PVdis) and PV ostium (PVost) at baseline and at 0.3 g tension. (B) Change of APD restitution curve slopes for PVdis and PVost at physiological stretch protocol (0–1.5 g).

Figure S3. Effect of pathophysiological stretch on the pulmonary vein myocardium action potential duration at 90% of repolarization (A) and functional refractory period (B).

Figure S4. Sarcomere length measured from immunofluorescent staining against α‐actinin staining as an average distance between α‐actinin striated bands within cells in pulmonary vein distal (PVdis) and PV ostium (PVost) at baseline (no stretch applied) and under pathological stretch (at tensions required to induce intra‐PV conduction block).

Figure S5. Combined effect of stretch and norepinephrine (1 μmol/L) superfusion on spontaneous electrical activity of the pulmonary vein.

Figure S6. Microreentry within the pulmonary vein.

Figure S7. Negative control for immunofluorescent staning experiments.

Figure S8. Coimmunoprecipitation Western blots, from left to right: inputs, immune complexes, supernatants, and laemmli/DTT elution.

Figure S9. Echocardiography data for 9‐ to 12‐month‐old Wistar (n=5) and spontaneously hypertensive (n=3) rats.

Figure S10. Spontaneously hypertensive rat pulmonary vein electrophysiological characteristics at baseline.

Click here for additional data file.