Cardiomyocyte Expression of ZO-1 Is Essential for Normal Atrioventricular Conduction but Does not Alter Ventricular Function

Abstract

Rationale:

Zonula occludens-1 (ZO-1), a plasma membrane-associated scaffolding protein regulates signal transduction, transcription and cellular communication. Global deletion of ZO-1 in the mouse is lethal by embryonic day 11.5. The function of ZO-1 in cardiac myocytes (CM) is largely unknown.

Objective:

To determine the function of CM ZO-1 in the intact heart, given its binding to other CM proteins that have been shown instrumental in normal cardiac conduction and function.

Methods and Results:

We generated ZO-1 CM-specific knockout (KO) mice using α-Myosin Heavy Chain-nuclear Cre, (ZO-1cKO), and investigated physiological and electrophysiological function by echocardiography, surface ECG and conscious telemetry, intracardiac electrograms and pacing, and optical mapping studies. ZO-1cKO mice were viable, had normal Mendelian ratios, and had a normal lifespan. Ventricular morphometry and function were not significantly different between the ZO-1cKO vs control (CTL) mice, basally in young or aged mice, or even when hearts were subjected to hemodynamic loading. Atrial mass was increased in ZO-1cKO. Electrophysiological and optical mapping studies indicated high-grade atrioventricular (A-V) block in ZO-1cKO comparing to CTL hearts. While ZO-1 associated proteins such as vinculin, connexin 43, N-cadherin, and α-catenin showed no significant change with the loss of ZO-1, Connexin-45 and Coxsackie-adenovirus (CAR) proteins were reduced in atria of ZO-1cKO. Further, with loss of ZO-1, ZO-2 protein was increased significantly in ventricular CMs in a presumed compensatory manner, but was still not detected in the AV nodal myocytes. Importantly, the expression of the sodium channel protein NaV1.5 was altered in AV nodal cells of the ZO-1cKO vs. CTL.

Conclusion:

ZO-1 protein has a unique physiological role in cardiac nodal tissue. This is in alignment with its known interaction with CAR and Cx45, and a new function in regulating the expression of NaV1.5 in AV node. Uniquely, ZO-1 is dispensable for function of the working myocardium.

Graphical Abstract

INTRODUCTION

The zonula occludens (ZO) family consists of ZO-1, −2 and −3 proteins (also known as tight junction protein (TJP)1, TJP2 and TJP3, respectively). ZO proteins are members of the membrane-associated guanylate kinase (MAGUK) family of proteins that were first identified in Drosophila Melanogaster¹, that assemble multiprotein complexes adjacent to the cell membrane. ZO proteins were initially found to be components of tight junctions² and later also found in adherens junctions (AJs)^3-6, and gap junctions (GJ)^{7, 8}. In cardiac myocytes (CMs) that do not have tight junctions, ZO-1 colocalizes with cadherins in AJs as they form components of the intercalated discs (ICD)^{9, 10}, and helps to maintain normal function of various ICD junctions. Furthermore, ZO proteins are also recognized to be important for signal transduction, modulation of transcription, and cellular communication. Recently, De Bortoli et al. identified pathogenic variants in ZO-1 that were associated with human arrhythmogenic cardiomyopathy¹¹.

ZO proteins carry three PDZ domains, named for the three proteins: Post-synaptic density protein (PSD5/SAP90), Drosophila disc large tumor suppressor, and ZO-1. In addition, they have SH3 and GUK domains. ZO-1 may complex with ZO-2 and ZO-3^{12, 13}. During development, ZO-1 and ZO-2 are expressed ubiquitously, while ZO-3 is restricted to epithelial cells¹⁴.

All three ZO proteins have been genetically ablated in the mouse. Global ZO-1–deficient mice showed 100% embryonic lethality by embryonic day (E) 11.5 with extensive apoptosis¹⁴, neural tube / notochord defects, with both embryonic and extraembryonic defects including defective chorioallantoic fusion. ZO-1 loss did not alter ZO-2 or ZO-3 expression in these mice. ZO-2 ablated mice showed abnormal gastrulation, cell proliferation and apoptosis with lethality around the time of implantation.¹⁵ Why ZO-2 loss led to these abnormalities led to proposals that leaky TJs, abnormal GJ and perhaps even perturbation of gene expression, given ZO-2’s interaction with transcription factors, was causal. In contrast, ZO-3 global deletion did not cause any changes in normal development, postnatal growth or function^{15, 16}.

ZO-1 binds to many proteins in multiple cell-types. These include connexins (Cx), N-cadherin, α-actinin, α-catenin, occludin, claudin, shroom-2, nephrin, F-actin and afadin, highlighting its role as an important cytosolic scaffolding protein^{17, 18}. In TJ, ZO-1 directly binds to adhesion molecules such as claudins and occludin, via its PDZ-1 and GUK domains, respectively^{19, 20}. In CM, the second PDZ domain of ZO-1 directly interacts with the GJ protein, Cx43^21-23 and Cx45²⁴.

Previously, we reported that the Coxsackievirus and adenovirus receptor (CAR), a transmembrane adhesion molecule localized at GJ, mediates atrioventricular-node (AVN) function²⁴. Disruption of CAR expression in mouse CM resulted in AVN block, with late onset of cardiomyopathy. This was associated with decreases in ZO-1 and Cx-45 expression levels, and loss of their localization at the AV-node²⁴. In that study, we found that CAR interacted with ZO-1 as had been previously demonstrated ²⁵, and that CAR binding to ZO-1, as well as Cx45 binding to ZO-1, was required for complex formation between CAR, ZO-1 and Cx45.

In separate work, we found also that the AJ / ICD-expressed protein Vinculin (Vcl) also colocalized at the ICD with Cx43 and ZO-1. When Vcl was ablated from CMs, ZO-1 expression was lost from the ICD and Cx43 became maldistributed and mice showed ventricular tachycardia and sudden death²⁶. We also showed that Vcl directly binds ZO-1²⁷.

We therefore hypothesized that ZO-1 would be required for AVN function by facilitating molecular interactions between CAR and Cx45, and perhaps other scaffolding or channel proteins, and that it would also be essential for maintaining normal working CM (e.g. ventricular CM) function, perhaps by maintaining localization of Cx43 and vinculin.

Somewhat surprisingly, the data demonstrates that absence of ZO-1 from CM did not affect ventricular structure or function, but led to varied degrees of AV block, including complete heart block. ZO-2 was seen to increase in the isolated ventricular CM suggesting that ZO-2 could substitute for loss of ZO-1, and preserve CM function, except in nodal tissue. Further, the qualitative expression of the sodium channel NaV1.5, that has been linked to conduction system disease, was altered in AV nodal cells. Thus, this data indicates that ZO-1 protein plays a critical role in the function of the cardiac conduction system but is dispensable for mechanical function of the intact heart.

METHODS

An expanded Methods section is available in the Online Supplement.

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

Within specific genotypes, animals, or samples from them, were randomly assigned to experiments. Studies were performed in a blinded manner. Unless specified in the figure legends, male mice were used.

Animal care.

All animal procedures were performed and approved by the University of California–San Diego Institutional Animal Care and Use Committee.

Gene targeting and generation of ZO-1cKO.

Techniques for generation of ZO-1cKO mice were as previously described^{28, 29} and detailed in Online Supplement.

Surface electrocardiography, conscious telemetry, transverse aortic constriction, echocardiography, intracardiac electrophysiology and optical mapping.

Procedures were as performed previously^{26, 30, 31} with EP study detailed in the Online Supplement.

Statistics.

Data were analyzed using GraphPad Prism 7 software (GraphPad Software, San Diego, CA, USA). All data are represented as mean ± SD. The data obtained from all analyses were evaluated using either an un-paired parametric t-test or 2-way ANOVA, as outlined, with P≤0.05 regarded as significant for all studies. Further details are provided in the Online Supplement.

RESULTS

Generation of ZO-1 cardiac myocyte specific knockout mice.

Given the embryonic lethality found in the global ZO-1 KO mice¹⁴, we generated mice with a ZO-1 floxed allele that targeted exon 3 of the ZO-1 gene (Online Figure IA and IB). Appropriately targeted R1 embryonic stem (ES) cells were used to produce mice which germline transmitted the targeted allele, and then bred into the Black Swiss background to generate ZO-1^f/+ mice. These mice were mated with α-Myosin Heavy Chain (MHC)-nuclear Cre mice^32,26 to generate CM-specific ZO-1 KO mice (ZO-1^f/f x αMHC-Cre, ZO-1cKO). Control mice utilized throughout the study were ZO-1^f/+ x αMHC-Cre (CTL).

To determine whether deletion of the ZO-1 allele caused protein reduction using this scheme, samples from the four cardiac chambers were dissected and Western blotting was performed on whole tissue protein lysates from the chambers, to detect ZO-1. Expression of ZO-1 was decreased 60–97% in all the chambers (Figure 1A, Online Figure IC). Reasoning that the residual ZO-1 in these samples could come from non-myocytes we isolated CM and prepared protein from ZO-1cKO and CTL hearts. Western blotting of these samples showed virtually complete loss of ZO-1 expression in the ZO-1cKO vs. CTL myocytes (Figure 1B, Online Figure IC). These biochemical results were confirmed by immunomicroscopy showing loss of ZO-1 protein from the ICD in the ZO-1cKO compared to CTL. Residual ZO-1 co-localized with endothelial cells that were present between myocytes (Figure 1C, D).

Figure 1. ZO-1 protein expression is virtually absent from CM in the ZO-1cKO, with only residual expression seen in non-CM cells in heart.

(A, B) ZO-1 expression was evaluated by Western blotting of protein lysates with an anti-ZO-1 antibody, using samples from (A) tissue samples discretely obtained from the four cardiac chambers and (B) Isolated adult CM lysates. GAPDH was used as a loading control. Only minimal ZO-1 expression was evident in CM samples of ZO-1cKO compared to CTL, indicating that residual expression in tissue was related to expression of ZO-1 in non-myocytes within the heart. WT samples of CM protein were also analyzed here as reference. Densitometric quantification of tissue samples showed protein expression in ZO-1cKO, relative to CTL to be: RA – 37%, LA – 37%, RV – 13% and LV – 39%, yet densitometric evaluation of Western blots of isolated CM protein showed only 3.3% residual expression in cKO vs. CTL (n=4 for CTL and ZO-1cKO). (See details of densitometry in Online Figure IC.) (C, D) Sections of left ventricle were evaluated using immunomicroscopy with antibodies to discretely detect (C) ZO-1 (red) and Desmoplakin (green), used as a marker of the cell-cell junction, which showed typical intercalated disc staining for ZO-1 in the CTL mice (arrows) but not in the ZO-1cKO mice, and (D) Residual ZO-1 (Green) localized with endothelial cells (marked by PECAM, Red) in ZO-1cKO heart, again supporting the concept that residual ZO-1 protein detected in the cKO heart was related to expression in non-CM. DAPI was used to mark nuclei blue. (Scale bar: 40 μm.)

ZO-1-cKO mice live a full life span, and show no significant basal histological or contractile abnormalities in the ventricles, nor abnormal responses to hemodynamic challenge with aortic constriction.

The ZO-1cKO mice were born, had a normal Mendelian distribution of genotypes, and were found to have a similar life span to littermate controls, to the age of 22-23 months. No obvious morphological differences were noted in the ventricles of ZO-1cKO vs. CTL, and histological evaluation of the ventricles by hematoxylin and eosin staining (Figure 2A) and using Mason’s trichrome (to detect fibrosis – not shown) did not detect any abnormalities in the ZO-1cKO ventricles.

Figure 2. Changes in atrial normalized weights were noted in ZO-1cKO vs. CTL mouse hearts.

(A) Representative H&E sections did not show alteration in histology of ZO-1cKO hearts vs. CTL. (Scale Bar: Ab and Ae: 2.5mm; Aa,c,d,f: 500uM.) (B) Normalized weights of RA and LA were significantly increased in ZO-1cKO vs. CTL Normality was evaluated using the Shapiro-Wilk test. Data are represented as mean ± SD. Unpaired two-tailed Student’s t-test with Welch correction was used to assess the p value by groups. (CTL n=11, M=6, F=5; ZO-1cKO n=10, M=5, F=5). No differences were seen in combined ventricular weights between groups (data not shown).

However, histological examination of the ZO-1cKO mice showed thicker atrial walls than CTL. Analysis was performed and showed both right atria (RA) and left atria (LA) had greater normalized weight in the ZO-1cKO vs. CTL (RA / Tibia Length and LA / Tibia Length) at 4–6 months of age (Figure 2B). Remarkably, assessment of the ventricles not only showed no histological differences between ZO-1cKO and CTL, but also showed no variance of ventricular normalized weights up to 9 months of age (Data not shown). Myocardial tissues from the various chambers were also examined using transmission electron microscopy (TEM) and did not show significant differences in the ultrastructure of specimens from the two groups. (Online Figure II) While minimal increase in interstitial matrix was noted between myocytes of the cKO atria as compared to CTL, there was no alterations of the CMs or ICDs in ZO-1cKO ventricles or atria.

To investigate cardiac function, we next evaluated the ZO-1cKO and CTL groups by echocardiography (Echo) at 4, 10, and 88 weeks of age. Echo parameters were assessed at each time point and did not show any significant differences (Online Figure IIIA).

Given the critical localization of ZO-1 expression in the ICD, we reasoned that the ZO-1cKO mice might not be tolerant of stress compared to CTL. Therefore, transverse aortic constriction (TAC) was performed to hemodynamically challenge the heart. Mice (11–12w age) were challenged with TAC and followed for 4w following surgery. No significant differences in echocardiographic parameters were detected in ZO-1cKO vs. CTL at 2, 3, or 4w post TAC (Online Figure IIIB). At study termination, pressure gradients (left and right carotid pressures) were measured and showed no differences between groups. (ZO-1cKO gradient = 66.8 +/− 21.7mmHg; CTL gradient = 64.6+/− 20.9mmHg (mean +/− SD) ) Likewise, at study termination, analysis was not significantly different between groups, except for higher atrial normalized weights in ZO-1cKO vs. CTL (Online Figure IIIC), in line with basal evaluation (Figure 2).

Abnormal rhythm in the ZO-1cKO mice.

Given the potential of a rhythm disturbance in the ZO-1cKO, we performed surface ECGs and conscious telemetry. ECGs in lightly anesthetized 11 to 12w mice demonstrated abnormal conduction in the ZO-1cKO mice, with indications of varied degrees of high-grade AV block (Figure 3A). Similar findings were detected in mice as young as 5w (data not shown). High-resolution signal averaging of these ECGs demonstrated no significant difference in the QRS duration between the ZO-1cKO and CTL mice (Figure 3B), but showed evidence of high-grade AV block, including complete heart block, in the ZO-1cKO. In the ZO-1cKO mice P-waves did not appear in the signal averaged ECG because of variability in the PR interval and the atrial rate was frequently decreased (Figure 3B), indicating the presence of sinoatrial (SA) abnormalities in addition to AV conduction delay. Next we performed conscious telemetry and saw that all CTL mice had normal sinus rhythm and normal AV conduction, however all ZO-1cKO mice had abnormal AV conduction. In addition, surface ECG and telemetric monitoring showed that P-waves were present prior to all QRS complexes in CTL mice, but were not always detected in the ZO-1cKO, suggestive that SA nodal dysfunction might also be present in the ZO-1cKO (Figure 3C, average view 3D).

Figure 3. AV–nodal block is detected in ZO-1cKO mice vs. CTL mice.

(A) Surface electrocardiograms obtained from lightly anesthetized 4-6m ZO-1cKO and CTL demonstrated complete AV dissociation in ZO-1cKO mice, with no abnormalities in CTL. Arrow indicates P-wave. (B) Signal-averaged beat representative of 50 high-resolution beats obtained from ZO-1cKO and CTL mice. P-wave (arrow) and QRS complexes are clearly shown in CTL mouse, but signal-averaged P-waves could not be displayed for the ZO-1cKO tracing because P-waves were not associated in a regular interval related to the QRS complex in the cKO mice. (C) Telemetric ECG analysis in conscious mice. All CTL mice had normal AV conduction. ZO-1cKO mice had high-grade AV block with none of the ZO-1cKO mice showing normal AV conduction. (D) Like the surface ECG, representative signal averaged beats (from 50 beats) is displayed with arrow = P wave. No P-Wave could be identified again for the ZO-1cKO samples. (n=8; M=4, F=4, for both CTL and ZO-1cKO)

Detailed electrophysiological study and optical mapping shows clearly abnormal AV conduction in the ZO-1-cKO mice.

Next we turned to more detailed analyses, first with intracardiac electrocardiograms obtained using an octapolar lead (EPA-800, Millar, Houston, Texas) via the right internal jugular vein. Lightly sedated CTL mice showed normal sinus rhythm, and no evidence of AV conduction abnormalities. In contrast, in ZO-1cKO, P waves occurred irregularly, at a slower rate, and P-waves were dissociated from the QRS complexes. (Figure 4A). At times in these recordings, atrial rates appeared slower than the ventricular rate. Averaged measurements of 20 continuous beats were used to measure P-P and R-R intervals from the various groups, and showed P-P intervals that were significantly increased in ZO-1cKO vs. CTL, but no significant differences in R-R intervals between groups (Figure 4B). To more accurately evaluate for AV block, we performed atrial overdrive pacing to increase the atrial rate and evaluate the subsequent ventricular conduction. CTL mice routinely showed that each atrial paced beat was associated with a corresponding ventricular depolarization. In contrast, atrial pacing in the ZO-1cKO was associated with AV block (Figure 4C).

Figure 4. Intracardiac electrophysiology shows abnormal AV conduction in the ZO-1-cKO mice.

(A) Native intracardiac electrograms were recorded from ZO-1cKO and CTL mice. Shown are typical baseline of CTL and ZO-1cKO mice. Upper panels in each display show native intracardiac recordings of atria (A), ventricle (V) and surface ECG (S) with ladder diagrams below for atrial (A), atrioventricular (AV) and ventricular (V) conduction. Normal conduction was evident in the control mice, but abnormal AV conduction was found in all the cKO mice. (B) P-P and R-R intervals were measured in the mice. P-P intervals showed significant differences in the ZO-1cKO vs. CTL mice. No significant differences were detected in R-R intervals between groups. Normality was evaluated with the Shapiro-Wilk test. Data are represented as mean ± SD. Unpaired two-tailed Student’s t-test with Welch correction was used to assess the p value by groups. Left - Averaged P-P and R-R intervals, Right - Representative Individual P-P intervals from three CTL and three cKO mice show consistent values in CTL and great variance in cKO mice.

(C) In CTL, atrial pacing (arrow=pacing spike) showed atrial capture (triangle) and subsequent ventricular depolarization. However, in ZO-1cKO mice atrial pacing was not associated with a ventricular depolarization. Corresponding intracardiac recordings (Upper panels) and ladder diagrams (lower panels) are displayed from representative animals. (n=11, M=6, F=5 for both groups)Figure 5

Given the telemetry and intracardiac recordings, we next used optical mapping to evaluate heart rhythm and action potential propagation in isolated hearts from 6–8m mice (n=4, each group). CTL hearts demonstrated clear sinus rhythm and normal conduction while ZO-1cKO hearts displayed a slow atrial rate and periodic escape beats. Optical mapping of the epicardial membrane potential revealed that escape beats had normal ventricular activation patterns suggesting appropriate Purkinje activation in both groups (Online Figure IV). ZO-1cKO beats that showed no evidence of atrial activation, showed a ventricular activation pattern consistent with a high junctional origin of the escape beat (Figure 5A). Atrial pacing at 150ms cycle lengths, uncovered AV nodal conduction abnormalities in the hearts from all ZO-1cKO hearts, while all CTL hearts maintained normal AV conduction (n=4 each). All ZO-1cKO demonstrated complete heart block at 100 ms cycle length atrial pacing. (Video demonstrations of the electrical propagation during intrinsic and atrial pacing are in Online. videos 1-4.) No difference was seen in total ventricular activation time between groups, but a trend towards significantly longer atrial activation was seen in the cKO hearts (Figure 5B). Similarly, there were no differences in ventricular conduction velocity (Figure 5C) measured during pacing on the anterior LV surface (BCL=150ms). Atrial conduction velocities measured during intrinsic rhythm showed significantly slowed LA conduction (Figure 5C). Finally, ventricular action potential duration was not different between groups (Figure 5D).

Figure 5. Optical mapping shows a slow atrial rate and periodic ventricular escape beats in ZO-1-cKO mice.

(A) Activation maps of isolated hearts showed anterior activation during intrinsic rhythm. CTL hearts demonstrated clear evidence of sinus rhythm and normal activation patterns. Isolated ZO-1cKO hearts demonstrated AV conduction abnormalities consistent with previously described ECG results. Beats with apparent AV conduction (ZO-1cKO, beat 1) had slow atrial conduction, long AV delays, and a ventricular activation pattern with dual breakthrough sites, suggesting Purkinje activation. Ventricular beats with no apparent atrial activation (ZO-1cKO, beat 2) had similar ventricular activation patterns suggesting a high junctional origin of the escape beat.(B) There was no difference in ventricular total activation time between the two groups, but a trend towards longer atrial activation time in the ZO-1cKO vs. the CTL hearts. (C) The ventricular conduction velocity (LV epicardial pacing; BCL = 150 ms) was not different between the two groups, but left atrial conduction was significantly slower for ZO-1cKO hearts during the intrinsic rhythm (p=0.023).(D) Action potential duration at 80% of repolarization (APD80) was not significantly different between the two groups (LV epicardial pacing; BCL = 150 ms).

Data presented as mean +/− SD. Statistical analysis was performed using the Shapiro-Wilk test to assess normality and an unpaired Student’s t-test to compare. (n=4 for both groups.) BCL, basic cycle length.

No morphological changes were detected in conduction system tissue.

To determine whether absence of ZO-1 causes morphological changes in the SA or AV nodes and associated conduction pathways, we crossed HCN4-GFP-ERT2 mice to the ZO-1cKO, to allow for GFP labeling of conduction system cells. These mice were termed ZO-1cKO-G and ZO-1CTL-G. SA and AV nodal tissue, and conduction system components, labeled by HCN4 nuGFP expression, did not appear varied in the ZO-1cKO vs. CTL (Figure 6).

Figure 6. General morphology of SAN and AV nodes were not changed in the ZO-1cKO hearts vs. CTL.

HCN4-ER-GFP was used to label conduction tissue in isolated and dissected hearts. (A - D) representative CTL heart; E - H representative ZO-1cKO heart, both mated to HCN4-ER-GFP. Dorsal anterior view (A, E), dorsal view (B, F), and right dorsolateral view (C, G) of the intact heart. After dissecting the junction of the right atria and right ventricle, the SA and AV nodes were seen from the dorsal view (D, H). SA and AV nodal tissue was viewed as GFP (green) labeled tissue (SAN, sinoatrial node; AVN, Atrioventricular node; RA, right atria; SVC, superior vena cava; LA, left atria; PA, Pulmonary artery; PV, Pulmonary Vein). (Scale bar: 2mm.)

Expression of ZO-1 associated proteins occurs in a differential manner in atrial vs ventricular tissue when ZO-1 is deleted from cardiomyocytes.

To assess how loss of ZO-1 might impact the localization and the expression of ZO-bound or complexed ICD or ICD-associated proteins in the CM we evaluated expression of a panel of proteins from tissue dissected from the four cardiac chambers of mice at 6m. Western blotting showed no difference in expression levels of vinculin, N-cadherin, Cx43, phospho-Cx43 (368), β & γ-catenin, Cx40 and Cx30.2 (Online Figure V). Immunomicroscopic analyses of Cx43 and desmoplakin (DSP) (Online Figure VI), N-cadherin, β & γ-catenin (Online Figures VII and VIII) qualitatively agreed with the Western blot result. In particular, recognizing that the Cx43 carboxyl-terminus appears to regulate GJ organization along with ZO-1³³, we noted that localization of Cx43 was essentially the same in the ZO-1cKO vs. CTL mice, in support of our finding preserved ventricular function in the cKO hearts.

Cx45 expression in whole atrial tissue was reduced by Western blotting in ZO-1cKO. (Figure 7 A, B), and immunomicroscopy showed alterations in Cx45 in AV nodal cells of ZO-1cKO vs. CTL (Figure 7 C). While there was no clear difference in ZO-2 expression level at the tissue level (Online Figure V), when ventricular CM were isolated, ZO-2 expression was found to be significantly increased in ZO-1cKO as compared to CTL (Figure 7D & E). ZO-2 protein is detected in working myocytes but not in AV node myocytes, and was still not detected in AV node when ZO-1 was deleted (Figure 7F, G and H). CAR protein was also reduced in atrial specimens from cKO vs. CTL detected by Western blot and immunostaining, without changes in ventricular tissue or cells (Online Figure IX, A, B, C). No significant qualitative difference in CAR expression was seen in AV nodal tissue. (Online Figure IXC)

Figure 7. Assessment of ZO-1 associated proteins shows alteration of expression of Cx45, ZO-2 and Nav1.5 in ZO-1cKO vs. CTL samples. (A, B, C) Cx45 was reduced in atrial tissue and AV nodal cells from ZO-1cKO vs. CTL hearts.

(A,B) Western blot (A) shows expression of Cx45 was downregulated in atrial tissue samples of ZO-1cKO vs. CTL mice. (B) Densitometric quantitation of Western blot of CX45, normalized to GAPDH (loading control) shows a significant reduction in ZO-1cKO vs CTL heart samples. Normality was assessed with the Shapiro-Wilk test. Data are represented as mean ± SD. Unpaired two-tailed Student’s t-test with Welch correction was used to assess the p value by groups. (n = 3 for both groups). (C) Cx45 protein was visibly reduced in AV node and working myocytes by immunomicroscopy. Yellow lines show separation of ventricle from AV nodal tissue in c and c’. HCN4 nuclear GFP was used to mark AV nodal cells. (CX45=gray, HCN4 nuclear GFP=green) (Scale bar: 20uM.) (D, E, F, G) ZO-2 was increased in isolated cardiac myocyte samples from ZO-1cKO vs. CTL hearts. (D) ZO-1 was virtually absent while ZO-2 expression was upregulated in isolated ventricular CM samples of ZO-1cKO vs. CTL evaluated by Western blot. (E) Densitometric expression level was determined for ZO-1 and ZO-2 normalized to GAPDH as a loading control. Densitometric quantitation of Western blot of CX45, normalized to GAPDH (loading control) showed a significant reduction in ZO-1cKO vs CTL heart samples. (n=3 for both groups). (F, G, H) ZO-2 protein localizes in the working CM (ventricle) (F) but was not detected in the AV node (G, H), and was still not detected in the AV node when ZO-1 was deleted from CM in ZO-1cKO hearts. Box in panels d and d’ of panel G is shown as a magnified view in all panels of H. (In ‘Merge’ panel of F, G and H: N-Cadherin=green, ZO-2=red, Nuclei DAPI=blue; then in panel G and H - HCN4 nuGFP=Pink and marks AV nodal cells;) (Scale Bar: 20uM in F and G, 5uM in H.)

Finally, we reasoned that a key interacting partner of ZO-1 could be the cardiac sodium channel Nav1.5, given that ZO-1 has a PDZ domain that could interact with Nav1.5, and since Nav1.5 interacts with other proteins also found in the ICD. As shown in Figure 7I, J and Online Figure X, loss of ZO-1 from AV nodal cells (marked by HCN4 nuGFP as well as contactin-2^{34, 35}), caused qualitative alteration in the expression of Nav1.5 is seen, without significant alteration in ventricular cells, as amplified by data in Figure 7K.

DISCUSSION

In this study we focused on the function of ZO-1 specifically in the cardiac myocyte by generating a ZO-1 CM-specific KO mouse model, using a floxed ZO-1 allele mated to a well-characterized αMHC-Cre transgenic mouse^{32, 36} that expresses Cre specifically in CM by E9.5. We found that mice with homozygous CM-specific deletion of ZO-1 are viable, have normal lifespans, and have no evidence of histological or morphological changes in the myocardium except for increased atrial weights. Further, normal contractile function was evident basally, and responses to hemodynamic loading provoked by TAC were similar to that observed in CTL mice. Most strikingly, we found electrophysiologic abnormalities in the ZO-1cKO when evaluated by surface ECGs, conscious telemetry, invasive electrophysiological study and optical mapping. Specifically, we noted defective nodal function and no indication of abnormal Purkinje system activation in ZO-1cKO. Together, these findings suggest unique physiological roles for ZO-1 in the conduction system, as distinct from its role in working cardiac myocytes. A model illustrating this unique role of ZO-1 in conduction system vs. contractile myocardium is shown in Figure 8.

Figure 8. Proposed model illustrating concepts of AV node dysfunction with preserved ventricular function in ZO-1cKO mice.

(A, C) CTL hearts have normal intercellular communication in AV node gap junctions (A) as well as transmission across ventricular intercalated disks (C). (B, D) When ZO-1 expression was deleted from CM in ZO-1cKO AV nodal tissue (B), Cx45 expression is reduced as ZO-1 is lost, though no increase in ZO-2 expression was seen. NaV1.5 expression was also altered in these AV nodal cells when ZO-1 was lost. Together, loss of ZO-1 leads to abnormal AV conduction. (D) In contrast, in the intercalated disk of ZO-1cKO ventricle, ZO-2 expression is increased allowing for ZO-2 function to replace ZO-1, and thus continued normal communication and function of KO ventricular tissue.

ZO proteins were initially detected in TJ² and AJs^3-6 in endothelial and epithelial cells, but have also been identified in CM gap junctions (GJ)^{7, 8}. Though the ZO protein family has three members - ZO-1, −2 and −3, only ZO-1 and ZO-2 are expressed in CM^{9, 37-39}. Data have shown that ZO proteins may regulate GJ size and distribution ⁴⁰ and connect AJs to GJs. While TJ are not present in CM, the area bridging two CM, the ICD, contains AJs and GJs, making the ICD essential for both strengthening the connection and intracellular communication between CMs⁴¹. The Cx-containing GJ allow for rapid conductance of action potentials and thus electrical coupling of CM, so that the working myocardium can contract as a syncytium^42-44. GJ also have an important role in transmission of electrical signaling in the SA and AV nodes⁴⁵. Of note is recent work which identified pathogenic variants in the ZO-1 gene associated with arrhythmogenic cardiomyopathy¹¹.

We confirmed ZO-1 protein localized in the ICDs (Figure 1C) and that it colocalized with Cx43 (Online Figure VI) in the ventricle of the mouse heart. Despite this, given the essential function of the ICD as described above, we were surprised to find that with the loss of ZO-1 expression from the CM ICD in the ZO-1cKO, ventricular contractile function was not altered, either basally or with hemodynamic stress (Online Figure III).

Prior work in our lab had studied the role of ICD and costameric proteins, including CAR and Vcl, a component of the integrin-talin-vinculin complex^{24, 26, 27}. Use of CM-specific CAR KO mice (CARcKO) illustrated that normal expression of this adhesion protein was essential for normal AV node conduction and cardiac function. Likewise, ablation of VCL expression from the CM in VCL-cKO, showed that VCL was necessary for preservation of normal ventricular function, and also that loss of CM VCL led to ventricular arrythmias. CAR was found to form a complex with ZO-1 and Cx45 via its PDZ binding motif, and also when reduced in the CM, led to reduction of both ZO-1 expression in the ICD and Cx45 expression, in the AV node. Similarly, loss of VCL from the CM in the VCLcKO caused loss of ZO-1 from ICD. VCL and ZO-1 were also shown by us to be direct binding partners²⁷. Based on these findings, we hypothesized that normal CM ZO-1 expression would be essential for both cardiac conduction and contractile function.

Surprisingly, we found no alteration in ventricular function here, nor did we find alterations in the expression of either CAR or Vcl expression in the ZO-1cKO ventricle (Online Figure V, IX). We posit that ventricular function was preserved in the absence of ZO-1, due to the increased expression of ZO-2 that occurred in ventricular myocytes.

While no VCL changes were noted in atrial tissue, CAR expression was reduced there, along with our findings of AV and SA nodal abnormalities. Further, expression of the CAR-linked connexin, Cx45, a key conduction system connexin^{46, 47}, was also reduced in cKO atrial tissue compared to control. It is important to note that nodal tissue is typically included in right atrial samples, as we evaluated the expression of these proteins.

Intriguingly, while ZO-2 expression increased in the working ventricular myocardium with ZO-1 loss, this was not evident in the AV node (Figure 7D - 7H). Therefore, the current data demonstrates that the decrease in AV node ZO-1 that was observed in CAR deficient hearts, and in the current ZO-1cKO, leads to AV conduction delay.

In addition to Cx45, we evaluated the expression of multiple connexins that have been associated with cardiovascular function and conduction, as well as other proteins expressed in the ICD. Expression of adherens junction proteins (N-cadherin, α-catenin, Y-Catenin as well as Vcl), and gap junction proteins (Cx40, Cx43 and Cx30.2) proteins were not significantly different in the ZO-1cKO vs. CTL (Online Figure V). While immunolocalization of all these Cx proteins could be instructive, it is worthwhile to note that combined knockout of Cx30.2 and Cx40 in mice did not alter AV impulse propagation⁴⁸.

Cx45 is of particular interest in conduction abnormalities since it is one of the main connexins in the SA node and the AV node, with limited expression in the atria and ventricles⁴⁹. Cx45 is strongly expressed in the early embryonic myocardium and is required for cardiac development^{47, 50}. Homozygous Cx45-deficient mice are embryonic lethal at around E10 due to heart failure with conduction block⁴⁷, heterozygous Cx45 KO mice display AV-septal defects and slowed AV-conduction⁵¹. In the adult heart, Cx45 is expressed mainly in the SA and AV nodes and His-Purkinje system, but has also been shown to be expressed in atrial and ventricular myocytes⁴⁹. It is not essential for survival of adult mice as inducible, CM-specific deletion of Cx45 leads to delayed AV conduction, but normal ventricular function⁵².

Cx45 has also been linked to nonimmune familial AV-block in man⁴⁶ and a disease-causing Cx45 mutant (p.R75H) was found in 2 unrelated families who presented with progressive AV block, and ultimately atrial standstill, without ventricular conduction abnormalities⁴⁶. Further, two groups have been shown that Cx45 interacts directly with ZO-1, but these studies were not carried out in CM^{53, 54}. Still, these results are in line with our findings since ZO-1 loss resulted in reduced expression of Cx45.

Importantly, we found that with loss of ZO-1 from CMs, qualitative expression of the sodium channel NaV1.5 was reduced in AV nodal cells, but not ventricular CMs. (Figure 7I-K and Online Figure X). These results suggest that ZO-1 is necessary for proper expression of Nav1.5 in the AV nodal cells. While variants / mutations in NaV1.5 has been linked to a variety of cardiac diseases, included amongst these are correlation with SA nodal abnormalities (sick sinus syndrome) and also other progressive cardiac conduction system disorders, suggesting that perhaps ZO-1 may be a component of these disease states^{55, 56}.

We respect that there are several limitations to our study. Experiments are required to further elucidate the basis for the unique findings in our model. Since we propose that ZO-2 is be able to replace ZO-1 function in the working CM, future studies where both ZO-1 and ZO-2 protein expression is deleted from all CMs will be necessary to discern the general function of ZO proteins in the ventricle. We are also aware that studies have begun to show how abnormalities of ICD proteins, particularly those that contain PDZ domains, can alter sodium and potassium channel function and predispose to arrhythmias and conduction disturbances^57-61. Though we have shown here that NaV1.5 expression is changed in AV nodal cells deficient in ZO-1, future functional studies are necessary. Also, since ZO-1 complexes with CAR and binds VCL, experiments are needed to fully explain how these three proteins function individually and together, to affect cardiac conduction and ventricular function.

In sum, we have identified a unique role for ZO-1 in cardiac conduction that has not been previously recognized. Future studies, currently underway, are necessary to fully define the role of ZO proteins in all CM in various regions of the heart.

Major Resources Table.

In order to allow validation and replication of experiments, all essential research materials listed in the Methods should be included in the Major Resources Table below. Authors are encouraged to use public repositories for protocols, data, code, and other materials and provide persistent identifiers and/or links to repositories when available. Authors may add or delete rows as needed.

Animals (in vivo studies) Genetically Modified Animals
	Species	Vendor or Source	Background Strain	Other Information	Persistent ID / URL
Parent - Male	Mouse	N/A	ZO-1 f/+, Cre
Parent - Female	Mouse	N/A	ZO-1 f/f or ZO-1 f/+
Parent - Male	Mouse	N/A	HCN4-GFP-ERT2
Antibodies
Target antigen	Vendor or Source	Catalog #	Working concentration	Lot # (preferred but not required)	Persistent ID / URL
For immunostaining
ZO-1	Millipore-Sigma	MATB11	1:50
ZO-2	Abgent	ABO11279	1:25
Desmoplakin	AbD Serotec	2722-5204	1:200
PECAM/CD31	R & D Systems	AF3628	1:100
Connexin 43	Invitrogen	71-0700	1:500
β-catenin	Abcam	ab16051	1:100
Y-catenin	ThermoFisher	PA5-17320	1:50
N-cad	BD Biosciences	610920	1:100
contactin-2	R & D Systems	AF439	1:50
donkey anti-Rabbit Alex-488	Jackson Lab	711-545-152	1:100
anti-Rabbit Alex-647	Jackson Lab	711-605-152	1:200
cy3 donkey anti-Rat	Jackson Lab	712-165-153	1:200
cy5 donkey anti-Mouse	Jackson Lab	715-175-151	1:200
cy3 donkey anti-goat	Jackson Lab	705-165-147	1:200
DAPI	ThermoFisher	D1306	10ug/ml
Fluoromount-G™	eBioscience	00-4958-02
For western-Blot
ZO-1	Millipore-Sigma	MATB11	1:1000
Desmoplakin	AbD Serotec	2722-5204	1:1000
Connexin 43	Invitrogen	71-0700	1:5000
p-Connexin 43	Cell Signaling Technology	3511	1:1000
Vinculin	Sigma	V9131	1;10000
N-cad	UIHB	MNCD2	1:1000
CAR	Santa Cruz Biotechnology Inc.	SC-15405	1:1000
β-catenin	Abcam	AB16051	1:1000
Υ-catenin	Santa Cruz Biotechnology Inc.	SC-7900	1:1000
ZO-2	Abgent	ABO11279	1:1000
Connexin 30.2	Invitrogen	40-7400	1:1000
Connexin 40	Invitrogen	36-5000	1:1000
Connexin 45	Invitrogen	41-4700	1:1000
GAPDH	Millipore-Sigma	MAB374	1:10000
donkey anti-Mouse	Jackson Lab	615-035-214	1:2500
donkey anti-rabbit	Jackson Lab	711-035-152	1:2500
donkey anti-rat	Jackson Lab	712-035-153	1:2500
donkey anti-goat	Jackson	Lab	705-035-003	1:2500

NOVELTY AND SIGNIFICANCE.

What Is Known?

• Intercalated disks (ID) that interconnect cardiac myocytes (CMs), contain three types of junctions, with two providing a means for strong cellular attachment (adherens junctions (AJ) and desmosomes) and a third (gap junctions (GJ)) directing essential electrical coupling and transmission of small metabolites between cells.

• There are three Zonula Occludens (ZO) proteins – ZO-1,2 and 3, with only ZO-1 and 2 expressed in AJ and GJ of the CM IDs.

• CM ZO proteins are important for GJ remodeling and control of GJ size.

What New Information Does This Article Contribute?

• CM-specific deletion of the mouse ZO-1 gene was performed and produced mice termed ZO-1 CM-specific knockout (ZO-1cKO) that are viable, have normal lifespan and normal ventricular contractile function.

• The ZO-1cKO mice showed varied degrees of atrioventricular (AV) block, including complete heart block.

• With loss of ZO-1 from CMs, ZO-2 was seen to increase in ventricular myocytes but not in atrioventricular nodal cells.

• ZO-1 loss from AV node tissue lead to altered microscopic expression of the sodium channel NaV1.5 in AV node tissue.

Interconnections of CMs via IDs is essential to maintain integrity of the myocardium and allow synchronous electrical activity and muscular contraction. Studies show that defective ID proteins can lead to cardiac contractile dysfunction or rhythm disturbances. ZO proteins are found in the ID, but their function in CMs is not fully understood. Given the known binding of ZO-1 to other CM proteins, we produced ZO-1cKO mice to explore how loss of CM ZO-1 would alter heart function. ZO-1cKO mice were viable, had normal lifespan and showed normal contractile function, even with hemodynamic challenge. Yet, electrocardiograms, intracardiac electrophysiology and optical mapping studies showed evidence of AV block, including complete heart block, in these mice. With loss of ZO-1 from CMs, ZO-2 expression increased significantly in ventricular CMs but not in AV node tissue, suggesting it could substitute for ZO-1 in working myocardium, but not in AV node. Further, in ZO-1cKO, qualitative alteration of the sodium channel NaV1.5, was seen in AV node. Thus ZO-1 protein appears dispensable for function of the working CM and has a unique physiological role in cardiac nodal tissue. This increased understanding of CM ZO-1 provides novel information relevant to cardiac conduction system function and disease.

Nonstandard Abbreviations and Acronyms:

AJ

adherens junction

AV

atrioventricular

AVN

atrioventricular node

BW

body weight.

CAR

coxsackievirus and adenovirus receptor

CARcKO

CAR cardiomyocyte-specific KO mice

CM

cardiac myocytes

CX

connexin

DSP

desmoplakin

E

embryonic day

ECG

electrocardiography

ECHO

echocardiography

EP

electrophysiology

GJ

gap junction

HR

heart rate

ICD

intercalated discs

PSD5/90

post-synaptic density protein

TJ

tight junction

VCL

vinculin

ZO

zonula occludens