A Peptide Mimetic of the Connexin43 Carboxyl-Terminus Reduces Gap Junction Remodeling and Induced Arrhythmia Following Ventricular Injury

Abstract

Rationale

Remodeling of connexin-43 (Cx43) gap junctions (GJs) is linked to ventricular arrhythmia.

Objectives

A peptide mimetic of the carboxyl-terminal (CT) of Cx43, incorporating a Post-synaptic-density/Disks-large/ZO-1 (PDZ)-binding domain, reduces Cx43/ZO-1 interaction and GJ size remodeling in vitro. Here, we determined: 1] Whether the Cx43-CT mimetic αCT1 altered GJ remodeling following left-ventricular (LV) injury in vivo, 2] If αCT1 affected arrhythmic propensity, and 3] The mechanism of αCT1 effects on arrhythmogenicity and GJ remodeling.

Methods and Results

A cryoinjury model generating a reproducible wound and injury border zone (IBZ) in the LV was used. Adherent methylcellulose patches formulated to locally release αCT1 (<48-hours) were placed on cryoinjuries. Relative to controls, Cx43/ZO-1 colocalization in the IBZ was reduced by αCT1 by 24-hours post-injury. Programmed electrical stimulation ex vivo and optical mapping of voltage-transients indicated that peptide-treated hearts showed reduced inducible-arrhythmias and increased ventricular depolarization rates 7-9 days post-injury. At 24-hours and 1-week post-injury, αCT1-treated hearts maintained Cx43 in intercalated disks (IDs) in the IBZ, whereas by 1-week post-injury controls demonstrated Cx43-remodeling from IDs to lateralized distributions. Over a post-injury time-course of 1-week, αCT1-treated IBZs showed increased Cx43 phosphorylation at serine368 (Cx43-pS368) relative to control tissues. In biochemical assays, αCT1 promoted phosphorylation of serine368 by PKC-ε in a dose-dependent manner that was modulated by, but did not require ZO-1 PDZ2.

Conclusion

αCT1 increases Cx43-pS368 in vitro in a PKC-ε-dependent manner and in the IBZ in vivo acutely following ventricular injury. αCT1-mediated increase in Cx43-pS368 phosphorylation may contribute to reductions in inducible-arrhythmia following injury.

Introduction

Gap Junctions (GJ) are fundamental to a stable heartbeat. The cardiomyocytes of the mammalian heart are connected by large numbers of GJs, providing the basis for conduction of action potentials from one cell to another^1-3 The main GJ protein present in mammalian ventricular muscle is connexin43 (Cx43)^4,5. In 1991, studies of infarcted ventricular myocardium in humans⁶ and dogs⁷ provided the first indication that Cx43 GJ distribution is profoundly altered in a narrow zone of surviving muscle bordering the necrotic injury or scar - the infarct border zone. Since publication of these results, a number of groups have confirmed the close linkage between GJ remodeling in the infarct border zone and the propensity of ventricles to develop reentrant arrhythmias^8-15. Typical of this border zone remodeling is a loss of GJs from intercalated disks (IDs) at end-to-end contacts between cardiomyocytes and redistribution of Cx43 to lateral domains of sarcolemma ^6,13,16,17.

We reported previously on a 25 amino acid (aa) peptide comprised of a cell-permeabilization sequence linked to the last 9 aas of the Cx43 carboxyl-terminus (CT)¹⁸. This Cx43-CT mimetic peptide, now referred to as αCT1¹⁹, was shown to bind to the second Post-Synpatic Density-95/ Discs-large/ZO-1 (PDZ2) domain of Zonula Occludens-1 (ZO-1)¹⁸, thereby inhibiting a protein-protein interaction occurring between ZO-1 and Cx43^18,20-22. In association with inhibiting this interaction, αCT1 effected remodeling of GJs in HeLa cells expressing Cx43 and cultured neonatal rat ventricular myocytes¹⁸. Specifically, it was determined that αCT1 treatment prompted increases in the size of membrane-localized GJs, increases in the p1 and p2 phosphorylation bands of Cx43 on Western blots and re-distribution of Cx43 from non-junctional to GJ-associated pools of the protein¹⁸. Consistent with these results in vitro, it was subsequently found that αCT1 increased the size of Cx43 GJs between epidermal cells proximal to skin wounds in vivo²³.

Since αCT1 inhibited remodeling of GJ size and distribution of Cx43 in vitro, we sought to establish whether αCT1 had similar effects on GJ remodeling in the myocardium in vivo. It was of specific interest to determine whether αCT1 would inhibit remodeling of GJs in the arrhythmia-prone tissues bordering ventricular injuries . To test this hypothesis, a cryoinjury model was developed to generate a uniformly sized injury and injury border zone(IBZ) in the left ventricle(LV) of mouse hearts²⁴. The cryoinjury protocol enabled avoidance of the variability in wound size and shape that often confounds interpretation of other mouse models of experimental injury, such as coronary artery ligation. Localized treatment of LV injuries with αCT1 resulted in decreased propensity to develop arrhythmia in response to programmed stimulation and increased conduction velocity. These physiologic changes were concomitant with stabilization of Cx43 at IDs in the IBZ, and with increases in phosphorylation at a serine at amino acid position 368 (S368) of Cx43 – a protein kinase C consensus site²⁵. Consistent with this observation in vivo, assays in vitro, indicated that αCT1 was sufficient to prompt dose-dependent increases in PKC-ε mediated phosphorylation of Cx43 at S368.

Methods

Methods and reagents are detailed in the Online Data Supplement located at http://circres.ahajournals.org.

LV Cryoinjury and Treatment

12-24-week female CD1 mice were given a left thoracotomy at the 4^th intercostal space to expose the anterior free wall of the LV. A cryoinjury was generated by epicardial exposure to 5-seconds contact with a liquid-N₂ chilled 3mm circular flat-tip probe. The injury was immediately covered with an adherent methylcellulose patch containing either αCT1(100μmol/L), reverse peptide (100μmol/L), vehicle (1xPBS) or DiI. The thoracotomy was closed, and the mice were allowed to recover prior to any further study.

Triphenyl-Tetrazolium-Chloride (TTC) Staining 48-Hours Post-Injury

Cryoinjured mouse hearts were harvested 48-hours after cryoinjury, stained in 1% TTC, washed in 1xPBS, and fixed overnight at 4°C. The epicardial area and transmural depth of injuries were imaged and measured as described in supplement.

Microscopy

Immunohistochemistry and analyses were performed as described in the supplement. Isolated adult feline cardiomyocytes were stained with Fluorescein-conjugated streptavidin, and imaged for peptide uptake. Hearts were harvested 24-hours, 48-hours, and 1-week, then bisected along the base-apex midline of the injury with half embedded in paraffin and half processed for frozen sectioning. Paraffin-embedded sections from the middle of injuries 48-hours post-injury were stained with H&E or antibodies against Cx43 and atrial-myosin-light-chain-2 (Mlc2a) to visualize the IBZ. Frozen sections from 24-hour and 1-week hearts were used for analysis of ID localization, and Cx43/ZO-1 and Cx43-pS368/total-Cx43 colocalization. Sections were stained with TRITC-wheat germ agglutinin (WGA), TO-PRO-3 nuclear stain, and antibodies to Cx43, ZO-1, and Cx43-pS368.

Western blotting

Western blot analysis was undertaken on hearts 2, 4, 6, and 48-hours post-injury and exposure to αCT1. Samples were processed as described in supplement. Peptide eluted from methylcellulose patches was detected in injured and remote myocardium on blots against biotin and Cx43. For assessment of Cx43-pS368 and total Cx43, samples were immunoblotted with rabbit pS368 Cx43 antibody and re-probed for total Cx43.

Electrophysiological Studies

Hearts were harvested, washed, and the aorta was cannulated for modified Langendorff procedure 7-9 days post-injury. The atrio-ventricular (AV) node was ablated. Then, hearts were subjected to premature and overdrive pacing protocols with simultaneous recording. Scores for severity of arrhythmia were assigned by 3 investigators blind to treatment as outlined in supplemental table 1.

Optical Mapping Studies

Hearts were harvested and prepared as above, except the AV node was left intact so that activation under sinus rhythm could be studied. Hearts were stained with Di-4-ANEPPS, treated with Cytochalasin-D and optically mapped as described in the data supplement.

In vitro kinase assay

PKC-ε phosphorylation of GST-Cx43-CT substrate at serine 368 PKC was evaluated under PKC assay conditions as detailed in supplement. Reactions were Western blotted for pS368 Cx43 and re-probed for total Cx43 as above. Assay was repeated 4 times.

Data Collection and Analysis

All data were collected blind to whether animals received treatment. Chi Square was used to determine significance (p<0.05) of arrhythmia frequency comparisons. ≥11 animals were tested in each treatment/control group in arrhythmia frequency tests. For determination of significance (p<0.05) of differences in arrhythmia severity a Kruskal-Wallis test with posttesting was used. For other comparisons, significance (p<0.05) was determined by one-way, two-way or repeated-measure ANOVA and posttesting. The ANOVA model used for each data set is provided in figure legends. Data from 3-9 animals (figure legends provide animal numbers for specific experiments) were pooled for each treatment/control group, tested for normality and presented as means±SEM. No strong evidence of divergence (p>0.05) from normality was found.

Results

Standardized LV Cryoinjury Model

Commonly used arterial ligation models for generating ventricular wounds can result in injuries with significant variability in size and shape. We developed a method that produced an injury of consistent size and shape (roughly hemispheric) on mouse LV (fig. 1A and B)²⁴. Our method was based on one described by Van den Bos and co-workers who used a liquid-N₂-cooled cryoprobe²⁶. We modified this protocol to include probe pre-chilling and non-transmural injury, as opposed to the more severe transmural injury they produced. Non-transmural injuries provided extended uniform regions of IBZ - a tissue of particular interest in our study. By varying exposure time to a liquid-N₂–cooled circular probe, it was determined that a 5-second probe application to the LV provided a repeatable acute cryolesion spanning approximately 60% of the free-wall - as assessed from TTC staining (fig. 1A and B).

Figure 1.

A&B)Whole mount and cross-section of a TTC-stained heart 48-hours after 5-second exposure to cryoprobe. C) Montage of H&E section through ventricle 48-hours after standard 5-second exposure. Inset: Cx43 immunolabeled (green) sister section at IBZ. D) Montage of section from 8-week scar immunolabeled for Mlc2a (red). Dapi nuclear blue signal. Inset: Cx43 immunolabeled sister section at IBZ. Inset dashed-lines represent border of injury/scar. Scale C=500 μm, D=25μm, C&D inset=10μm.

48-hours after LV cryoinjury, hematoxylin&eosin histochemistry (fig. 1C) and Cx43 immunolabeling (inset fig. 1C) showed a discrete border between necrotic tissue and viable myocardium. In hearts 8-weeks after cryoinjury, a narrow, but definite sector of IBZ could be observed (fig. 1D). This 10 to 20 cardiomyocyte-wide IBZ adjacent to the injury could was also marked by increased immunolabeling for ventricular-myosin-light-chain-2 and atrial-myosinlight-chain-2 (Mlc2a fig. 1D). Similar to reports on Cx43 at infarcts^6,13,16,17, Cx43 showed disorganized/lateralized patterns of distribution in the IBZ (inset fig. 1D).

Localized Delivery of αCT1 to Cryoinjured LV

Figure 2A illustrates the Cx43-CT peptide-αCT1 (alpha-Connexin-Carboxyl-Terminal 1). As previously reported¹⁸, αCT1 has a two-part design comprising an amino-terminal (NT) antennapedia cell-permeabilization domain linked to the CT-most 9 aas of Cx43 (RPRPDDLEICT). We have previously demonstrated that αCT1 inhibits Cx43/ZO-1 interaction via binding to the ZO-1 PDZ2 domain and not other ZO-1 PDZ domains¹⁸. For a control, we used a peptide (Reverse fig. 2A) consisting of the antennapedia permeabilization sequence and the Cx43-CT 9 aa sequence reversed (IELDDPRPR-CT). In previous work, we showed that this control peptide is inactive with respect to inhibiting Cx43/ZO-1 interaction¹⁸. 30μmol/L αCT1 was rapidly taken up into neonatal¹⁸ and adult ventricular cardiomyocytes (fig. 2B and C).

Figure 2.

A) Schematic representation of αCT1 and the inactive reverse control peptide (Rev).B&C) Streptavidin-fluorescein-labeled adult cardiomyocytes exposed to biotin-αCT1 or vehicle for 4-hours. D-G) Cryoinjured mouse heart 2-hours post-surgery and application of a DiI impregnated methylcellulose patch. The injury is outlined by dashed-line on the intact heart and transected view (F&G). H) Streptavidin and Cx43 blots from myocardial lysates of injured and remote myocardium at 2, 4, 6, and 48-hours post injury and treatment (n=4 mice), and lysates from remote myocardium spiked with 2.5, 12.5, and 25ng of αCT1. Scale C=5μm, G=2mm.

In a previous report, αCT1 was delivered to skin wounds in a 25% pluronic F127 gel²³. We found that gel would not remain localized to the LV cryolesion during surgery. To achieve sustained localized delivery of peptides to injured LV for up to 48 hours, we reformulated αCT1 in a “dry” methylcellulose patch (patch) that adhered stably to cryoinjured tissue. Figure 2D-G shows a cryoinjured heart harvested 2-hours after application of a methylcellulose patch containing DiI. Whole mount images of the epicardial surface show injury location (fig. 2D) and the location of the applied patch (fig. 2E). A transverse cut at level of red line on 2D demonstrates local diffusion DiI(red) into underlying LV in the region of injury (fig. 2E and G).

Samples were harvested from injured and remote ventricular myocardium at 2, 4, 6, and 48-hours following application of the patch containing αCT1 (100μmol/L) to cryoinjured hearts. αCT1 was detected on Western blots from 2, 4, and 6-hour LV injury samples after application, but by 48-hours, αCT1 was minimally detectable (fig. 2H). No peptide was found in myocardial samples remote from cryoinjury at any point after treatment (fig. 2H).

αCT1 decreases Cx43/ZO-1 colocalization 24-hour post-injury

Analysis of ZO-1 interaction with Cx43 was performed on IBZ tissues 24-hour post-injury. Initial attempts to dissect sufficient IBZ tissue from cryoinjuries for immuno-precipitation of Cx43/ZO-1 were unsuccessful. However, in Hunter et al., a method of quantification of ZO-1 colocalization with Cx43 GJs was shown to correlate with biochemical assays of Cx43/ZO-1 interaction¹⁸.

Colocalization analysis was undertaken in IBZ at 24-hours post-injury in vehicle (fig. 3A, 3B, 3B’), reverse (fig. 3C, 3C’), and αCT1-treated (fig. 3D, 3D’) hearts. Cx43/ZO-1-association levels in the IBZ varied significantly (p<0.05) between αCT1, Rev and Veh groups in a multiple comparison test by one-way ANOVA. Rev and Veh controls did not differ significantly, and when combined showed significantly higher levels of Cx43/ZO-1 colocalization (p<0.05) in comparison to the αCT1 group in posttesting(fig. 3E).

Figure 3.

A) Low magnification of an IBZ showing a representative field. IBZ is identified by dashed-line in A. B-D) Images from the IBZ of Veh, Rev, and αCT1-treated hearts stained for Cx43 (green) and ZO-1 (red). B’-D’) High magnification insets from images B-D. White represents a mask of Cx43 and ZO-1 signals of half-maximal intensity or above that colocalize within pixels. E) Graphical representation of the percent of colocalization between Cx43 and ZO-1 for animals from each treatment/control group. n≥5 (mice/group). Treatment vs control p value on E is from a One-Way ANOVA in which αCT1 group was compared to a combined control comprising Veh and Rev mice. Scale bar B=10μm.

αCT1 increases GJ intercalated disk localization in the IBZ

We have previously reported in echocardiography studies of cryoinjured hearts that the αCT1 effect on LV dilation was most pronounced 1-week after cryoinjury²⁴. We examined the cellular distribution of Cx43 24-hours and 1-week post-injury to determine if there was a concomitant change in GJ localization at intercalated disks(ID) following cryoinjury and treatment. Figures 4A-F shows IBZ (fig. 4A-C) and ventricular tissue remote from injury (fig. 4D-F) in αCT1 (fig. 4A,D), Rev (fig. 4B,E), and Veh (fig. 4C,F) hearts 24-hours post-injury. Cx43 is immunostained green, with cell membranes stained red by TRITC-WGA to assist discrimination of end-to-end contacts (IDs) between cardiomyocytes. Dashed lines mark the boundary between the cryoinjury necrotic area and the IBZ (in figs. 4A-C&G-I).

Figure 4.

A-C) Representative confocal images of IBZ 24-hours post-cryoinjury from αCT1(A), Rev (B), and Veh (C) groups stained for Cx43 (green) with nuclear TO-PRO-3 (blue) and cell membrane labeling WGA (red). D-F) Images from remote myocardium of the same hearts respectively. G-L) Representative images of IBZ from 1-week cryoinjured hearts (G-I), and remote myocardium (J-L). Dashed-lines in IBZ images denote injury border. Arrows in G denote GJs at IDs, and arrowheads in H and I denote lateralized Cx43 GJs. M) Graph of percent Cx43 ID localization in the IBZ for αCT1, Rev, and Veh hearts (#:p<0.00005 24hr/1wk, ‡:p<0.0001 24hr/1wk, ø:p<0.001 αCT1/Rev or αCT1/Veh). N) Graph of Cx43 ID localization in the remote myocardium. n=4 mice/group/timepoint. p values for posttest comparisons within IBZ and remote ventricular regions were generated from two-way ANOVAs (treatment × time). Scale bars C and I=10μm

At 24-hours, differences in Cx43 distribution between treatment and controls were not easily visualized. Consistent with this, statistical comparisons of Cx43 levels at IDs at 24-hours in the IBZ or remote myocardium indicated no significant differences between the αCT1 group and the Rev or Veh groups (fig 4M&4N).

By 1–week (fig 4G-L), differences in GJ remodeling between treatment and control groups were more readily observed. Figure 4G-I shows magnified images from IBZ of αCT1 (fig. 4G), Rev (fig. 4H), and Veh (fig. 4I)-treated hearts 1-week post-injury. Corresponding images of remote myocardium from the same hearts are also shown (fig. 4J-L). Arrows in figure 4G denote immunolabeled ID-associated Cx43 GJs in the IBZ of a αCT1-treated heart. Arrowheads in figure 4H and 4I denote lateralized GJs in IBZs from controls.

Pairwise comparison indicated that Cx43 localized to IDs was significantly decreased in both Rev and Veh IBZs relative to the αCT1 group (fig. 4M Ø:p<0.001). The fraction of Cx43 at IDs in αCT1 group decreased by only 4% (p>0.1), between 24-hour and 1-week timepoints (fig. 4M), indicating maintenance of Cx43 distribution in IDs. By contrast, ID-localization of Cx43 decreased approximately 55% in Rev and Veh control groups over the first week (fig. 4M #:p<0.00005, ‡:p<0.00008). No significant treatment- or time-dependent effects on Cx43 localization at IDs occurred in remote myocardium (fig. 4N), suggesting that αCT1 effects on GJ remodeling were localized to IBZ tissues.

αCT1 inhibits arrhythmias in the cryoinjured heart

IBZ GJ remodeling is associated with increased arrhythmogenecity following infarction²⁷. Accordingly, our next step was to determine if there was a difference in arrhythmic propensity following treatment. As before, mice received standardized cryoinjury and patches, containing αCT1, Rev, and Veh. At ~1-week (7-9 days), cryoinjured and uninjured control hearts were isolated and kept viable by retrograde aortic perfusion. There was no notable difference in survival between the groups up to ~1 week post-injury.

Once stable, isolated hearts were subjected to two induced-arrhythmia protocols-premature, s2-s3 pacing and overdrive pacing. Uninjured hearts demonstrated minimal propensity to develop induced arrhythmias, after stabilization in the bath. However, 75% of cryoinjured hearts developed arrhythmias when subject to either premature or overdrive pacing. Arrhythmias ranged in severity from ectopic beats to ventricular fibrillation (fig. 5A-E).

Figure 5.

A-E) Representative tracings from premature ventricular pacing protocol on isolated perfused hearts illustrate no arrhythmia (A), 3 spontaneous PVCs (B), resolving tachycardia (C), sustained tachycardia (D), and fibrillation (E). The green numbers in figures A-D label the s1, s2, and s3 stimuli. The blue arrows in figure A denote the stimulated ventricular action potential. F) Numbers of hearts displaying arrhythmias (dark red and blue colors) that were unsustained (left-hand bar graph) or sustained (right-hand bar graph) in αCT1, Rev and Veh groups following pacing. Lighter red and blue colors within bars indicate number of hearts within groups in which arrhythmia was not induced by pacing. G) Graphical representation of the median severity of arrhythmia for the three treatment groups (p<0.02 αCT1/Rev, p<0.02 αCT1/Veh). n≥11 (mice/group). p values for comparisons of frequency and severity of induced arrhythmia were generated from Chi square and Kruskal-Wallis tests and posttests respectively.

Figure 5A shows a recording from a cryoinjured heart stimulated using the premature pacing protocol with a s1 interval of 200ms, s2 of 80ms, and s3 of 30ms. Each of the stimuli are followed by an action potential (AP, blue arrows fig. 5A). Figure 5B shows a recording of a cryoinjured heart where the stimulus is followed by 3 ectopic APs. Tachycardia was defined as more than 3 consecutive ectopic APs.

We further classified the tachycardia by whether it was unsustained or sustained irrespective of additional stimuli (fig. 5C and D), or if it progressed to fibrillation (fig. 5E). For the premature pacing protocol, αCT1-treated hearts showed a significantly decreased number of sustained (or more pronounced) arrhythmias compared pairwise to either the Rev (p<0.007) or Veh control groups (p<0.03)(fig. 5F). All hearts were ranked for severity of arrhythmia, according to a 10-point scale (supplemental Table 1). The severity of arrhythmia was significantly reduced in αCT1-treated hearts compared to either Rev (p<0.02) or Veh (p<0.02) groups (fig. 5G).

αCT1 Increases Rate of Depolarization of Cryoinjured hearts

It has been shown that GJ remodeling can potentiate arrhythmias by decreasing conduction velocity through the IBZ¹³. Previously described methods were used for optical mapping of electrical activation in cryoinjured hearts^28-32. Figure 6A is a light microscope image of a frame from the ultrafast digital camera taken prior to imaging of Di-4-ANEPPS fluorescence. Notice that the cryoinjury stains with decreased intensity compared to surrounding ventricle.

Figure 6.

Resting light image from αCT1-treated heart (A) with corresponding APs (B) and dV/dt (C) for selected points (1-7 on A). Representative isochronal maps of αCT1 (D), Rev (E), and Veh (F) treated cryoinjured hearts 1-week post-injury. Black circles denote the approximate position of the IBZ, and color scale with 10 colors each corresponding to 1ms. G) Graph of the time vs % LV depolarization for each treatment group (‡:p<0.05 αCT1/Rev, ø:p<0.05 αCT1/Veh). n≥5 (mice/group). p values for posttest comparisons between groups were generated by repeated measures ANOVA.

Figure 6B shows optically recorded action potentials (APs) corresponding to points 1-7 on figure 6A, and 6C shows the first derivatives of these APs, which in turn, were used to generate isochronal maps. Figures 6D-F are isochronal maps from αCT1 (6D), Rev (6E), and Veh (6F) treatment/control hearts. Black circles on maps indicate the cryolesion, as determined by light microscopy (e.g., fig. 6A).

Blind assessment of activation sequences suggested that depolarization often originated in the IBZ, and this tendency was reduced in the αCT1-treated group. Consistent with this, in 50% of peptide-treated hearts, earliest activation breakthrough coincided with the cryo-injury. By comparison, in Rev and Veh control ventricles the spatial coincidence between breakthrough and injury was higher at 71% and 75% respectively. This difference in breakthrough location tendency between treatment and controls was not significant.

Isochronal maps indicated that conduction velocity may be increased in αCT1-treated hearts. However, irregularities in wave fronts caused by the cryoinjury complicated the identification of primary vectors necessary for calculation of longitudinal and transverse conduction velocities (figs. 6D-F). To address this, we took a simple approach of calculating the time required to depolarize given fractional areas of the ventricular epicardium.

The rate of ventricular depolarization was significantly faster in αCT1-treated hearts than either Rev (‡:p<0.05) or Veh (ø:p<0.05) controls (fig. 6G). For example, the time required to depolarize 50% of the epicardium for the αCT1 group was interpolated from figure 6G as 3.45ms. By comparison, it took 4.94ms and 5.17ms to depolarize 50% of the ventricle in Rev and Veh control groups.

αCT1 causes an acute increase in Cx43-pS368 levels in the IBZ

Reports by others have indicated that phosphorylation of a serine at amino acid residue 368 (Cx43-pS368) of Cx43 is associated with maintenance of the connexin at IDs in hearts subjected to low-flow ischemia³⁴. We sought to determine whether Cx43-pS368 levels were also associated with increased Cx43/ID localization following αCT1 treatment.

Visual assessment of myocardium indicated Cx43-pS368 in immunolabeled sections at 24-hours post-injury (figs. 7A-F and supplemental fig. IA-I). The insets in figures 7A-F show representative GJs in IBZ and remote myocardium from each group. The white overlay on the green Cx43 signal in these insets represents overlap of Cx43-pS368 with a pan-Cx43 marker (total Cx43). The IBZ (fig. 7A-C) showed an elevation in Cx43-pS368 24-hours post-injury compared to remote myocardium (fig. 7D-F, arrows= colocalization). The largest increase in IBZ Cx43-pS368 occurred in IBZs from αCT1-treated hearts (fig. 7A). The increased ratio of immunolabeled Cx43-pS368 to total Cx43 in αCT1-treated IBZs at 24 hours post-injury was significant in ANOVA posttesting (p<0.04) relative to Rev and Veh control groups combined (supplemental figure IJ).

Figure 7.

A-C) Representative montaged microscopic fields from IBZ of αCT1, Rev, and Veh group hearts stained for total Cx43 (green), Cx43-pS368 (red), and nuclei (dapi). Asterisks in B-C denote intersection of images from adjacent microscopic fields. High magnification insets (bottom right A-C) illustrate regions labeled for total Cx43 that colabel with antibody to Cx43-pS368. Lower magnification insets (A’-C’) from above images illustrate single channel signal for Cx43 (green) and Cx43-pS368 (red). Dashed-lines in A-C and A’-C’ denote IBZ/injury transition. D-F) Corresponding images of remote myocardium from hearts with high magnification insets, and arrows denoting Cx43-pS368/Cx43 colabeling. G) Western blot showing a representative set of samples probed for Cx43-pS368 and total Cx43 at 0 (uninjured), 4, 6, 24, 48, 168 hours post-cryoinjury and treatment with αCT1 or control. H) Graphical representation of the average ratio of Cx43-pS368 to total Cx43 for all time points in each treatment group compare to uninjured hearts for IBZ and remote myocardium. p<0.03 αCT1/control in IBZ overall and p<0.04 αCT1/control 6 hour IBZ. No significant treatment (ns) effect occurred in remote myocardium. n=3 mice/group/time point. p values for comparisons within IBZ and remote ventricular regions are posttests generated from two-way ANOVAs (treatment × time). Scale bar=20μm.

To further quantify this effect, we performed Western blot assays of Cx43-pS368 normalized to total Cx43 levels and GAPDH in the IBZ and remote ventricular myocardium at 4, 6, 24, 48-hours and 1-week after cryoinjury and treatment (fig. 7G). Pairwise comparison indicated that normalized Cx43-pS368 levels were significantly lower in control versus αCT1-treated IBZs (p<0.03) over the 1-week time course (fig. 7H). Moreover, the increase in Cx43-pS368 in treatment vs control was significant at 6-hours after the cryoinjury (fig. 7H p<0.04). It was concluded that a localized application of αCT1 delivered just after cryoinjury results in increased levels of IBZ Cx43-pS368 over a 7-day period compared to controls.

No significant treatment effect was observed on normalized Cx43-pS368 in ventricle remote from the injury between groups over the post-injury timecourse. However, remote myocardium did show a time-dependent decrease (p<0.00002) in Cx43-pS368 over the 7 days that was similar in both αCT1-treated and control groups (fig. 4H). This time-dependent effect was not seen (p=0.65) in IBZ tissues (Fig 4H), indicating that the cryo-injury may have a regional effect on sustaining Cx43-pS368 level independent of treatment.

αCT1 increases Protein Kinase C epsilon (PKC-ε) mediated phosphorylation of GST-Cx43 at S368

As αCT1 increased levels Cx43-pS368 following injury in vivo, we next determined whether peptide affected Cx43 phosphorylation via modulation of PKC-ε, the kinase responsible for Cx43 phosphorylation at serine 368 (fig. 8). Standard reactions incorporating purified enzyme (human PKC-ε) and substrate (GST-Cx43-CT – aas 255-382) were undertaken in vitro in the presence of αCT1, isolated ZO-1 PDZ domains and control peptides.

Figure 8.

A-C) In vitro assays of PKC-ε catalytic competence for phosphorylating GST-Cx43-CT at serine 368 in the presence of αCT1, PDZs and control peptides. A) Blots of Cx43-pS368 (upper panel) and total Cx43 (lower panel) in mixtures (from left to right) including a no kinase control that includes substrate (GST-Cx43-CT aas 255-382), but no PKC-ε enzyme (-PKC-ε), and then blot lanes of reaction solutions containing both substrate and enzyme, as well as increasing concentrations of αCT1 (i.e., at 5, 10, 20, 50, 100, and 200μmol/L). B) Blots of Cx43-pS368 (upper panel) and total Cx43 (lower panel) in reaction mixtures (from left to right) containing PKC-ε and GST-Cx43-CT, as well as lanes for increasing concentrations of GST-PDZ2 (i.e., at 0, 27, 54, 108, 270, and 540nmol/L). Right hand most lane on (B) is -PKC-ε. C) Blots of Cx43-pS368 (upper panel) and total Cx43 (lower panel) in reaction mixtures (from left to right) containing both PKC-ε and GST-Cx43-CT, 100μmol/L inactive reverse control peptide (rev – 1^st and 6^th lanes), 100μmol/L αCT1 (2^nd and 7^th lanes), and 100μmol/L αCT1 plus 100nmol/L PDZ2 in lane 5. The 4^th and middle lane on (C) is -PKC-ε. Each assay was repeated 3 or more times.

Assays of Western-blotted Cx43-pS368 to total Cx43 indicated that αCT1 caused a dose-dependent increase in phosphorylation of Cx43-CT by PKC-ε, relative to baseline (fig. 8A, -PKC-ε). By contrast, increasing concentrations of PDZ2 had no effect on PKC-ε-mediated phosphorylation of GST-Cx43-CT (fig. 8B). The reverse control peptide also did not increase PKC-ε-phosphorylation of GST-Cx43-CT (fig. 8C). Interestingly, although PDZ2 had no direct effect, addition of PDZ2 (but not PDZ1 or PDZ3) with αCT1-containing reactions caused a reduction in Cx43-pS368 below that of reverse control and αCT1 minus control levels (fig. 8C). Thus, αCT1 alone appeared sufficient to increase PKC-ε activity in vitro, but this activity could be abrogated by PDZ2.

Discussion

Here, we show that treatment of a standardized cryoinjury of the mouse LV with a peptide (αCT1) incorporating the last 9 amino acids of the Cx43-CT decreased stimulated arrhythmias such as ventricular tachycardia or fibrillation. Associated with these effects on cardiac electrical function, the peptide reduced colocalization between ZO-1 and Cx43, and gap junction (GJ) lateralization in arrhythmia-prone tissues of the IBZ. However, αCT1 effects on GJ remodeling may not have resulted solely from affects on ZO-1/Cx43 interaction. Instead, a novel target for αCT1 was identified–protein kinase C epsilon (PKC-ε). αCT1 was sufficient to increase PKC-ε phosphorylation of a consensus target on the Cx43-CT - the serine residue at amino acid position 368 (S368). Consistent with the in vitro results, levels of phosphorylated S368(pS368) in IBZ were elevated significantly above controls within hours of exposure to αCT1. While ZO-1 was not necessary for enhanced PKC-ε activity, PDZ2 inhibited αCT1 enhancement of pS368 levels, indicating potential regulatory interplay between PKC-ε and ZO-1 at the Cx43-CT.

Our observation that αCT1 enhancement of Cx43-pS368 in the IBZ was associated with stabilization of Cx43 at the ID parallels observations made by others in ischemia models. Ek-Viktorin and co-workers reported that in response to transient ischemia, Cx43-pS368 remained stabilized at IDs, even as total Cx43 underwent remodeling to lateral myocyte surfaces³⁴. Kardami and colleagues showed that ischemic pre-conditioning(IPC) or FGF-2 treatment increased PKC-ε-mediated Cx43 phosphorylation of serines at positions 262 and 368, and inhibited remodeling of Cx43 away from IDs in response to an ischemic insult³³. This dual S368 and S262 modification was referred to as “the P*Cx43 state”. They concluded that the “P*Cx43 state” was necessary for pre-conditioning, as mutant mice incompetent to undergo these Cx43 phosphorylations did not develop cardioprotection in response to IPC or FGF-2.

A number of other chemical agents have been reported to have effects on myocardial Cx43-S368 phosphorylation. Miura and co-workers showed that the delta-opioid ligand induced PKC-ε-mediated phosphorylation of Cx43 at S368, and that activation of the delta-opioid receptor is an IPC-like adjunct mechanism of infarct size limitation³⁵. Dhein and colleagues determined that a short peptide (AAP10) prevented ischemia-induced Cx43 dephosphorylation and remodeling of Cx43 GJs away from myocyte IDs³⁶. In a parallel to the results with αCT1 as reported here, AAP10 was found to attenuate ischemia-induced slowing of activation³⁷. A 6 amino acid peptide related to AAP10 called rotigaptide has been reported to suppress dephosphorylation of Cx43-pS368 and also a second serine at position 297 on Cx43³⁸. Moreover, similar to αCT1, AAP10 and rotigaptide have been shown to improve cardiac function in animal models, significantly extending time to ischemia-induced asystole and demonstrating anti-arrhythmic effects^38-40.

It is presently unclear if rotigaptide is efficacious as an anti-arrhythmic agent in humans. Phase IIa clinical trials involving intravenous infusion of rotigaptide in myocardial infarction patients were begun in 2005, but terminated before efficacy end-points⁴¹. The provision of a sustained, localized concentration of αCT1 to the injury and IBZ via the adherent methylcellulose patch was necessary for the efficacy of our approach. In light of this experience with αCT1, a revised clinical protocol, involving focal delivery of rotigaptide directly to infarcts, may be worth considering.

AAP10 and rotigaptide promote maintenance of PKC-dependent phosphorylation of Cx43 by inhibiting dephosphorylation³⁷. This suggests a difference in mechanism to αCT1, which prompts increased levels of pS368 via upregulation of PKC-ε catalytic activity. Also unlike αCT1, Kjolbye et al found that rotigaptide had no effect on GJ remodeling from IDs following ischemic insult⁴². These workers reported the anti-arrhythmic mechanism of rotigaptide was that it suppressed transition from concordant to discordant alternans following ischemia, perhaps via promoting GJ coupling-a conclusion supported by electrophysiological studies of communication between myocyte pairs⁴³.

Other workers have reported that Cx43-S368 phosphorylation is associated with reduced cell-to-cell electrical conductance by GJ channels⁴⁴. Consistent with these data, Burt, Lampe and colleagues determined that S368 phosphorylation prompts a 2-fold reduction in electrical conductance of unitary Cx43 channels³⁴. It was also shown that this decrease in single channel electrical coupling was associated with a paradoxical increase in selective permeability of GJs. The authors concluded, that in addition to anti-arrhythmic benefits, the formation of distinct tissue compartments defined by Cx43 phospho-status might have implications for repair following myocardial infarction. The localized increase in pS368 in the IBZ and improvement in cardiac function that we observe in response to αCT1 may be consistent with this hypothesis.

αCT1 appears to induce a preconditioned-like state in injured myocardial tissues via local enhancement of PKC-ε phosphorylation of Cx43-S368. The PKC-ε consensus domain on Cx43 that includes S368 also incorporates sequences at the CT of Cx43 involved in ZO-1 binding^45,46. Physical interaction between Cx43 and PKC-ε has been confirmed in co-immunoprecipitations from myocardial tissues and is increased by IPC and pharmacophores activating PKC-ε³⁵. These studies raise the prospect that αCT1 induction of a preconditioned-like state could involve modulating normal processes of co-operativity or competition between Cx43-interacting proteins. Consistent with this, we found that ZO-1 PDZ2 inhibited αCT1 enhancement of pS368 by PKC-ε. It also remains to be determined whether αCT1 enhances the actions of PKC-ε by allosteric interaction with PKC-ε or by making the Cx43 CT more available for S368 phosphorylation.

The CT of PKC-ε displays significant homology with the Cx43-CT. This similarity includes a class II PDZ-binding domain (x-Φ-x-Φ where Φ= hydrophobic aas and x=any aa) Asp-Leu-Met-Pro, a similarity that to the best of our knowledge has not been remarked upon previously in the literature. The PKC-ε-CT is involved in intramolecular interactions, which are necessary for catalytic competence⁴⁷. It remains to be shown that PKC-ε binds ZO-1 or other PDZ-containing proteins via its CT. Indeed, we could find no evidence that over range concentration that PKC-ε was effected by PDZ2. Interestingly though, the combined presence of αCT1 and PDZ2 consistently decreased pS368 levels below that even of vehicle control and the control peptide. Further work will be required to determine the mechanism of this combinatorial effect that results in a loss of ability of PKC-ε to phosphorylate Cx43 at S368.

A caveat here is that cryoinjury has limited clinical relevance. In the clinic, cryoablation is regularly used in the left atrium and regions with proximity to the atrio-ventricular node. The strength of our cryo-model lies in its ability to provide a consistent IBZ that is not complicated by topological heterogeneities associated with ischemia reperfusion or ligation injuries. We suggest that our model may provide an approach for identifying molecular targets in injured myocardium that cause arrhythmia and LV dysfunction. It is encouraging that αCT1 reduces inducible arrhythmia following cryoinjury in mouse hearts. Tests of inducible arrhythmia have been widely used as a prognostic indicator for morbidity and mortality in patients who have recently suffered from a myocardial infarction⁴⁸. This being said, in future studies it would be useful to investigate propensity to develop spontaneous arrhythmia and long-term survival. The findings with the standardized cryoinjury model will also need to be verified in ischemic injury before definite conclusions can be drawn on possibilities for ameliorating infarction-induced arrhythmia. To this end, studies of αCT1 in myocardial infarction in large animal models should be informative.

Non-standard Abbreviations and Acronyms

GJ

Gap Junctions

Cx43

connexin43

ID

intercalated disks

ZO-1

Zonula Occludens-1

PDZ2

second Post-Synpatic Density-95/ Discs-large/ZO-1 domain

CT

carboxyl-terminal

aa

amino acid

IBZ

injury border zones

LV

left ventricle

S368

Cx43 aa serine 368

Cx43-pS368 or pS368

Cx43 phospho-isoform, p-serine368

TTC

Triphenyl-Tetrazolium-Chloride

WGA

wheat germ agglutinin

PKC-ε

Protein Kinase C epsilon