Mice aged 3 – 4 months were used for these experiments. All studies were performed in accordance with the regulations of the Institutional Animal Care and Use Committee of the New York University School of Medicine (New York, NY). The C57BL/6J strain was used for all experiments except for immunoblotting of ventricular lysates and quantification of mRNA expression, for which CD-1 mice were used. Mice were anesthetized with inhaled isoflurane (4 vol% induction, 1.5 vol% maintenance; Baxter, Deerfield, IL) and immobilized on a heating pad set to 37°C. Electrocardiographic signals from limb leads were monitored during the experiment and recordings were obtained prior to the initiation of pacing, during pacing (to document 1:1 capture) and following cessation of pacing.

Following recording of the baseline electrocardiogram, a 1 cm midline incision was made in the epigastric region. A custom-designed UE-GM1 cardiac stimulating electrode with a 200 μm monopolar platinum tip (Frederick Haer & Co., Bowdoinham, ME) mounted on a micromanipulator (Fine Science Tools, Inc., North Vancouver, BC, Canada) was inserted through the diaphragm directly into contact with the surface of the beating right ventricle. Electrode contact was ensured by monitoring the electrocardiographic activity. Pacing was performed with a Model 2352 Programmable Stimulator (Medtronic, Minneapolis, MN). Output was set at twice the stimulating threshold in all animals, with a pulse width of 1.0 ms. Pacing cycle lengths were set and maintained at 10–15% above the underlying heart rate, to ensure 100% capture for 6 hours. To serve as controls, age-, sex- and strain-matched mice were prepared and anesthetized with the pacing electrode placed into contact with the heart identically to the paced cohort, but the programmable stimulator was not switched on ("sham-paced").

Electrocardiographic signals were recorded as previously described. Electrocardiographic intervals were calculated from leads I, II and III recorded at the beginning of each study and after the completion of the pacing protocol.

Echocardiography was performed according to a previously described protocol [14] on paced and sham-paced mice (n = 6 in each group), prior to and during or immediately following six hour pacing and sham protocols. Mice were imaged using a Philips HDI 5000 echocardiography machine equipped with a 15 MHz linear probe.

Programmed electrical stimulation (PES) was carried out in paced and sham-paced mice at baseline and after the pacing protocol as described previously [13]. The PES protocol consisted of trains of eight beats at pacing cycle lengths of 100 ms and 80 ms, followed by single extrastimuli for the determination of ventricular effective refractory period (VERP) and double extrastimuli to assess for inducible arrhythmias. PES data represent the average of two trials at separate sites on each heart.

Following the pacing procedure, hearts destined for immunostaining were rapidly excised and frozen in Tissue Tek OCT compound (Sakura Finetek USA, Inc., Torrance, CA). Five micron-thick sections were cut in an HM 560 cryostat (Microm, Walldorf, Germany) at -20°C, placed onto Superfrost/Plus microscope slides (Fisher Scientific, Pittsburgh, PA) and fixed in acetone. We selected frozen sections to evaluate the left ventricle from basal, mid-ventricular, and apical regions. We defined the basal region as those sections proximal to the papillary muscles; mid-ventricular regions were those sections of LV with clearly identified papillary muscles; and apical regions were distal to the papillary muscles. We also evaluated apical and basal sections of the right ventricle.

Sections were blocked and then double-stained with a custom-made rabbit polyclonal anti-Cx43 antibody [15] at 1:1000 and wheat germ agglutinin to visualize myocyte borders as previously described [16]. For visualization of cadherin staining, frozen sections were incubated with a rabbit anti-pan-cadherin antibody (1:5000; Sigma), followed by a Texas Red-conjugated goat anti-rabbit secondary antibody (Jackson ImmunoResearch Laboratories) and a FITC-conjugated anti-Cx43 antibody. Stained sections were visualized on an Axioskop 2 Plus fluorescence microscope. Images were collected using uniform exposure settings for each staining run on an Axiocam camera with AxioVision 4.30 software (Carl Zeiss, Munchen-Hallbergmoos, Germany).

Images of immuno-stained sections from paced and sham-paced hearts were obtained from blinded slides and digitally archived for offline analysis. Blinded image files were uniformly thresholded by eliminating signal-free areas above and below the distribution of intensity values using the histogram function on Adobe Photoshop. Digital image processing was then performed according to previously established techniques [17], again in a blinded fashion, with NIH Image J to determine Cx43 and cadherin signal area as a percentage of total tissue area, as well as Cx43 plaque size and the number of Cx43 plaques per 40× field.

For the evaluation of total protein levels by immunoblotting, endocardial ventricular tissue from paced and sham-paced hearts was prepared by Dounce homogenization in lysis buffer supplemented with Complete protease inhibitor cocktail (Roche, Mannheim, Germany). Endocardial samples were prepared by mounting the excised LV free wall in an OCT block, separating the endocardial, mid-myocardial and epicardial regions by sectioning on an HM 560 cryostat and homogenizing as above. The inner third of the LV free wall was considered the endocardial region, middle third mid-myocardial and outer third epicardial. Protein concentrations were determined by Bradford assay performed in triplicate and equal loading was confirmed with coomassie staining. Proteins were electrophoresed on 10% SDS-PAGE gels and transferred onto nitrocellulose blots (Bio-Rad Laboratories, Hercules, CA). Immunoblots were blocked followed by incubation with appropriate primary antibodies directed against Cx43, cadherin (see above), Cx45 [18], Cx40 (Alpha Diagnostics, San Antonio, TX) and GAPDH (Chemicon/Millipore, Billerica, MA). HRP-conjugated secondary antibody (Santa Cruz Biotechnology, Santa Cruz, CA) was then applied, followed by HyGlo chemiluminescent processing (Denville Scientific, Metuchen, NJ) and autoradiography. At least two separate experiments were quantified by scanning the autoradiograms on a Bio-Rad Gel Doc GS 800 and calculating band intensity using Quantity One software (Bio Rad, Hercules, CA). Connexin band intensities were normalized to the relative intensity of the corresponding GAPDH band for each sample. Results were expressed as a percentage of matched sham-paced controls.

To determine sarcolemmal and non-sarcolemmal Cx43 concentrations, fractionation of samples was performed as described [15,19]. Briefly, heart samples were Dounce homogenized, centrifuged at 500 × g for 10 minutes to remove insolubles and layered over a 45% sucrose cushion. After centrifuging at 7000 × g for 20 minutes, the supernatant (non-sarcolemmal fraction) was separated from the cloudy layer immediately overlying the sucrose (sarcolemmal fraction). Protein concentrations in each fraction were determined by Bradford assay performed in triplicate and equal loading was confirmed with coomassie staining. The resulting fractions were analyzed by SDS-PAGE and Western blotting.

For immunoprecipitation, 50 μg of total heart lysate was incubated with polyclonal anti-Cx43 antibody. After addition of protein A-agarose-immobilized protein beads (Roche) to the samples, the protein A suspension was centrifuged at 5000 × g and the supernatant was removed. The protein A beads were washed in IP buffer and resuspended in loading buffer prior to incubation at 100°C and analysis by SDS-PAGE. The resulting blots were incubated with a monoclonal antibody directed against ubiquitin (FK2, Biomol) and processed as above.

To isolate total RNA, the LV free walls of paced and sham-paced hearts were excised, mounted in OCT and snap frozen in liquid nitrogen. Endocardial and epicardial thirds were collected by sectioning through the LV free wall as described above. RNA was isolated using Trizol (Invitrogen, Carlsbad, CA) according to the manufacturer's recommendations. RNA quality was verified with the Agilent 2100 Bioanalyzer (Agilent Technologies, Santa Clara, CA). The concentration of RNA was determined by the Quant-iT RiboGreen RNA Assay Kit (Invitrogen, Carlsbad, CA). Real time quantitative PCR was performed using the ABI Prism 7700 Sequence Detection System (Applied Biosystems, Foster City, CA). Cx43 primer sequences, located in the 3' untranslated region, were as follows: forward primer sequence was 5'-GTGCCGGCTTCACTTTCATTAAG-3'; reverse primer sequence was 5'-ACTGACCTCGCGGAACC-3'; probe sequence was 5'-TTTCTCTCCACGGGTCT-3'. All data were normalized to Cyclophilin A (primer sequences as per [20]) and expressed as fold change compared to the sham-paced controls. Negative controls were performed for each of the samples, in which reverse transcriptase was not added prior to RNA quantification. Results per animal represent the mean of six measurements of each endocardial and epicardial sample.

Data are expressed as mean ± SEM. Quantitative immunofluorescence data were compared between groups with ANOVA using StatView (SAS Institute, Inc., Cary, NC). Data from electrocardiography, echocardiography, PES, immunoblot densitometry and qRT-PCR were compared between groups with unpaired two-tailed t-tests (Microsoft Excel). Electrocardiographic indices, echocardiographic measurements and ERP values obtained during or after the pacing protocol were compared to baseline measurements with paired two-tailed t-tests (Microsoft Excel). P < 0.05 was considered statistically significant.

For ease of pacing and reliability, we used the subdiaphragmatic approach, which allows for placement of the stimulating electrode on the heart without necessitating sternotomy, thoracotomy, mechanical ventilation or manipulation of the vasculature [13]. For the investigation of probe placement after the pacing protocol was completed, a midline sternotomy was performed at the end of four consecutive experiments. In all four animals examined in this fashion, the stimulating electrode traversed the diaphragm and made contact with the right ventricular (RV) surface without puncturing the heart (Figure 1A). Local structures (including the abdominal viscera, diaphragm and lungs) were not disrupted by the electrode. Thus, the pacing model employed in this study can be considered as single-lead RV epicardial pacing.

Since our objective was to model altered activation rather than induce tissue damage [21], we limited the stimulator output to approximately twice the stimulating threshold (mean stimulator output was 0.366 ± 0.011 mA). The average threshold prior to pacing was 0.165 ± 0.007 mA (n = 65). In a subset of eight mice, the threshold was determined after pacing. In this group the average post-pacing threshold was 0.162 ± 0.026 mA (p = NS compared to pre-pacing values). The unchanged threshold values after pacing underscore the stability of the experimental preparation.

To determine whether short-term pacing influenced synchrony of cardiac contraction, we examined mice prior to and during pacing using echocardiography. The septal-to-posterior wall motion delay (SPWMD), an echocardiographic index of dyssynchronous contraction [22,23], increased significantly from 7.8 ± 2.5 ms before pacing to 27.9 ± 2.4 ms during pacing (p < 0.01; n = 6 each; Figure 1B and 1C). Delay in onset of systolic wall thickening measured from the septal to the posterior walls also increased significantly during pacing (12.8 ± 1.1 ms at baseline vs. 31.6 ± 1.4 ms during pacing, p < 0.001). As expected, surface electrocardiograms demonstrated substantial widening of the QRS complex during pacing, suggesting aberrant intraventricular electrical conduction (Figure 1D). Fractional shortening was marginally decreased from 44.2 ± 2.0% at baseline to 37.2 ± 2.8% during pacing (p < 0.05; Table 1).

Although global ventricular function was slightly diminished during pacing, fractional shortening was not significantly altered after cessation of pacing in the paced mice compared to matched sham-paced controls or compared to their baseline values. Similarly, left ventricular dimensions and wall thicknesses were not significantly changed in paced mice compared to matched sham-paced controls or baseline values (Table 2). Thus, short-term pacing was associated with dyssynchrony during pacing and a slight decrement in ventricular function that normalized immediately after cessation of pacing.

To characterize the effect of short-term pacing-induced dyssynchrony on gap junction distribution in the mouse heart, we paced mice for 6 hours (3–4 half-lives of Cx43) [24-26] and assessed Cx43 localization using immunofluorescence staining. In sham-paced mice, Cx43 immunosignal was arrayed in heterogeneously distributed membrane-associated plaques throughout the myocardium (Figure 2A and 2B). In the sham-paced hearts, Cx43 signal area as a percentage of total tissue area was greater in the endocardial region of the left ventricular (LV) free wall (9.12 ± 1.20%) than in the epicardial region (6.82 ± 0.40%; p < 0.05; Figure 2M). After 6 hours of pacing, however, Cx43 immunosignal area was reduced specifically at the endocardial region of the LV free wall (6.76 ± 0.60%) in comparison to the endocardium of sham-paced controls (p < 0.05), without a detectable change in epicardial Cx43 signal (6.41 ± 0.30%; Figure 2C and 2D and Figure 2M; data are from n = 6 sham and 6 paced hearts).

Post-hoc analysis showed that the Cx43 signal area was decreased significantly only at the basal LV endocardial segments in paced hearts compared to sham-paced hearts (6.38 ± 0.50% in paced vs. 9.36 ± 1.28% in sham, p < 0.01). There were trends at the apical (6.76 ± 0.89% in paced vs. 7.97 ± 0.96% in sham) and mid-ventricular (7.15 ± 0.69% in paced vs. 9.10 ± 1.59% in sham) LV endocardial segments toward decreased Cx43 signal area in paced hearts, although these were not statistically significant.

There was no difference in Cx43 signal area between sham (3.40 ± 0.25%) and paced (3.91 ± 0.45%) RV sections. Specifically at the RV apex where the stimulating electrode was placed, there was no statistically significant difference in Cx43 signal area between sham (3.51 ± 0.46%) and paced hearts (4.16 ± 0.53%; Figure 2P and 2Q).

Trends in endocardial gap junction plaque sizes mirrored changes in Cx43 immunosignal area, although differences in plaque size among groups were not statistically significant (Figure 2N). There was no significant difference in the number of gap junction plaques per 40× field in sham and paced endocardial and epicardial regions (Figure 2O). Thus, the epicardial-to-endocardial gradient of increasing Cx43 signal area observed in the LV free wall of sham-paced hearts, which has been described previously by other investigators[17], was eliminated after 6 hours of pacing due to a focal decrease in Cx43 immunosignal area at the endocardium of the LV free wall.

Adherens junctions, like gap junctions, are concentrated at the intercalated discs of adult cardiac myocytes. Since the expression pattern of Cx43 immunosignal was altered by short-term pacing, we investigated whether immunosignal area of cadherin, a critical component of the adherens junction with a half life of 5–6 hours [27,28], was also changed. Unlike the epicardial-to-endocardial gradient of increasing Cx43 immunosignal area, there was no difference in cadherin immunosignal area between epicardial and endocardial regions in sham-paced control hearts (Figure 2E and 2F). Furthermore, cadherin signal area in both epicardial and endocardial regions remained unchanged after 6 hours of pacing (p = NS; n = 6 sham and 6 paced hearts; Figure 2G and 2H). This suggests that the reduction in endocardial Cx43 signal area with pacing occurs in the absence of changes in the distribution of the adherens junctions.

To determine whether short-term pacing was associated with redistribution of Cx43 signal away from the intercalated disc, we performed colocalization analysis of Cx43 and cadherin immunosignal (Figure 2I–L). The percentage of Cx43 immunosignal colocalizing with cadherin in the paced hearts was unchanged in comparison to sham-paced controls. This suggests that short-term pacing in the mouse heart does not affect the process whereby Cx43 protein is directed to sites of adherens junction aggregation in the sarcolemma [29].

To further assess the effect of short-term pacing on targeting of Cx43 protein along the sarcolemma, we used a subjective scoring system in which the degree of Cx43 immunosignal at the lateral myocyte borders in sections from paced and sham-paced hearts was compared. Paced and sham heart sections were blindly assigned scores ranging from 0 (no lateralization) to 3 (extensive lateralization). We found that sham sections demonstrated extensive subjective lateralization of Cx43 signal (2.75 ± 0.25) and that the extent of lateralization did not appear significantly different in the paced hearts (2.40 ± 0.40, p = NS; n = 4 sham and 5 paced).

Since Cx43 immunosignal area was reduced at the LV endocardium after 6 hours of pacing, we investigated the effects of short-term pacing on Cx43 mRNA levels in the LV free wall endocardial and epicardial regions of paced and sham-paced control hearts. As shown in Figure 3, qRT-PCR demonstrated a 2.5 ± 0.1-fold down-regulation in expression of Cx43 mRNA at the endocardial region of the paced hearts compared to the sham-paced group (p < 0.001; n = 3 sham and 3 paced). Furthermore, a 2.0 ± 0.1-fold reduction in Cx43 mRNA expression was observed at the epicardial regions of the paced hearts compared to sham-paced controls (p = 0.001; n = 3 sham and 3 paced). Thus, six hours of pacing at rates just above those of sinus rhythm resulted in significant reductions in Cx43 mRNA levels at the LV free wall.

Since short-term pacing in the murine heart resulted in down-regulation of Cx43 mRNA levels in the LV and decreased immunosignal of Cx43 at the endocardium, we investigated whether endocardial Cx43 protein abundance was affected by pacing. Surprisingly, there was no significant change in Cx43 protein expression in the paced LV endocardial region compared to sham-paced controls (n = 11 sham and 11 paced; Figure 4A). Densitometry values of the slower migrating (upper) Cx43 bands expressed as a ratio to the lower-most band was unchanged in paced compared to sham-paced samples, suggesting that the overall phosphorylation status of Cx43 was unaffected by short-term pacing. Expression levels of cadherin, Cx40 and Cx45 by immunoblotting were also statistically unchanged after pacing in the LV free wall endocardial myocardium (n = 7 sham and 7 paced; Figure 4B). Thus, despite significant reductions in Cx43 immunosignal area and mRNA abundance in the paced hearts, overall Cx43 protein expression levels are preserved.

Since total endocardial Cx43 protein levels in the paced hearts were unchanged despite reduced mRNA levels and immunosignal area, we considered that pacing might have resulted in redistribution of Cx43 from the sarcolemma into non-sarcolemmal pools. Fractionation of heart samples demonstrated significantly reduced Cx43 abundance in the sarcolemma-enriched fraction isolated from paced hearts compared with sham-paced controls (40.4 ± 8.7% decrease in the paced hearts; p < 0.05; n = 11 sham and 12 paced hearts; Figure 5A). In contrast, Cx43 levels in the supernatant (non-sarcolemmal fraction) increased by 104.0 ± 35.5% (p < 0.05; Figure 5A). These findings suggest that the intracellular distribution of cardiac Cx43 may be altered by short-term pacing with a reduced proportion that is localized to the sarcolemma.

With reduced Cx43 mRNA levels despite preserved total protein abundance and an apparent redistribution of Cx43 immunosignal from the sarcolemma, we predicted that the process of degradation of gap junction moieties might be disrupted, resulting in an accumulation of ubiquitinated Cx43 in the paced hearts. We tested this prediction by immunoprecipitating Cx43 in lysates from five sham-paced and five paced hearts, followed by gel electrophoresis and blotting for ubiquitin. We observed a substantially increased expression of ubiquitinated forms in the paced hearts (Figure 5B). These findings suggest that while Cx43 immunosignal area and mRNA levels are decreased, overall Cx43 protein levels in the paced hearts are preserved, possibly due to a combination of decreased production, altered intracellular distribution and disrupted degradation of Cx43.

We next investigated whether pacing-induced changes in Cx43 expression were associated with persistent alterations in electrocardiographic and electrophysiologic indices. After cessation of pacing, we did not detect significant changes in electrocardiographic parameters in the paced mice when compared with baseline values or sham-paced controls (Table 3). There was also no significant difference between paced and sham-paced mice in refractoriness as determined by PES after the pacing period (Table 3). Furthermore, PES with single and double extrastimuli did not result in sustained ventricular arrhythmias in any of the mice tested. Thus, there was no effect of short-term pacing at rates just above sinus in wild type mice on electrocardiographic or electrophysiologic parameters. These data suggest that the observed pacing-induced changes in Cx43 distribution are not sufficient, in the absence of other insults to the myocardium, to cause a demonstrable arrhythmogenic effect.