Ischemic Preconditioning Protects Against Gap Junctional Uncoupling in Cardiac Myofibroblasts

Abstract

Ischemic preconditioning increases the heart's tolerance to a subsequent longer ischemic period. The purpose of this study was to investigate the role of gap junction communication in simulated preconditioning in cultured neonatal rat cardiac myofibroblasts. Gap junctional intercellular communication was assessed by Lucifer yellow dye transfer. Preconditioning preserved intercellular coupling after prolonged ischemia. An initial reduction in coupling in response to the preconditioning stimulus was also observed. This may protect neighboring cells from damaging substances produced during subsequent regional ischemia in vivo, and may preserve gap junctional communication required for enhanced functional recovery during subsequent reperfusion.

Keywords:

• Cardiac myofibroblasts
• connexin43
• coupling
• gap junction
• ischemia
• preconditioning

INTRODUCTION

Gap junction plaques contain aggregates of protein units known as connexins. The connexins form channels that link the cytoplasms of two cells (1). These pathways provide for the cell-to-cell diffusion of small (< 1200 Da) hydrophilic molecules, including ions, amino acids, nucleotides, and second messengers (e.g., calcium, cAMP, cGMP, and IP₃). Connexins are thought to affect cell growth control and embryonic development, as well as the transmission of electrical signals between cells. In heart, gap junction channels form the low-resistance pathways that are essential for impulse propagation and thus for coordinated contractions (2). In adult mammalian myocardium four different connexins, connexin43 (Cx43), Cx40, Cx45 and Cx37, have been identified (3). In the rat and human heart, the most widespread gap junction protein is Cx43.

In the normal heart, two-thirds of the cell population is composed of nonmuscle cells, the majority of which are cardiac fibroblasts (4, 5, 6). Although the main role of fibroblasts, consisting of synthesis and maintenance of the mechanical scaffold for cardiomyocytes, is undisputed, their involvement in electrophysiological processes within the working myocardium is less clear. Electrical coupling between cardiomyocytes and fibroblasts was demonstrated several decades ago in cell cultures (7) and, more recently, in intact tissue (8). Fibroblasts are coupled with gap junction channels containing Cx43. In cell cultures, it has been shown that single fibroblasts are capable of synchronizing contraction among individual cardiomyocytes (9, 10, 11) and that these contractions are accompanied by synchronous membrane potential fluctuations in the interconnecting fibroblasts, suggesting the presence of electrical communication (7, 10, 11). External stress causes cardiac fibroblasts to change their phenotype (12, 13). These altered cells are referred to as myofibroblasts because they express several smooth muscle markers, including α -actin, myosin heavy chain-B, and tropomyosin (13, 14, 15). The characteristics of myofibroblasts in myocardial infarcts have not been extensively investigated. Investigation of the coupling between myofibroblasts is important in order to understand their contribution to the healing of myocardial infarct and in order to design therapeutic interventions aimed at improving postinfarct cardiac recovery.

Ischemic preconditioning (PC) is the phenomenon by which a brief period of ischemia followed by reperfusion greatly increases the heart's tolerance to a subsequent longer ischemic period. Several triggers and mediators of PC have been identified (16), whereas the final end effector is still unknown. Recently, it has been hypothesized that the protective effect of PC against lethal injury secondary to ischemia-reperfusion can at least be partially explained as a consequence of its effect on gap junction-mediated intercellular communication (17).

Cx43 contains multiple serine, threonine, and tyrosine phosphorylation sites, and phosphorylation of Cx43 is thought to be an important regulatory mechanism for gap junctional communication. Studies in cultured rat neonatal cardiomyocytes have identified Cx43 phosphorylation sites for several kinases, including protein kinase (PK) A, PKC, PKG, and mitogen-activated protein (MAP) kinase (18, 19, 20). It is believed that activation of PKC plays a major role in PC (21). PKC is involved in signaling complexes with many proteins, among them Cx43. Cx43 may be phosphorylated by PKC-α, -γ, -δ or -ε (22, 23, 24), but the involvement of a particular PKC isoenzyme might vary, even between systems that express the same complement of isoenzymes (23). Activated PKCε colocalizes with Cx43 in rat cardiomyocytes and might modulate gap junction transmission characteristics and intercellular communication (25, 26). In noncardiac cells, PKC activation has been reported to lead to rapid inhibition of gap junctional communication (27, 28, 29, 30). In cardiac cells, however, it has been found that PKC activation increases gap junction electrical coupling (18).

Schwanke et al. (31) provided strong evidence for the involvement of Cx43 in the protective effect of preconditioning. This study convincingly shows that a 50% reduction in Cx43 content is sufficient to abolish the infarct-sparing effect of PC. Recent studies (32, 33, 34) suggest that PC modifies the phosphorylation status of Cx43. Schulz et al. (32) concluded that PC increases the colocalization of protein kinases with Cx43 and preserves the phosphorylation of Cx43 during sustained ischemia in in vivo pig hearts. Jain et al. (34) found that inhibition of K_ATP channel during PC caused marked dephosphorylation and internalization of Cx43, whereas PKC inhibition allowed Cx43 to become dephosphorylated but prevented internalization of Cx43. They suggested that K_ATP channel activation either directly or indirectly inhibits activation of phosphatases that dephosphorylate Cx43 during ischemia. This would result in less dephosphorylated Cx43 in gap junctions that could be acted on by PKC and subsequently become internalized, thereby preserving functional Cx43 in preconditioned hearts since preconditioning resulted in a marked preservation of Cx43 in gap junctions. A recent study in isolated rat hearts and in situ pig hearts found no effects of ischemic preconditioning on the ischemia-induced changes in electrical impedance, suggesting that intercellular electrical coupling is not important for protection (35). A very recent study by Li et al. (36) found that PC protected wild-type isolated cardiomyocytes, obviously without forming gap junctions, but not Cx43-deficient myocytes, suggesting a volume regulating role of Cx43.

Little is known about the consequences of metabolic coupling between cardiac myofibroblasts under normal or pathologic conditions such as myocardial ischemia-reperfusion. The purpose of the present study was to investigate the effect of PC in gap junction communication in cardiac myofibroblasts and to determine if any alteration in communication could be explained by changes in the phosphorylation status of Cx43 in cardiac myofibroblasts.

METHODS

Myofibroblast Isolation and Culture

All aspects of animal care, handling and experimental procedures were conducted according to the European convention. Rat neonatal fibroblasts were isolated from the hearts of 1–2 day old pups by enzymatic dissociation. Briefly, hearts were excised into a calcium free HEPES buffer (120 mM NaCl, 5.4 mM KCl, 5 mM MgSO₄ ×7H₂O, 20 mM glucose, 5 mM pyruvate, 20 mM taurine, 9.1 mM HEPES, pH 7), minced finely with scissors and washed free of blood. Minced tissue was incubated at 37°C with collagenase (0.5 mg/ml) and hyaluronidase (0.3 mg/ml) in calcium free HEPES buffer for 15 min. Supernatants were removed by aspiration and inspected for the presence of dissociated cells. The incubation with enzyme solution was repeated until the tissue was completely digested (3–4 digestions). At the end of this isolation procedure a mixture of myocytes and cardiac fibroblasts was present in the final cell suspension. Preplating for 45 min to 1 hr separated the two cell populations. Then unattached cells, mostly myocytes, were removed by aspiration and washing. Cells were washed once more to ensure the removal of all unattached or weakly attached cells and then returned to culture under fresh maintenance medium consisting of DMEM + 5% fetal calf serum. Cells were maintained in culture until confluent, at this point > 95% of the cells present are fibroblasts. Confluent cells were detached by trypsinisation and an aliquot taken for counting. Cells were spun at 1400 rpm for 10 min and the supernatant removed and replaced with fresh maintenance medium. Cells were plated onto uncoated 35 mm petri dishes and allowed to attach for two hours, washed once and supplied with fresh medium. Plating densities were 2 × 10⁵ cells/dish. Cells plated on this density reached approximately 90% confluence at 3–4 days in culture.

Simulated Ischemia and Preconditioning

The simulated ischemia was comprised of a glucose-free HEPES buffer modified by the addition of 20 mM lactate (Sigma) and 12 mM KCl (37, 38). In addition, the pH was adjusted to 6.2 and 0.75 mM sodium dithionite was added prior to use. Addition of the O₂ scavenger, dithionite, resulted in an immediate drop in pO₂ in the solution from 25 kPa to 0.00 kPa (37°C). The partial pressure of oxygen was analyzed with a blood-gas analyzer (Rapidlab 865, Chiron Diagnostics, Essex, UK). After 10, 30, and 45 min the O₂ tension in the solution was measured to 0.04, 7.07, and 11.07 kPa respectively. The simulated ischemic buffer was applied to cells in minimal volume. Cooper et al. (39) found that a duration of exposure to simulated ischemia corresponding to more than 50% reduction in cell viability, was 4 h. This was called lethal ischemia (LI). A PC stimulus of 30 min exposure to simulated ischemia followed by 30 min recovery in maintenance medium prior to lethal ischemia resulted in protection by PC. The experimental protocol is shown in Figure 1.

Figure 1 Schematic diagram of experimental protocols. PC, simulated ischemic preconditioning; LI, simulated lethal ischemia. Shaded area represents exposure to simulated ischemia.

Determination of Cell Viability

To determine cell viability we used the conversion of 3-(4,5-dimethylthiazol-2-yl)-5-(3-caboxymethoxyphenyl)-2-(4-sulfophenyl)-2H-tetrazo-lium, inner salt (MTS) to its formazan derivative by the action of respiratory chain dehydrogenases. The formation of a soluble purple formazan product is widely used as an assay of both cell proliferation and cell survival both in vitro and in infarct size studies in isolated hearts. The intensity of color produced is proportional to the number of actively metabolizing cells present. It was possible, using this technique, to determine cell viability in terms of dehydrogenase enzyme activity and to recover cells for the determination of total cellular protein.

At the end of the experiments cells were exposed to MTS substrate for 2 h. Supernatants containing formazan dye were collected and absorbance determined at 490 nm. Total protein was estimated in cell lysates using a microtitre Lowry estimation. The Lowry assay modification used was checked for reproducibility and linearity in the presence of lysis reagent. Results are expressed as formazan absorbance at 490 nm per mg of protein (mean ± S.E.). Single factor ANOVA was performed for each group of experiments. Tukey's test of pair wise differences was used as the post hoc test.

Gap Junctional Intercellular Communication (GJIC) Assayed by Dye Transfer

The cells were exposed as indicated in the text and legends. After exposure, single cells were pneumatically injected with the Li-salt of Lucifer yellow CH (10% w/v in 0.33 M LiCl, Sigma) through glass capillaries using a Nikon Diaphot microscope and an Eppendorf micromanipulator coupled to a pressure control unit (Eppendorf). The capillary entered the cell close to the nucleus and the dye was injected continuously for 2–3 seconds. The capillaries (GC120F, Clark, Pangbourn, UK) were prepared by a vertical puller (L/M-3P-A, List-electronic, Darmstadt, Germany). The cells were routinely fixed in formaldehyde (2% in PBS) 7–9 min after injection. The spreading of Lucifer yellow into the neighboring cells was assessed by counting the number of fluorescent cells using the same microscope with a modified fluorescein filter block (435, 510, and 520 nm filters).

The data are presented as mean ± S.E. of normalized values. The corresponding control in each set of cell cultures is defined as the 100% level. Statistical significance was examined using ANOVA combined with Fisher's post hoc test.

Immunoblot Assay of Cx43 Expression

Cell cultures were washed once in cold PBS and then harvested by scraping into 150 μ l of 2× sample buffer containing 20% glycerol and 6% sodium dodecyl sulfate in 0.12 M Tris at pH 6.8 with protease inhibitors (Complete®, Roche). Samples were processed for Western blotting immediately by the addition of 5% β -mercaptoethanol and 5 μ l of 0.8% bromophenol blue followed by heating in a boiling water bath for 10 min. Samples were then applied to electrophoresis gel immediately or stored frozen at −80°C. Aliquots containing 14 μ g of total protein were analyzed by 10% SDS polyacrylamide gel electrophoresis using BioRad miniprotean apparatus.

A rabbit polyclonal antibody (Zymed, 71-0700) directed against epitopes in the C-terminus of Cx43, detecting both phosphorylated and nonphosphorylated Cx43, and a mouse monoclonal antibody (Zymed, 13-8300) directed to sequence 360-376, were used. A phosphorylation in the latter epitope abolishes the recognition of Cx43 by 13-8300, and this phosphorylation is probably the first phosphorylation occurring on Cx43 in some cell systems (40), but not in all (41). The probed membranes were incubated with, respectively, HRP-conjugated sheep anti-mouse (Amersham, NA931V)- and goat anti-rabbit (Zymed, 62-6120)- secondary antibody and visualized by chemiluminescence (Amersham ECL) using a Kodak Image Station 1000 for image capture.

Cx43 signal was quantified by densitometric measurements by ADOBE photoshop version 6.0. The values are expressed as mean ± S.E. Differences between groups were analyzed with ANOVA and Student-Newman-Keuls method as post hoc test.

Immunofluorescence Microscopy

For staining of Cx43, cardiac myofibroblast cultures were fixed in 4% paraformaldehyde for 5 min at room temperature. After rehydration with a brief wash of phosphate-buffered saline (PBS), pH 7.4, they were exposed to blocking buffer composed of PBS containing 0.1% Triton X-100, 3% normal goat serum, and 1% BSA for 30 min before being incubated at 4–6°C overnight with a 1:200 dilution of rabbit antisera against Cx43 (71–0700). The cells were rinsed 6–8 times in PBS, and bound primary antibodies were detected by incubation for 2 h at room temperature with a 1:500 dilution of Alexa Fluor 488 goat-anti-rabbit IgG. The cells were rinsed in PBS, and for nuclear staining cells were incubated and mounted with 4′, 6-diamidine-2′-phenylindole dihydrochloride (DAPI II, Vysis) at a concentration of 0.125 mg/ml. The immunofluorescence-labeled myofibroblasts were examined using a Nikon Eclipse E600W (Nikon Instruments Inc., Melville N.Y., USA) equipped with epifluorescence, and the images captured using a Nikon CCD camera (ER3339, Applied Imaging, Newcastle, England) and analyzed by CytoVision (Applied Imaging, Newcastle, England).

Chemicals

Chemicals for buffer solutions were purchased from Merck (Kebo labs, Oslo, Norway) and type 2 collagenase was obtained from Worthington Biochemical Corporation (Lakewood, USA). Cell culture media were supplied by Gibco RL (Gaithersburg, USA) and Sigma (St.Louis, USA). Fetal calf serum was purchased from PAA (Linz, Austria). MTS nonisotopic cell proliferation assay was purchased from Promega (Madison, USA). Research Biochemicals International (Natick, USA) supplied chelerythrine. Stock solutions of the PKC activator PMA (phorbol 12- myristate 13- acetate), obtained from Sigma, was dissolved in dimethyl sulphoxide (DMSO). Affinity purified, rabbit polyclonal antibodies (71-0700) and mouse monoclonal antibody (13-8300) against Cx43 were purchased from Zymed Laboratories Inc. (San Fransisco, CA, USA). Labeled goat anti-rabbit IgG Alexa Fluor 488 was purchased from Molecular Probes (Eugene, OR, USA).

RESULTS

Validation of the Preconditioning Protocol

To confirm that the PC protocol was effective, we compared cell survival in preconditioned (PC'd) and nonpreconditioned (non-PC'd) cells exposed to 4 h of ischemic solution (lethal ischemia). PC resulted in a 4.7 ± 0.4 fold increase in cell survival (p < 0.01 vs. non-PC). Treatment of PC'd cells with chelerythrine (2 μ M), a specific PKC inhibitor, added prior to PC reduced cell survival to 1.4 ± 0.1 fold that of cells treated with lethal ischemia alone, not significantly different from non-PC'd cells, but significantly different from PC (p < 0.01). Thus PC protects against cell death through activation of PKC (Figure 2).

Figure 2 Summary of results obtained using simulated ischemic preconditioning (PC) to elicit protection against simulated lethal ischemia (LI). The PC stimulus consisted of 30 min exposure to simulated ischemia followed by 30 min recovery in maintenance medium under normal cell culture conditions to mimic reperfusion. PC'd myofibroblasts showed a significant increase in cell survival compared to cells exposed to LI alone (p < 0.05). In addition the graph shows the effect of the PKC inhibitor chelerythrine (Chel) (2 μ M). The PKC inhibitor was an effective inhibitor of protection when administered together with the ischemic solution under the PC'd period (p < 0.05 vs. PC).

Dye Transfer

Gap junctional intercellular communication (GJIC) was assessed in dye transfer assays in which single cells were microinjected with the membrane-impermeant fluorochrome Lucifer yellow. Cardiac myofibroblasts showed a relatively high dye transfer of Lucifer yellow, averaging 18–22 cells/injection.

The intercellular communication was reduced to 19 ± 3% of control values after LI (Figure 3). PC, on the other hand, resulted in dye transfer that was 36 ± 9% of control values (Figure 3, p < 0.05 vs. non-PC), indicating significantly less uncoupling in PC'd cells. The functional differences in dye coupling between non-PC'd and PC'd cells were apparent already at 1 and 2 h of ischemia. After 1 and 2 h of ischemia, the intercellular communication in non-PC'd cells was reduced to 15± 2% and 17± 2%, respectively, while PC'd cells demonstrated dye transfer that was 34± 4% and 36± 4% of control values, respectively. Interestingly, dye transfer assessed in the PC'd cells just before the LI period (PC stimulus) was only 38± 7% of control (Figure 3). Thus, after PC stimulus, LI did not cause a further decrease in GJIC.

Figure 3 Effect of simulated ischemic preconditioning (PC) and prolonged lethal ischemia (LI) on the response of gap junctional intercellular communication (GJIC). Neonatal myofibroblasts were cultured for up to two weeks with 3–5 passages before dye-transfer study was performed. All of the interventions were performed from the same batch of cells and five 35 mm dishes were included in each intervention from the same cell batch. A total number of five different cell batches were used. The mean number of injections per dish were 25. Thus, each circle or square represents around 125 injections. The asterisks represent mean ± S.E. of GJIC from the five different experiments in each group. PC'd myofibroblasts showed a significant increase in GJIC compared to cells exposed to LI alone (p < 0.05). Interestingly, cells that have undergone PC stimulus only decreased their GJIC level compared to nontreated cells (p < 0.05).

To investigate whether activation of PKC would cause a decrease in GJIC in the cardiac myofibroblasts, we exposed the cells to PMA, an activator of PKC. Data presented in Figure 4A shows that GJIC was dose-dependently decreased by activation of PKC. PMA at 1nM blocked communication by 50% after 10 min exposure. At 100 nM PMA only 2–3 cells received Lucifer yellow from the microinjected cell. In order to shed light on the possible involvement of PKC on Cx43 phosphorylation, increasing concentrations of PMA were given to monolayers of cardiac myofibroblasts and Cx43 phosphorylation was observed by immunoblots. Figure 4B shows that PMA did not cause any clear induction of Cx43 phosphorylation. However, there was an increasing blurring of the Cx43 band pattern with increasing concentrations of PMA, and a slight decrease of total Cx43 signal. PMA-induced blurring of the Cx43 band pattern has previously been observed in primary embryonic fibroblasts (23, 42).

Figure 4 (A) Effect of PMA on gap junctional intercellular communication (GJIC) in myofibroblasts from neonatal rat hearts. Cells were exposed to three different concentrations of phorbol esters for 10 min before assessing GJIC by microinjection and dye transfer. All of the interventions including the control were performed from the same batch of cells and five 35 mm dishes were included in each intervention from the same cell batch. A total number of six different cell batches were used. The mean number of injections per dish were 25. Thus, each circle or triangle represents around 150 injections. The asterisks represent mean ± S.E. of GJIC from the six different experiments in each group. (B) Changes in Cx43 band pattern in myofibroblasts from neonatal rat hearts after exposure to PMA. The cells were exposed to three different concentrations of PMA for 10 min.

Preconditioning and Dephosphorylation

To determine whether decreased GJIC during ischemia and protection against decreased GJIC by PC was associated with changes in Cx43 phosphorylation, cardiac myofibroblasts were subjected to the ischemic solution followed by immunoblotting using a polyclonal anti-Cx43 antibody that binds to all isoforms of Cx43 and a monoclonal anti-Cx43 antibody that only binds to unphosphorylated Cx43 according to Nagy et al. (40). In untreated rat cardiac myofibroblasts the polyclonal anti-Cx43 antibody detected major bands at 41, 44 and 46 kDa (Figure 5, lane 1). The two upper bands represent phosphorylated forms of Cx43 and the lower, 41 kDa band, represents the nonphosphorylated form of Cx43. The 41 kDa band may also contain certain phosphovariants (43). For convenience we will refer to the fast migrating 41 kDa band as P0.

Figure 5 Immunoblots of rat myofibroblasts exposed to 4 h of lethal ischemia (lane 2 and 4) or no exposure (lane 1 and 3) and probed with polyclonal anti-Cx43 antibodies (detecting all isoforms of Cx43), and monoclonal anti-Cx43 antibodies (detecting a site-specific nonphosphorylated isoform of Cx43).

When the cells were exposed to LI, an intense band at 41 kDa was observed at the position of the P0 isoform of Cx43, and with a faint band at the 44 kDa position (Figure 5, lane 2). However, in some experiments the dephosphorylation seemed more complete, and the 44 kDa band was not found. The monoclonal anti-Cx43 antibody recognized the P0 isoform of Cx43 only when the cardiac myofibroblasts had been exposed to ischemic solution for 4 h (Figure 5, lane 4).

To investigate and quantify the phosphorylation status of Cx43 under the same interventions as described for the dye transfer experiments, monolayers of cardiac myofibroblasts were exposed to simulated ischemia for 30 min followed by recovery in normal buffer for 30 min (PC stimulus) and lethal ischemia in the presence or absence of PC. Thereafter immunoblots and densitometric analysis were performed (Figure 6). Neither Cx43 content nor phosphorylation status of Cx43 in cells that had undergone PC stimulus differed when compared to untreated cells (control) as analyzed by densitometric measurements (Figure 6). Densitometrical analysis of the polyclonal antibody showed that LI in the presence or absence of PC resulted in an increase in dephosphorylated status of Cx43 (Figure 6) which was statistically significant when compared to control level (Figure 6, upper bar chart). However, in the presence of PC the P0 band for the polyclonal antibody was clearly not as dense as in the absence of PC (Figure 6, upper bar chart, p < 0.05). As shown by the monoclonal antibody, no or little dephosphorylation occurred in its Cx43 epitope for cells that underwent PC stimulus (Figure 6, lane 3, monoclonal panel). Only a faint band appeared after LI in the presence of PC, while in the absence of PC, LI provoked massive dephosphorylation in the 360–376 amino acid region. The LI-induced increase in the P0 band occurred in combination with an LI-induced increase in total Cx43 content (Figure 6, lower bar chart).

Figure 6 Immunoblot of Cx43 in myofibroblasts subjected to nontreatment (control), 30 min of ischemic buffer followed by 30 min maintenance buffer (PC stimulus), 30 min of ischemic buffer followed by 30 min maintenance buffer and 4 h of ischemic buffer (PC) and ischemic buffer for 4 h (LI) for both monoclonal and polyclonal anti-Cx43 antibodies. Dephosphorylated (upper bar chart) and total (lower bar chart) Cx43 content were quantified by the polyclonal immunoblot (signal intensity normalized to a control value of 1.0, n = 4). ^#p = 0.038, *p = 0.002 PC vs LI.

To investigate the dynamics of the phosphorylation status of Cx43 during the LI period, monolayers of cardiac myofibroblasts were exposed to 1, 2, and 3 h of ischemic solution in the presence or absence of a PC stimulus prior to the prolonged ischemic exposure (Figure 7). Exposure to ischemic solution for 30 min did not seem to result in detectable difference compared to control cells (Figure 7, lane 1 and 2, polyclonal panel). However, after 1 h in ischemic solution, Cx43 became dephosphorylated as indicated by an increase in P0 in the panel for the polyclonal antibody (Figure 7, lane 4, polyclonal panel). A marked reduction in P0 at all time points of ischemic simulation, was observed for those cardiac myofibroblasts that were PC'd, as seen in the panel for the polyclonal antibody. The dynamics in phosphorylation status at the 360–376 amino acid region were also investigated (Figure 7, lane 4, monoclonal panel). After 1 h of ischemic solution, we observed a faint band at 41 kDa, representing a dephosphorylation in the 360–376 amino acid region, while in preconditioned myofibroblasts no dephosphorylation at the 360–376 amino acid region occurred until 3 h of ischemic solution. At this point, a 3-fold and 10-fold increase in the P0 band appeared in the PC'd and non-PC'd cells, respectively, (upper bar chart). The total amount of both phosphorylated and nonphosphorylated Cx43 increased in the non-PC'd cells after some lag time (lower bar chart). The band for the monoclonal antibody, that appears at 3 h of ischemia in Figure 7, would therefore, at least partly, be due to an increase in the total amount of Cx43. There is a 10-fold increase observed for the monoclonal antibody at 3 h of ischemia and a 1.5-fold increase in Cx43 content (Figure 7, bar charts). This indicates an increase in the phosphorylated isoform above the increase in total Cx43. Thus, PC tended to preserve Cx43 phosphorylated isoforms at the 360–376 amino acid region in this experiment.

Figure 7 Immunoblot from rat myofibroblasts subjected to selected intervals of ischemic buffer exposure in the presence or absence of preconditioning stimulus (PC stimulus) and probed with polyclonal and monoclonal anti-Cx43 antibodies. Upper bar chart represents dephosphorylated Cx43 after 1, 2, and 3 h of ischemia in PC'd cells (gray bars) and non-PC'd cells (black bars) (densitometry of the bands from the monoclonal antibody). Lower bar chart represents total amount of Cx43 and dephosphorylated Cx43 after 1, 2, and 3 h of ischemia in PC'd cells (gray bars) and non-PC'd cells (black bars) (densitometry of the bands from the polyclonal antibody).

As the amount of Cx43 was increased by LI, but coupling was reduced, the distribution of Cx43 was investigated by immunofluorescence. As seen in Figure 8, Cx43 was mainly found intracellularly after LI, and the distribution was clearly altered compared to control cells. The LI-induced increase in Cx43 could not be abolished by the protein synthesis inhibitor cycloheximide (10 μ g/ml; not shown).

Figure 8 Immunofluorescence microscopy of cardiac myofibroblasts showing LI-induced changes in Cx43 localization/distribution. Control myofibroblasts (A, B) and myofibroblasts subjected to LI (C). Negative control were performed by omitting the primary antibodies and subsequent labeling with Alexa Fluor 488 secondary antibodies (B). Nuclei were stained with DAPI. (See Color Plate I at the end of this issue).

COLOR PLATE I See R. SUNDSET et al., Figure 8.

To investigate if there was a change in re-phosphorylation after LI with or without PC stimulus, cells were harvested for WB analysis after 30 and 60 min in normal media after 4 h exposure to the ischemic solution. The results presented in Figure 9 indicate that both groups clearly rephosphorylate after 30 min, and there seemed to be no difference in rephosphorylation between the two groups.

Figure 9 Immunoblot of Cx43 in myofibroblasts subjected to no treatment (control) or to lethal ischemia in the presence or absence of preconditioning with reperfusion for 0, 30, or 60 min. The blots were probed with the polyclonal anti-Cx43 antibody.

DISCUSSION

The results of this study indicate that ischemic preconditioning leads to less reduced communication between cardiac myofibroblasts after sustained ischemia. Furthermore, we found that preconditioning decreased communication prior to the sustained ischemia. The present study also shows that preconditioning of myofibroblasts reduced the increase in total Cx43 content and tended to reduce the extent of dephosphorylation during prolonged ischemia.

When the myocardium is exposed to severe and prolonged ischemia, electrical coupling between adjacent cardiomyocytes is impaired, propagation of electrical impulse is slowed and eventually blocked. However, there is some controversy on how fast the gap junction channels shut down in cardiomyocytes. Studies of intercellular resistance and whole tissue resistance have suggested that gap junction channels in the myocardium closed approximately 15 to 20 min after the onset of ischemia (44, 45), electrically segregating injured myocytes from nonischemic myocytes. Ruiz-Meana et al. (46) demonstrated, however, that cardiomyocytes can communicate after 45 min of ischemia through cell-to-cell transport of Lucifer yellow. In the present study, using cardiac myofibroblasts, 60% reduction in cell communication was observed after 30 min superfusion of ischemic buffer. After 1 h of ischemia 85% reduction in Lucifer yellow transfer was present and remained at that level during the subsequent 3 h, suggesting that a certain level of communication persists in myofibroblasts during sustained ischemia. We note that similar controversies exist also for ischemic effects in other cell types. In astrocytes Li and Nagy (47) found that after initiation of hypoxia, coupling was reduced by 77% after 15 min, 92% after 30 min and 97% after 1 h. Cotrina et al. (48) found that within 15 min metabolically inhibited astrocytes reduced coupling by 70%, and it remained at that level during the subsequent 2 h. The reasons for divergent conclusions in the ischemic-induced inhibition of gap junctional communication are unclear, but might be due to different ways of measuring intercellular communication (e.g., microinjection of Lucifer yellow, scrape loading, dual voltage clamp, FRAP, electrical conductance in whole tissue), different ways of inducing ischemia (e.g., hypoxia, metabolic inhibition), different species origin (rabbit, rat, pig), or different cell types (myofibroblasts, cardiomyocytes, astrocytes).

It has been shown that PC delays electrical uncoupling during ischemia (34, 45, 49). In this study we found that PC'd myofibroblasts diminished the reduction in cellular communication upon LI (Figure 3). Furthermore, in the cells that have undergone PC stimulus only, we observed a 62% decrease in dye transfer. In a recent study Miura et al. (33) found that, in isolated rabbit hearts, PC inhibited Lucifer yellow transport by 40% prior to the sustained ischemia. Inhibition of gap junction communication prior to the LI could be cardioprotective, because it prevents the propagation of Na⁺ and Ca^{2 +} overload and thus cell damage in an ischemic situation (50, 51, 52). The diffusion of additional apoptotic factors may also be prevented by PC-induced decrease in GJIC (53).

While almost all Cx43 is phosphorylated in isolated normoperfused rat hearts, it becomes progressively dephosphorylated during ischemia. Such dephoshorylation corresponds to electrical uncoupling as observed by Beardslee et al. (44). Jain et al. (34) found that in PC'd rat hearts in vitro a delayed electrical uncoupling occurred related to diminished dephosphorylation during prolonged ischemia. Schulz et al. (32) also concluded that PC attenuated the ischemia-induced dephosphorylation of Cx43 in pig hearts in vivo. Miura et al. (33) found that PC accelerated the appearance of dephosphorylated Cx43 in rabbit hearts 10 min after the onset of ischemia but preserved phosphorylation of Cx43 during a longer period of ischemia. Jain et al. (34) and Schultz et al. (32) used confocal laser microscopy to investigate the phosphorylated status of cardiomyocytes while Miura et al. (33) used Western blot of myocardial tissue including both myocytes and fibroblasts. In the present study, we investigated the phosphorylation status of Cx43 in cardiac myofibroblasts to see whether the differences in dye transfer in myofibroblasts during ischemia in the presence or absence of PC could be related to changes in phosphorylation status of Cx43. Some studies (32, 33, 34) indicate that PC reduces dephosphorylation of Cx43. The molecular explanations may include specific modulations of both phosphatase and kinase activities. For the latter, PKC is a main candidate for the following two reasons. Firstly, PKC is known to phosphorylate one or more serines in the 360–376 region of Cx43, and secondly, there are strong indications that PKC is involved in the protective effect of PC. Our study may support this explanation. The present study indicates that PC influences the phosphorylating status of Cx43 during prolonged ischemia.

Densitometric analysis of immunoblots showed an increased amount of Cx43 in myofibroblasts exposed to LI. No increase in Cx43 expression was seen in PC'd myofibroblasts. The LI-induced increase in Cx43 observed in this study could be due to either an upregulation in synthesis or a decrease in internalization/degradation. The LI-induced increase in Cx43 was not abolished by blocking protein synthesis with cycloheximide, indicating an accumulation of Cx43 due to a disruption in degradation of Cx43. Immunofluorescence studies showed that the LI-induced increase in the level of Cx43 was localized intracellularly and thereby not involved in GJIC. From the representative pictures presented in Figure 8, there was clearly a more disorganized labeling of Cx43 in the ischemic cells. The well-defined perinuclear labeling and the distinct membrane-bound labeling between neighboring control myofibroblasts was not observed in myofibroblasts exposed to LI. The patchy disorganized LI-induced intracellular labeling is probably due to a disrupted Golgi apparatus and/or a failure in degradation of internalized Cx43. Jain et al. (34), analyzing ventricular lysates from rats, observed equal amounts of Cx43 between non-PC'd and PC'd cells. The half-life of Cx43 is in the order of 1–2 h, even in the intact heart (54). Modulation of Cx43 synthesis and degradation therefore allows rapid changes in the amount of the protein. Increased VEGF expression has been reported in experimental models of myocardial ischemia and infarction (55, 56, 57, 58). A study by Pimentel et al. (59) indicates that an important component of the autocrine myocyte response to VEGF secretion is upregulation of Cx43. A recent study by Azzam et al. (60), using human fibroblast cultures, found that Cx43 was upregulated by stresses like irradiation (α -particle and γ -rays), t-butyl hydroperoxide, and hyperthermia. Thus, it could be that the upregulation of Cx43 is due to an increase in VEGF expression or VEGF related signal transduction pathways caused by ischemic stress which might be prevented by PC.

As mentioned above, Miura et al. (33) found that PC inhibited Lucifer yellow transport in rabbit hearts during the ischemia. In the present study, a decrease in dye transfer in PC'd cells occurred prior to the LI. By using an antibody from Transduction Laboratories, Miura et al. (33) showed that PC accelerated the appearance of nonphosphorylated Cx43 after 10 min of ischemia but reduced the extent of dephosphorylation of Cx43 during a 30 min period of ischemia. It has been shown (61, 62) that PC causes a partial loss of the adenine nucleotide pool so that the PC'd myocardium has less ATP at the onset and early phase of sustained ischemia compared to non-PC'd myocardium, though the rate of decline in ATP level during ischemia is slowed by PC. In the present study, we used a polyclonal antibody from Zymed and we were not able to find an early increase in the P0 band during ischemia in the PC'd myofibroblasts. On the contrary, the PC'd myofibroblasts showed less dephosphorylation after 1, 2, and 3 h of sustained ischemia (Figure 7). Also, at these early time points the dye transfer of Lucifer yellow was significantly higher for PC'd cells versus non-PC'd cells which might be a result of different phosphorylation status of Cx43. Immediately after the PC stimulus (Figure 7, lane 2) or just prior to the LI in the PC'd cells (Figure 6 lane 2), we did not find the phosphorylation status of Cx43 to be different compared to the control cells. The lack of change might indicate that ATP depletion by PC stimulus alone was not large enough to induce overall dephosphorylation of Cx43. Thus, our results suggest that the subclosure of gap junctional communication in PC'd cells prior to the sustained ischemic period is not triggered by an early dephosphorylation.

As detected by the polyclonal antibody, a considerable degree of dephosphorylation occurs in Cx43 already after 1 h of ischemia, irrespective of PC or not. In contrast, the monoclonal antibody detects only a very faint band at 1 and 2 h, but a stronger band at 3 h of ischemia (Figure 7). PC seemed to delay the recognition of nonphosphorylated Cx43 by both antibodies and particularly the monoclonal antibody, indicating that PC might induce a resistance towards dephosphorylation in the epitope for the monoclonal antibody. The monoclonal antibody in this study detects the nonphosphorylated form of the Cx43 in the 360–375 amino acid region. This region contains six serines (DQRPSSRASSRASSRPRP). Lampe et al. (63) reported that serine 368 is a major site of PKC phosphorylation in Cx43. Zymed claims that the monoclonal antibody (13-8300) binds to Cx43 if serine 368 is nonphosphorylated. However, Cruciani and Mikalsen (41) suggested, on the basis of peptide competition experiments, that the recognition site of this antibody includes serines 364/365. Phosphorylation of serine 364 occurs at least in some cell systems (64). In addition to serine 368, PKC may also phosphorylate serine at position 372 (65). It cannot be excluded that PMA-induced phosphorylation of Cx43 also may take place elsewhere in the molecule (66). Taken together, our results might indicate that PC either activates kinases or inactivates phosphatases, which lead to a more sustained phosphorylation of the 13-8300 epitope. Subsequently, this may affect the regulation of Cx43 channels. The precise residues that are dephosphorylated during ischemia and the residues that are rephosphorylated or newly phosphorylated during PC remain to be elucidated.

In conclusion, ischemic preconditioning improves cell survival in cardiac myofibroblasts. One proposed mechanism of protection in ischemic preconditioning might be to decrease cell-to-cell communication prior to the sustained ischemic period and to protect neighboring cells during the sustained ischemic period with an outcome of better cell-to-cell communication immediately after the longer ischemic period. Activation of PKC and phosphorylation status of Cx43 in relation to ischemic preconditioning and ischemia as well as regulation of the total amounts of Cx43 are complex events that require further clarification.

ACKNOWLEDGMENTS

We gratefully acknowledge technical assistance of Thomas Andreasen, David Johansen and Mona Nystad. This work was supported by a Research Fellowship (122570/310) from the Research Council of Norway.